| Mark A. Gulbrandsen |
To be able to run continous class A power into 2 ohms and also being able to go higher class A power into 4&8 ohms the 1.5kva would in theory be needed... also to have a little headroom. Since I have room for at least 4 extra pair of OP devices I want to extend my Class A power to about 100+watts. Keep in mind that the KSA-100 does only 80 watts in class a then goes ab. The clip level comes out to about 128 watts running a single channel. Am not after overall higher power just to be able to get Class A out if it to full power... yes, I plan on running my pair of mono's on 240V!
My KSA-50 has a 1.5 kva unit... sure I got it surplus but it also allows me to expand it if I want to.
Also as for those wanting the original xformer doesn't it make more sense to get the one that already has all the needed secondaries?
Mark |
|
|
| AndrewT |
Hi Mark,
you are rightly looking at the severe loading to assess the transformer needs.
I will not say you are wrong, but I will propose some numbers to show an alternative way of assessing the VA requirement.
130W into 8r
240W into 4r
400W into 2r
seem reasonable targets for supply rails that hold +-52Vdc when fully biased to ClassA.
For a full 100W of ClassA into 8r you need 5Apk and Iq>2.5A
a sensible Iq is 2.6A giving a dissipation of 270W.
The transformer to support this is 130W of ClassAB +270W of Pd =400W, the transformer~=600VA (1.5*[130+170]).
Staying with the same 100W ClassA and now 240W ClassAB into 4r the total is now 510W giving transformer ~=770VA
Now using 400W ClassAB into 2r + 270W pd=670W. transformer~=1000VA. This is pretty extreme but the output is capable of driving it so some may want to design for this loading.
If you raise the Iq to 3A then Pd=312W and 4ohm loading becomes 312+240 leading to a VA~=830 (Iq=3A gives 126W of ClassA into 7r).
From the above, 750VA to 1000VA seems to straddle the 800 to 900VA that Flod has identified.
I still think he is in the ball park and if I were building a 2ohm capable 100Klone I would go for about 900VA to 1000VA. |
|
|
| BobEllis |
Don't forget the main filter bank charging currents. Big capacitor banks mean short charging cycles, and very high peak currents.
I ran a PSU Designer simulation of an Aleph-X PSU (120,000 uf - 0R1 - 120,000 uf, IIRC) and the RMS current in the transformers was ~twice the amp's draw. Seems like a fudge factor of at least three times expected power used in the amp would be required to allow some headroom. A factor of four would give you a transformer running at half its rated power. I'm with Mark, 900 VA seems small, 1500 VA seems reasonable. |
|
|
| AndrewT |
Hi Bob,
that comparison of rms secondary current and amplifier DC current fits exactly the model most transformer manufacturers state.
A transformer with rectifier feeding a capacitor input smoothing filter should be derated to 70%. This takes account of the pulse currents and the excess heating that occurs.
Now apply the square root of 2 rule for peak voltage.
The result is that the ACrms current = 2 times the rectified and smoothed DC current.
eg. 500VA 25-0-25Vac transformer has a continuous AC current of 10Aac.
The DC supply rails after the smoothing caps will be @ about 34.7Vdc and the maximum continuous DC current is 5A. Check the power available, 5*2*(34.7+0.7) = 354W ~=0.7*500VA.
A 900VA transformer will give 630W of PSU capability. i.e. +-52Vdc @ 6A continuous. Iq uses about 30% to 50% of this, leaving the remainder to satisfy the ClassAB requirement.
Keep in mind, I am not stopping anyone from using a larger transformer, it has some advantage, but simply showing that the original Krell and Flod's replacement can support a very large power output (240W into 4r and sustained peaks of 400W into 2r). |
|
|
| jacco vermeulen |
You guys are so funny.
Any toroidal can deliver twice the nominal peak current without a sweat.
Toroids in the 750 to 800VA were common for all High End US manufacturers during the KSA100 era, whether Krell, Threshold or the rest of the bunch.
Overhere that was, and stil is, regarded as serious hardware from the squeemish European view.
The French AudioAnalyse A100 was produced during the years of the KSA100, same price range, and did 100 true watts in Class A on 300VA toroids.
The 1988 upward Krell models had a standard range 750VA Holden&Fisher toroidal.(or 800VA, blame my dodgy memory)
Even a power factory like the ML23.5 only had a 1200VA H&F toroid per channel, and the 23 did 450 watts in 8 Ohms. (you really do not wish to hear about the price tag of H&F toroids)
Dan d´A and his little brother were spenderish, not cost cutters.
Remember that a KSA50 was supposed to double the power in 4 and 2 ohm loads.
Hence the 50 watts in 8, a KSA50 can do a clean 200 in 2 Ohms.
Same for the KSA100, called a 100 because it does/did 400 watts in 2 Ohms.
If you've got a big hunk of a toroid lying around, use it.
But don't say it's needed, a new 1000VA is costly, and anything over 1kW is like tossing cash to a coke head.
I've got some custom 3kW toroids lying around, anyone in the mood to spend big bucks ?
Pavel,
please beam me up, i need a tranquiliser. :clown: |
|
|
| BobEllis |
| quote: | Originally posted by jacco vermeulen
You guys are so funny.
... anything over 1K is like tossing cash to a coke head.
|
Isn't DIYing a project like this about serious overkill? :devilr: |
|
|
| jacco vermeulen |
Sure Bob, why do you think i own silly toroids.
But anyone thinking Big, think serious Big all the way.
(i'm a cat, i've got a little pecker. :clown: ) |
|
|
| zox2003 |
Hi,
I asked this question before but nobody was able to answer it straight and 100%.Let me try again:
I have 30 pieces of ST 60N10 MOSFET (200 W).Here is a data sheet (please read-pdf):
http://www.st.com/stonline/books/pdf/docs/3105.pdf
Can we use this MOSFET to build PA Amplifier?
What schematic should I use to build one(post it please)?
How much power should we expect from one pair of these 200W MOSFETS on +/-42V and 4 ohms?
Thanks everybody for help.
Best Regards |
|
|
| steenoe |
Since this thread is growing just as fast as the other Krell thread I had to give up reading it all (at page 15 or thereabouts:) )
Simple question: How are the boards comming along?
Or to quote Lt. Ellen Ripley " Who do you have to ****, to get on this ship":D Slightly rewritten, without permission and all that ;)
Steen:cool: |
|
|
| jacco vermeulen |
Steen,
i doubt that you look like Sigourney, probably more like the Alien character, droolin and all.
Gewa has the KSA100 link on his posts.
(personally i wouldn't mind impersonating The Supreme Commander if an androidal like Ripley posed that question. :clown: )
www.diyaudio.com/forums/showthread....282#post1044282 |
|
|
| AndrewT |
Hi Zox,
this thread is about a Krell KSA100 Klone.
FETs for output devices are not appropriate.
Go look at the Quasi threads.
Or start your own.
Can you delete your post? |
|
|
| steenoe |
| quote: | | i doubt that you look like Sigourney, probably more like the Alien character, droolin and all. |
Yep, I drool bigtime, when I hear "100 watts class A" allright ;) Might even give this one a shot. Can you imagine the heatsinks for an amp like this Krell, if you dont want them propellers running:D They would be worthy of a big NP amp then.
BTW the line I quoted is from Alien 4. Those French directors has a way with words, the 4'th was directed by Jean-Pierre Jeunet.
The best Alien film is by the way (and without question!!) the first one. Directed by Ridley Scott, The film was just as scary as if it was made by Hitchcock, himself:eek
Steen:cool: |
|
|
| zox2003 |
Andrew and others,
Please -If I try to start new thread and try to post it I receive message "Please search forum first".I did search forum and tried to re-post message and I received same message.
What is going on-never had that before...
