Ampeg PF350 - IRS20957s no oscillation

[Mojoampguy]

I have repaired the SMPS in this class D bass amp. Unit powers up, fault light goes out, but no audio output. This is due to IRS20957s gate driver NOT starting up oscillation. All voltages look good, except for CSD which sits at 3.2v. It need to be greater than 5v to start up. External mute circuits are not active,.

I’m stuck.

Schematic is attached. IRS20957s is U8 in power amp section. I have changed this device 3 times already. All connecting components appear good.

[Audiorepair]

Where did you buy the 3 IRS devices?

[Mojoampguy]

Bought from Farnell components in Uk. Reputable supplier. I realise there are fakes about.

[Chris56000]

Hi!

. . .Thank you for providing the schematics and layouts – it helps us enormously when diagnosing a fault like this!

. . .Apart from some strange Asian semiconductor type nos, it's relatively straightforward to diagnose this!

. . .First of all, is the delayed power–on reset releasing the "MUTE" line? The +55 V main h.t. supply is connected to a reset circuit consisting of C127, R86 and R89, the voltage at the positive terminal of C127 should be about 5.7 V and Q9 should be conducting with its base approx. 0.6 V.

. . .If Q9 is not being turned on, check the +55 V h.t. supply is reaching that voltage, and replace C127, Z2 and Q9 if you're in any doubt about these components – a BZX84–C5V1 will do for Z2 if the package matches!

There is also a 5 V "P_MUTE line needed to turn ON transistor Q11 to release the "MUTE" line once the main power supply circuit has started up – this comes from the 5 V "Reference Voltage" output of the '3525 SMPS controller in the power supply circuit – if the '3525 isn't running correctly, Q11 will not turn ON and the "MUTE" line will remain low – this is a safety protection feature designed to turn OFF the drive to the Class D power MOSFETs until the power supply is running correctly!

. . .Also make sure you have no extraneous DC at the mid–point of the Class D output stage due to defective MOSFETs or a low or missing ± 55 V h.t. supply – problems here will trip the power muting circuit via the d.c. protection transistors Q15, Q16 and Q17 in the power amplifier circuit!

. . .There are also three temperature sensors in your amplifier that need a quick check – TS1 is a normally closed bimetallic thermal switch fitted to the main heatsinks, an PTC thermistor TH2 in the temperature protection circuit, approx. 50 Ω at room temperature , and a DO220 package thermal device TH3 called an "AIRPAX 67L050" which should be a short–circuit cold – this becomes high resistance when its tab reaches 50°C or greater, when any of these sense an over–temperature fault, Q12 in the power supply circuit turns ON, shutting off the main '3525 SMPS controller by way of the '555 monostable U4 which trips the "/SHUTDN" input at pin 10 of the '3525!

. . .Therefore, the first thing to do is check both the +55 V, –55V and the ±15 V supplies as all these are sensed, followed by the power–on–mute circuit, then the output stage itself for DC faults, then the over–temperature protection, roughly in that order!

Chris Williams

[Mojoampguy]

Lots of great input there Chris. I check all this when the Grandaughter baby sitting duties allow???

One thing I did not mention yet. +/-15v in spec. The HT voltages are measuring +/-73v at the moment, but this May again be due to the IRS20957s not oscillating bringing o/p Mosfet load into play.

Also, I have struggled with this for months on and off, your suggestions are well worth checking. I can’t even remember what I’ve checked previously, at this stage.

Thanks again for your input.

Dave

[Mojoampguy]

Hi Chris,

I made all the measurements on the Mute/ startup circuit, and Temp sensors. I had done this in the past. All appeared correct. Do you agree.

See my readings in scrawny hand written RED next to your recommendations.

Note my comment that +/-55v read +/-72v ( possibly due to o/p Mosfets not presenting load, due to lack of oscillation of the IRS20957s )


The only ODD reading is 3.2v on CSD. I will replace Capacitor C61 10uF connected to it. It’s SMD, I don’t have test equipment to measure it’s ESR.

Thanks again for the suggestions. Any more ideas VERT welcome.

[Chris56000]

Afternoon Dave!

