Layout is coming along nicely. Still need to go through a few rough iterations before I can post a preliminary layout picture. Quick poll about a few things though:
1) If one of the dimensions have to go up, should it be length or width?
2) Is it necessary for the capacitors to be so large? This is in particular referring to the bias transistor's 100nF CE cap, the input shunt and the 62pF's. Isn't standard 7.5 and 5mm lead spacings big enough? Only axial caps need that big spacing, and the ones I used were still much smaller.
3) Since the power dissipation between resistors is closer in this version I've decided to use the same footprint for all the resistors. Admittedly some of the low-power ones' size can be reduced, but a lot of people are presumably going to use those big Dales throughout. Tip: the PRP resistors from pcX are available in nice small packages and offer very good quality at a price not much higher than Dale.
4) Is it OK if I reduce the resistor length Al's used by 3mm from 19 to 16mm? That is still long enough for 0.5W resistors, including Dale. Same thing goes for the diodes and zeners, their footprints can be much smaller even to accommodate 1.3W ones.
5) Is the heatsink arrangement used for the four cascode transistors on the original Krell acceptable (i.e. four separate ones) or should I try making them all face in the same way for a larger shared sink?
6) Would people like the idea of having both the value as well as reference designator on the silkscreen for each component?
7) Still one zener or two in series for lower power dissipation (then 0.5W ones can be used without fear) and higher accuracy? 2x20V will work well; 1V higher than 39V will not make a difference.. besides a 39V zener is likely to drift anyway.
8) Has anybody yet made sense of the silver mica cap next to the feedback electro's on the picture of the Mk2 board vs the schematic? By looking at the routing of some of the components I suspect that there are small changes between the two. Another reason for suspicion is the power dissipation of the cascode transistors: for the schematic we have, one of each pair (those connected to the rails) dissipates almost a Watt, but the other one less than 50mW. The second one hardly needs a heatsink, but on the picture all four have equal sized, big, sinks. I'm probably going to service a Mk2 within the next two weeks or so, and then I'll take a good, long look at its circuitry.
1) If one of the dimensions have to go up, should it be length or width?
2) Is it necessary for the capacitors to be so large? This is in particular referring to the bias transistor's 100nF CE cap, the input shunt and the 62pF's. Isn't standard 7.5 and 5mm lead spacings big enough? Only axial caps need that big spacing, and the ones I used were still much smaller.
3) Since the power dissipation between resistors is closer in this version I've decided to use the same footprint for all the resistors. Admittedly some of the low-power ones' size can be reduced, but a lot of people are presumably going to use those big Dales throughout. Tip: the PRP resistors from pcX are available in nice small packages and offer very good quality at a price not much higher than Dale.
4) Is it OK if I reduce the resistor length Al's used by 3mm from 19 to 16mm? That is still long enough for 0.5W resistors, including Dale. Same thing goes for the diodes and zeners, their footprints can be much smaller even to accommodate 1.3W ones.
5) Is the heatsink arrangement used for the four cascode transistors on the original Krell acceptable (i.e. four separate ones) or should I try making them all face in the same way for a larger shared sink?
6) Would people like the idea of having both the value as well as reference designator on the silkscreen for each component?
7) Still one zener or two in series for lower power dissipation (then 0.5W ones can be used without fear) and higher accuracy? 2x20V will work well; 1V higher than 39V will not make a difference.. besides a 39V zener is likely to drift anyway.
8) Has anybody yet made sense of the silver mica cap next to the feedback electro's on the picture of the Mk2 board vs the schematic? By looking at the routing of some of the components I suspect that there are small changes between the two. Another reason for suspicion is the power dissipation of the cascode transistors: for the schematic we have, one of each pair (those connected to the rails) dissipates almost a Watt, but the other one less than 50mW. The second one hardly needs a heatsink, but on the picture all four have equal sized, big, sinks. I'm probably going to service a Mk2 within the next two weeks or so, and then I'll take a good, long look at its circuitry.
1. More width is ok.
2. 7.5mm should work but I think the woder spacing was for WIMA caps.
3. I get the Dales for 10 Cents each now. Be sure the holes are large enough for the Dales!!LAst board the holes were tight.
