| dinu |
I'm trying to put together an a75 and am truly baffled. To start with, is it just me or is the hookup for the fan printed wrong on the board? R18 is hooked up to drain of Q1 and should be from base/gate to ground. The led seems wrong too.
THE BIG PROBLEM
The big issue is that i'm getting alot of AC ripple on the regulated supply. aprox 250mV :confused: SEE THE PICTURES i ATTACHED
IT GETS WEIRDER
The chassis is set up in a way that places the psb and the input board as well as the rectifiers and fuses on a 3mm aluminum plate which sits on the transformer at one end, held down by the transformer bolt, and two 70mm stainless steel pins on the other side. The plate is very firmly grounded to the rest of the chassis with a nive, heavy gauge ground bus wire BUT THERE'S STILL A 250mV AC POTENTIAL FROM THE PLATE TO GROUND. AND THE 70MM PINS GET REALLY HOT :hot: :hot:
Do I need to keep the plate a good distence away from the transformer? I even grounded the transformer bolt with a wire to earth ground. Is this the same 1/4v I see in the regulated supply? Should I stick to buying SONY?
PLEASE HELP. |
|
|
| Magura |
You can't place an aluminum plate on top of the transformer and ground the plate. That gives you a shorted turn for the transformer. DO NOT RUN THIS SETUP, as it may blow.
Magura :) |
|
|
| BobEllis |
Ditto - use insulated standoffs to prevent the shorted turn if the plate needs the support. the voltage you see on the plate is the result of the current flowing plate-leg-chassis-bolt. This current is probably coupling to your regulated supply, causing the ripple.
and yes, if you are using the Audioexpress boards, there is an error in the fan circuit. It will apply full rail voltage to your fan. Poof! (I ended up with 60 volts on a 12 volt fan that screamed for around 20 seconds) there is an unmarked hole in one of the traces around the fan regulator pass transistor that allows the proper connection to be made.
I assume that you found the silk screen errors on the main board.
I found the LED circuit correct.
EDIT: nice looking amp. |
|
|
| Alain Dupont |
dinu,
In my case I have used 4 CRC's
5 * 5600 MicroF + 0.5 Ohms 30watts + 5 * 5600 MicroF
added some 100 micro F plus 0.1 micro F,
gives 2*40 Volts and the ripple is down to 80 milli volts...
To see the details picture #5
http://www.passdiy.com/gallery/a75-p11.htm
Regards.
PS: For the fan I have made a temperature regulated supply in 12 VDC
and didn't use the original pcb {with the pb}... it's DIY ; but don't hesitate
to ask!
Alain. |
|
|
| dinu |
Wow, what a great community of helpful people! Thanks alot guys.
Now I can't wait to get home and rework.
As for the silkscreen errors, a few yanked out hairs later, I think I got them all. The article posted on the passdiy site seems to include the corrections. I just wish audioxpress made that clear with the instructions in the article they send with the boards.
The way the article shows the LED connected right to those tabs gets you + rail and - rail. (100v potential) anyway, I was thinking of just using a high value voltage divider to the unreg side so you can see the LED dim as the caps discharge on shut down. (OPINIONS?)
THANKS AGAIN :D |
|
|
| dinu |
Alain,
Just saw your post. I've been following your A75 and, like everyone else was very impressed. That tap problem got everyone pretty fired up. As for that 12v temp control. I'd love to see it. I was just thinking of just placing a thermal switch on the fan to kick on at about 65c.
Here's another pic of the face plate. Like the meters?
By the way, what do you think of your amp sound wise. |
|
|
| BobEllis |
that would work fine - just pick a resistor that will drop the voltage at the 20 mA or so you'll need. Check the power in the resistor and use one rated for at least double the dissipation. You could also go rail to rail. Either way it will take a few minutes to dim out. If it stays lit a long time, you might want to put bleeder resistors across the PS output - maybe 5K 2W. The idea is to bleed it off fast enough that the charge will be gone before you can unscrew the case top and get to it.
the LED on the PSU board is in series with a couple of resistors that limit the current and put the LED pins at close to ground portential. If you measure across the terminals without an LED you will see 100V, but as soon as you put the LED into the circuit, the current flows and voltage drops. I ended up using 15K 1/2W resistors to keep my blue LED intensity reasonable. It's about a 30 second fade out here.
I like my A75s - they aren't as pretty as Alain's but the sure sound nice.
Hi Alain. I figured I'd see you here, too. ;) |
|
|
| dinu |
Bob,
I didn't even hook up the led when I saw that 100V. What you're saying makes sense though. I'll try hooking it up as is. Worse case, I'm out a 50cent LED. I did place bleeder resistors on the caps. At 5k they do take awhile.
Thanks for the help, by the way. |
|
|
| Alain Dupont |
dinu,
the A75 is working with a pair of Electrostat Martin Logan Aerius i
and the preamp a CCS-X-Bosoz
I have listened to a lot of Class A amplifiers and others, especialy here in Montreal there is a "Festival du Son" each year ; to listen to the world best Hi-Fi set-up's
I am pretty happy with this, and enjoy every second of music! :)
To be honnest, the only one that impressed me last year was a pair of Passlabs XA250!!! I think they are not cheap, but sooooo good...
I am working on a Tube SE 6C33C-B 2*20watts and a pair of
back loaded horns with Fostexe's...
PS: BobEllis is a talented builder, but too modest!:angel: :)
For the temperature regulation I'll check my notes.
You have done a great job! :)
Alain. |
|
|
| Alain Dupont |
dinu,
Welcome to the A75 Club !!!!!!! one of the best DIY designs !
Thanks to the Master.
Hey Bob, pleased to see you here! ; thats a lot of watts gathered here!
Pure Class A ; wouahhhh ! Bob is alway there to help and a real friend.
I can tell that 2 friends have listened the A75 and all were pleased
with the details, the stereo image is wide and in case of big orchestral
pieces it's a marvel! {Compared to Audio Research D100 and other's}
The only drawback in summer ; you must use the clim.
Enjoy!
Alain. |
|
|
| ckwong99 |
Dear All,
Start to make my A75 before, have few questions consult to your preempt.
1) I have 2 300W 2 x 40 VAC transformers not sure it can be applicable for using on output stage supply ?
2) Additionally, I have made one +/- 68VDC regulated supply, does it sutiable for input stage supply, so I can't adopt the voltage doubler design ?
3) In hand I have 14 pairs of Hitachi J56 & K176 MOSFET want to substitute IRF power stage . Thereby two technical problem in my mind , ONE) I only make an A75 with 14 pairs output each channel use 7 pairs, is it possible ? TWO) Don't know they can directly use in the circuit not need to do revise ?
Thanks for your instructions.
CK |
|
|
| dinu |
| Bob, I read your thread when building your a75. I'm not sure I understand what you are talking about with the coupling caps. And also, I saw nothing about an extra diode with the audioXpress package. Could you enlighten me? |
|
|
| BobEllis |
ck - I modded my A75s to use a 42 -0- 42 C transformer I had. It works fine.