HELP!!!!!! |
|
|
| zox2003 |
My mistake-I found why I could not start new thread...
Thanks to everybody... |
|
|
| z2003 |
Hi,
I asked this question before but nobody was able to answer it straight and 100%.Let me try again:
I have 30 pieces of ST 60N10 MOSFET (200 W).Here is a data sheet (please read-pdf):
http://www.st.com/stonline/books/pdf/docs/3105.pdf
Can we use this MOSFET to build PA Amplifier?
What schematic should I use to build one(post it please)?
How much power should we expect from one pair of these 200W MOSFETS on +/-42V and 4 ohms?
Thanks everybody for help.
Best Regards |
|
|
| jacco vermeulen |
| quote: | Originally posted by steenoe
from Alien 4. |
Yeah, Ripley resurrected, with a bit of bug.
Ron Perlman, the ugly guy, is one of my fav actors=> La cite des enfants perdus
You should see other French movies like "Delikatessen"
Your bud Magura supplying the heatsinks again for a passive cooled K-100 ? |
|
|
| PWatts |
Z2003: start a new thread asking that question. A new amp using MOSFET's under this thread is unrelated, hence the apparent lack of help. This thread is purely for the resurrection of the Krell KSA100 (MK2), and nothing else.
For those uncertain about it, here are the current status:
- Board layout was made to be a drop-in replacement for commercial Mk2 owners or Mk2 upgrade for Mk1 owners
- Prototype Mark built is working fine
- Final board gerbers has been sent to Mark for GB manufacturing
- Boards are according to Mk2 spec, with some refinements including separable frontend/driver supplies, numerous on-board decoupling caps, 3 different FET types, multiple TO92 types for the LTP
- Board looks are to be blue soldermask, white silkscreen, gold-plated traces
- No heatsink is required for the FET's, a small piece of sheet aluminium is sufficient for the predrivers and a larger ones as on the original is the minimum recommended for the drivers
On a side note, I've just completed Jozua's rebuilt MK2 amp. Will post pictures tomorrow of the measurements made with a) his commercial Mk1, b) his commercial stock MK2, c) the rebuilt Mk2, d) the KSA50 clone I built for him. The frequency spectra are very interesting!
Even though we know it, it was also neat to observe how the spectra changes as the amp warms up. Even after 10minutes under load, the THD is still higher than it should, with the harmonics sticking up high. After about 30mins everything has settled, and yields the optimal results.. always nice to verify stuff we think we hear :D |
|
|
| steenoe |
| quote: | | This thread is purely for the resurrection of the Krell KSA100 (MK2), and nothing else. |
Oh, I thought I just resurrected Ripley for a moment:D
Other than that PWatts, thanks a lot for filling in:)
There is a saying that goes something like this: "You cant hold good men down"
BTW Jacco, I love Ron in that movie too;)
Steen:cool: |
|
|
| Zen Mod |
| quote: | Originally posted by AndrewT
Hi Zen,
when you read the article that goes with the JLH80W, you see that he chose the parallel pair of 4u7F plastic caps first and then calculated the resistive components to suit.
The problem is that he chose too high a turn over frequency resulting in -1db @ 27Hz. The lower leg resistor value should be raised by a factor of 3 to 10 times. Then the gain is way below JLH's target, but much closer to the gain many others use today.
But the really attractive use for this technique (and still keep the low value NFB resistors) would be dedicated use for mid driver or treble driver when the high pass turnover point can be tolerated at 100Hz or even much higher. |
Tnx Andrew,but I'm already aware of that;
what I intended is to point on 150K feedback resistor,nothing more;
rest of it I covered in previous post ,quoted above pic |
|
|
| AndrewT |
Hi Zen,
I have not tried to measure the effect but output noise will change as the NFB values are scaled up or down.
Do you have any figures? |
|
|
| Zen Mod |
| quote: | Originally posted by AndrewT
Hi Zen,
I have not tried to measure the effect but output noise will change as the NFB values are scaled up or down.
Do you have any figures? |
agree with you,but I don't have any figures;
I will -in Mark's boots-just change few bits'n'pieces and listen what I have with this......
maybe,one day-too short in time right now.....but....when I lay my dirty hands on any iteration of KSA pcbs........:smash: |
|
|
| LuckyLyndy |
Zen Mod,
Mark is on work assignment in Wyoming, and will be back next week.
So, this is why he was working so hard to get it up and running this past week.
Zen, thanks again for the trannie help, but I tried to email you from your earlier reply and it was bounced back, so I went through the forum mail, and then Mark said you tried me, and got the same problem. It sounds like EdcorUSA trannies will let me ship back that underpowered transformer, and will consider building one that meets the requirements of running those OD3 gas tubes...
However, I guess I didn't get that info fast enough to them, and because I teach elementary kids, 8 year olds, I did not get a single break till this afternoon to call Edcor, and by then, the shop was closed.
I'll just have to put off any building this weekend, suffer with my modified Grounded Grid, wimpy? Krell Klone 50, and ProAc 2.5 clones
(somebody in Australia said boomy bass?-they must have built it wrong--great bass). Perfect bicycle weather in Utah...leaf colors way up...and I have some fresh music to spin...:cool: |
|
|
| Zen Mod |
| quote: | Originally posted by LuckyLyndy
Zen Mod,
Mark is on work assignment in Wyoming, and will be back next week.
So, this is why he was working so hard to get it up and running this past week.
Zen, thanks again for the trannie help, but I tried to email you from your earlier reply and it was bounced back, so I went through the forum mail, and then Mark said you tried me, and got the same problem. It sounds like EdcorUSA trannies will let me ship back that underpowered transformer, and will consider building one that meets the requirements of running those OD3 gas tubes...
However, I guess I didn't get that info fast enough to them, and because I teach elementary kids, 8 year olds, I did not get a single break till this afternoon to call Edcor, and by then, the shop was closed.
I'll just have to put off any building this weekend, suffer with my modified Grounded Grid, wimpy? Krell Klone 50, and ProAc 2.5 clones
(somebody in Australia said boomy bass?-they must have built it wrong--great bass). Perfect bicycle weather in Utah...leaf colors way up...and I have some fresh music to spin...:cool: |
if blablabla@neobee.net isn't functional to you,you can always use mail via board(btw-you didn't enable this function to ya) or you can use sasica5ATgmail.com........
bicycle weather......one day I'll put few pages on my little web with my bicycles.........and one day I hope that I'll have more time to spend on them...
btw-this days I'm looking american chopper series season one in late night hours......pretty amusing stuff ;)
nice off topic in pretty pale thread :devilr: |
|
|
| Flodstroem |
| quote: | That seems like a pretty small toroid to me for this amp... I was thinking more in the area of 1.5KVA per channel... Its hard to believe that it was only 800 to 900 VA as they were quite large toroids. As for the front end supply we don't know yet what the specs and type of supply it was...
Flodstroem, perhaps you should ask Avel about the KMA-100 toroid instead. Then all secondary windings necessary will be present. As long as there is 115+115 I'm fine and you guys can fight it out ver other odd voltages you may need. Otherwise I'll skip this GB sonce its smaller than I wanted.
|
This transformer was exactly the one that was fitted in the mk-II model. Krell used a smaller one in their mk-I model. Its obvious that the mk-I transformer was a little bit to weak for the purpose.
Regarding physical format: Krell used a very special core in this transformer and I think it was capable to feed the amp with at least 1-1,2kW of continuos power if needed (this will give you ca 2.4kWatts in a stereo unit which is the upper limit for my house main fuses rated at 10Amps/2.3kW). Also it had a lot of windings including an electro static shield and the whole thing was then potted in a steel can.
Also I must add this: the only chance we had to recognize that this model was used in the KSA 100mk-II was by the model number for the mk-I given her by PWatts (thanks Pierre for that) and for the research they did at Avel Lindberg.
