Normally a Stax would sport a vintage BJT types configuration of Vbe controlled CCS, current mirror loaded differential error amp with trimmer Norton ref, and some MKT caps around decoupling. Nice classic gear, you do well keeping it going strong.
each heatsink features a 2SD736A and a 2SB700A, which seem to be complementary, and also a 2SA814. i could post the PS schematic if anyone would like, but i know it's slightly OT so i'll hold off for now.
Ikoflexer,
Don't wait, just go and get your 5k PCBs done so that the guys can have them.
I have just finished building the 5k. Here is a picture.
Regards,
Bill
Don't wait, just go and get your 5k PCBs done so that the guys can have them.
I have just finished building the 5k. Here is a picture.
Regards,
Bill
An externally hosted image should be here but it was not working when we last tested it.
I was lucky to get these tall and slim heatsinks so I can make the regulators small. Input +/-26V. Output +/-12V. They are currently running at 150mA. The heatsinks are a bit hot on the CCS, but I can still stick my fingers on them for as long as I like, not too hot.
Bottom view:
The thick wires are of 16swg, more than enough for the rails on an power amp! The rest are of 20swg, vs the normal 22swg. I tried to make it of very low resistance and tried to reduce the inductance as well by using really short tracks, easier to solder too. No components at the bottom. No jumpers are needed, bottom or top.
The thick wires are of 16swg, more than enough for the rails on an power amp! The rest are of 20swg, vs the normal 22swg. I tried to make it of very low resistance and tried to reduce the inductance as well by using really short tracks, easier to solder too. No components at the bottom. No jumpers are needed, bottom or top.
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
I have not heard it yet.
As you can see in the photo below, I was testing it with 680R resistor load.
The output caps are just 2.2uF MKT for the time being.
The scope showed a thin flat line across all frequencies, up to at least 25MHz that my scope can detect, no ripples found.
Note: due to the compression of JPEG, it is funny the line looks like having a bit of ripples. In fact, they are the artifacts of the compressed image.
As you can see in the photo below, I was testing it with 680R resistor load.
The output caps are just 2.2uF MKT for the time being.
The scope showed a thin flat line across all frequencies, up to at least 25MHz that my scope can detect, no ripples found.
Note: due to the compression of JPEG, it is funny the line looks like having a bit of ripples. In fact, they are the artifacts of the compressed image.
An externally hosted image should be here but it was not working when we last tested it.
The good news is, the CCS seems to be stable! I have built 5a, 5d, why would I build 5k? I was happy with everything else but was a bit worried about the stability of the CCS. Now it appears that the problem has been fixed. I have turned the regs on and off twice, each time about half an hour. After about 5 minutes, the heatsinks felt the same amount of heat. I would not be bothered to measure the actual CCS current.
Builders aware:
I matched the 2sk170 between the rails. I built a special jig using opamp socket for selecting the 2sk170. However, the J310 has different leg layout. I could not use that jig to test it and I was too lazy to build another one. So the J310s was randomly selected.
This seems to create some descrepancy between the CCS current of the pos and neg rails. I felt the neg heatsink to be a bit hotter. This was also reflected in the trimpot setting of the Vref, and the difference was quite large! I initially got +16.X VAC and -13.Y VAC!
So don't be lazy like me. Match all your jFETs if you have two polarities.
With the heatsinks restricting access I think I am going to have a hard time replacing the J310!
I matched the 2sk170 between the rails. I built a special jig using opamp socket for selecting the 2sk170. However, the J310 has different leg layout. I could not use that jig to test it and I was too lazy to build another one. So the J310s was randomly selected.
This seems to create some descrepancy between the CCS current of the pos and neg rails. I felt the neg heatsink to be a bit hotter. This was also reflected in the trimpot setting of the Vref, and the difference was quite large! I initially got +16.X VAC and -13.Y VAC!
So don't be lazy like me. Match all your jFETs if you have two polarities.
With the heatsinks restricting access I think I am going to have a hard time replacing the J310!
Last edited:
I won't post my subjective opinion on the sound of the regulators again, as I expect it to be as excellent as its previous incarnations, which I already reported.
By the way, the out voltages have been very stable.
By the way, the out voltages have been very stable.
Last edited:
When I built it it was intended for +/-15V rails. I just changed my mind to have them on +/-12V so that I could perhaps mix with my DCB1 like crossover filters.
So I have to replace the 27k with something smaller, perhaps something like 22k.
I wonder how much difference can it allow between the upper resistor and the lower resistor - 30% difference OK?
So I have to replace the 27k with something smaller, perhaps something like 22k.
I wonder how much difference can it allow between the upper resistor and the lower resistor - 30% difference OK?
Last edited:
Hi Bill, good job! And I appreciate you taking the time again to let us know about it.
About the Vref upper and lower difference, I don't know if we can go by percentages. One hard constraints is the Vceo of the pnp to its left.
About the Vref upper and lower difference, I don't know if we can go by percentages. One hard constraints is the Vceo of the pnp to its left.
Iko, can a mismatched J310 cascode top fet in the mini CCS create the discrepancy that Bill experienced, or it could be something else? The 2SK170 is mainly responsible for the running current there, yes? I would be more suspicious of different LED Vf and CCS Mosfet Vgs.
IMHO just the J310 mismatch shouldn't be responsible for such a difference between + and - rails. Instead of matching devices I would just use the resistors to adjust the current and output voltage as desired.
Thanks, guys. I was too excited yesterday after the build so I made a number of mistakes in my thoughts / reports. First I could use the same jig to test the J310 as the drain is always on the side for those jFETs. Secondly, the CCS current is determined by (Vf_d1 + Vf_vbeQ4) / R1 so the J310 is not playing a role there. The 5d schematic was in my mind at the time.
Yes I knew I could replace R1 to adjust the CCS current but with the highly compact PCB and the tall heatsinks it is not going to be great fun, but it can be done.
I guess the mismatch of the Vref was partly due to the CCS current but also partly due to the Idss of the J201 too. So matching the jFETs is still a good idea.
Is it possible that the MOSFETs have a large discrepancies? Ikoflexer, are you sure that eBay source is reliable selling geniune products not those cheap ones made from southeast Asia? I bought 20-25 pieces from that guy. Actually, Farnell is not too expensive on that parts. Should I try the Farnell ones?
Yes I knew I could replace R1 to adjust the CCS current but with the highly compact PCB and the tall heatsinks it is not going to be great fun, but it can be done.
I guess the mismatch of the Vref was partly due to the CCS current but also partly due to the Idss of the J201 too. So matching the jFETs is still a good idea.
Is it possible that the MOSFETs have a large discrepancies? Ikoflexer, are you sure that eBay source is reliable selling geniune products not those cheap ones made from southeast Asia? I bought 20-25 pieces from that guy. Actually, Farnell is not too expensive on that parts. Should I try the Farnell ones?
J201 and its current set 1k2 resistor most likely govern Vout differences if fixed a priori. Trimmer for that 1k2 IMHO.
- Status
- Not open for further replies.