1 - That should give you ~54 volt rails. 300 VA is a little small, unless you reduce your bias to areound 1.5 A total given the higher voltage. It ought to get you going, be sure to provide adequate ventilation for the transformers
2. 68 volts is suitable for the 55 volts you can get with your output stage tranny- watch the dissipation on the 10K resistors from the cascode gate to ground. You can increase this value a bit to reduce dissipation or just use a 1W resistor in this spot.
the dissipation in the input differential will be high - over 1/2 watt. The IRFd210/9210 will take it, but with reduced life. I ended up bending legs and using IRF610/9610s in their place. They don't get quite so hot and sound a little better.
3. I think your hitachi devices are at least 150 watts. NP said that 6 pairs of IRFs should be plenty, so that is what I used. The catch for you is that the Hitachis turn on at lower voltages than the IRFs. You need to rework the Vgs multiplier to avoid driving your output stage full on. Turn it into a Vbe multiplier substituting something like a MJE15030 for the IRF610. You may also want to look for the Hafler DH-500 schematic (or another lateral mosfet amp) and copy that biasing scheme. The connections are the same in a Vgs multiplier, so you can just change the resistors and transistor.
Enjoy the A100 (depending on your heat sink capability you should be able to get 100 watts class A with 54 volt output rails, and at least 125 in AB.
Dinu,
I should have said DEcoupling caps - use around 100 uf from each rail connection to ground, right at the board power connection. Although my oscillation went away with a 2 uf film cap, I used 220 uf, 2 uf and 100nf on each rail to be sure. |
|
|
| dinu |
| By any chance, has any of you put in any kind of output protection on your A75? Or maybe you could make some suggestions. I have a nice pair of Joseph Audio speakers I'd like to keep as is.....just in case.:dodgy: |
|
|
| GRollins |
Although a bipolar Vbe multiplier can probably be made to work, I think you'll find that you get better bias stability if you use a MOSFET as a Vgs multiplier. The thermal characteristics will match those of the output devices more closely. A lateral device would be easier to adapt to this purpose than the IRF part (I think the A-75 uses an IRF510 or 610, but I'm quoting from memory and could be wrong).
Bob's suggestion about looking at the Hafler DH-200 or DH-500 bias circuit is a good one, but the parts might not be available these days. Wouldn't hurt to look for similar parts, though, if that should turn out to be the case.
Grey |
|
|
| ckwong99 |
Thanks to all your inputs :>
1) I have planned to use x2 300W transformer for it , does it get enough ?
2) Actucally, I don't know how to rework the bais part circuit. So do you know I need not make any change, just adjust the bais voltage around 1.7V - 2V for the PWD MOSFET operate ? Can this master circuit support I do that ?
CK |
|
|
| BobEllis |
CK
one transformer per channel will be adequate if yo don't bias too highly - 1.5A total will around 160 W drawn, a little over half the transformer's rating (a DIY rule of thumb is never use a component at more than half its rated power) This would give you 36 watts Average into 8 ohms class A, 72 Peak - more than enough for most home use. You'll still have around 150 W average in class AB.
The bias circuit must change to use your hitachi outputs. A Vgs Multiplier multiplies the gate-source voltage of the IRF610 by a number greater than 1 depending on the ratio of pot setting and the fixed resistor above it (R60). Since Vgs on an IRF610 is on the order of 4 volts, you're going to have too much bias voltage for your outputs. That is why I suggested using an MJE15030 for Q11. Your reference voltage (Vbe) will be around .6 V so you can set up your multiplier to give you a reasonable bias voltage. You can probably use the circuit just changing Q11, but be sure to use a multi-turn trimpot for P3 or you will drive yourself nuts trying to set the bias.
I tried to attach the DH-500 schematic but it is too big. Email me if you want a copy.
Actually, all the pots should be multi-turn types. I wish I had spent the extra $3, it would have saved an awful lot of aggravation.
Grey -
The hafler DH500 uses a bipolar (NP2222A) in its bias circuit. I suppose it is relying on the negative temperature coeficient of the lateral devices. It does not use source resistors, either.
Dinu,
Nope, I did not add any protection to my A75. I might have, had I ended up with more room in the case. there are several threads here that deal with protecting against output faults. The DH500 schematic also has a protection circuit based on an op amp and a few general purpose transistors that does turn on delay and DC detection.
As for the extra diodes, I cannot remember the connection but I think than they were on Q9 and Q10. The purpose was prevent input stage damage if overdriven, IIRC. Perhaps AudioXpress will provide the update info to you - they should have included it. |
|
|
| Alain Dupont |
Hi,
Here the note from Nelson about these... |
|
|
| dinu |
| thanks alot Bob. That helps. If you could email me the DH500 schematic it would help alot. I read alot of the threads and find the technical explanations a bit confusing sometimes. Are there any good books that can help with these apparently fundamental EE principles. Much to the amusement of some of you out there I'm an ME. Basic electrical design is all that was required and I was only too happy to fulfill the minimum at the time:apathic: |
|
|
| Alain Dupont |
And the explanations...
Alain. |
|
|
| dinu |
| Good morning Alain. Thanks for that but it's hard to make out which transistors those are and what kind of diode. Could you clarify? Thanks |
|
|
| dinu |
| Guess I shouldn't respond so quick. Thanks Alain. That written explanation was crystal clear. |
|
|
| BobEllis |
Dinu,
send me an email though the system here, and I'll shoot you the schematic.
you're not the only non-EE here. Jacco Vermullen (sp?) and I are Naval Architects - ME-s who only worry about mild steel structure. |
|
|
| dinu |
| :cannotbe: I see Q9 and 10 but the other two are Q3 and Q6. Is that right? Where's Q6? :confused: |
|
|
| Alain Dupont |
Dinu,
Well the camera is handy this morning...
Alain. |
|
|
| dinu |
| Thanks alain. Picture's pretty foggy but it Looks like Q3,4,9 and 10. Right? |
|
|
| dinu |
| ahhh yes. Q6 is mislabeled as Q4. Duh.:rolleyes: How long does it take?:headbash: |
|
|
| Alain Dupont |
dinu,
Welcome to the A75, it's not for everyone!
But if you manage to make it, wouahh ...
I was helped by Bob, and many others, but I lost everything
while crashing my HD... Happy to put some papers a side...
And since the A75 there has been the Bosoz, Ono, Aleph-X, and tube SE
Exept for the fan regulation I kept all prints, but managed to loose this one.
Best regards.
Alain. |
|
|
| dinu |
No problem. My first build was the zen V7 (induction loaded output) It has three 1KVA transformers in it. It must weigh damn near 90 pounds. I thought THAT was tricky. I'll finish the A 75 if it kills me but I hope it's worth the effort.
I was just talking to Bob about the BOSOZ. It's built but still not working right. I think it's the volume control. |
|
|
| BobEllis |
| Rod Elliot's site has a couple versions of temperature driven fan controllers. |
|
|
| GRollins |
I thought I had remembered the Haflers using MOSFETs in the bias circuit but didn't have the schematics with me at work to look them over. I'll scratch around and see if I can find them. It's been quite some time since I've sat down with them.