Avel do not have Krell-models connected to unique numbers, they only have model numbers to reference. To find the transformer used in the KMA-100 Avel need to have the transformer model number at first.
And finally, whats your price limit for a transformer for this amp, I think $ 543 (exclusive freight and customs) for a stereo unit will be in the upper most level for most of us if choosing the originals.
PWatts:| quote: | On a side note, I've just completed Jozua's rebuilt MK2 amp. Will post pictures tomorrow of the measurements made with a) his commercial Mk1, b) his commercial stock MK2, c) the rebuilt Mk2, d) the KSA50 clone I built for him. The frequency spectra are very interesting!
|
I could hardly wait for the pics.
Regards :cool: |
|
|
| gaborbela |
HI
I just follow these project until today . But now I decided to go on and build one KSA -100 to even I'm a owner the KSA-50 clone .
I went to the Wiki page were is the list for the interest on the PC board to ad my name but some how I'm unable to put my name into the list properly .
It always comes at the last row outside , not between the lines .
I would like to ask for help if someone can fix that to me .
My user name
gaborbela - Canada and I'm up for 2pc board
Thanks
Regards |
|
|
| rolandong |
| quote: | I just follow these project until today . But now I decided to go on and build one KSA -100 to even I'm a owner the KSA-50 clone .
I went to the Wiki page were is the list for the interest on the PC board to ad my name but some how I'm unable to put my name into the list properly .
It always comes at the last row outside , not between the lines .
I would like to ask for help if someone can fix that to me .
My user name
gaborbela - Canada and I'm up for 2pc board
|
I signed the wiki before. Now I cant find my name. Who trashed the wiki ? |
|
|
| Flodstroem |
I also signed on the wiki earlier in this thread and suddenly I was deleted. It was because the whole side was changed to a new side I think. All of us had to sign once again, so must you.
Need help, give me a tip.
Regards :cool: |
|
|
| Mark A. Gulbrandsen |
Thefirst wiki was only there to determine how much interest there was. The new wiki is permanent and you will need to sign back up. if you can't edit the wiki please make a post here as to how many boards you need and I will collect those names and edit them into myself. We should be able to nail down the exact pcb price as soon as I am back in town middle of next week.
Mark |
|
|
| TomWaits |
I don't quite get the wiki and I don't want to muck it up. Could someone tack me on for 2 boards? I'm not lazy just a little numb these days. :xeye:
Cheers,
Shawn. |
|
|
| GeWa |
Shawn, you are wiki'ed for two boards.
Regards |
|
|
| TomWaits |
| quote: | Originally posted by GeWa
Shawn, you are wiki'ed for two boards. |
:cheerful: Jah,
Shawn. |
|
|
| LineSource |
| quote: | Originally posted by PWatts
Z2003: Even though we know it, it was also neat to observe how the spectra changes as the amp warms up. Even after 10minutes under load, the THD is still higher than it should, with the harmonics sticking up high. After about 30mins everything has settled, and yields the optimal results.. always nice to verify stuff we think we hear :D |
This KSA100 klone may be a good design to experiment with using On-Semi's ThermalTrak MJL3281A / MJL1302A output transistors with integrated on-die thermal diodes, plus the Leach Vbe multiplier circuit. Having the thermal compensation diodes on the output silicon could improve the warm-up characteristics.
Is anyone planning on using On-Semi ThermalTrak MJL3281A / MJL1302A on their KSA 100? |
|
|
| PWatts |
OK here's a sobering thought: apparently the "upgrade" I made from the commercial Mk2 was a step backwards, so much so that the owner now prefers both his Mk1 and KSA50 clone above it :bawling: Before the upgrade the Mk2 was hands-down the winner. Apparently now the bass is untight and doesn't extend as low as either of these. Can anyone offer insightful (non-black magic) advice as to what could be the reason? None of them out of a purely engineering standpoint should have any adverse effect, and many of them were used in either the KSA50 clone or upgrades to the Mk1. This may perhaps not be directly related to the KSA100 clone project, but may be insightful when it comes to component selection.
Here're the changes I made between the two listening sessions.. I list every single one even though many are negligible:
- Changed the transformer primaries from 110V to 220V since it was imported from the USA. The previous listening test used a 3kVA stepdown transformer.
- Changed the feedback capacitor to a 1000uF Black Gate. The original used 500uF, but going up on the capacitance isn't a problem. I used the same cap on both the KSA50 clone and when I upgraded the Mk1 last year, with improved results
- Changed the MJE15030/31 drivers and vBE multiplier to 32/33's since I didn't have any of the others and it proved successful in the KSA50 project and Mark also used it in his prototype
- Changed the output transistors to MJ21194/21193. This is the only major upgrade IMO but it cannot make such a large difference. I used the same ones on the KSA50 clone
- Reduced the bias from 110W to 80W to make it run cooler in the hot South African climate. I did the same on the Mk1 without any adverse effect since the speakers are an easy 6ohm load, and the KSA50 clone is of course even lower.
- Added a 2uF MKP cap on the mains voltage at the transformer primaries at the wire harness in the front
- Added 100nF caps across the AC at the rectifiers (I did that with the Mk1 too)
- Changed the hookup wire with very expensive silver-coated OFC from van den Hul (on owner's request)
- Changed the input coax to van den Hul (ditto) - the same coax is used on the outside going to the preamp, and was also used in the KSA50 clone.
- Replaced the old trimpots. These, the driver transistors and the BG cap were the only changes I made to the PCB itself.
- Changed the fuse to a 6.3A one.
I forgot the flashdisk with the measurements at work where I took them, but they don't reveal anything much.. I never measured it before I changed the primary voltage, dropped the bias and changed the cap though, only after that. The other set of measurements were taken after the other changes listed here. Frequency response is, just like the other two amps, basically flat down to DC.
So.. any ideas? Since I don't have the amp with me anymore and it's far too heave to lug around, the first steps will be basic. First off the amp will just be "played in" for a while, since the BG caps apparently takes a long time until it reaches proper performance but I don't believe any of it. Then take out the cap on the mains (since I know some people experienced it to be a step backwards in some instances; perhaps the 2uF is a bit high too), and then up the bias back to 100W. If desperate the 100nF's too. If that doesn't help, it's a massive operation to revert back to the MJE15030'31's and MJ15003/15004's. |
|
|
| AndrewT |
Hi all,| quote: | | here's a sobering thought: apparently the "upgrade" I made from the commercial Mk2 was a step backwards |
We really need to get a handle on this or there may be others very dissapointed with the outcome or not know any better for lack of comparson.
I posed these questions earlier and no-one came forward with suggestions.
Please, have a long hard think and let's try and get this right first time!
| quote: | Originally posted (331) by AndrewT
there seems to be a dual identity regarding KSA50 Klone bass developing.
Some are saying that the 50 has really strong deep bass and others are consistently saying the 100 beats it every time.
Is the power supply the difference between the two opposing sides, or is it in the implementation/selection of the PCB components?
Whichever, we need to understand the mechanism to ensure that all the builders of the 100 get the full benefit of this renowned bass ability.
|
| quote: | Originally posted (361) by AndrewT
Hi,
jacco said quote:
"A bunch of folks here built a bunch of different KSA50s (Klones), judging one variant doesn't speak for another. "
That was the point of my earlier question.
Which components control the bass quality in the final build of a KSA50 Klone? That knowledge can and should be used to ensure the 100Klone build gives the result that most would expect.
We could start by comparing the differences between the common parts of the two 100s and how they are different from the 50.
The LTP seems similar,
The VAS seems similar,
The FET stage (mk2) is different,
The driver stage is doubled,
the driver bias is reduced,
the output stage is doubled and the bias is +40% (reduced per transistor),
the transformer is bigger,
the capacitance is bigger?,
the layout is different,
What else can be added to the list?
