Just keep reading/searching, these articles are treasure:
FIRST WATT ARTICLES
Old threads here (about Pass projects) are also full of useful info...
I know that building stuff is more fun than reading, but this way you'll know what are you doing
FIRST WATT ARTICLES
Old threads here (about Pass projects) are also full of useful info...
I know that building stuff is more fun than reading, but this way you'll know what are you doing
Thoughts on possible rebuild...
Hi everyone,
Partly due to the problems discussed above, and partly to feed my diy addiction, I'm considering rebuilding the amp. The idea is to reuse the heatsinks and chassis (with a little rebuilding) to keep costs within reason. I'd be interested in hearing what people think about the following ideas.
1. The PSU is presently in a separate case. It's a capacitor multiplier, and works fine, but for physical convenience I'm thinking about building it onto the back of the amp chassis. I'd want to reuse the trafo, but a CRC or CLC could probably be tried with a little thought, and wouldn't be too much work if I'm rebuilding anyhow. Not that there's anything obvious wrong with the cap multiplier - it's just me being hit by what Neal Stephenson called the "imp of the perverse"... Anyone have a good reason to try something else?
2. I'll probably do PCBs - laser printer and chemicals at home, not a professional job. Nonetheless, it has occurred to me that it would be a good idea to use one of the "universal mounting specs" from the diyaudio store page (probably the F5 one) so other forum members could have go if they fancied to. Does anyone have any observations about things that could/should be included?
3. I have no particular reason to complain about the IRFP9140 and 9240 I used, but many people have praised amps using Toshiba outputs. A quick look at the 2SJ201 datasheet with my untrained eye suggests it might be a drop-in replacement (and Zhou Fang apparently has them available). Would anything else need to be changed? What about 2SJ313? Paralleled, maybe? Other suggestions?
Well, still at the planning stage. I'll be interested to hear any and all input.
Regards
Nigel
Hi everyone,
Partly due to the problems discussed above, and partly to feed my diy addiction, I'm considering rebuilding the amp. The idea is to reuse the heatsinks and chassis (with a little rebuilding) to keep costs within reason. I'd be interested in hearing what people think about the following ideas.
1. The PSU is presently in a separate case. It's a capacitor multiplier, and works fine, but for physical convenience I'm thinking about building it onto the back of the amp chassis. I'd want to reuse the trafo, but a CRC or CLC could probably be tried with a little thought, and wouldn't be too much work if I'm rebuilding anyhow. Not that there's anything obvious wrong with the cap multiplier - it's just me being hit by what Neal Stephenson called the "imp of the perverse"... Anyone have a good reason to try something else?
2. I'll probably do PCBs - laser printer and chemicals at home, not a professional job. Nonetheless, it has occurred to me that it would be a good idea to use one of the "universal mounting specs" from the diyaudio store page (probably the F5 one) so other forum members could have go if they fancied to. Does anyone have any observations about things that could/should be included?
3. I have no particular reason to complain about the IRFP9140 and 9240 I used, but many people have praised amps using Toshiba outputs. A quick look at the 2SJ201 datasheet with my untrained eye suggests it might be a drop-in replacement (and Zhou Fang apparently has them available). Would anything else need to be changed? What about 2SJ313? Paralleled, maybe? Other suggestions?
Well, still at the planning stage. I'll be interested to hear any and all input.
Regards
Nigel
Hi Juma,
Sure. Not only longish, but oldish too....
Schematic is almost the same as the one you posted in post #53 (which I have reposted below for convenience - hope that isn't a breach of forum etiquette). Only difference is R7 and R16 are 0R56 rather than 0R5.
Power supply is the cap multiplier supply from ESP audio project 15. In use it's giving +/- 18.7 VDC. (It's the same PSU discussed above in post #37. I can't believe it's two years ago already... ) This could be changed in a rebuild, of course. (I recall you made a couple of suggestions about this earlier in the posts around #37).
By Iq I presume you mean current through the output device. (Right?) I just measured 0.55V across R7, so it's almost exactly 1A.
Looking through the thread now I see Nelson suggested 2SJ200... so apparently I don't remember everything in the thread either...
Cheers
Nigel
Sure. Not only longish, but oldish too....
Schematic is almost the same as the one you posted in post #53 (which I have reposted below for convenience - hope that isn't a breach of forum etiquette). Only difference is R7 and R16 are 0R56 rather than 0R5.
Power supply is the cap multiplier supply from ESP audio project 15. In use it's giving +/- 18.7 VDC. (It's the same PSU discussed above in post #37. I can't believe it's two years ago already... ) This could be changed in a rebuild, of course. (I recall you made a couple of suggestions about this earlier in the posts around #37).
By Iq I presume you mean current through the output device. (Right?) I just measured 0.55V across R7, so it's almost exactly 1A.
Looking through the thread now I see Nelson suggested 2SJ200... so apparently I don't remember everything in the thread either...
Cheers
Nigel
Attachments
OK, memory refreshed ! 