. . .See how you get on first with my initial suggestions – it is very possible that the ±73 V you're reading on the main h.t. supplies is because they're not being loaded by the output stages as you suggest, however I would also, in addition, check the temperature sensor voltage from U10 (LM19CIV) pin 2 – this should be about 1.6 V at normal room temperature – if this is low or missing,  check R27, C19, Z3 (5V1) and U10 by substution!

It's possible there could be a fault in U12 – this is a standard SOIC-14 LM324 despite its peculiar Asian type number!

. . .Don't forget to double– check your output MOSFETs and the components associated with the driver ICs themselves, in case you made an accidental slip–up – you mentioned you'd tried three!

. . .I have read your notes and figures appended to the circuit diagram extract you posted by the way, on the face of them, there doesn't seem to be anything obvious in the protective circuits that's shut it off that I can see!

. . .There aren't any microcontrollers, front panel circuits, etc., etc., not shown on the diagrams you provided fitted to your amplifier by any chance?

. . .Have you tried lifting up the "CSD" pins from the print and using a 10k resistor to the +15 V supply to see if this brings the driver stages on? If this trick enables your amplifier I can only suggest going through all your work again – it's possible there might be a solder whisker (easily done with surface–mount!), dry–joint, etc., that's the driver ICs have detected and are shutting off internally!

Chris Williams

[Mojoampguy]

While I’m waiting for parts, I can confirm U10 pin 6 is 1.6v as required.

Mosfets.appear good.

P.S I’ve a attached a good Application note document on the IRS20957s. It describes the shutdown protection design. I’m no Design engineer so I struggle to interpret some of it. You may find in useful Chris56000?

Thanks for the continued help. I don’t like to give in , but this one is driving me crackers.

[Chris56000]

Afternoon Dave!

. . .It makes perfect sense to me, I wonder it the internal excess current protection is being activated – if you're only monitoring the CSD pin with a meter, which would not necessarily show up the rapidly fluctuations of the CSD pin if the IC is tripping, but you would read a low average voltage!

. . .An oscilloscope would show it up tho!

. . .I would take a look at the associated diodes and low–value resistors associated with the Output MOSFETs, these are R141, R142 and R143 and R187, and diodes D28, D31, D32, D33 and D47!

. . .Did you use exact replacement MOSFETs? MOSFETs have an input capacitance Ciss and an Input Gate Charge Qg, and if these are markedly different from the original OEM fitted MOSFETs, the IRS20957s might not be able to drive them at the frequency the Class D output stage is switching at, therefore this could trigger the internal excess–current protection in the driver I.C.!

Chris Williams

[Mojoampguy]

Checked CSD signal with scope, Static 3.2v dc. No changes. I seen this happen on at least 3 different pieces of musical equipment powered class D speakers, same scenario. It’s driving me mad.

All components recommended  rechecked . All OK

Mosfets were replaced with same models irfb4227’s. Probably bought off eBay, As I do one off repairs and Farnell, RS, Mouser charge too much for delivery. Release I risk Fakes….

Any recommendations on measurements around irs 20957 particularly OCSET. But resistors here seem fine.

Thanks again

[Chris56000]

Evening Dave!

. . .Here are some more circuits for you to refer to – the first is a Behringher B215 that uses the same circuit and the International Rectifier IRS20957s driver IC Reference Design for this device that goes into a lot of detail into its operation!

Chris Williams

[Mojoampguy]

Thanks Chris 56000. I’ll check these out.

[Chris56000]

Morning Dave!

. . .I have been studying the I.R. Reference Design Books, and in the light of the experiences you have been seeing with your Ampeg and the other similar units also giving you trouble, I can come to no other conclusion that the output  MOSFETs may be fakes or otherwise sub–standard that don't have a sufficiently low dynamic "on" resistance!

It only takes a few microseconds at the most for the internal excess current protection to sense this, which it does by comparing the Vds of the conducting MOSFET with the voltage set on the OCP set pin, and if the Vds is higher than that permitted by the OCP comparator within the '20957, the IC will switch off!

. . .There are two ways to prove whether your MOSFETs are at fault, unfortunately both methods involve some expenditure but that will be repaid by the time saved from an eventual successful resolution!

