Ampeg PF350 - IRS20957s no oscillation

[Chris56000]

An update on this for the OP and others :–

I am now looking at the unit at home for the OP foc, and it looks like the fault is related to the switch–mode power supply coming on at too high a voltage that is causing the IRS20957s to trip out into a protective mode, so the ±73 V the OP said he measured  at switch–on is not correct!

I am continuing to check the Power Supply and it's regulation and feedback components and I will report back in due course!

Chris Williams

[jboy32]

Good morning Chris,

Thanks for the update, from my side, I did check the power supply part with a focus on the KA3525A that is the IC regulating high +/- voltages, and didn't didn't find any issue their, could it be the KA3525A himself ... I was thinking to play with the PWM on KA3525A to adjust the high voltages in case that is the root cause. I tested the amp by powering with a VARIAC to move from 230Vac to 190Vac and get +/- 55 Vdc. all other voltages still good as they are regulated. Even with that no change on IRS20957s.

In my case by changing C61, IRS20957s seems not to be in fault now as I have +5,576 Vdc taken to GND. I'm still trying to understand why I have the same voltage on VS on IRS20957s than on HO. It seems that something is wrong on the output.

When I plug a 8R resistor on the output,

VS Pin 13 moves from 2,71Vdc to 0Vdc
VB Pin 15 moves from +9,75Vdc to 7,7Vdc
CHS Pin 16 moves from 3,7Vdc to 1Vdc

I followed with an oscilloscope the input signal and it goes to Pin 3 of IRS20957s.

It is a REV H of the AMP with some differences with the schematic that we have, so maybe on the REV H +/- 75Vdc is the voltage expected ...
In the differences :
1. no 12V regulator, U7 is missing
I did test by adding one but no changes.

2. on the output, there are 4 additional components that are not on the schematic.
R94 = 10R
R98 = 10R
C84 = ?
C97 = ?

3. D34 is an ES1D diode and not an ER1D diode

they make a bridge (R98+C97) from D to S on Q13
they make a bridge (R94+C84) from D to S on Q10

I did replaced IRS20957s but maybe I have a fake IRS20957s. I don't know if I have a DC voltage on VS pin13 because I removed Q10 and 13 ... I have tried with the 8R resistor on the output that moves VS to 0Vdc and replaced Q10 and Q13 but again they died after few seconds ...

In your case with Q10 and Q13 in place are they safe?

I have ordered ES1B diodes to replace them also, for D32 and D33, I will get them next week.

I add a link to an interesting video in fixing a PF350

[jboy32]

Good afternoon,

Some update on my investigations, I did not yet received the ES1B Diodes.

I put a 440Hz signal on the input of the amp to follow it. it comes to Pin 3 of IRS20957s with a square so seems to be good, I was thinking that ADTL082JRZ IC could be the culprit.

IRS20957s might be a fake IC that I ordered, so I'm going to buy some from some websites that I hope certified they are true.

I found lot of information there that I'm still trying to understand, situation is very similar. https://www.diyaudio.com/community/threads/behringer-1400w-class-d-for-b1800d-repair-help.247903/

[jboy32]

Hello

Some update on my investigations.

I changed :
 - D32 D33 and D34
 - IRS20957S (other supplier)
 - IFRB4227 x2 (other supplier)

Still in fault but with only IFRB4227 on HO in shortcut, so currently only keeping the IRFB4227 on LO in place.

What I have seen now is that CSD is oscillating from 2,5vdc to more that 7vdc a a frequency of 1hz ...
Btw, I did some progress to learn how it works.

If :
1. I remove C61, I have a stable DC 5,5 Voltage on CSD pin 2 but not enough to get the IRS20957 starting to oscillate ..
2. I inject 440hz on signal on the input, I increase gain around 90% and volume around 90%
-> CSD increase from 5,5dc to over 8,5V dc and stable
-> Pin 3 has a square signal at 440hZ around 4,3Vdc
-> LO start to get 440Hz oscillation ... on pin 10

Regarding the forum and the schematic, I have seen that normally pin3 of IRS20957 should get a square signal between 340Khz and 400Khz.
Should it be without an input signal I don't know it seems yes.

So now I'm looking at the ADTL082A that is supposed to produce the PWM at 370Khz and 430Khz that might be dead ... I checked all components around and they are good.
I would expect that without input signal when I power up the amp, I should get a voltage on pin 3 of IRS20957 with a square frequency of 370Khz / 430Khz and on top of that if there is something on the input that will influence the signal ...
 

[Mojoampguy]

JBOY32.

The way this part of the circuit should work in normal operation, in my understanding is this:

With no signal fed from the preamp. i.e nothing plugged into the amp.  The output of the irs20957s should normally be a fast uniform square wave. In the range you suggest. Approximately 400khz.

When an audio signal is presented to the TL081, this op amp (and feedback from the o/p of the Mosfets ) effectively operate as an integrator - net result, it varies the pulse width of the TL081 output. So now you should see the 400khz signal with varying pulse width feeding the input to the irs20957s and subsequently the o/p from the Mosfets.

You need to have the high side and low side Mosfets in circuit to operate normally.


I have not looked at this for some time, but I think this is generally correct.

[jboy32]

Good afternoon

I followed your recommendations by sitting an IRFB4227 on the HO. So getting all new and replaced component in place.

In parallel, I was thinking that it could be good to reduce the input voltage of the amp, I have a VARIAC (0Vac to 300Vac) and moved from 235Vdc to 190Vdc main input so, I have +/- 56 VDC like the original schematic, and other low voltages are pretty closed to what is expected.

You know what ? it works !   

I have 9,31 stable dc voltage on pin 2 CSD and 2Vdc on Pin3 with 400Khz PWM ! with a load of 8R connected to the output and no input signal.

I plugged into a CAB ang got a nice sound, for the moment I didn't go forward, because I don't know if it is because of the low input voltage that finally adjust rails on +/- 56Vdc or I did something else.

I'm just afraid that if I move from 190Vac to 235Vac so getting +/- 75Vdc on rails that could destroy again the power amp part.

If that is the case we need to investigate on the SMPS power supply. That in my case was also burnt at the beginning, and I replaced some components there.

From the power amp part the only thing I changed :
- IRF20957s x1
- IRFB2742 x2
- C61 with another 10uF capacitor

If only an Ampeg tech could confirm that rails voltage on this release should be +/- 55Vdc or  +/- 75Vdc with REV H that could help to investigate on the SMPS.

Julien

[jboy32]

Good morning,

Update on the situation, I did some tests yesterday evening by increasing progressively main input voltage so increasing in parallel rail voltage from:
190Vac Main input -> +/- 56Vdc rail Working and powering off then on = ok
200Vac Main input -> +/- 59Vdc rail Working and powering off then on = ok
210Vac Main input -> +/- 62Vdc rail Working but when powering off then on -> one IRFB4227 (Q10) shortcut

I replaced again Q10 IRFB4227 and it is working by reducing main input voltage to 190Vac.

So it makes me thinking that we need to get +/- 55 Vdc on the rails.

Possible investigations :
U1 : KA3525A (I already checked all components around they are all good)
T1-1 : Transformer for rails voltages damaged ?
D16 : MUR1620CT damaged ?
D19 : MUR1620CTR damaged ?