i'm on a roll... the kgsshv

[rille]

Been kind of quiet around here. Anybody else finish their kgsshv yet?

Not yet. I'm still struggling with the power supply

[deepak]

That and a bigger trafo? I dunno.

The start-up surge on this thing is quite intense.

A loud buzz/thump can be heard, while the livingroom lights go dim for a second.

I mentioned this earlier during the build, another more experienced board member dismissed it

saying that he wouldnt worry about it, or something to that effect.

So I figured why bother and didnt implement a soft start. I will now.

BTW, the trafo never ran really hot, it was just warm to the touch with the amp turned on for an extended period.

I couldnt find any cheap Hammond trafos on ebay, usually they have them for about $50.

New, they are $105 on up, so I'm not going that route for an older, noisier design.

This fried SumR toroid only cost ~$100 with the core band insulation added.

Craig Sawyers talked of some super-duper trafo that has every turn layer insulated

which he sourced for his T2 build, I might look into that as well as a fully encapsulated unit.

So it's back to SumR? Makes me cringe, although I will spec something a bit more substantial this time around.

I have room for it in this chassis.

EDIT: Another thing. I'm not actually blaming SumR here due to the fact that

I had previously shorted the power supply twice in the past,

possibly weakening the trafo insulation in the process.

I am aware that they are getting a rep for "donut fail".

Since I dont have a failure analysis lab at my disposal,

nor want to pay to have that done, I can only guess.

I used to have the same thing happen on my DIY Pass Pearl phono. There is no power switch so after plugging in the power cable there was a semi loud humm (you could hear it for about 2 seconds then disappeared). Strangely switching to a higher gauge power cable solved that issue, which was really an accidental finding when I needed to use a longer cable to reach the outlet.

[cetoole]

I used to have the same thing happen on my DIY Pass Pearl phono. There is no power switch so after plugging in the power cable there was a semi loud humm (you could hear it for about 2 seconds then disappeared). Strangely switching to a higher gauge power cable solved that issue, which was really an accidental finding when I needed to use a longer cable to reach the outlet.

Higher gauge meaning thinner wire, and therefore more resistance?

[FrankCooter]

Sorry to hear about your amp Livewire. I was tempted to go to the recent San Diego meet primarily to hear it. I'm not familiar with the specifics of this design, but perhaps you'll find something useful in a few general comments.

1) Are you sure the transformer failure is a root cause rather than a symptom of another failure? Considering your previous problems with the regulator, you might want to have a closer look. How close are you running your filter capacitors to their voltage limits? After a major short, I'd probably change them. Use the highest voltage rating available. I've had issues with Ebay imports not meeting their voltage ratings. Pull the transformer out of the circuit, if it seems OK with a meter, try it with a dummy load and a Variac to bring it up slowly. If it's OK,put it back in the circuit and try a simple RC filter in place of the regulator and again bring it up slowly with the Variac. If you're OK, the problem is in the regulator, which is much more likely than the transformer. If You've still got a problem, I'd be looking for a short somewhere in the amp itself.

2) I'd definitely want both soft start and overcurrent protection in the power supply, If ordering a new transformer, I'd spec it for 3x the normal current draw if you have room in the chassis.

3) You might want to check out Antek for reasonably priced toroids. You'd probably have to stack a pair to get all the voltages you need, but you'd still be substantially under a $100.

Hopefully, someone with specific knowledge will give you some better info, Good luck! Looking forward to hearing your amp at a future Ca. meet.

[livewire]

Thanks for your comments Frank. I do appreciate your input.

I have not had a chance to open it up and poke around with my DMM yet.

As far as the filter caps go, they are running right on the rated voltage.(450 VDC)

I bought them from Digikey, I dont trust ebay imported crap. Yes, a lot of it is fake.

When I previously shorted the psu output, I pulled the caps and tested them.

They checked ok (static capacitance on a RCL meter & had stable DC output under load).

The amp has been running fine since then for about the last month and a half.

I have around 100 hours on it.

2) Will do.

3) Thanks for the tip!

Antek it is, $54 for a beefy dual 475 volt secondary toroid. Excellent!

[FrankCooter]

Usual "rule of thumb" for electrolytic capacitor operating voltage is 80% of voltage rating. If you're running 450V rails, you are really hammering the first capacitor of the filter. In this situation I'd want at least 525V rated caps.

[livewire]

Looking at the schematic below, there are two caps in series between each rail and ground.

This would effectively divide the incoming voltage load between them, yes?

kgsshvps2.pdf

[livewire]

To clarify, 2 each 680uf @ 450VDC before the shunt & series regulators,

and 2 each 100uf @ 450VDC after the series regulator, at the output.

[Pars]

Barring any other comments, yes, it should divide the voltage across the two caps. That is what the 300K resistors are for, to ensure that it is divided.

[kevin gilmore]

Yes a pair of 450 volt caps in series on about 625 volts of raw unregulated DC.

Has to be this way as 650 volt electrolytics are very custom items and i have

not seen any in quite a long time. In fact the 500 volt electrolytics (550 surge)

are very hard to find these days.

The resistors balance the leakage currents. Which is why the center tap is a

reasonable idea.

[rille]

My power supply is still not working. Could any of the changes between Version 3 and 4 help? If I change to the simple version with zener diodes I would only need the diodes? Also the cap in parallel (C5, C6)?