Grey |
|
|
| BobEllis |
| I think you'll be impressed Grey - the way I see it, if the wiper on the pot goes open, you end up with max bias - good for the service department. |
|
|
| dinu |
Morning everyone. The original problem I started the string with is definitely fixed. THANK YOU ALL. The power supply has less ripple than I can measure with the MM, fan runs nice and quiet and the LED works and is not too bright. I put a ferrite band around the transformer which also seems to suppress any stray magnetic flux. Front end goes in tonight. I'm trying to decide if I need bother with heat sinks on the front end 220's. The board is right in front of the fan which is moving a decent amount of air past it. I'm guessing it should keep them pretty cool although there's no such thing as too much heat sinking.
What's the consensus on R81? I was thinking of using a 3 position switch. (ON, 500ohm, and off) That should give me (full, 50%, and 100%) feedback. I'd love to hear what you all found sounds best? |
|
|
| BobEllis |
Even with the fan, I think I'd use sinks on Q9/10
I'd originally planned to do the same thing and switch R21/22 using the folded cascode or not, but never got around to it. I use the cascode and R81=0, and am happy with it.
I look forward to hearing the results fo your experiments |
|
|
| ckwong99 |
Hello Bob,
Comes up all your advice seems just replace IRF610 by Vbe multiplier and uses 20 turns VR is very simple.
Yesterday in my bench found two N Channel MOSFET 2sk216. Its characteristics is similiar to output stage K176. I think it can help to improve the bias voltage stablity.
How to modify the voltage multipler resistor network ?
Thanks too much
CK |
|
|
| BobEllis |
CK,
I'm sure someone could calculate ideal resistor values, but since like the original your Vgs transistor is similar to your output devices, I'd try the circuit as is. Just swap in your 2sk216 and see how it works with no output stage connected.
If you cannot adjust the bias voltage low enough to almost cut off your output stage, I'd use a smaller R80 as NP recommends in the article. Being able to get up to 4 volts ought to be plenty, although I have not worked with lateral fets. I suggest looking for an amp design that uses them and see what bias voltage is used. |
|
|
| GRollins |
Jeez, I hate the search engine here. It's bad enough that I lose threads, but even knowing what I'm looking for doesn't seem to help.
Anyway, life being what it is, I had cause to dig out a Hafler DH-200 schematic the other day on a service-related sort of thing and Bob's right--it's a bipolar bias circuit. In fact, the whole front end is bipolar, which isn't at all what I had remembered. I thought the front end differential was bipolar and everything else was MOSFET from there on. Oh, well, point to Bob. Thanks, man.
Grey |
|
|
| PierreQuiRoule |
| quote: | Originally posted by dinu
What's the consensus on R81? I was thinking of using a 3 position switch. (ON, 500ohm, and off) That should give me (full, 50%, and 100%) feedback. I'd love to hear what you all found sounds best? |
dinu,
Your "switch" idea sounds like a good one.
You are unlikely to get a consensus on R81: the A75 sounds pretty darn good in all of its configurations. The flexibility offered by the design (e.g., percentage of global feedback, folding cascode) allows you to voice your end-to-end system sound in accordance with your personal taste and listening habits.
In my case, with 50% global feedback and stock configuration (no folded cascode), the sound is detailed, neither bright nor dark.
With 0% global feedback and folded cascode, the sound is darker - as in less bright. It is also more "laid back" - details don't draw as much attention, the image is recessed a feet or two, the whole spectrum is seamlessly integrated. The result is very pleasant. For now I am sticking with this... until I move on!
Pierre
PS: I would be curious to hear the opinion of others!
Source: Sony DVP-S9000ES
Preamp: passive or BOSOZ
Amp: A75 with reduced bias (1 Amp per side)
Speakers: DIY MTM towers (PHL 1340 and Aurum Cantus G2)
Music: Classic rock, jazz (trio), classical piano, and opera. |
|
|
| dinu |
Thanks Pierre. I haven't had a chance in a good three weeks now to even touch it. Did you try 100% feedback? Am I seeing a pattern? 0% = less pronounced detail, more reserved sound stage. 50% feedback = slightly brighter sound with a bit more forward soundstage, Can I take a leap and assume what 100% will yield? Possibly very bright detail and sound staging further forward in the room? I too would love to read more oppinions.
PS
I previously asked about output protection. I ordered and received the Velleman kit from Audioxpress. Have yet to assemble, and now that I have it, I realized I have very little room for it. Typical ! |
|
|
| PierreQuiRoule |
As requested, here are some personal observations pertaining to the A75 sound using folded cascode and a balanced line stage. Beware that all this is very subjective. For the following, I listened to Bill Evans (Analog Production SACD), Angela Hewitt (Hyperion CD), Artur Schnabel (NAXOS CD), Eagles (DCC CD), and Diana Krall (Verve SACD).
With 100% global feedback: detail, precision and holographic spatial cues convey an impression of listening "into" the music. On Diana Krall, I notice a hint of sibilance; on other albums, a crisp top end. Good rythm and excitement.
With 0% global feedaback: the top end is smooth, the sound stage wide open, and the overall listening experience relaxing. A word that springs to mind is "natural".
Both settings sound nice on their own right. Please note that the differences discussed here are subtle, and that I find them to be challenging to describe. Upon casual listening, one might find that the two settings sound very close.
If you build and listen, please share your findings!
Pierre |
|
|
| dinu |
| Well, that clinches it Pierre. I have to make R81 a 3 position switch that I can access externally. It seems like each feedback setting would be suited to a perticular type of listening. Thanks alot for the input. |
|
|
| dinu |
Guys, HELP.
I'm almost done with my A75 but am having problems. I tested the front end the way described in the article and it works fine. The problems arise when I plug in the output stage. Voltage on R7 and 8 drops to about 350 mV. No matter what I do to P1 and P2 that figure does not increase much. I get alot of DC offset at the output. Like almost a volt in un balanced mode. I also get more than 11V of AC output signal. I see nothing on the source resistors. What the :bawling: :bawling: :smash: :smash:
I hope you guys can help. Thanks for looking. Dinu |
|
|
| BobEllis |
Are R9 and R10 still showing ~5V?
What voltage do you get at the drive terminals? AC and DC -
You won't see any voltage on the output source resistors until you bring the bias up with P3.
what are gate-source and source-drain voltages on Q3, Q6, Q9 and Q10?
Check all your connections to the output section. What are your rail voltages?
Did you disconnect the output stage and verify that the front end still works? It may have been damaged somehow.
What input voltage gives you 11V out? (is it amplifying properly?)
that ought to give you something to start with |
|
|
| dinu |
Are R9 and R10 still showing ~5V?
~~Yes. I get about 5.25V
What voltage do you get at the drive terminals? AC and DC -
~~When it's all connected I see .01V on -DRV and 3.2V on +DRV That's DC. They both have about 11vac.
