The ultimate DIY? A Stax SRM-T2!

[starcat]

On 7/25/2011 at 7:45 AM, Kerry said:

Got the Amp and PS done, just working on the umbilical cords. I'm going to make them 30" which allows for both side x side or stacked configurations.

 

Amp...

 

 

PS...

 

So sad, that all those pics are missing... any chance to resubmit or relink those somehow?

[starcat]

T2 PSU is now live

Got a new LV transformer with the proper secondary 30VCT and not 30-0-30, changed both 4700uF 25V caps with new ones rated 50V. Other involved caps were 50V already. The 7812 and 7912 regulators I left. 

Powered the 12V supply first, then the positive rails, then the negative. All voltages look good except for the -260VDC I am measuring -135. Do I miss something?

Edited June 19, 2021 by starcat

[Kerry]

If the -560v is correct, then the 300v supply coming off of it must be the problem. The -560 + 300 voltages gives  -260v. 

[starcat]

For -500 and -560, I am measuring -501.14 and -562.1, both are absolutely stable. 

The -260 measures -134... -135 and fluctuates in the .1 range

Checked all resistor values, orientation of diodes, etc. The 2SC3675 I measured before soldering. The IXCP10M90S and FQPF8N80C I didn't check as they were straight off Mouser. Not measuring any short for any of the heat sink mounted devices.

Any hints as to where I might start looking first? 

Edited June 19, 2021 by starcat

[JoaMat]

Familiar with the diode test method? See this post.

Seems the regulator isn’t doing anything but passing all unregulated voltage through.

My bet is on a faulty mosfet... or something else.

[starcat]

Nope, just reading... but not getting any wiser. Faulty FQPF8N80C mosfet or IXYS regulators, desolder and testing in the DY294 or measuring while on the pcb?

Finishing the teflon output wiring in the mean time. 

Edited June 20, 2021 by starcat

[Kerry]

Possibly Q18, but most likely Q19 or the regulator circuitry itself. 

You could check the voltage across D12. Should be around 30V. 

[starcat]

On 5/5/2013 at 1:28 PM, Lil Knight said:

I am using the quad TKD pot. Wondering which pot you are having problem with?

Did you made any intermediate board fitting the TKD pot to the original PCB? 

[jamesmking]

15 hours ago, starcat said:

Nope, just reading... but not getting any wiser. Faulty FQPF8N80C mosfet or IXYS regulators, desolder and testing in the DY294 or measuring while on the pcb?

Finishing the teflon output wiring in the mean time. 

 

You may have done some of these tests already but for the -260V output (i.e. 300+ above the -560V rail,

-1 inspect carefully for solder shorts, joints missing solder etc.

0. I would check the AC voltage going in is what is expected.

0.5 check dc voltage after bridge rectifier and smoothing cap is what is expected.

0.6 check the input side of the pass transistor Q19, if Q19 only has say 145V on the input side then it cant be expected to output 300V... 

1. I would connect to a variac and see if the psu is actually regulating at 134V or if it varies with input voltage

2. I would check the voltage reference is outputting the correct voltage

3. If the op amp is in a socket I would swap it out for a known good one

4. I would check the voltage drop across the zeners diodes and check the value of the resistors in the voltage divider that sets the output voltage

5. if you have a peak dca75, de-solder  the transistors and run an identify on each. I have found that most of the failed transistors I have tested on the dca either show up as back to back diodes, shorts or open, or crazy low gain compared to the spec.

5.1 although not as reliable, you can use the dca in circuit without desoldering the transistors first.  Make sure the circuits are not powered and the reservoir caps are fully discharged before running any in circuit tests. if you run dca75 identify tests on the transistors on a known good rail and compare to the corresponding transistors in the bad rail. Any major differences may identify that component as a possible issue, since the rails have identical topology and the low voltage behaviour of all the transistors in the rails should be approximately the same. Do expect some identification fails (device not recognised or no device connected) since you are testing in circuit. But at least it gives you a point of comparison and can be quicker and safer than desoldering...

good luck

 

I have not used the DY294 in circuit, I certainly would not advise using any of the breakdown modes in circuit because the voltages used could easily over volt components. The peak dca75 identify function seems to be quite "gentle" and I have used it in circuit on various golden reference lv and hv boards, blue hawaii amp boards etc with no ill effect.