What components in the difference list can be eliminated from the bass investigation? Any?
What components are left that may have an effect on the bass quality? |
|
|
|
| PWatts |
Excellent comments. From what has happenned in this case, it would seem apparent that various tweaks that may offer improved performance in some cases may not necessarily hold true.
Based on these observations I think the window for investigation can be made smaller than the ones jacco mentioned earlier, since many stayed the same during the conversion process such as power supply, layout, front-end, number of driver/output transistors, output bias. They may of course have an influence, but in this case they all stayed the same, yet the bass had changed. Since the change is so profound, these may be the starting point for many seeking the difference between the 50 and the 100 too. Differences with transistor proximity may also influence the mixed comments even with the same transistors used.
First off I'd suggest somebody do a side-by-side comparison of the different driver and output transistors. Everybody says "I'm using xxx and it sounds so much better than yyy", but did anybody ever make an observation of comparing some directly in the same amp? Another possibility that obviously only holds true in certain instances might be that the higher fT (4MHz instead if 2) of the 21193/94's may not be as happy as the 15003/4's with the fairly long leads going to the transistor bases on the original heatsink: almost 20cm for the rear ones.
Same goes for the driver transistors.. I can't see why but maybe the higher-gain models don't work too well with the drivers and/or vBE multiplier.
PSU and capacitance would IMO be secondary, since the Mk1's bass is apparently not as good as the Mk2's, yet they use the same power supply - only difference is the capacitors and transformers are newer models, but the same values and VA ratings.
I know many of these should make no difference in a theoretical sense, but since there was a change, and a big one at that, it would appear that further investigation may be warranted. Unless somebody finds out what's going on, people wishing to play it safe may want to stick as closely to the original as possible. |
|
|
| Zen Mod |
| quote: | Originally posted by PWatts
OK here's a sobering thought............................. |
Black Gates;
they realy need long burn in......I mean longgggggggggggggggg.
some ppl use AC jig for that,for sorta pre-burning.
you can try temp substitute with any decen cap,I can bet that's culprit |
|
|
| PWatts |
| Well on the Mk1 that very same BG model in the same application offered an improvement straight out the package.. :confused: |
|
|
| AndrewT |
Hi Pwatts,
is there room to parallel in the 500uF to the back of the 1mF black gate? |
|
|
| awpagan |
before you try to analyse too soon.
if you used new BG's.
they don't sound that good new, they have to broken in.
and they do take a while.
obersvations from many posts in these forums from BG devotes.
allan
edited too late already stated by others
|
|
|
| PWatts |
Well no.. with crocodile leads I guess I can tap the 825ohm's one lead and ground and thus have a parallel connection but that'd probably do more harm than good. Too little space to get in there with a soldering iron due to all the heatsinks, and only the front is accessible. Removing the boards is a massive task since 8 wires for each channel have to be soldered loose, and then on again, and possibly once more to remove it again later.
All in all much easier to just give the amp high-amplitude input without a speaker connected to give the BG lots of cardio workout ;) |
|
|
| awpagan |
| quote: | Originally posted by PWatts
All in all much easier to just give the amp high-amplitude input without a speaker connected to give the BG lots of cardio workout ;) |
wrap some water pipes around it.
Good Hot water service, or for the coffee:D
allan |
|
|
| K-amps |
| Reduction of bias plus serial trafo wiring could explain perhaps the softer bass, due to softer supplies as a result of these two... |
|
|
| AndrewT |
| quote: | Originally posted by K-amps
Reduction of bias plus serial trafo wiring could explain perhaps the softer bass, due to softer supplies as a result of these two... |
We are all at a loss, but I would not expect the 240/120V conversion to be significant.
The impedance reflected from primary to secondary should remain about the same for the two different voltage set-ups.
I guess the reflected impedance may be slightly lower for the 240Vac wiring but this will depend on the house and distribution board wiring.
Now, bias setting, that could be VERY significant. This has been reported by many others with different ClassA and ClassAB amplifiers. |
|
|
| jacco vermeulen |
| quote: | Originally posted by LineSource
using MJL3281A/MJL1302A |
Moi.
a: i've already got more than a few
b: i've still got enough beautifull Avid-Thermalloys, for which i still owe Mark a big favor.
c: i hate to cut those up again for 1/2-duty on Jappon devices.
I've got 4 spare KSA50 output boards as my fellow Dutch KSA50 GB member on second thought did not require the second batch Pinky boards.
If there's a way of getting another 8 KSA50 output boards or the KSA50 Gerbers i'm a happy camper.
Mr Watts Sir,
Given that Mark's reaction to his K-100 was written in very BOLD lettering, i suggest to set yours to comply with his first to see if there's a difference.
The 32/33s are far superior to 30/31s, 110/115 to 220/230 should be irrelevant.
i'd definitely rip the BG out.(no criticism of BGs intended) |
|
|
| AndrewT |
Hi Jacco,| quote: | Originally posted by jacco vermeulen
....I've got 4 spare KSA50 output boards .......If there's a way of getting another 8 KSA50 output boards or the KSA50 Gerbers i'm a happy camper........ |
What if you can't make up the shortfall.
I'll take the four output boards.
Just throw then over the water and they might land on my doorstep. |
|
|
| AndrewT |
Hi Linesource,| quote: | | On-Semi ThermalTrak |
how many would you (or anyone else) use in the Leach style Vbe multiplier?
Leach went for three series diodes and changed to four to suit physical constraints.
I have found that four (1n4004 or 1n4148) diodes overcompensates (reduces bias as sink heats up) but it is a nice, easy & free, protection mechanism.
On-chip diodes will be more effective (both in speed and deltaT) and so I guess three will be sufficient.
Which three (from 8 or 12) should be chosen? |
|
|
| Mark A. Gulbrandsen |
I run large cinema sub woofer amps on 240 instead of 120 and it's always with a definate improvement in cleanliness on the extreme lows.
I think BG's are almost a joke.
I'd run the bias back up where it belongs and also let the amp break in for several days without shutting it down. Also let it play music even when no one is around.... I've found bias level to have the largest overall efect on how an amp actually sounds.
I used the 21193/94 combo in both the 50's I've built and in the 100. Keep in mind that when I get back home that I am going to revamp the 100 power supply and install that 1.5 kva toroid I have and have another listening session. In theory it should improve alot over how it sounds now... which is alot better than my 50 does.
Mark |
|
|
| Mark A. Gulbrandsen |
| quote: | | Having the thermal compensation diodes on the output silicon could improve the warm-up characteristics. |
As it is the 100 stabilizes in about 20 minutes. If one uses a smaller "Krell" size heatsink for the drivers it would stabilize even faster. Mine is quite a bit heftier than theirs was. With out the VBE multiplier on the actual O.P. device heatsink it's very stable over the long haul. I did not note any more than a few MV drift after a half hour warm up once things are set up... thats pretty good if you ask me....
Mark |
|
|
| jacco vermeulen |
I was planning to tell my grandchildren i used to talk to a Pink Mouse, in 20 years from now.
And when they run to their mommy, screaming that grandpa is foaming and hallucinating again, i'll boomerang the little angels unconscious from behind with a couple of output boards. |
|
|
| AndrewT |
Hi Jacco,
Is that a NO then?
Mark,
there's those Mega (MV) again.
I would not call that stable, dangerous more like!| quote: | | If one uses a smaller "Krell" size heatsink for the drivers it would stabilize even faster |
I'm not sure this follows. The lower case temperature might imply getting there sooner and less change when it does get there. The smaller sink has to let the case temp rise higher and this may take longer and will involve a bigger change in parameters while it is heating up.