4. Smaller resiistor foorprint should work ok.
5. They should face the same way so they can be on one sink. Sturdier this way. Trasistors supported by their leads and carrying a heatsink are not very sturdy.
6. Refrence designator is sufficient since some people drift away from the actual values anyway.
7. What ever on the zener... either is fine.
8. The mica cap could have been added to some amps to cure parasitic oscillation if an individual amp exhibited it.
Hope some of this helps.....
Mark
2. 7.5mm should work but I think the woder spacing was for WIMA caps.
3. I get the Dales for 10 Cents each now. Be sure the holes are large enough for the Dales!!LAst board the holes were tight.
4. Smaller resiistor foorprint should work ok.
5. They should face the same way so they can be on one sink. Sturdier this way. Trasistors supported by their leads and carrying a heatsink are not very sturdy.
6. Refrence designator is sufficient since some people drift away from the actual values anyway.
7. What ever on the zener... either is fine.
8. The mica cap could have been added to some amps to cure parasitic oscillation if an individual amp exhibited it.
Hope some of this helps.....
Mark
Frequency/gain compensation in the NFB line.
One lead of the silver mica is connected at the connection of the Roederstein electrolytic + pole and the feedback resistor lead.
The other silver mica lead can only go to the feedback resistor's other connection.
Which makes sense:
- that's where a silver mica is good for
- the technique is in line with the trend of the MKII years.
(the other bypass cap is a Roederstein MKC1860, btw. Metalised polycarbonate, not polystyrene. 10% accurate/100C max)
KSA PCB Design Consideration
Pierre
Personally I dont like PCB's that are too small and where you struggle to fit parts. Rather err on a size too big than too small. It also makes it so much easier to position and mount heatsinks on the PCB.
Maybe somebody (Mark G.?) can source a standard heatsink in bulk for which the board can be custom designed and that could possibly be sold with the board? It just looks so much better when a board is custom designed for a particular heatsink.
Who is designing the smaller PCB's that operate the relays in the Mk2 version?
Jozua
Pierre
Personally I dont like PCB's that are too small and where you struggle to fit parts. Rather err on a size too big than too small. It also makes it so much easier to position and mount heatsinks on the PCB.
Maybe somebody (Mark G.?) can source a standard heatsink in bulk for which the board can be custom designed and that could possibly be sold with the board? It just looks so much better when a board is custom designed for a particular heatsink.
Who is designing the smaller PCB's that operate the relays in the Mk2 version?
Jozua
A custom heatsink would be great, esp if it can be bolted down as with the original.
No need to go ultra-small on the board, but smaller reduces chance of oscillation and other nasty parasitic effects. Not too likely on this board though.
I also had the problem with tight resistor holes. I even had to shave some Holco H4P resistors with a sharp knife to make them fit - well at least they're damn sturdy now hehe.
I also figured the mica to be an additional feedback pole, was hoping for a confirmation. Judging from its position it's very likely. Will include it in the design, easy to leave out. COG ceramics are even more linear than mica btw.
The other feedback isn't possible to be polystyrene anyway; 100nF polystyrene would be one huge bulky affair. Polyprop/polyester/polycarb's the only choice there. I personally prefer Wima/Rifa MKP's for the 100nF's, mica for the 62pF's and polystyrene for the input shunt.
Mark, do you have any idea of how many people would be on the group buy list?
No need to go ultra-small on the board, but smaller reduces chance of oscillation and other nasty parasitic effects. Not too likely on this board though.
I also had the problem with tight resistor holes. I even had to shave some Holco H4P resistors with a sharp knife to make them fit - well at least they're damn sturdy now hehe.
I also figured the mica to be an additional feedback pole, was hoping for a confirmation. Judging from its position it's very likely. Will include it in the design, easy to leave out. COG ceramics are even more linear than mica btw.
The other feedback isn't possible to be polystyrene anyway; 100nF polystyrene would be one huge bulky affair. Polyprop/polyester/polycarb's the only choice there. I personally prefer Wima/Rifa MKP's for the 100nF's, mica for the 62pF's and polystyrene for the input shunt.