Diode testing the transistors is also an option but I have found transistors that have failed but pass the multimeter diode check because they have failed in such a way that they now act as diodes with zero gain. In this respect peak dca75 identify is better. But multimeter diode check is better than nothing. NOTE any depletion mosfets may diode test as shorts or near shorts since depletion mosfets are fully on by default unlike enhancement mosfets...

 

 

Edited June 20, 2021 by jamesmking

[starcat]

Thanks, James! Checked 1. 

Is 2 and 3 a possibility when I am getting stable -560V out of the same PSU? It is the -260 that is flaky, pointing to the 300V as Kerry suggested.

Edited June 20, 2021 by starcat

[jamesmking]

11 minutes ago, starcat said:

Thanks, James! Checked 1. 

Is 2 and 3 a possibility when I am getting stable -560V out of the same PSU? It is the -260 that is flaky, pointing to the 300V as Kerry suggested.

I mean just for the -260V / 300V rail. I entirely agree the rest of the psu looks like its working including the -560V rail the +300V rail sits on. 

Edited June 20, 2021 by jamesmking

[starcat]

Hmm, diode testing Q16-Q19 doesn't show anything suspicious, they measure like Q11-Q14. 

Going further...

[starcat]

Tested with the Peak DCA 

4 hours ago, jamesmking said:

5.1 although not as reliable, you can use the dca in circuit without desoldering the transistors first.  Make sure the circuits are not powered and the reservoir caps are fully discharged before running any in circuit tests. if you run dca75 identify tests on the transistors on a known good rail and compare to the corresponding transistors in the bad rail. Any major differences may identify that component as a possible issue, since the rails have identical topology and the low voltage behaviour of all the transistors in the rails should be approximately the same. Do expect some identification fails (device not recognised or no device connected) since you are testing in circuit. But at least it gives you a point of comparison and can be quicker and safer than desoldering...

With the Peak DCA75 measured without desoldering, Q18 show as unrecognized more often than not. Changing the pins and it shows up. It measures like Q16, Q17 however (when it gets recognized). 

FQPF8N80C Q19 gets recognized the same way as Q14.

BR3 must be ok as the -560V is output correctly and stable as well.

Edited June 20, 2021 by starcat

[Craig Sawyers]

You can't use a Peak DCA75 to test in-circuit. It says so in the manual.

[jamesmking]

25 minutes ago, starcat said:

Tested with the Peak DCA 

With the Peak DCA75 measured without desoldering, Q18 show as unrecognized more often than not. Changing the pins and it shows up. It measures like Q16, Q17 however (when it gets recognized). 

 

22 minutes ago, Craig Sawyers said:

You can't use a Peak DCA75 to test in-circuit. It says so in the manual.

 

You can and I have/do. I agree its not 100% reliable in circuit and I know it was not designed to be used testing in circuit. In some cases it will not be consistent when identifying a transistor because the circuit puts it close to the edge case for a recognition and probably has zero input protection but I still have and do find it useful in circuit.

Its useful when you have two identical or near identical circuits. You can make comparisons and it at least narrows down what you might have to desolder. In some cases the circuit around the transistor effect the behaviour of the transistor so its miss identified or is right on the border of identification (as stated in the post above). But quite often it will identify correctly and I have never had it identify a bad transistor in circuit as good... its just another tool to be used on special occasions when you know what you are doing and you have to think about the results. For example when testing one of two individual transistors wired as a Darlington pair can identify the single transistor as a darlington. But again if it does the same identification with similar gain on both the good and bad rails then those transistors are likely to be good etc etc. I prefer testing this way than desoldering all the transistors and checking them all.

But different people go about problem solving and troubleshooting in different ways. When I know it identifies a transistor correctly and consistently in one circuit and identifies the corresponding transistor in the other rail as two back to back diodes or says the gain in the good rail is say 100 and the gain on the other transistor is 2 then that transistor is suspect and a candidate for pulling and further testing.... (all of which I have experienced)

Its more limited when you can't make comparisons between known good and known bad, but if it correctly identifies a transistor type and pin out then a quick look at the schematic for possible other transistors nearby it could be identifying instead then you can be reasonable certain that transistor is ok.