Either 2sj200/201 or 2sj313 (x 3 paralleled instead od one irfp9140, with 1R/1R2 in each Source) can be considered an upgrade (compared to IRFP9140).
I would go for 2sj313 for two reasons - subjectively, it sounds better to me and objectively, it offers better ratio of parasitic capacitance and transconductance. Also, with Iq of 400mA (per each 2sj313) they work in their most linear region. The downside is more complicated mounting - you'll need 6 devices per channel instead of two (yes, it brings some advantage to use them in Aleph CCS too since it also works as an amplifier when you cross over the SE class A boundary).
Either 2sj200/201 or 2sj313 (x 3 paralleled instead od one irfp9140, with 1R/1R2 in each Source) can be considered an upgrade (compared to IRFP9140).
I would go for 2sj313 for two reasons - subjectively, it sounds better to me and objectively, it offers better ratio of parasitic capacitance and transconductance. Also, with Iq of 400mA (per each 2sj313) they work in their most linear region. The downside is more complicated mounting - you'll need 6 devices per channel instead of two (yes, it brings some advantage to use them in Aleph CCS too since it also works as an amplifier when you cross over the SE class A boundary).
Hi juma,
Either 2SJ201 or 2SJ313 appear to available from Zhou Fang, and not too pricy, not sure about 2SJ200. I rather like the idea of paralleling the 2SJ313s, although one clarification please:
6 per channel means 3 for Q3 and 3 for Q6, symmetrically, right? "Crossing the SE class A boundary" means crossing into class AB? (I thought that didn't work on SE... maybe I misunderstood) ? Or some other stuff from Nelson's articles (that I should know already)?
No thoughts on the PSU, or on boards? The only disadvantage of paralleling all the 2SJ313s is that a board made to the F4 size won't fit on my heatsinks, so my idea of making a design useful to others maybe won't work so well... On the other hand maybe a little thought will show a solution...
Regards
Nigel
Either 2SJ201 or 2SJ313 appear to available from Zhou Fang, and not too pricy, not sure about 2SJ200. I rather like the idea of paralleling the 2SJ313s, although one clarification please:
6 per channel means 3 for Q3 and 3 for Q6, symmetrically, right? "Crossing the SE class A boundary" means crossing into class AB? (I thought that didn't work on SE... maybe I misunderstood) ? Or some other stuff from Nelson's articles (that I should know already)?
No thoughts on the PSU, or on boards? The only disadvantage of paralleling all the 2SJ313s is that a board made to the F4 size won't fit on my heatsinks, so my idea of making a design useful to others maybe won't work so well... On the other hand maybe a little thought will show a solution...
Regards
Nigel
Right - 3 up and 3 down.
Usually SE cl. A operation is limited by Iq delivered by CCS. The Aleph CCS (Q6) is different - its Id is not constant, it varies with output signal (brought through C3 and R22, taken from Rsense=R19). That way the Q6 also acts as an amplifying device, contributing to output current and simulating some sort of push-pull operation. The limit and percentage of output current contributed by Q6 is set by R19 and R22.
Usually SE cl. A operation is limited by Iq delivered by CCS. The Aleph CCS (Q6) is different - its Id is not constant, it varies with output signal (brought through C3 and R22, taken from Rsense=R19). That way the Q6 also acts as an amplifying device, contributing to output current and simulating some sort of push-pull operation. The limit and percentage of output current contributed by Q6 is set by R19 and R22.
When it comes to PSU, my favorite for small/medium power Cl. A amps is a cap-multiplier. Before you ask - no value is critical
An externally hosted image should be here but it was not working when we last tested it.
Hi Guys,
I've been thinking a little more about rebuilding this, especially about juma's comments about 2SJ201 or 2SJ313 (paralleled). It looks like I'll be doing another project first, so there's plenty of time to think some of these things through.
I rather like the idea of paralleling the 2SJ313, but they may not be available when I need them (Seems Zhou is running out, and isn't selling any right now...) So I looked at a few datasheets, and am wondering about IRF9510 and IRF9610. Numbers I've found are the following (capacitances from the tables, transconductance approximate values from the graphs for 0.4A or so):
2SJ313: Ciss = 210 pF, Coss = 90 pF, Crss = 45 pF, Yfs = 0.7 S
IRF9510: Ciss = 180 pF, Coss = 85 pF, Crss = 30 pF, Yfs = 0,7 - 0,8 S
IRF9610: Ciss = 170 pF, Coss = 50 pF, Crss = 15 pF, Yfs = 0,7 S
When juma spoke of parasitic capacitance, which of the three numbers is the issue? Crss? In any event, IRF9510 looks pretty close to 2SJ313 by that measure, unless I misunderstood something. (ALways a possibility...)
So:
1. If the issue is linearity, then which of the graphs on the datasheet should I have been looking at?
2. Would paralleling one of the IR mosfets (or Fairchild equivalent) be an interesting idea?
3. Did I miss some other interesting alternative?
Any input very welcome, as always.
Nigel
I've been thinking a little more about rebuilding this, especially about juma's comments about 2SJ201 or 2SJ313 (paralleled). It looks like I'll be doing another project first, so there's plenty of time to think some of these things through.