Mark, do you have any idea of how many people would be on the group buy list?
PWatts said:
Sure,
just thought of mentioning it, there are some reading this who can't tell one from the other. Just the other day i dreamt of opening the front door to trip over a 1uF styroflex cap.
Probably a cosmetic revision for which Krell may have sent an update to service mechs. , but not an entire new service manual for just 1 cap. Krell had a number of revisions and optionals in those days, from exchanging PCBs to converting to KMAs.
I've been wanting to experiment with those teflon REL-caps for the Miller caps but they're just too damn expensive and bulky.
What exactly is the difference between KSA and KMA anyway? I know of some refinements to the power supplies.
From the looks of the input stage on the picture it doesn't gel if the transistor pinouts have the collector in the centre (as 2S series do), but it's hard to judge without having the bottom side as well and the component values.
What exactly is the difference between KSA and KMA anyway? I know of some refinements to the power supplies.
From the looks of the input stage on the picture it doesn't gel if the transistor pinouts have the collector in the centre (as 2S series do), but it's hard to judge without having the bottom side as well and the component values.
Off topic but here goes:
Changes in the powersupply and all the main boards were swapped.
Example:
In 1990 Krell had a marketing offer for the KSA250.
For just the transport costs Krell converted a KSA250 to a monaural MDA500. Of course, unless you had 2 KSA250s you had to buy a matching second MDA.
(With the KSA250, Krell converted back to BJT only, the previous used FETs. Dan d'A personally did the KSA250 introduction overhere, unfortunately i couldn't get in, invitation only)
Changes in the powersupply and all the main boards were swapped.
Example:
In 1990 Krell had a marketing offer for the KSA250.
For just the transport costs Krell converted a KSA250 to a monaural MDA500. Of course, unless you had 2 KSA250s you had to buy a matching second MDA.
(With the KSA250, Krell converted back to BJT only, the previous used FETs. Dan d'A personally did the KSA250 introduction overhere, unfortunately i couldn't get in, invitation only)
KMA==2 Bridged channels of KSA?
Based on what I've seen on other threads here, it seems (at least for some models) that a KMA-xxx is based on a bridged configuration using a stereo chassis of about 1/3rd the nominal power, ie a KSA-50 can be turned into a KMA-160...
Theoretically bridging should give a 4x increase into the same load (ignoring losses), but perhaps the specs are being convervatively stated...or the rail voltage is reduced to allow the idle current to be increased, giving more class A output...
Hopefully someone knows the details...
Stuart
Based on what I've seen on other threads here, it seems (at least for some models) that a KMA-xxx is based on a bridged configuration using a stereo chassis of about 1/3rd the nominal power, ie a KSA-50 can be turned into a KMA-160...
Theoretically bridging should give a 4x increase into the same load (ignoring losses), but perhaps the specs are being convervatively stated...or the rail voltage is reduced to allow the idle current to be increased, giving more class A output...
Hopefully someone knows the details...
Stuart
Regarding those FET´s
If you design the board and footprints for the FET´s With thru hole including an SMD-footprint for SOT-223 pattern on the back/solder side of board, then this should allow member to choose their own type of FET´s (ZVN/ZVP 2110G has standard FET leg-configuration and same leg spacing as an TO-220). Only check for that you leave some board spacing for the "Drain"-tab
to the proper direction (including some Cu for to solder the Drain to the board).
If using the same philosophy for placing those FET´s in a straight line as for the cascaded TO-220 transistors it should be easy to mount a thermal plate/strip on those to.
Regards
If you design the board and footprints for the FET´s With thru hole including an SMD-footprint for SOT-223 pattern on the back/solder side of board, then this should allow member to choose their own type of FET´s (ZVN/ZVP 2110G has standard FET leg-configuration and same leg spacing as an TO-220). Only check for that you leave some board spacing for the "Drain"-tab
to the proper direction (including some Cu for to solder the Drain to the board).
If using the same philosophy for placing those FET´s in a straight line as for the cascaded TO-220 transistors it should be easy to mount a thermal plate/strip on those to.
Regards
Hi,
7.5mm and 5mm pitch restrict capacitors to polyester.