. . .The first method is to order the recommended MOSFETs used in the I.R. Reference Designs from Mouser, Digi–Key or RS, OR a direct equivalent to them – Ciss, Qgs and Rds (on) MUST all match the specifications given for those used in the I.R. Reference Design ;

. . .The second method is to invest in one of these :–

https://www.ebay.co.uk/itm/275400928273?mkcid=16&mkevt=1&mkrid=711-127632-2357-0&ssspo=oon53xzNRgC&sssrc=4429486&ssuid=qNpFa7hUTgy&var=&widget_ver=artemis&media=COPY

. . .and set it up using two external supplies from another source, then try fitting the MOSFETs you've already fitted in your Ampeg & other faulty units – if this eBay test board fails to start up on these MOSFETs, you will have proved the replacement MOSFETs you previously ordered were substandard or fake!

Chris Williams

[Mojoampguy]

Struggling to appreciate all points in this design note -

One question it did raise for me was, “Maybe the problem is OVer Voltage Protection being triggered”,  as the schematic shows +/-55v dc for supply and I’m reading +/-72v. I’d previously postulated that this was maybe due to Lack of of Mosfet Load???

Maybe to fault is in the SMPS section? I’d convinced myself not, as I thought SMP feedback was derived from +15v supply that is in spec. As you know know much about SMPS maybe you can comment?

[Chris56000]

. . .The ± 55 V supplies are of the "semi regulated type" which rely on feedback from the +15 V rail and also on the MOSFETs taking a normal load current as there is no feedback from the + 55 V line directly to control them!

. . .You can make up a load consisting of 2 × 24V 21W HGV type indicator/reverse lamp bulbs wired in series across the +55 V and –55 V supplies, two bulbs in series per supply – if this drops the power supply enough you can try your Ampeg again – if the IRS20957s was turning off solely due to overvoltage it will come on!

. . .If you have a quantity of suitable resistors on hand you can experiment with the values applied to the '20957 used in the "IRAUDAMP6" schematic (page 36 of the "IRAUDAMP6" booklet) – this Reference Design is rated to operate from ± 73.5 V supplies, so you can see from this book that the ±73 V from your Ampeg's power supply isn't too much of a problem if the component values are right!

. . .A suitable alternative SMPS to try your Ampeg with can be obtained from eBay here :–

https://www.ebay.co.uk/itm/272847383409?mkcid=16&mkevt=1&mkrid=711-127632-2357-0&ssspo=oon53xzNRgC&sssrc=4429486&ssuid=qNpFa7hUTgy&var=&widget_ver=artemis&media=COPY

. . .and if you message the supplier, he can supply it adapted to give ± 55 V supply if desired!

Chris Williams

[Mojoampguy]

Still No oscillation on the IRS20957s!

I have again replaced both Mosfets irfb4227’s and IRS20957s purchased from Mouser ( so definitely Not fakes).

No change. CSD voltage sat at 3.2v. Probably due to elusive fault.

See attached file with measured voltages on all pins. Note values in () are relative to Vss (-73v).

I subsequently took a tip from Chris56000 to pull up CSD to VDD via 10k resistor. No joy. CSD does rise to 4.5v but needs to exceed +5v to start oscillation. This leads me to believe that the device is protecting itself.


Points I notice are

1. VB- Vs = 7.5v - This below is Undervoltage threshold. Needs to be between 10-14v. I do not know why as I have confirmed 100K resistor from +73v rail is in spec and connects to all appropriate points.

2. CSH voltage is 1v. Again too low.

At this stage I seek further help. It’s driving me mad!!!!

[Chris56000]

Afternoon Dave!

. . .I would replace all of the ES1D gate circuit diodes and the '4148s associated with them, but it might be worth removing the '20957 and the MOSFETs again, and making sure that you've not got some hidden or latent PCB defect, possibly associated with your previous attempts – this i.c. can sense a fault on its external circuit quicker than you can see it with a meter or oscilloscope!

. . .A few meter checks with the i.c. and MOSFETs removed might show up something o/c, low or high resistance, or s/c that the i.c. has detected as a fault – I agree with you that it has gone into an internal protection mode!

. . .There are any number of faults that.can occur on a circuit like this that can't be drawn on a diagram, and it is very possible something has cropped up associated with your previous work that's consistently been missed!

My late Dad always used to say "check and re–check the work you've done!"

Chris Williams

[jboy32]

Hi Dave and Chris

I followed closely your discussion on this situation. I have exactly the same situation on an AMPEG PF 350. I have already fixed the SNPM part with Q1 an Q2 burnt. + some other diodes and resistors around.