You won't see any voltage on the output source resistors until you bring the bias up with P3.
~~yes but I see nothing even when I turn up P3.
what are gate-source and source-drain voltages on Q3, Q6, Q9 and Q10?
~~Q3 4.8 Vgs and 4.1 Vds
~~Q6 4.26 Vgs and 4.1 Vds
~~Q9 4.66 Vgs and 23.9 Vds
~~Q10 4.03 Vgs and 22.1 Vds
Check all your connections to the output section. What are your rail voltages?
~~I've gone over these several times. the P channel drains are connected to - unreg. N channel to +unreg. -Drv to gate resistors of P channel and +drv to N channel gate resistors. output is off the source of all output transistors. The reg supply puts out + and - 56V and unreg puts out + and - 47V.
Did you disconnect the output stage and verify that the front end still works? It may have been damaged somehow.
~~I've disconnected several times to see if it still works properly and it seems to me like it does. It passes all the criteria. almost 10V across the Zeners, 5V on R9 and 10, 900mV bias at R7 and 8, and I can get the DC offset to nearly zero. Plug in the output and it all falls apart.
What input voltage gives you 11V out? (is it amplifying properly?)
~~input voltage is about 320mV. I don't have an oscillator so I used an ipod with a 1Khz tone I downloaded as an MP3. I cou;dn't see any reason why this wouldn't work. Only problem is that 300mV is about all the signal I can get from the ipod. I am an amateur ya know
that ought to give you something to start. |
|
|
| BobEllis |
| My guess is a bad output transistor or two. If the DC voltage on the drv terminals is as adveertised without the output stage and drops when you hook it up, you either have a bad 9240 or the gates are accidentally shorted to ground. |
|
|
| dinu |
| Yeah, I was afraid of that. I reallydidn't want to start taking apart the output stage but I guess I have no choice. Do you think that the fact that -drv shows such little voltage is a sign that the problem lies with the P channels? Well, if I build this thing enough times I'm just bound to get it right. WheI actually get to do this is another matter altogether. I just got a new puppy recently too so that's a huge drain on what was already nearly non existant free time. Thanks again for your help Bob. I'll let you know what develops. |
|
|
| BobEllis |
well, if connecting the output stage causes the -drv voltage to drop to 0, it sure seems like the problem is connected to the -drv.
the puppy could actually help - it might keep you away long enough to spot a mistake. Could you have made a mistake in the feedback switching setup that you mentioned trying?
good luck! |
|
|
| dinu |
Good call on the puppy. I don't know why, but frustration at something not working properly just makes me try the same thing again hoping that, by some act of divine kindness, it will work; Trying the same thing over and over and expecting a different result. The name for that is "insanity".
Well, I'll just have to take the P channel apart. R81 is open right now. I figured I'd adjust everything first and then hook up the feedback switch.
Talk to you later, |
|
|
| dinu |
| Before I get back to the output stage I still have to work out a bug in one of the front end channels. I can't get the bias to stay up. Adjusting P1 and P2 brings up the bias on R7 and 8 as well as the voltage on the drains of Q2 and 5. The drain of Q6 and gate of Q10 come up very slowly after the amp is turned on. They wander up to about 39-40V and then stop. As you bring up R7 and 8 they're constantly dropping. At full clockwise, the bias eventually settles to about 350 mV. I've matched the input fets Q1,2 and 4 to 5. The 220's aren't matched but I've checked that they work. Please help. |
|
|
| BobEllis |
| quote: | Originally posted by dinu
Before I get back to the output stage I still have to work out a bug in one of the front end channels. I can't get the bias to stay up. Adjusting P1 and P2 brings up the bias on R7 and 8 as well as the voltage on the drains of Q2 and 5. The drain of Q6 and gate of Q10 come up very slowly after the amp is turned on. They wander up to about 39-40V and then stop. As you bring up R7 and 8 they're constantly dropping. At full clockwise, the bias eventually settles to about 350 mV. I've matched the input fets Q1,2 and 4 to 5. The 220's aren't matched but I've checked that they work. Please help. |
OK, I am confused. Where is the 350 mV measured? R7/8?
Are you getting 5 V across R1, R6, R 9 and R10?
Do you mean Q6 Drain and Q10 Source go to -40V referenced to ground and drift towards ground? With no signal it should be around 5V below the rail.
What is the voltages on Q9/10 relative to the rail?
check Q11 - |
|
|
| dinu |
Hi Bob,
I can't get a good grasp of wht the circuit is doing. It does different things at different times. I just turn it off and then back on again and sometimes get different results. Well, most of the time, these are the values I get. Rails are at about 57V. The most voltage I can get across R7 and 8 is about 350mV no matter how far I turn up P1 and 2. Once it all stabilizes, I see 6V across R9 and 10. R1 and 6 start at about 4.5V and slowly climb up to about 6. Q5's drain starts at about 30V from rail and slowly climbs to settle at about 10. Q9 and 10 both have the same values from their respective rails. Both drains see full 57V, the gates are at about 6.7, and the Sources are at about 10. There is no voltage across +and - DRV. I'll just replace Q11 again to make sure but I've had it out and it tested fine at about 20mA. Is it possible for a FET to be alittle broken? I don't know if that sounds dumb but is breakage in a FET always an absolute thing?. I sure hope you have some ideas Bob. The puppy's been eyeballing this thing as a new chew toy and I'm pretty close to letting him have it. (it's going to the dogs):devilr: |
|
|
| BobEllis |
no voltage across the DRV terminals means either there is no current flowing in that leg or Q11 is a dead short (Turned on completely or blown) With P3 set at the midpoint do you measure about 2.5K gate to source on Q11? With power on and p3 set fully CCW, what is the voltage across R80? (looking for circuit continuity.)
Moving on - the gates of Q9/10 should be 9.1V below the rails. Check the voltages across Z1/2.
Check also across R17-20. Any voltage there indicates a blown FET.
If the Q5 drain (and therefore Q6 Drain and Q10 source) is 10V below the rail (10V across S-D of Q6) Q10 cannot turn on when its gate is below its source. therefore you'll get no current in the Vgs multipler - you can skip that testing above.
If You have nearly 6Vgs on Q6 and you get 10V across it, there is a problem - it doesn't seem to be turning on properly.
Not sure of all the failure mechanisms, but I seem to remember NP saying that damage can be cumulative, so it sounds like they can be partially damaged.
Since you seem to have a problem with at least Q6, I'd start by replacing all of the 610s and 9610s, as well as Z1-2.
Did you add back to back zeners on the inputs to protect against static damage? If not, replace the input differential devices. They seem to be drawing too much current - as indicated by the 6V across R10. This may be due to something funky with the VAS, so you might try just replacing the To-220's first.
Replace all the semiconductors is probably not what you wanted to hear, but that's what I would do next.
Please state what the voltage reference is - I assumed that when you described the Q9/10 voltages they were referenced to their respective rail, but that is not clear. If you reference all to ground, a rail or say that Q10 source to drain is 47 volts, it becomes clear what you are measuring.