Very few polypropylene fit the 5mm pitch and a few more in 7.5mm. Many are only available in 10mm pitch and bigger.
For on board electrolytics allow 5mm and 7.5mm pitch.
Try to get the LTP pairs touching face to face and include the fourth pin hole for 2sa/c types.
Keep the zeners to 400mW or 500mW singles. Duals will need twice as much current to get around the knee of the IV curve.
1.3W types will need three times the current (have pity on the dropping resistor and internal temperatures).
As an aside, I built my first Leach amps using 120MHz to 280MHz devices (other than the 35MHz output devices) and I plan to do the same to my KSA50 Klone when I get around to it. It sounds wonderful and no alteration to the Miller comp cap was required.
7.5mm and 5mm pitch restrict capacitors to polyester.
Very few polypropylene fit the 5mm pitch and a few more in 7.5mm. Many are only available in 10mm pitch and bigger.
For on board electrolytics allow 5mm and 7.5mm pitch.
Try to get the LTP pairs touching face to face and include the fourth pin hole for 2sa/c types.
Keep the zeners to 400mW or 500mW singles. Duals will need twice as much current to get around the knee of the IV curve.
1.3W types will need three times the current (have pity on the dropping resistor and internal temperatures).
As an aside, I built my first Leach amps using 120MHz to 280MHz devices (other than the 35MHz output devices) and I plan to do the same to my KSA50 Klone when I get around to it. It sounds wonderful and no alteration to the Miller comp cap was required.
Stuart Easson said:
Stuart,
that's about right.
You can also tell by the amplifier case dimensions.
The KSA models had/have different sizes, the KMA model had the same dimensions as the one that would deliver the KMA juice in bridged mode.
The KMA models retained the balanced option, the main difference in layout.
Good show, Andrew.
You better think GB for those Zetex MOSFETs.
At the moment I haven't got the LTP pairs back-to-back but side-to-side. With thermal paste and a piece of heatshrink it will also work well. Back-to-back will route harder.
I got some Rifa 100nF MKP caps at RS with 5mm pitch that I used in my Klone, but yes I know they're usually quite larger. Support for 5, 7.5 and 10mm it will be then. If there's space I'll try for larger as well. The Miller caps need to be no larger than 5 or 7.5mm though.
I agree with Andrew, in my experience higher BW transistors provided better quality sound with more detail. This was very apparrent with upgrading from MJ15003/4 to MJL3281/1302.
I don't think that two zeners in series would be a problem (If I remember the Leach also did it that way), but OK I'll make it simpler and closer to the original by keeping it a single one.
BTW Andrew, how would you compare the Krell to the Leach? I've never done a side-by-side comparison, but the Leach is a bit soft in the bass.
I got some Rifa 100nF MKP caps at RS with 5mm pitch that I used in my Klone, but yes I know they're usually quite larger. Support for 5, 7.5 and 10mm it will be then. If there's space I'll try for larger as well. The Miller caps need to be no larger than 5 or 7.5mm though.
I agree with Andrew, in my experience higher BW transistors provided better quality sound with more detail. This was very apparrent with upgrading from MJ15003/4 to MJL3281/1302.
I don't think that two zeners in series would be a problem (If I remember the Leach also did it that way), but OK I'll make it simpler and closer to the original by keeping it a single one.
BTW Andrew, how would you compare the Krell to the Leach? I've never done a side-by-side comparison, but the Leach is a bit soft in the bass.
Common observation "the Leach is a bit soft in the bass."
Many builders have found increasing the bias up to as much as 200, depending on the heat sinking capacity quite elegantly solves this complaint. Your woofers/crossover also have a part to play.
For the purist, the MJ15024/25 compares quite favorably to the plastic case MJL3281/1302 while retaining superior heat transfer ability.
Two zeners in series allows more accurate matching with fewer components to buy for that perfect voltage control, quoth the Prof.
IMHO nothing compares to a KSA50 which has a voice all its own.
HTH
Many builders have found increasing the bias up to as much as 200, depending on the heat sinking capacity quite elegantly solves this complaint. Your woofers/crossover also have a part to play.