I'm not a tech amp but that is my hobby  and have some electronic skills.

On the Power side, Q10 and Q13 were shorted, so I change them but again they go to shortcut. So I have been looking to IRS20957s and changed Q10 and Q13 it and got the same results.

For the moment Q10 an Q13 are off the board until I understand why they are burning. Without them the amp power up without the Fault Light with sound on the headphone.

I have also +/- 74Vdc on the rail where based on the schematic I should get +/- 55Vdc. I was thinking the same as without any load or because IRS20957s do not start to oscillate that is why I'm getting a high voltage ...

On Pin 2 of  IRS20957s voltage is not stable. I read some other topics related to IRS20957s but without real fix.

I noticed on my amp based on the revision that +VCC_12 coming on pin 11 of IRS20957s is not produced as what I have in the schematic. I have not U7 that seems to regulate voltage from a floating +18Vdc to 12Vdc ...

I measured this and having +14,97VCC referenced to -74Vdv.

In my case also I have a DC voltage on the output 2,7 Vdc (GND ref), I measured with and without IRS20957s. I replaced 2 times IRS20957s but still the same situation. So that is not good.

I add the voltages measured on IRS20957s if that can help to investigate without Q10 and Q13 installed.

Did you make some progress there ? Many thanks.

[Mojoampguy]

Hi JBoy32,

You appear to be in the same position as me. Still not resolved. If we finally find a resolution, we will post details here for others to benefit from.

[jboy32]

Hi Mojoampguy, for sure if we fixe it we will update a solution here 

I was thinking to investigate also on the rail +/- 74vdc vs +/- 55 on the original schematic. I fixed the SMPS by replacing what was burnt or without the right value but maybe I missed something around.

I was also looking to maybe change the diodes D32/D33 ES1D and D34 ER1D.

[jboy32]

Hi @Mojoampguy

Did you have a look to this discussion ? https://www.eevblog.com/forum/repair/irs20957s-based-amplifier-sound-loss/ ?

In my case on the revision that I have REV H01, U7 do not exist, it seems that on your picture you have it may be a good point to investigate there, to get exactly 12V?

The floating voltage that I have is +14,97Vdc so >12Vdc expected ... So I don't know if on the new revision U7 that is a 7812 regulator is an add on or if it has been removed.

[Mojoampguy]

Yes I had seen that post. 

Mine is the newer Rev H version like yours with no U7 12 v regulator. The voltages I measured are shown in an earlier post from me above. I too measured ~ 15v on that pin. We are chasing a similar fault, I suspect?

. I ran out of ideas after several months working on this. Kindly Chris said he’d take a look, so I’ve posted the amp down to him. Fingers crossed.

[jboy32]

Good afternoon Dave & Chris,

I did some progress during the week end 

I have been focusing on the CSD pin2 oscillation.
Mute should be 0V dc
Unmute should be around 5Vdc and stable taken to the GND.

I checked all the parts of the mute circuit and found that C61, 10uf was not very stable while measuring outside the board. So I replaced by a 10uf SMD capacitor and now I have a stable voltage on Pin 2 of IRS20957s = 5,537Vdc taken to the GND.

All other voltages on IRS20957s are the same, just getting now +1Vdc on LO Pin 10 taken to COM. Q10 and 13 are still off the board. I'm thinking to put new one in place but was is weird are the voltages on pin 12,13 and 14. I'm thinking that I should get 0Vdc on VS pin 13. So without fixing this zone it may burn again Q10 and Q13. Have have still 8 in spare. Maybe some other 10uf are in trouble in this area I'll check.

Julien

[Mojoampguy]

In my case CSD is always sat at 3.2v. To unMute it must be greater than 5v.

I replaced the mute circuit 10uF cap twice, no change in my case. But that was with irs20957 and Mosfets in place. I believe iCSD is essentially going into protect. Don’t know why? Possibly because rails are in fact over voltage?

Best of luck.

[jboy32]

Good morning 

I'm now more focused on the rail voltage, and looking to KA3525A and oscillation,
I understand that this IC is supposed to drive the output voltage on the rails so trying to understand how it works and checking al parts around. Maybe replacing it, very cheap.