Good luck |
|
|
| dinu |
Hi Bob,
well,across gate to source on Q11 I get about 1.5K with P3 at half point. I also get no voltage across R 80 when P3 is full CCW. I decided I'll replace transistors just to make sure. I need to go get a couple but while I was looking I noticed that the 610's have an Rds of 1.5 Ohm not <1 as stated in NP's article. Looking in the parts box, I found some 9640's (125W, -200V Vdss, .5 Ohm Rds) I thought I might pair them with 510's. (43W, 100Vdss, and .54 Ohm Rds). I wonder if the Rds figure really matters. The other channel seems to work fine with 610's. Anyway, I'll do what little bit I have time for and let you know. Again, I can't thank you enough for the help. |
|
|
| BobEllis |
You're welcome.
No voltage across R80 means no current - but we expect that with Q10 shut off.
ok, you've got continuity across P3.
I don't think Rds matters since this is not a switching application.
did any of the gate stoppers (R17-19) show any leakage? |
|
|
| dinu |
Hi all,
I'm moving along with this thing at a snail's pace. The front end seems to work fine now and am trying to deal with whatever damage I did to the output awhile back. With regards to the output, an interesting question was posed to me. Why the need for all the fets in parallel? Would much fewer and higher power fets not have worked as well. Were these high current fets just not available when the A 75 was designed? I'd love to hear if anyone did anything like this and if so, what they used. |
|
|
| BobEllis |
NP has said several times that the A75 will do just fine with 6 pairs of outputs. IIRC, the original choice of 12 pairs had something to do with what heat sinks were on hand.
I used 6 pairs of IRFP240/9240 without issue (once I got a few mistakes out of the way.) |
|
|
| dinu |
I was actually thinking of as few as 2 to 4 pairs. I saw a reply by Nelson to another post where he suggested IRF540 and 9540's. Those have a peak current of 30 amps. One could probably get away with just a few of those and a much lower value source resistor. Suggestions?
I also have to admit I tried cheating and didn't match the output fets. I used 12 IRF230-9231 pairs per channel. I was hoping that the source resistor would take care of any reasonable Vgs difference. Well, I'm still trying to figure it all out but it didn't work. So now I either pull them all out of the board and test them all or just start fresh with the output.
While I'm at it, anyone have some matched sets they'd like to sell? I'm desperate and probably a pretty easy sell. (sounds like a bad personal ad).
The battle continues. |
|
|
| BobEllis |
Current rating is only part of the issue. The biggest obstacle to going to two pairs is the heat. With 2 pairs and the design bias of 2.2A you've got about 42W in each device. with Normal thermal resistances junction to case and case to sink, your junctions will be 95 degrees hotter than the heat sinks. Makes it tough to keep them below 100C unless you keep them in the freezer. ;)
4 pairs should be doable, if you have enough heat sink - you'll need bigger than you could get away with for 12 devices.
One of my sets of lost outputs was due to misreading the labels I'd put on my outputs as I measured Vgs. One device with significantly lower Vgs than the rest ended up taking almost all of the current, at least for a while. I noticed that one part of the heat sink was significantly hotter than the rest but it didn't register until the smoke came out, taking several others with it. of course, I was using smaller than recommended Source resistors.
In other words, I suggest that you match your outputs. You can simply measure the voltage across the Source resistors in place as you bring it up. If you got lucky, you'll find that they are all close (within .01 or .02V). If they are way off, pick the 6 closest devices and disconnect the rest. Even if your output stage is no longer connected to the front end, it would be a good idea to match them on the sink. |
|
|
| dinu |
Hey Bob,
All the output fets are on the sinks and disconnected from the front end. When I did have them connected I could turn P3 as much as I liked, I still saw no drop on the source resostor. Of course, I wasn't looking at all the resistors so I bet a couple of them had a drop, and a good one at that. I would like to symplify the design as much as possible so I think I'll try the 4 pair approach. I just need to find four matched sets. It should be alot easier a task than 12.
Interesting side note, I made the swith for the feedback. With r81 open I see the output square off almost like it's clipping. I don't get why. (This is just the front end with no ouput devices). |
|
|
| BobEllis |
I remember NP saying that the front end cannot drive the zoebel. maybe that's what is causing the clipping.
If your outputs are from the same lots, then you may find that you have pretty good matching. I bought 50 of each and sets of 12 weren't great matches, maybe within .02-3, but got 6 or 7 sets of 6 within .01V. |
|
|
| dinu |
Yup,
Even a blind squirrel finds a nut now and then. I matched the output fets on the board as you suggested and kept only 6 pairs. It's a damn good thing too because there would have been no way the heat sinks were big enough for 12 pairs at 170-200mV each. As is, I am able to bias to almost 250mA on the source resistors and still keep the temperature well in check. I still have a few details like hooking up these cool old VU meters. I know they serve no purpose other than to look cool but that's sometimes purpose enough.
I do have more ripple on the power supply than I'd like and need to take care of that somehow because I hear it in the speakers. I might need to use shielded mic cable on the inputs. I also made the switch that selects between 100%, 50% and 0% feedback. This was not such a hot idea for one reason. You can only adjust the DC offset for one feedback setting. As soon as I flip the switch to another feedback setting, the DC is way off again. I'll have to pick one and stick eith it.
I want to thank you all very much. This thing would have been flung off the deck long ago if you all weren't there to reel me back in. I'll be sure to follow up with some pics as soon as I'm done and hopefully add it to Nelson's gallery. Hope it makes the cut. |
|
|
| BobEllis |
Congratulations. I'm glad we have another member of the A75 club. Other than the hum, how does it sound?
Have you measured the rail ripple? This amp should be relatively immune, you may be picking up hum from AC mains routing. Is it 60 Hz or 120 Hz? For reference I used only 45,000 uf per rail, and have no hum issue. (at least once I got my grounding squared away) |
|
|
| dinu |
| I have 120000uF on each channel and about 200mV of ripple on the unregulated supplies. Off the top of my head 120Hz sounds right. I also see a real lot of gain (40dB). R 21 and 22 are 100 ohm each as listed in the article. My assumption was that any ripple on the input may be greatly amplified. Have you tried different values for R21 and 22? In the article, Nelson mentions that at 10K Ohm you'd see about 20dB gain. That sounds much better to me. I should be able to get about 4V out of a preamp and that would clip the amp anyway so I don't really see needing more than 20dB. I also assumed that with higher values for R81 I would have less gain. That was not the case. Anyway, I'm having fun experimenting now. I still am very grateful for any tips you think may help me. |
|
|
| BobEllis |
If you are getting a gain of 40 db you are missing something in the feedback circuit - you aren't getting any. I think Nelson meant to say open loop gain "of" this stage rather than "at". Not that it really matters, closed loop gain should be around 20V/v (26 DB) single ended. Check all resistors in the feedback loop, including R2 and R4.