For the purist, the MJ15024/25 compares quite favorably to the plastic case MJL3281/1302 while retaining superior heat transfer ability.
Two zeners in series allows more accurate matching with fewer components to buy for that perfect voltage control, quoth the Prof.
IMHO nothing compares to a KSA50 which has a voice all its own.
HTH
Well, A KMA 160 is a bridged KSA-80 which is an entirely different beast from the KSA-50. The KSA-80/KMA 160 is a true balanced differential power amp from input to output. Perhaps you meant the KMA-100....? That was just a KSA-100 monaural with regulated front end/driver supplies similar to the regulated suplies in the KSA80. It was also smaller, more managable and weighed in at around 70 lbs.
I can look at extrusions but that does get expensive. I will check some of the smaller extrusion places first like M&M Metals. They are reasonable to deal with on small quantities. I'll see whats out there..... At least I have facilities to cut them to the required length. They could be shipped out with each board.
As far as a group order I would say not more than 50 boards. But then the KSA-50 thread suprised us all.
I quite agree! Its a very special power amp. Nothing else out there is quite like it.
Mark
Hi,
I cannot compare Leach clone to Krell Klone. I could be quite a while before I start the Klone, so don't wait for me.
The Leach is just as strong in the bass as any other of my good bass amplifiers. I use a 80mS to 90mS high pass filter on the input. 120mS to 150mS on the NFB and 200mS on the PSU. The Klone obviously omits the input filter and could omit the NFB filter if you go for direct coupled, but then you really do need matched high gain transistors that run at the same temperature (close coupled and heatsinked?).
I cannot compare Leach clone to Krell Klone. I could be quite a while before I start the Klone, so don't wait for me.
The Leach is just as strong in the bass as any other of my good bass amplifiers. I use a 80mS to 90mS high pass filter on the input. 120mS to 150mS on the NFB and 200mS on the PSU. The Klone obviously omits the input filter and could omit the NFB filter if you go for direct coupled, but then you really do need matched high gain transistors that run at the same temperature (close coupled and heatsinked?).
Great, seems like we're getting somewhere. The heatsinks would be a welcome bonus, but I'm concerned about cost - not only the materials, but also the shipping for overseas builders since a padded envelope is unlikely to be feasible.
What's interesting about the KSA50 being so great, is that its schematic is one of the oldest of Krell - has anybody ever made a side-by-side comparison between the KSA100 Mk1 and Mk2, or even better, a KSA50 klone adapted for 100W? Maybe the KSA100 is a bit disappointing compared to the 50? Just playing devil's advocate, but if nobody knows what to expect it's worth checking into.
Regulated supplies are indeed a good idea, but these can be incorporated off-board. I've done some simulations with elementary regulators for the front end. At high frequencies, they rule. However, with low frequencies (at the input, not the rails), the ripple is higher. Going regulated is also likely to increase noise, so adequate post-filtering is also necessary. A properly filtered two-stage zener-follower worked very well, much better than LM317/337. IMO the best and easiest would be to use a separate small 50VA or so transformer driving the front end and driver stage, and keep the board the same. At higher loads that works much better than trying to subregulate them from the main rails. Noise can be adequately addressed with proper RC/LC filtering.
What's interesting about the KSA50 being so great, is that its schematic is one of the oldest of Krell - has anybody ever made a side-by-side comparison between the KSA100 Mk1 and Mk2, or even better, a KSA50 klone adapted for 100W? Maybe the KSA100 is a bit disappointing compared to the 50? Just playing devil's advocate, but if nobody knows what to expect it's worth checking into.
Regulated supplies are indeed a good idea, but these can be incorporated off-board. I've done some simulations with elementary regulators for the front end. At high frequencies, they rule. However, with low frequencies (at the input, not the rails), the ripple is higher. Going regulated is also likely to increase noise, so adequate post-filtering is also necessary. A properly filtered two-stage zener-follower worked very well, much better than LM317/337. IMO the best and easiest would be to use a separate small 50VA or so transformer driving the front end and driver stage, and keep the board the same. At higher loads that works much better than trying to subregulate them from the main rails. Noise can be adequately addressed with proper RC/LC filtering.
,,,,,,,,
or 