I get ~60V from the positive rail. I changed all semiconducters already

[Kerry]

Check the 10V reference it should be at -10V with respect to the output. Also, what are the voltages at the emitters of Q4, Q6, and Q7 as well as the collectors of Q6 and Q7.

[livewire]

My power supply is still not working. Could any of the changes between Version 3 and 4 help? If I change to the simple version with zener diodes I would only need the diodes? Also the cap in parallel (C5, C6)?

I get ~60V from the positive rail. I changed all semiconducters already

Sounds just like what I went through...

Maybe Kerry can help you solve it.

Read back to my post #833 (@ page 42?) for info on the simple psu. Click on the attached schematics.

KG did mention that something's not right with the 10 volt reference diode, but didnt explain what's up.

[rille]

Check the 10V reference it should be at -10V with respect to the output. Also, what are the voltages at the emitters of Q4, Q6, and Q7 as well as the collectors of Q6 and Q7.

At the LM4040 I can measure 10V. I will measure at the other points tomorrow.

Read back to my post #833 (@ page 42?) for info on the simple psu. Click on the attached schematics.

Yes, I read it. The only question is if I should leave the cap in parallel to the zeners. I guess yes because I saw it in other power supplies.

KG did mention that something's not right with the 10 volt reference diode, but didnt explain what's up.

Thats the problem. Generally some more information and help would be nice

[Kerry]

The emitters of Q4 and Q5 should sit just above (about .6V) the 10V reference (base of Q4) and the base of Q5. Q4 and Q5 bases (440V) and collectors (225V) should be very close to each other in voltage. Q6 and Q7 are there to split the difference between the output and ground and should likewise sit just above their bases at about 225V. Collectors of Q6 and Q7 should both be below 1V.

When you do the measurements, get the voltages of the bases as well for Q4 - Q7.

The circuit that Kevin designed should work fine. I wouldn't give up just yet.

EDIT: I was trying to clarify the values for Q4 & Q5 collectors.

Edited May 23, 2011 by Kerry

[Kerry]

One other thought. The hFE for each pair (Q4/Q5, Q6/Q7, and Q8A/B ) should be matched when you are selecting transistors. I don't think this will cause what you are seeing, but I know it will throw the voltage off a bit.

Edited May 23, 2011 by Kerry

[rille]

The emitters of Q4 and Q5 should sit just above (about .6V) the 10V reference (base of Q4) and the base of Q5. Q4 and Q5 bases (440V) and collectors (225V) should be very close to each other in voltage. Q6 and Q7 are there to split the difference between the output and ground and should likewise sit just above their bases at about 225V. Collectors of Q6 and Q7 should both be below 1V.

When you do the measurements, get the voltages of the bases as well for Q4 - Q7.

One other thought. The hFE for each pair (Q4/Q5, Q6/Q7, and Q8A/B ) should be matched when you are selecting transistors. I don't think this will cause what you are seeing, but I know it will throw the voltage off a bit.

Thanks for your help

I measured Q4-7 (all voltages to GND):

Q4 (B/E/C): 46,15/46,60/30,1

Q5 (B/E/C): 54,7/47,3/17

Q6 (B/E/C): 27,8/30,4/0,6

Q7 (B/E/C): 27,8/16,8/0,02

Output voltage was ~56V. The voltage is not stable and fluctuates a little bit.

I didn't matched the transistors in the power supply. Only in the amp section.

[kevin gilmore]

There is nothing wrong with the lm4040. Its just that it is 90% or more

of the total noise in the circuit. The linear lt1021 is a much better part.

Its also 5 times the cost.

I'm trying to figure out what is going on.

But i have a new camera toy to play with.

[spritzer]

Leica M9?

[wink]

But i have a new camera toy to play with.

.

[rille]

Was anybody successfull to build the PSU version 3 at the corresponding PCB? I have still problems, livewire had problems too. Perhaps there's something wrong with the PCB? Or this version was not the best solution. Version 4 has some changes.

[rille]

So long story short, I left the crippled battery circuits in place and cut some traces

to disable current flow through them and soldered in the zener strings(3ea@150V-5W) as shown in the schematics below.

After having no success at all I tried this today with 3 x 160V zeners. It's still not working. The output voltage raises slowly. After 10 min it reaches 285V.

Is it normal that the voltage across the zener diodes drops instantly if I want to measure it? Also the voltage across D7-D10 is IMHO too low. I get only 98V and not 580 V as expected.

[Kerry]

The voltage should not drop significantly when you measure the zeners. Most DVM are about 10M resistance. If you are getting the proper current from Q8 and Q17 then you should not notice a drop.

If the input voltage is correct >600VDC, then seems like Q8 and Q17 are not set correctly or working properly. Double check that R1 and R39 are both 3K resistors. Also try to measure the voltage drop across them. It should be about 4V. This would equate to approximately 1.3mA of current. My guess is that it will be way low for both, which would point to bad current sources.

[rille]

The voltage is much too low, only some mV. The strange thing is that in the working part it's also too low. I really have no idea why they are defective. I changed Q8 and Q17 already. I bought them from littlediode. I hope they don't sell faked parts.

Is there any alternative for the 10M90S which is better available and not so expensive in Europe? They cost > $10 and it takes two weeks to get them

[kevin gilmore]

If there was a replacement for the 10m90s i would be using it.

It is a very sturdy part, i've yet to be able to blow one up.

And I have tried. People in europe should get together and

have someone in the usa buy 100 or more of them.

Pretty sure there is significant stock in iceland