I will come down on the side that the bigger sink will stabise more quickly (to within an acceptable tolerance). |
|
|
| Mark A. Gulbrandsen |
| quote: | | The smaller sink has to let the case temp rise higher and this may take longer and will involve a bigger change in parameters while it is heating up. |
I thought it the other way round myself. Sonce a smaller sink will reach temperature alot faster stabilization might occur quicker. The larger/beefier sink I am using only gets to about 35deg. C. with the bias at 2/3 reccomended level, not very hot.
Mark |
|
|
| AndrewT |
Well Mark, there's only one way to satisfy our (let's keep it personal since no else joined in) curiosity.
Measure it with your bigger sink,
then measure it with a smaller sink and post the two sets of results.
Starting bias and output offset,
bias and offset at 5minute intervals,
final bias and offset.
Try to arrange for ambient temperature and supply voltage to be similar for both tests.
Then we can try to identify how long it took each version to reach, within some selected tolerance (not too wide), the finish line. |
|
|
| Jozua |
Hi
Some quick feedback re the sound quality following the work done by Pierre.
I have removed the 2uF cap from the mains voltage at the transformer primaries at the wire harness. That made a 20-30% improvement in restoring the bass tightness the Krell. However I still find it lacking in realism, overall transparency and extreme dynamics. The attached picture refers to the cap change on the PSU.
I am hoping that the above sound attributes could be due to the fact that some burn-in is still required. For the meantime the amp stays as is. If there is no improvement my gut feeling is that if one might need to revert back to the original PCB drivers?
At the moment I am playing the amp without a cover and despite the lower class A rating the output transistor heat sinks still do get pretty hot..
I keep the group updated.
Jozua |
|
|
| jacco vermeulen |
Me recollects Mark posted he's a hands-On type, and prefers to leave the calculating to nerds and terds.
I remember doing civil engineering classes called System Dynamics, some 15+ years ago.
Part of the course was Dynamo training, the oldest simulation software.
As a civil engineer you'd have to be familiar with Dynamo, given that Great Britain was the first old world country to introduce it in the Euro zone.
Although Dynamo already was quite outdated when i got to know it, it didn't/doesn't do graphic displays, it did help me to understand the dynamic thermal behaviour of heatsinks better.
www.vensim.com/sdmail/sdsoft.html
Surprise:
an ideal heatsink (one that halves thermal resistance with 2 times the size and mass) takes exactly the same time to reach 1/2 the temperature difference with regard to ambient than a heatsink that has a thermal resistance which is twice as high.
Which isn't that difficult to understand, as in the end state both heatsinks have the same amount of stored heat with regard to ambient temperature and both have the same heat flow rate to ambient.
(of course this is not the case if the mass/heat capacity of the heatsink is low compared to the mass of the heat dissipating device)
Stuffing a thermal differential equation in a simple software package is easy.
In real mode, a heatsink that has 1/2 the thermal resistance has more than twice the mass.
This results in the bigger heatsink having a larger amount of stored energy when it reaches the steady state end temperature than the smaller heatsink.
So the bigger heatsink takes longer to reach the end state temperature.
There is a second way if you're that curious: you could do the measuring yourself and/or stuff a dozen command lines in a simulation software package.
Let's not interrupt your tête à tête, i'm just a terd passing by. :clown: |
|
|
| ANDYLASER |
| With regard to main transformers, please can someone give me the required specs. I know a company in the UK that do very high quality work and could ask them for a quote. |
|
|
| AndrewT |
Hi Joz,| quote: | | I have removed the 2uF cap from the mains voltage at the transformer primaries at the wire harness. That made a 20-30% improvement in restoring the bass tightness the Krell |
this result surprises me.
I would have expected interference suppression (at the mains side) to have more effect on the mid or treble frequencies. |
|
|
| Mark A. Gulbrandsen |
I've got the quote back and the DFM is fine but I still have to call Advanced today for the exact pricing on 200 boards, they only gave me vague pricing on boards at the 50 board quantity and 4 week delivery.
Since there are several of us that appreciate the look of seeing all of the gold plating what I am going to do is order 50 boards that have no solder mask at all but have gold plated traces and white silkscreen and the rest with gold traces and full soldermask... for those that still can't solder. These will be scheduled for 4 week delivery or what ever the cheapest lead time is when I call. I am pretty sure they can just pull 50 boards and skip the soldermask step and go directly to silkscreen. I will post the prices and info later today or first thing tommrrow.
Once this is posted I will give payment instructions on the WIKI and on this thread so you can make payments, it will be the same as the KSA-50 G.B.
I am only going to accept PAYPAL on this G.B. so if you are not set up with PAYPAL you may want to do so or ask a friend. If you are in the U.S. I will accept all other forms of payment as well however my Credit Union does not accept out of country checks of any sort. Since my Credit Union has been soo nice to me over the years I have no intention of opening a bank account just to satisfy a few out of country bank checks. If there is a volunteer out there that would be willing to launder the funds from those out of country individuals with real bank accounts and then forword that payment to me via PayPal would also be ok. Perhaps I can get LuckyLyndy to do this……?
Mark |
|
|
| Jozua |
AndrewT
I did the initial listening test with a transistor pre-amp that was fired up cold with the amp. The pre-amp has now been on one and half days and listening to the amp this afternoon I could detect a further improvement in speed and low end extention. I now suspect that the cold-start of the pre-amp could have had a impact on the overall sound.
It is summer in Cape Town and I was very surprised how hot the large pcb driver heatsinks became. My gut feeling is that if the case is not a constraint nothing can be lost by using larger heatsinks.
For the moment I intend to leave the PSU cap out.
Jozua |
|
|
| mpmarino |
| quote: | | ... for those that still can't solder. |
oooh! that hurts :eek: :) |
|
|
| still4given |
My wife prefers looks over function too. You are in good company. :D
Blessings, Terry |
|
|
| PWatts |
Ouch indeed.. I've a $1000 solder station and years of SMD experience and will still take the mask without thinking twice: the trouble without solder mask often isn't during the board population, but afterwards during testing, mounting etc. Besides, after two years the dust will have covered all those pretty traces anyway, especially with fan cooling.
People who wish for the "all-gold" look should just keep in mind that the silkscreen can only be applied to bare board or solder mask, not directly onto the copper (gold). With all the traces, pads and copper pours on the top layer, this will imply severely restricted and cut off silkscreen, i.e. the PSU decoupling caps' silkscreen will only be semicircles.
It would therefore perhaps look better to just leave the silkscreen out altogether on these particular boards instead of the partially complete one - after all, the elite people boasting proper solder skills should have no trouble coping without a silkscreen either :clown:
A bit delayed, but here are the spectra I promised of the three amps: Mk1, Mk2 original, Mk2 upgraded. All are of the left channel, fed with 1kHz 1Vrms input sinewave into a purely resistive 8ohm load.
Apologies for the low resolution; I had to make it very small to fit within 100kB. Also note that the vertical scale for the Mk1 (top) isn't the same as the other two. I'm not going to elaborate too much about the details, people can make their own conclusions. My KSA50 clone plots are on totally different scales etc. but what was very interesting to note was that its low-power spectra is significantly better than the KSA100's. Much lower noise floor and very low harmonics. When the power is increased though, the harmonics rise sharply.
The KSA100 (commercial) may perhaps not measure as good as the KSA50 at low power, but its big advantage is that its distortion characteristics remain remarkably consistent as the output power increases - this is very different with amps with small power supplies and many class-D amps in particular where the distortion rises rapidly with output power increase. This is also an explanation for the legendary bass grip, since the amp doesn't buckle under heavy load. I actually needed to do some burst tests as well to investigate transient power response, but that will have to wait until the Mk1-Mk2 upgrade is complete. |
|
|
| jleaman |
| quote: | Originally posted by PWatts
Ouch indeed.. I've a $1000 solder station and years of SMD experience and will still take the mask without thinking twice: the trouble without solder mask often isn't during the board population, but afterwards during testing, mounting etc. Besides, after two years the dust will have covered all those pretty traces anyway, especially with fan cooling.
|
id have to say that smd soldering is my favorite. I used to hate it but i challenged my self to solder lots of smd stuff and learn how to do it. Example Take something that is totally broken that is not worth fixing and trying to remove the smd part off of the boards to learn how to.
and to say, i love smd part's.
only advice i have to share is take your time and do one at a time.
so far the briangt/jleaman SMD mini aleph's are my favorite. Color and size.