The ripple on your unregulated supply should be no problem. Check that your ground connections are tight (voice of experience), and that the star ground is not the junction between positive and negative rail caps. A lot of current flows there and could induce noise in the signal grounds. I did a simulation of the AX supply I intend to build and each cap in the CRC sees at least 24A peak current. :bigeyes:
Another thing to check is the mains wiring to the power switch/triac. My first A75 was sloppy and the leads were not tightly twisted together and against the chassis. The triac switching noise sounded like 120 Hz, with an edge, until I dressed the leads properly. |
|
|
| PierreQuiRoule |
If I may, please allow me to throw in my 2 cents...
Bob is right: your gain is high. With R21 and R22 open (the amp run stock), I measured 20 dB. With folded cascode (R21 and R22 at 100 ohms) the gain goes down (to 6 dB if I remember correctly).
If you are using the PCBs from the Old Colony Sound Lab, you will have noticed mistakes on the silk screen. My notes show that R24, R26, R80 and R81 on the schematic respectively correspond on the PCB to R22 (next to Q11), R24, R21 (next to P3 and Q11) and R26 (next to C8).
As for R81, it does not affect the gain. It does, however, affect the DC offset. As R81 is reduced from infinity down to 0, the output stage is gradually included in the feedback loop, which in turn corrects DC offset at the output. This does not change the gain because the output stage does not provide voltage gain.
Hope this helps. Let us know...
Pierre |
|
|
| dinu |
Hi Pierre
I always appreciate any help I can get. I did use the audioxpress boards but am getting some strange results. I am getting the 40dB gain with R21 and 22 set at 100 Ohm. In the original article, Nelson suggests that you could achieve 20dB of gain with a 10K Ohm resistor for R21 and 22. Have any of you tried this? and, if so, to what end?
Despite all else I think I need to check the component placement again. I am not sure that I identified all the silkscreen problems correctly. As a side note, I'm surprized that they don't include a little note with the boards warning people of these mistakes. |
|
|
| BobEllis |
Dinu,
I'm sure the last thing you want to hear is pull the boards, but you should trace the feedback paths to ensure that you have the proper resistors in place.
The amp has two gains - open loop and closed loop. Open loop is what it would do without feedback. Closed loop is what it does with feedback. As long as the open loop gain exceeds the closed loop gain, then closed loop is what you see in operation. As I said earlier, I think Nelson meant to say that the VAS stage would have 20 db of gain with R21/22 approaching infinity. This would mean that there is more gain to throw away with feedback, resulting in lower measured distortion. You should have the same 26db of gain whether or not you use R21/22. therefore cehck the feedback resistors.
This amp does have 6 db less gain if you use single ended inputs and don't flip the switch, but that's something different. |
|
|
| rolandong |
Hi BobEllis,
I noticed that pcb layout for A75 power supply posted from passDiy shows R18 connected accross GND and +U where it should be accross GND and base of Q1 from the schematic diagram.
Also, the schematic shows R3 and R4 wired to secondary taps A&D while the layout shows directly to C4&C5.
can you pls kindly help ?
regards,
roland |
|
|
| BobEllis |
Roland,
You are correct, the layout is in error. If you are not using a fan you can leave out Q1 and the associated components. If you are using the audioexpress boards, there is an unmarked hole in the trace leading to Q1's base. Use that hole and the ground hole to connect R18 or you will have a fan operating on full output rail voltage. A 12V fan working on 60V really runs fast, for about 30 seconds :xeye:
I didnt notice the R3-4 connection issue - the board works fine as is, but connecting per the schematic would reduce the peak currents in C4-5. As is it reduces the heat load for the regs. |
|
|
| dinu |
| I've had a bit of time this week and have been able to get it working, with one exception. I get a 60Hz hum on the output. If I ground the pos input pin the hum just gets louder. I'm guessing I have a grounding issue. All but the earth and chassis ground on the power supply are connected to 2 star grounds that are connected together with 12 gauge wire. The earth ground and chassis ground from the power supply are both connected to the chassis, not in the same point. I don't want to start guessing at my problem so I hope one of you may be able to help. |
|
|
| PierreQuiRoule |
Dinu
It is good to hear that you are making progress. There is good information on Aleph 30 grounding in this thread, including two posts by Nelson Pass. In a nutshell, one wants grounds that are subject to heavy currents and thus voltage ripple (main caps, bridges and transfo) to exchange current between themselves, and not through signal grounds.
Pierre |
|
|
| dinu |
| I replaced the input wires with a pair of cabls I got from a really great electronics supply place in brooklyn. They have conductive plastic covers and a drain conductor. It helped but I still hear a hum. Would it be a good idea to isolate power ground and signal ground through a bridge rectifier? If so, Any ideas? |
|
|
| BobEllis |
| quote: | Originally posted by dinu
I've had a bit of time this week and have been able to get it working, with one exception. I get a 60Hz hum on the output. If I ground the pos input pin the hum just gets louder. I'm guessing I have a grounding issue. All but the earth and chassis ground on the power supply are connected to 2 star grounds that are connected together with 12 gauge wire. The earth ground and chassis ground from the power supply are both connected to the chassis, not in the same point. I don't want to start guessing at my problem so I hope one of you may be able to help. |
That's not quite the way NP had the grounding scheme. The mains earth ground connects to the chassis - always. However, the amplifier's grounds connect to the chassis through the diodes (D7, D8) and resistor (R17) on the PS board. Lift your star gound from the chassis and let the connection to the chassis be through the ground leads on your PSU board. The mounting hole near D7 and D8 is connected to the chassis grougn side of the circuit, so if you use condictive standoffs you should be OK. I mounted all teh potentially high current grounds at one end of the bolt used as my star and the signal grounds at the other.
Check your lead dress. Are all AC mains leads twisted together and close to the chassis? Are the rail leads twisted tightly together? Are the input connections as far as possible from the mains leads? How about the feedback and output drive connections to the main board?
When you say grounding the positive input makes the hum louder, is the negative input grounded and the amp set for single ended operation? Are you sure it is really grounded and you aren't acting as an antenna? |
|
|
| rolandong |
Hi Bob,
thanks for the reply.
How could I add thermal compensation diode to Q1 to make a variable speed fan ?
roland |
|
|
| BobEllis |
I don't think a diode would give you much change. You might try a negative temperature coefficient themistor for R19 if you can find one.
The question is why would you want to vary the fan speed? In class A the heat load is highest at idle and drops until you get into class AB. Chances are you won't ever make enough power to increase the heat load further. |
|
|
| rolandong |
I intend to use it with jens super leach board front end.
thermistor is okey then, thanks ! |
|
|
| BobEllis |
| This would probably be a better choice to provide temperature control. |
|
|
| dinu |
Hello all.