 

 

 

 

Edited June 20, 2021 by jamesmking

[Craig Sawyers]

Well I wouldn't even try. But then I'm a stickler for doing things by the manual, and believe in the maxim RTFM. I think that doing inter-comparisons between bits of circuit is likely to confuse rather than illuminate.

The best way to test the circuit is to run it and very carefully measure voltages (because they are high, and deliver substantial currents) and compare with what is expected. Just don't slip with a probe, because it might spoil your day.

To keep from generating frightening sparks and dead silicon, I have often soldered on stiff short lengths of wire as temporary test points and use those to clip to. Once the fault is diagnosed, they can be desoldered.

[jamesmking]

1 minute ago, Craig Sawyers said:

Well I wouldn't even try. But then I'm a stickler for doing things by the manual, and believe in the maxim RTFM. I think that doing inter-comparisons between bits of circuit is likely to confuse rather than illuminate.

The best way to test the circuit is to run it and very carefully measure voltages (because they are high, and deliver substantial currents) and compare with what is expected. Just don't slip with a probe, because it might spoil your day.

To keep from generating frightening sparks and dead silicon, I have often soldered on stiff short lengths of wire as temporary test points and use those to clip to. Once the fault is diagnosed, they can be desoldered.

another possibility is to either change the voltage set resistor or parallel it so that the voltage is lower. I do this on initial testing of high voltage rails. I generally parallel to get about 50 - 100V output, use a variac and check it regulates close to the expected output voltage. This reduces the voltages and if something does go wrong there is more chance of more components surviving... but as I say each to their own problem solving style.

[Kerry]

Don’t worry about Q16 and Q17 since you are getting about 425v out of the regulator. 

If you are getting 30v across D12 then it says that Q18 is most likely working. 

If you report back on that it will be easy to figure out next steps. 

[starcat]

Ok, measured some voltages

Q15/C3675   B -500.79   C -227.5  E -501.31

 

Q20/C3675   B -562.1   C -120.74   E -562.4

Q19/FPQ   G -126.27   D -132.84   S -137.05

Q16/IXYS   G -132.19   A -132.03   K -132.49

Q17/IXYS   G -132.35   A -132.11   K -132.12

Q18/IXYS   G -531.1   A -131.92   K -528.00

 

ref102   pin2/v+ -531.2   pin6/vout -531.9/-532

op27   pin2/-in -539.6   pin3/+in -540.2   pin4/-v -562.4   pin6/out -561.5   pin7/+v -531.1

 

across D12 zener 31V

right of R45 -531.1   R44 -130.27   R43 -130.67   D10 -136.51

R48 400k  at one side -562.4 and -130.05 on the other

 

-560V rail output measures -562.4V and stable

-500V rail output measures -501.35V and stable 

-260V rail output measures -135V and fluctuates in the .1 range

Edited June 20, 2021 by starcat

[JoaMat]

May I suggest a diode test comparison Q20 against Q15 to exclude the 2SC3675? If not already done.

[starcat]

5 minutes ago, JoaMat said:

May I suggest a diode test comparison Q20 against Q15 to exclude the 2SC3675? If not already done.

Inserted above Q15 voltages referenced to ground. I did a diode test with both when PSU was off and they both measured the same. Will post asap.

[JoaMat]

Great!

Looking at your measurements of ref102 I suspect you might have the problem there. You should have a potential at Out (pin 6) of 10 volts above -562V - that will be -552V and not -532V.

You have a spare ref102?

[Kerry]

^ Agreed

[starcat]

4 minutes ago, JoaMat said:

Great!

Looking at your measurements of ref102 I suspect you might have the problem there. You should have a potential at Out (pin 6) of 10 volts above -562V - that will be -552V and not -532V.

You have a spare ref102?

Measured at pin6 again: -531.7

No spare atm but I could take one off the working PSUs (and turn that one off).

Edited June 20, 2021 by starcat

[Kerry]

That will do. Take it from one of the positive rails. 

while you’re at it, double check the value of R72 and confirm continuity of the ground pin for the reference.