Jase |
|
|
| PWatts |
I hear you - if it was just for myself I'd have made it a full-out SMD project using 0603 components throughout except for a few 1210's for the higher power. With proper via'ed pads all the semiconductors up until the drivers could be SMD types, although alternative choices available in SOT-223 or similar might be needed.
Especially with through-hole plated boards SMD is just so much easier to remove the components with a dual-tip soldering iron, fine angled tweezers and a variable speed&temp air blower.
But for nostalgic purposes and increased spectrum of resistor types to play with through-hole is the best choice for a GB - I did fit in 1206 SMD pads for the power LED's though;) |
|
|
| jleaman |
Yup i have had this soldering iron.. for about 5 years now and love it.
But just bought this blue weller off ebay for 60$ in march. The Weller 921 is ment for smd and small stuff so i now have 2. |
|
|
| Mark A. Gulbrandsen |
The main problem with SMT is that many parts are ONLY available in large quantities. To build a pair of mini boards boards would have had to order 500 SMT IRF input devices... and then there's the matching problem too, without a special jig to hold them its rediclous. I'll stick with through hole at least as long as its available. The surplus store that I frequent has finally gotten a rather large qiantity of SMT resistors and caps in many sizes and most values... But no semi's yet unless its an NTE.
As for the soldering comment that was meant as a tease, I just forgot the winkie face.....;) Sorry yous guys took it the wrong way. In reality any of you that can solder SMT could easily solder this board without any mask. I had all the semi's on and off one of the proto board three times while I was troubleshooting with no problem and no traces pulling up... I think that says alot for the quality of the board house and PWatts excellent design work. I have a 50.00 Weller station and with the smallest tip and smallest diameter silver bearing solder it works extremely well. Sometimes throwing too much $$ at something doesn't work either.
Don't know yet about the silkscreen but will later today. H-P silk- screened all their boards in either white or black print with no problem what so ever on top of some of the traces. But I also have no problem leaving the silk- screen out completely, assembly is very straightforword.
Jozua, You may want to consider building the 12B4 line stage... its the only one I've used that can get the full value back out of the Krell.
Mark |
|
|
| jleaman |
I like smd because for me it is easy. I like how clean the parts look and how small they are. Especially that the mini a's are so small and compact.
I was one shown and i proved it too. You can go out and buy a 1000$ soldering iron, and claim your the best soldering person. I know people that use a 5$ radio-shack soldering iron and can do just as good. It's not always the tool its how you can use it.
i only bought another one because the smd one can't heat larger things as fast and i don't want to burn it out. |
|
|
| Zen Mod |
Mark:| quote: | [i].............................
Jozua, You may want to consider building the 12B4 line stage... its the only one I've used that can get the full value back out of the Krell.
Mark [/B] |
nah.....
toobz are no more in fashion......
besides-they aren't so linear as SS.......
:devilr: |
|
|
| mpmarino |
| quote: | | Sorry yous guys took it the wrong way |
Not at all Mark. It was taken as a tease.:)
I'm looking forward to this build. I have too many amps around here and this may be the last amp build for a while (ya right). I think I'm gonna go 'balls to the walls' on this one... I'm even shopping for a (semi) full size vertical mill to help me make the chassis. Oh boy, here comes the endless money pit of machine tooling:rolleyes: ...oh well. |
|
|
| LineSource |
| quote: | Originally posted by AndrewT
Hi Linesource, how many would you (or anyone else) use in the Leach style Vbe multiplier?
Leach went for three series diodes and changed to four to suit physical constraints.
I have found that four (1n4004 or 1n4148) diodes overcompensates (reduces bias as sink heats up) but it is a nice, easy & free, protection mechanism.
On-chip diodes will be more effective (both in speed and deltaT) and so I guess three will be sufficient.
Which three (from 8 or 12) should be chosen? |
Hi Andrew,
I've used four series diodes in order to cover two PNP plus two NPN for symmetry.
I typically put a micro-switch that selects between a LOW and a HIGH class-A bias point resistors on my PCB right at the Vbe transistor for low inductance, and use a plastic chop-stick to slip in to push the micro switch. I think having the on-transistor diodes would provide some adantages in pulling-off this two-bias point design.
Dynamic plateau bias as used on the big Krells is more complex than I need, but I like saving on Global Warming when I can.
John Curl also uses a MOSFET in his VAS stage and commented on how the MOSFET helps isolate the back EMF from inductive speakers running high currents from coupling back into the front end gain stage. I think this diff-MOSFET VAS helps the Mark II KSA/KMA 100 have superior sonics to the KSA50, and suspect many KSA50 folks will convert to the KMA100 II klone, perhaps with a 2-bias point idea like I mentioned above to save some heat.
It would be nice to find a good price on a KL-271 heatsink profile for purists who will not accept fan noise. It typically cost $65. |
|
|
| Mark A. Gulbrandsen |
I agree its the person not the tools.... I've done some miracle soldering on 16 pin I.C.'s with a 150 watt soldering gun several times:hot:.
The verticle mill is a great idea... We have one at work that I use ALOT. I think I do more of my stuff on it than the company stuff. Get a really good vise, at least a 6" capacity precision type. Don't be afraid to spend at least $300 to $500 on a vise. Without one clamping things down gets hopeless but is still sometimes the only way to hold a piece.
Mark |
|
|
| Flodstroem |
See my post #693, its located here:
http://www.diyaudio.com/forums/show...0&pagenumber=70
Specs. for a suitable one is depending on what quality and size of transformer you are in the need of. you can give me a hint (are you going to use only ±52 DCV rails or are you planning to feed the LTP by a regulated voltage).
If using one xformer for both channels it will be BIG (ca 1.8-2.2kW). I plan to use one for each channel.
Regards :cool: |
|
|
| Harry3 |
"Apologies for the low resolution; I had to make it very small to fit within 100kB. Also note that the vertical scale for the Mk1 (top) isn't the same as the other two. I'm not going to elaborate too much about the details, people can make their own conclusions. My KSA50 clone plots are on totally different scales etc. but what was very interesting to note was that its low-power spectra is significantly better than the KSA100's. Much lower noise floor and very low harmonics. When the power is increased though, the harmonics rise sharply.
The KSA100 (commercial) may perhaps not measure as good as the KSA50 at low power, but its big advantage is that its distortion characteristics remain remarkably consistent as the output power increases - this is very different with amps with small power supplies and many class-D amps in particular where the distortion rises rapidly with output power increase. This is also an explanation for the legendary bass grip, since the amp doesn't buckle under heavy load. I actually needed to do some burst tests as well to investigate transient power response, but that will have to wait until the Mk1-Mk2 upgrade is complete."
----------------------------------------------------------------------------
Hi Pwatts
Does this mean that bi amping two KSA50 amps might give better results than one KSA100?
Also what power supply did you use for the KSA50 as you mention that the distortion increases with increase in power? |
|
|
| PWatts |
Harry3,
The KSA50 supply was a bit lacking, I had to work with what I was given. The caps were 55000uF per channel per rail, trafos were 500VA. Rails were a bit low at 35VDC; the trafo's were initially bought for a Stan Curtis amp. Higher voltage rails would probably improve the spectra, but the THD was still below 0.2% while pushing 30W into the load, but would have improved with the same power but with a smaller input signal and lower load impedance. The spectra of it driving 5W was remarkable and much better than the KSA100, so for an active biamped system the KSA50 should work superbly on the tweeters.