The A75 is done and now at my buddy's house in Florida. It sat there for a very long time until I finished putting the output stage back together with some new Irf240's and complement. Fire it up and it works beautifully. I also included that feedback swith. I found a nice 3 position and made a selector for 0, 50%, and 100% feedback. I definitely preferred the no feedback setting for most music but it seemed like rock and other music with a decent amount of distortion sounded very harsh without any correction. It was almost grating. The other problem was that switching the feedback setting also switched the DC across the output. You would lose the carefully adjusted setting. All in all, I woul not make the selector again. For my listening style, I would set R81 to 500R and be very happy. My buddy's a different story. He listens to his German punk as well as alot of Classic Rock. As he was playing the Puhdys on my Joseph Audio RM25's I wanted to stab my ear with a fork. He loved it though. The ultimate insult was that he bought a pair of, in my opinion, totally inadequate speakers for such a beautiful amp. They are hooked up to a pair of Vienna Accoustic Schonbergs. He thought they looked cool. I was there when he bought them and despite my every effort to get him to stretch the wallet for a pair of Beethoven Grands....well the picture says it all. He had to buy me a few drinks afterwards to calm my nerves. I'm getting bent out of shape just thinking about it.
Anyway, it's new project time. I'm building an A75 for myself this time and intend to go all out. A full complement of Riken resistors are on their way as I live and breathe. They are being accompanied by a real nice pair if IXYS rectifiers and Mills resistors for the sources.
My intention is to drive 60V output rails off of a pair of 45V 1KW LoNo Plitron transformers.
A separate 60V 120W Plitron will provide the power for the front end.
I want to use the same regulator setup as the original design minus the voltage doubler. I figure I can get 77V if I just change R13 to 15K and R10 to 2K.
The output stage will consist of 8 pairs of IRF9240's and IRF240's.
Heat sinking will be substantial since I want nothing to do with a fan.
I don't want to use the Old Colony boards but would rather not have to go lay the whole thing out myself. Have any of you gone through this exercise already or have any ideas where I can get some really nice boards. I want to really tart this thing up. (Gold plated traces?).
BOB ELLIS WHERE ARE YOU?
GAME ON!!
Here's the Pic. |
|
|
| Alain Dupont |
dinu,
The only ones I know having built a A7 Is Peter Daniel, Bob Ellis and me...
The others can be seen on: http://www.passdiy.com/gallery-misc.htm
You'll have to ask.
Good luck with your A75 project.
PS: Pict missing?
Regards.
Alain. |
|
|
| dinu |
| I guess there's a file size restriction. Send me an email and will forward. |
|
|
| Alain Dupont |
Dinu,
Your email is blocked, send it to me at
aldupon AT sympatico . ca
Replace the At by @ and remove the spaces
I strech it to the right size and will send-it back to you.
Nice explanations about feedback, I didn't use it only the small 33pF cap
i am prowd to say that at the last happening in Montreal. my A75 and CCS-X-Bosoz
got me some nice compliements, and served on about 8 speaker pairs, from
Fostex, and some others 2 3 ways. Giving a "good" sound.
Follow the link :
http://www.freewebs.com/ulzog/journediy.htm
Picts 2,3,4 are from my Bosoz
7 is the Bosoz/Ono PS
8 The A75
Best regards.
Alain. |
|
|
| BobEllis |
I haven't thought about laying out an A75. It shouldn't be too hard to do one channel per board in the free version of Eagle. Like a lot of things there is a steep learning curve at first. I doubt I'd go with gold plating, but if you must...
If you post your first attempt at a layout I am sure the knowledgeable will guide you to a better one. I might be interested in a few, not that I don't already have too many boards. I've got the group buy thing down to a science (let my teenager pack orders) if there is enough interest.
I've been very happy with the service of Advanced Circuits (www.4pcb.com) for the group buys I've done. I Highly recommend them. Let's talk before you order boards.
As for the regulator, are you going to use a separate 60V transformer to avoid the doubler? Watch the power dissipation in R13 - You might want to increase it to >20K. Something I learned when I bumped the rails up on mine.
Jens Rasmussen improved on the version of the A75 regulator I was laying out, which turned into a group buy. The pass devices are darlington connected for greater current gain, there is an option to use a zener, LM329 or LM4040 voltage reference and short circuit protection. I have several boards available, as well as parts kits for15V nominal operation.
On my website under PSU2.2 is a spreadsheet that will calculate component values for you (usable for the straight A75 reg, too) as well as a bit of a tutorial on scaling the voltage.
Enjoy your project. |
|
|
| dinu |
My intention was to use the oroginal psu board for the regulated supply minus the voltage doubler. I think a dedicated 60v transformer would get me the voltage I need. If you still have some PSU 2.2 boards left that might be even better. Do you think its a better choice for what will probably need to be a 77v supply?
I wanted 2 separate boards for the front end that can include the coupling caps to the power supply and those extra diodes. The output would have its own boards. Do you think it would be a good idea to mount the output power supply caps right on the output boards? |
|
|
| BobEllis |
A dedicated 60V transformer would work fine. i used a dedicated transformer in mine since doubling 42VAC would have made caps expensive and the dissipation in the MJEs unacceptable.
The PSU2.2 would theoretically have lower noise and better regulation, but I don't have the equipment to prove it. I have about 20 extra boards on the way. A 15v version did sound a lot better than 7815s in my crossover.
Output caps on the output boards would work fine. Was it Peter Daniel who used a whole bunch?
When laying out your front end boards It would be a good idea to allow more room for heat sinking the FETs (compared to the original layout) and use IRF610 and IRF9610 for the input devices. They sound fine in mine and the IRFDs would fry at your proposed rails.
Also, you'll want to size R11 and R12 to give about 5 mA through the zeners and watch their power dissipation, too. They'll need to be 1W. You could cut down the current a bit, at the expense of a little noise. The LM4040-10 could get by with a lot less than 1 mA, at a buck apiece it's not too expensive. Of course using it would mean recalculating the ccs resistors R's 7-10. |
|
|
| dinu |
I downloaded Eagle this weekend and was able to get quite a bit of the schematic done on a train trip from Boston to New York this weekend. I intend to mount all the 220's which will include Q1,2,4 and 5 outboard of the board on a heatsink or the chassis' bottom plate which is 1/8" aluminum. I remember the biggest broblem with the A75 was the very delicate bias adjustment and therefore will try to find a nice multi turn trim pot. Any suggestions are welcome. For the most part, the pad spacing for all the resistors are going to have to accomodate 1/2 watt Rikens which have an almost 10mm body. Some of the resistors like R11,12,25, and 26 are going to be 1 watt Rikens with a 14.3mm body to allow for the additional dissipation due to the higher voltage rails.
Bob, save me one of those boards, I'd like very much to try it.
Alain, thanks very much for compressing my pic but it is still too big. It's getting kicked back. |
|
|
| BobEllis |
The Bourns 3296Y is either included in Eagle or in one of the downloadable libraries (bourns.lbr). It's a bit hard to find using search. search on r-trimm32*. The 3299W has the pins in a line, if you prefer that arrangement.
Trim_1234-S64YW appears to be a generic pad for either type, but check pins and dimensions.
How about setting up a photbucket account? upload your images there, then copy the [img] tag that they generate and past it in your post. Works like a champ and is free. Still not a bad idea to keep file sizes reasonable though.
 |
|
|
| Alain Dupont |
Dinu,
Here it is, all my excuses ; a better size...