I rather like the idea of paralleling the 2SJ313, but they may not be available when I need them (Seems Zhou is running out, and isn't selling any right now...) So I looked at a few datasheets, and am wondering about IRF9510 and IRF9610. Numbers I've found are the following (capacitances from the tables, transconductance approximate values from the graphs for 0.4A or so):
2SJ313: Ciss = 210 pF, Coss = 90 pF, Crss = 45 pF, Yfs = 0.7 S
IRF9510: Ciss = 180 pF, Coss = 85 pF, Crss = 30 pF, Yfs = 0,7 - 0,8 S
IRF9610: Ciss = 170 pF, Coss = 50 pF, Crss = 15 pF, Yfs = 0,7 S
When juma spoke of parasitic capacitance, which of the three numbers is the issue? Crss? In any event, IRF9510 looks pretty close to 2SJ313 by that measure, unless I misunderstood something. (ALways a possibility...)
So:
1. If the issue is linearity, then which of the graphs on the datasheet should I have been looking at?
2. Would paralleling one of the IR mosfets (or Fairchild equivalent) be an interesting idea?
3. Did I miss some other interesting alternative?
Any input very welcome, as always.
Nigel
Hi juma,
I may well do that, but I'd still like to understand the stuff I posted above...
Cheers,
Nigel
I may well do that, but I'd still like to understand the stuff I posted above...
Cheers,
Nigel
when paralleling 2sj313, what kind of matching is necessary? i have not found any info on this except that i needed to match the paralleled devices but it benefits only for heat sharing between devices. but i m kind of not sure.
is that correct?
also, would two sj313 in parallel at 500mA be ok for power dissipation with +-15V rails or is it pushing it?
is that correct?
also, would two sj313 in parallel at 500mA be ok for power dissipation with +-15V rails or is it pushing it?
Last edited:
nevermind about heat...juma said in his ''f5 with 2sj313/k2013'' thread:
''It's very silent amp (even without cap multipliers) but you have to be aware of the heat
I did play with the value of the bias current and, to my ears, these MOSFETs sound best at Id between 400 and 500 mA. If you are brave and lower a PS voltage a bit (say +/-16V) at Id=600mA you'll have additional advantage of zero tempco.
Also, in this case you don't have to worry about low Vds becaus parasitic capacitances are very low (compared to usual HEXFETs) and you just want to drive a single pair.
If you do it please post your results, it would be great to see a lineamp getting hotter than many power amps... ''
still, is there drawback 2 vs 3 outout fets paralleled? i guess not really exectpt for possible higher voltage/bias...?
''It's very silent amp (even without cap multipliers) but you have to be aware of the heat
I did play with the value of the bias current and, to my ears, these MOSFETs sound best at Id between 400 and 500 mA. If you are brave and lower a PS voltage a bit (say +/-16V) at Id=600mA you'll have additional advantage of zero tempco.
Also, in this case you don't have to worry about low Vds becaus parasitic capacitances are very low (compared to usual HEXFETs) and you just want to drive a single pair.
If you do it please post your results, it would be great to see a lineamp getting hotter than many power amps... ''
still, is there drawback 2 vs 3 outout fets paralleled? i guess not really exectpt for possible higher voltage/bias...?
mosfets in parallel must match themselves. the CS fets dont have to match the driving fets, Mr.Pass suggested in one of his article.If you had searched a bit you would have foun!
As to how exactly it affects performance, i dont know.Hi Guys,
I'm still collecting parts for my rebuild of this circuit. I've decided to rebuild the PSU also, probably in the chassis, and I brought back a nice toroidal transformer from a recent visit home to the UK, which at 250VA should be enough for a stereo build (I hope). I also bought some thin film SMD resistors and stuff for the front end around the BF862s.
I've been thinking about pisesd's question about matching. I bought exactly the 12 2SJ313s I needed, not having thought about matching too much, and not wanting to buy too many, since they aren't that cheap. I'm hoping I can match adequately within the 12 I have, but am wondering how closely the triplets should match. Anyone have ideas?
Cheers
Nigel
I'm still collecting parts for my rebuild of this circuit. I've decided to rebuild the PSU also, probably in the chassis, and I brought back a nice toroidal transformer from a recent visit home to the UK, which at 250VA should be enough for a stereo build (I hope). I also bought some thin film SMD resistors and stuff for the front end around the BF862s.
I've been thinking about pisesd's question about matching. I bought exactly the 12 2SJ313s I needed, not having thought about matching too much, and not wanting to buy too many, since they aren't that cheap. I'm hoping I can match adequately within the 12 I have, but am wondering how closely the triplets should match. Anyone have ideas?
Cheers
Nigel
first of all, what i would do is maybe substitute only the mosfets, with best matching with what you have, and see i'd hear an improvement. You could also build with pairs and not triplets to get better matching...?
or maybe you could build an f5. Buzzforb matches the fets specifically for this amp
or maybe you could build an f5
. Buzzforb matches the fets specifically for this amp
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.