The KSA50 would IMO be a great amp for a biamped system driving the tweeters and mids. It will still struggle more with the woofers than a KSA100 driving the lot (with most speakers anyway).
I'd think a KSA50 driving the top end and a KSA00 for the woofers will be a good combination, but it will have to be evaluated to see how well they integrate with each other. A 3-way with the KSA100 driving only the woofers shouldn't have incompatibility issues as long as you account for the gain difference in mind.
I suppose the silkscreen-on-copper will depend on the board house, but it would be worth checking out before having the boards made. It would anycase perhaps be a good idea to leave out the silkscreen for the Golden Girls boards; many of the newer high-tech boards don't use it anymore besides for a logo or something - although it's mainly due to the dense component population that don't allow any space for Refdes and other silkscreen patterns.
Of course a $5 iron can do SMD with a little experience, and it's fine for a little project here and there, but use it for a minimum of 4 hours a day as part of your day job and you'll soon yearn for something better.
The assembly part is not that hard, removing is where it gets tough. For 2-lead stuff you can get away with one iron if you're very careful or preferably two irons, but I still have to see how one can remove a $500 160pin PQFP with it damaging NEITHER the chip nor the board. You also don't need a special holding jig for matching resistors; I use two sharp isolated tweezers connected to the multimeter. As long as you don't go higher than several megaohms or below 100pF it works fine for most matching/identification purposes.
If you buy from large agents you often need to buy MOQ's, but from the smaller vendors (here at least), you can buy as little as 10 pieces and often they give it for free since it's so inexpensive. Surplus stores are even cheaper, but at less than $0.01c per resistor it's hardly an issue.
For most projects I'd recommend the blue Weller stations as mentioned by jleaman. At less than $100 used on eBay they're cheap, and with it I've assembled huge boards with 0402 components and 0.5mm pin spacing chips without problem before I was given the larger station.
And yes.. the tools don't make the man - I've seen enough students screw up very expensive boards using that same station since they've no SMD solder experience. And then it's large stuff like 1206 and SOIC packages. Not good for their project evaluation too :whazzat: |
|
|
| jacco vermeulen |
Yeah Well, so what if i can't solder.
I just hit that CS8414-CS with a hammer.
Mind your own business ! :clown: |
|
|
| PWatts |
| There must be a curse with the Cirrus stuff.. somehow their products gave me the most migraines when soldering :clown: |
|
|
| jacco vermeulen |
| quote: | Originally posted by PWatts
big advantage is that its distortion characteristics remain remarkably consistent as the output power increases |
Pierre,
That is indeed how the distortion spectrum of the KSA-100MKII was described in mid 80s test articles.
The difference between the 50 and the 100 correspond quite well with mosfet characteristics imo.
I'll skip posting German Lingo Blah Blah, here's a picture of the KSA-100 from the article.(the PAM-5 on the left sucked big time, also demonstrated by the accurate article. Objective German testing went down the drain in the 1990s ) A French article on the KSA-100MKII at the time derived at the same conclusion.
The Accuphase P500, with twice the output power, finished after the KSA-100. On certain loudspeakers, like electrostats, i found the KSA way superior. |
|
|
| LuckyLyndy |
| quote: | | The Accuphase P500, with twice the output power, finished after the KSA-100. On certain loudspeakers, like electrostats, i found the KSA way superior. |
Jacco,
Is this also true of Magnepans, which are a bit of a different beast then true electrostats? I assume magneplanars are also a bit power hungry. I wonder how the sound/dynamics comparison goes between
SS amps like the Krell vs. the Tube (Hurricanes), and the Class D (UCD modules, ICE, TriPath) with the bigger Magnepans.
Lyndon |
|
|
| jacco vermeulen |
The true strength of a class A amplifier such as the KSA-100 is in the mids and high frequencies.
The biggest plus of a class AB spaghetti wire amp like the P500 is in bass reproduction.
Magnetostats require a lot of power but are overall less demanding than electrostatic loudspeakers.
The Maggies i've heard are not at the level of (some) electrostats when it comes to holographic imaging and mid/highs.
But still very strong areas compared to many dynamic loudspeakers, plus heaps of bass power.
I think i've even heard a combo of the largest Maggies/KSA100 once, Magnepan used to be very popular in these parts.
In my experience, tube amps can be as good or better than SS, but at a cost.
For a diy type it is easier to reach heaven's gate on silicate than glass. A reason for me not to mess with tubes in power amps anymore. I could have bought a set of Jadis goldplated armored vehicles with the pecunia i've poored into solid state over the years, but diy is my main therapy.
My sentiments towards class D is more outspoken than to BGs, cold and sterile.
You prefer a short response instead of my usual jabbing ?
Yes !
btw, the loudspeakers used for auditing in the KSA-100 era were generally in the $+10K league.
All the CBS artists on this club who describe class A amplifiers such as the KSAs as fossiles/historical artifacts seem very unwilling to mention the kind of loudspeakers they have. |
|
|
| Mark A. Gulbrandsen |
| quote: | | All the CBS artists on this club who describe class A amplifiers such as the KSAs as fossiles/historical artifacts seem very unwilling to mention the kind of loudspeakers they have. |
Of course all those artists have since become fossils in their own right!
Mark |
|
|
| Zen Mod |
| quote: | Originally posted by PWatts
......................
The KSA50 would IMO be a great amp for a biamped system driving the tweeters and mids. .................................... |
certainly......
for normal tweets - with just one pair of outputs,or even without-with slightly adjusted driver's loads resistors...... |
|
|
| AndrewT |
Hi,
I had originally thought of using the KSA50Ks as active tweeter only drivers.
But after hearing all the tributes began to think they would be wasted on that duty and thought that mid/treble duty through a passive crossover may get best use from the Klones.
However, these recent postings make me wonder if my first instinct is the one to follow.
No conclusion, until I actually try it for real. |
|
|
| Mark A. Gulbrandsen |
| quote: | | but I still have to see how one can remove a $500 160pin PQFP with it damaging NEITHER the chip nor the board. |
Agreed! That is a job for the right tools. I have however cliped out pin by pin other smaller SMD chips knowing well that they had failed and soldered in a new chip... Its not all that hard. When I worked at CANON we repaired the circuitry in the CANON cameras AE-1 and A-1. The A-1 had a large(for the time) SMD microprocessor that I believe was 36 pins.... This was all mounted to a flex board which made it all the more difficult ro remove and replace. This mp was too small to snip all the leads so it had to be unsoldered carefully and the new replacement soldered in. I lasted almost 4 years there before it got to be too much for my eyes:xeye: .
Mark |
|
|
| still4given |
Well, I have built a few amps by etching my own boards so I know I can solder without solder mask, but I must say there is some degree of confidence when using boards that have soldermask installed. If my vote counts for anything, I vote for soldermask.
Blessings, Terry |
|
|
| LineSource |
| quote: | Originally posted by AndrewT
Hi,
I had originally thought of using the KSA50Ks as active tweeter only drivers. No conclusion, until I actually try it for real. |
How about building a few KMA 100 revII's with a 2-level Class-A bias selector switch on the PCB? This strategy would give a pretty universal amplifier solution from one common design. I think a well designed KMA100 would have better sonics than the KSA50 based upon my past experience on my Apogees. This is why I favor a single PCB KMA100 design suitable for a compact monoblock. I'm interested how the ears on this tread think they compare.
I've become speaker centric over the years..........
I currently combine the active crossover discrete opamps plus two different wattage Class-A power amps onto one PCB and in one chassis. I use the same amp topology, but with a different numbers of outputs and bias. With PASS JFET opamps in front of the power amp, my amplifier input stage performs better with MAT02-MAT03 matched bipolars instead of the matched JFETs that are more commonly used to interface between separate boxed components, power supply grounds, and long cabling.