Alain.
DINU A75 |
|
|
| dinu |
| Thanks Alain. Now I'm official. |
|
|
| Alain Dupont |
Dinu,
You could send a few pictures and a story to Nelson....
to get in the passdiy gallery
webmaster AT passdiy . com
Regards.
Alain. |
|
|
| dinu |
| Unfortunately, that's the only pic I have of the amp except for a couple that I took with my camera phone. |
|
|
| dinu |
Hi all,
For the next week until the first, Im in Romania, also working on a new board layout for the A75. The new layout will include film caps in critical locations and a board layout that utilizes all TO220 packages that are populated at the board edge so they can all be mounted on a heat sink.
Til then, |
|
|
| dinu |
Hi all,
I'm back at it. Over the break I started thinking about some power and current values for my design. I figured that using a 45V transformer on the output PS would yield about 63V rails. Each channel will use 16 pairs of Irf240's and complementary 9240's. Bias at about .170A. When I started looking at various speaker loads I came to an unsettling realization. Consider the following.
Supply Vac (transformer) 45.00
Supply Vdc 63.64
Output max Vrms 42.17
Output max Vpk 59.64
VALUES AT 8,4,2 OHM LOADS
Max power output Wrms 222.31 444.61 889.22
Max power output Wpk 506.25 1012.50 2025.00
Max Current Irms 5.63 11.25 22.50
Max Current Ipk 7.95 15.91 31.82
CLASS A BIAS
# of output pairs 16.00
Bias current / pair 0.17
Total bias current draw 5.44
Power dissipation 346.20
Total Power used Wrms 568.50 790.81 1235.42
Total Power used Wpk 852.45 1358.70 2371.20
Total Current draw Irms 11.07 16.69 27.94
Total Current draw Ipk 13.39 21.35 37.26
These values are per channel. I intend to use a 1.5kW transformer per channel.
QUESTUION
1 How do I determine the proper amount of capacitance for the power supply so I can successfully drive the 2 OHM load (If I ever see one).
2 I want to use an array of 3 main breakers in parallel that will allow me to set the max current draw based on the value of the speakers being used. Do I set the breaker values based on Peak current or RMS? Is this atall a good idea? |
|
|
| BobEllis |
The easy answer is "how much capacitance can you afford?" ;)
The downside of lots of capacitance is that the charging times are short resulting in high peak currents - watch your rectifier current and dissipation. They will need significant heat sinking. The high power discrete devices that NP suggested would make spreading the heat load easier. If you find that recommendation again, please shoot me a link, I lost it when my hard drive crashed.
The other thing that you may not have factored in is that you will see tremendous ripple at 22A out (and your rectifiers will see roughly double this current).
Use PSU Designer from Duncan Amps to simulate your PSU and watch the interrelationship between capacitance, load current and rectifier current. Be sure to enter your proposed transformer's regulation in the transformer properties, the default is a high voltage transformer with relatively poor regulation/high internal resistance. Also watch the filter charging current. You'd be surprised at how high it can be, both average and peak.
Circuit breakers and fuses have varied response times ranging from super fast acting to slow blow - see their data sheets. This means that you could tailor the protection to trip on instantaneous peaks or allow them to pass but blow on sustained output beyond a certain point.
For practical home use however, the fuses can be quite small. My subwoofer amp is a bridged Hafler DH500 (specs say 800W RMS capable bridged), driven with 6 db of boost at 20 Hz. I have blown one 3A output fuse in over 20 years. It was due to a shorted speaker cable, not high power listening.
Not that having a fuse will do much to protect the amp if there is an internal fault. Those afflicted with audiophilia nervosa will object to fuses. I don't have any in my A75s, but I did fairly extensive testing before I connected them to a speaker. Your modulated arc welder version probably could benefit.
Be sure to plan room for a soft start circuit. With that big transformer and lots of capacitance you'll need it.
Also, consider the safe operating area limits of your output devices. With high voltage rails it may not be safe to draw 2 amps from each device. Be sure to derate for the elevated temperature. IIRC, the X1000 uses twice as many output devices as you propose. If you are planning to use IRFP devices, they are a bit less robust. Figure 3 IRFP for every 2 IRF.
Matey, does ye really wanna design for 900WRMS into 2 ohms? Arr! :Pirate: With the proposed transformer you'd be dissipating at least 800Win the output stage and power supply and drawing close to 20 Amps from the wall per channel. Your transformer would be right at its limit. 3 KVA would be better (or a 1.5KVA per rail). :bigeyes: Think 250 pound amp... |
|
|
| dinu |
Anything worth doing is worth overdoing. I meant 1.5KVA per channel and realistically, I'll never have 2R speakers. I have no plans to bridge to drive a sub or anything like that and am very unlikely to listen to any music that loud. This is a matter of having a ridiculous amount of overhead.
You got me worried about my rectifiers. I bought 2 IXYS FRED units from partsconnexion rated at 22A and 600V each. From what you're saying, I'm really pushing them...right? There are 68A units available for not much more money. I can use the 22A units on the front end PSU.
I'm under the impression that that thermistor used on the original design IS a soft start. I'm betting I'm wrong. What do you suggest? |
|
|
| BobEllis |
How much you are pushing the rectifiers depends on how much capacitance you use and how much power you push. PSUD will show you the turn on surge if you do without a soft start. Check the data sheets to see what their surge current limit is.
Using the "standard" derating factor of two and rectifier sees double the RMS current rule of thumb, 22A rectifiers should do for 5.5A rms out. My guess is you're about to place an order for the high current ones. ;)
The thermistor is for inrush limiting. However, at your proposed power, I don't know of a thermistor rated to carry the current. Higher current thermistors also tend to have lower cold resistance, so the effect is lessened.
I'd plan on enough heat sink to keep 500W cool, which means a lot of metal anyway. You can use the AXE-1.2 spreadsheet (link in the AX wiki) to help you calculate the thermal resistance needed. Watch the output device junction temperatures.
The input capacitance of 16 devices means the amp will be a bit slower than the original. You may want to increase the VAS bias and/or add a driver stage to provide enough current to drive the extra gate capacitance (especially if you discover thermal considerations make you want more output devices.)
BTW, 5.4A bias is .3375A per device (~21W), not .17 (total bias/number of pairs). With standard mounting (mica/grease) you'll need a K <.05 C/W to get the junction temperatures below 100C. Half the bias reduces your heat sink requirements tremendously and still allows plenty of class A power at 8 ohms. |
|
|
| dinu |
| My intention was to bias each device to .17A each. Am I doing this right? .17a*32devices=5.44A. |
|
|
| BobEllis |
Not quite - You have 16 pairs. The bias goes through the 16 N channel devices and the same current continues through the P channel devices to the negative rail. Total bias is Number of PAIRS x per device bias. 16 x.17. Your 5.4 A is the output current you can deliver in class A, though.
Perhaps a new thread? Higher power A75 or A75 modifications? |
|
|
| dinu |
| Definitely need a new thread. Current topic has little to do with the thread title. |
|
|
|