I thought of this when Mark G mentioned future plans to build a KMA160. I was thinking that stacking the active crossover into the amp chassis and bi-amping might make a hi/lo bias KMA100 design a better system solution than a more complex amp topology like the KMA160 and separate Xover box. |
|
|
| TomWaits |
Do you like PRP resistors ?
Based on the volume of boards signed up for in the wiki it looks as though we may be able to save $$ even on nice resistors?
An email I received from PRP today, "PRP has a minimum line order of $45.00. We can accept 20 line items at 10 pcs each if each line equals $45.00
Best Regards, Teri"
So it doesn't really work for one for 1-10 qty but if there are 100 or more boards, then this approach would certainly be an option.
Just thought I'd run it by yee all.
Shawn. |
|
|
| Jozua |
Hi
There are two issues that I think we should clarify.
a) Publish a parts list for the project.
b) PCB heatsinks.
Mark are you able to get suitable heatsinks in bulk or must we start looking at alternative options? At this stage this is the biggest concern to me.
If need be I can explore the possibility of machining the heatsinks in South Africa and then supplying to the group?
Jozua |
|
|
| PWatts |
| I've posted a full BOM here a few weeks ago but it sort of went by unnoticed by most. I'll do a few updates to it tonight and repost it. |
|
|
| still4given |
| quote: | Originally posted by PWatts
I've posted a full BOM here a few weeks ago but it sort of went by unnoticed by most. I'll do a few updates to it tonight and repost it. |
Hi Mark,
Maybe we can start a build wiki and post the BOM there so all can find it. Posts get buried quickly in a popular thread such as this.
Blessings, Terry |
|
|
| Mark A. Gulbrandsen |
| quote: | | If my vote counts for anything, I vote for soldermask. |
I am planning on both types to be available. I have not gotten my quote back fomr Advanced as of yet so I will call and holler at them today.
Mark |
|
|
| Flodstroem |
| quote: | | I think a well designed KMA100 would have better sonics than the KSA50 based upon my past experience on my Apogees. This is why I favor a single PCB KMA100 design suitable for a compact monoblock. |
Do you know where to find a schematic for the KMA100 revII ?
Regards :cool: |
|
|
| TomWaits |
I couldn't turn my back on yet another heat sink tunnel surplus item! :) This one weighs a ton, has a fan on each end and holds 2x10 TO3's. It is already machined out and the spacing between the TO-3's is generous and the unit is very robust. Each half measures 16.5cm wide by 32.5cm long. This too could easily chill a mean KSA100 down. :cold:
Planning in advance,
Shawn. |
|
|
| Harry3 |
I'm still not clear about the dual driver transistors. What is the advantage?
Do they need to be very well matched so they turn on/ off at the same time?
Also are we certain the cct we have is indeed the last version for the KSA100. |
|
|
| Zen Mod |
| dunno where I found this.........:xeye: |
|
|
| LineSource |
| quote: | Originally posted by Flodstroem
Do you know where to find a schematic for the KMA100 revII ?
Regards :cool: |
I think we can DIY a KMA100 by adding a driver board voltage regulator to this KSA100-II design and putting everything in a good monoblock package. |
|
|
| jacco vermeulen |
| quote: | | dunno where I found this |
Eeeh, on the KMA thread ? :clown:
Chok-olate,
any thoughts on why that thing has such a bulky servo ? |
|
|
| jacco vermeulen |
| quote: | Originally posted by Harry3
I'm still not clear about the dual driver transistors. What is the advantage?
Do they need to be very well matched so they turn on/ off at the same time?
Also are we certain the cct we have is indeed the last version for the KSA100. |
Harry,
The main drawbacks of the antique MJ/MJE devices are the low current amplication factors and low Ft values at high output currents.
At lower output currents their hFe and Ft are pretty good.
One reason why a bunch of the American 80s designs had big arrays of parallel output devices, such as the big Threshold and Rowland Research Model 7 amplifiers. And Krells.
Not just for flashing El Major hardware, for a reason.
At lower output currents, the later Motorola devices reach Ft values of 7 to 8 MHz, with hFe values that are as good as Jappon competitors. See a bunch of posts by one of the Motorola TO3 experts, DJK.
With fewer parallelled output devices the driver section needs to compensate for the lower hFe of the output devices at high output current levels.
As the available driver devices in the 1980s weren't that great at higher load levels either, splitting the load across 2 drivers keeps their hFe value more stable.
The idea of course is that the driver section keeps floating, at the output level where they start switching i doubt that it's a big thing if they don't switch simultaneous.
Dissipation of the drivers gets pretty nasty when the output stage has to drive really low impedance loads, AND doesn't have sufficient high hFe.
As you can see from the schematic the load on the drivers is reduced by halving the bias load, the KSA100 drivers have 75 ohms emitter resistors instead of the 25 on the KSA50.
(40 % more rail voltage, bias current reduced to 1/3d)
A way around the issue is by using more modern output and driver devices, which retain their hFe values up to higher output currents.
Or like in the old days, use a sheetload of TO3s in parallel.
The main reason for using TO3s is because they may be easier to get in some places and/or they are less expensive.
A suggestion might be to match the combined hFe products of driver and output device section.
I was hoping we could chat through another year or so on the different KSA-100 versions. :clown: |
|
|
| jacco vermeulen |
A thing to be highly aware of is that in low impedance loads the drivers get a serious beating.
If the voltage on the output stage rises, there's more output current going through it, means more current having to flow into the base of the output devices.
Suppose the output voltage doesn't go higher but the load impedance drops. Also means more current, again the base of the output devices need more current.
Then, if the output current becomes so high that the hFe value of the output devices decreases, again there needs to flow more current in their bases.
On top of that, in both cases the voltage drop across the emitter resistors of the output devices goes up because of the higher current going through them.
That means that also the voltage drop across the emitter resistor of the drivers increases !
Means there's also more current going through those resistors when the output stage is driven, NOT less.
And all of that current has to be supplied by the driver(s)
So why were these Krell designs so good in driving low impedance loads ?
One reason => Because they have a very sturdy driver stage compared to the less capable varieties.
In relative terms, the KSA-100 is a good example of an amplifier that can be seen as one with an output stage after the output stage.
The 4 MJE drivers are good for a total Pd of 200 watts, an output stage with a Pd of 200 watts is good enough for an average 50 watts/8 amplifier.
Less robust amplifiers that have to drive a low impedance load can burn their driver section before the output stage blows. |
|
|
| AndrewT |
Hi,
Jacco has made the case well for a generous driver stage.
One point I would like to add.
When the output stage is working into a reactive, the driver also feels the effect of that reactive load.
A condition to design for that is fairly extreme (but achieveable) is full output load current when the Vce is =PSU rail voltage.
For the KSA100 into 4ohm this would be about 40Vpk/4r = 10Apk @ Vce=50V giving 500Wpk through those four devices.
The drivers each see half this divided by the output gain @ 2.5Apk (10/4).
The driver dissipation is about 49V*5/40~=6Wpk.
A single driver would be dissipating 12W for the same 4ohm load.
The numbers for 2ohm (and god help us, 1ohm) get much worse. |
|
|
| AndrewT |
Hi Tom,
that tunnel leaks.
It will work much better if you can "tunnel" it.
I note that the two sinks are spaced quite far apart and the fins on the outside are in passive air.
Can I suggest you make a sheet metal tunnel and bolt both sinks inside it but much closer together. A small gap between the outside fins and the tunnel inner wall will work quite well.
Remove the fan that is sucking and place the pushing fan at the bottom.
Now you might get 5pairs up and running (but how do 5pairs fit on twin drivers?). |
|
|
|
