Class D PA amplifier repair, need help

[JHAnd]

Hi!

I'm a beginner but want to learn. I've mostly worked on guitar amps. And within that category mostly tube amps with large components and once a solid state amp. Now I'm faced with a Class D PA amp, small SMD parts... no tubes in sight... and I find myself a little lost, so in need of some guidence.

The PA system is the type that has a pre amp + power amp located in the subwoofer and you can connect two satellite speakers to the subwoofer through speakon cables to get a complete stereo system. Link to product: https://www.dbtechnologies.com/en/products/es/es-503/

Problem is there is seems to be no output from the power amp, so the subwoofer and satellite speakers are not outputting any sound. (The Line output seems to work fine and is outputting sound). So I suspect something is wrong with the power amp section.

I have access to a second system that is fully functioning... So my plan is to power both systems up simultaneously and start to compare voltages at various points to see how the faulty one differs.

However I'm a bit worried though that the faulty one might become even more broken if it is powered on for long while testing, since I don't know what's wrong with it to begin with. Also I've read that class D amps might need to have a load attached to prevent damage. So I plan to have the speakers attached when measuring.

I have access to a variac, multimeter, sig gen and oscilloscope. I have asked the manufacturer for a schematic but they will not supply me with one.


So I'd like to ask a few questions to get started...

1. How do I power the faulty board up in a safe way?

2. Looking at the pictures of the board below, would you say it is safe to power the board up without any load attached?

3. Where should I begin looking for faults?

4. What are the most likely failure modes?


Advice and thoughts from more experienced folks is greatly appreciated!


Thanks!

[Vicus]

What IC are U11 and U6?

[BreakingOhmsLaw]

Allright, here we go.
Let's take a look at that PCB. There's a primary side, which has line voltage on it. Then there's a switched power supply (SMPS), that creates the DC for the Class D amp.

  Commandment
  Rule number one of all repairs:
THOU SHALT CHECK VOLTAGES! (Imagine this spoken by a loud voice from the sky for dramatic effect).

You have no signal, so likely there's no power at all. Could be the SMPS. Let's check.
Two difficulties:
1: You don't have an isolating transformer (and no, your variac likely isn't one), this is dangerous.
2: The heat sink is not attached, so U8, MOS1 & MOS2 are not cooled during measurement.
You want to measure the voltage over the 6 electrolytic capacitors to check if the SMPS is working. You can switch the unit on briefly to avoid overheating the now uncooled parts. Make sure not to touch any spicy parts on the primary side. Best to cover it with copious amounts of an insulating material  (Kapton tape, iso tape, whatever you have at hand).

If there is no voltage across those caps, we need to work on the primary side. I didn't see any fuse at first glance. Likely broken parts are the bridge rectifier BR1 on the left, the controller U2. Not sure what U3 is. A MOSFET gate driver perhaps? Please read off the part number of U2 and U2. And lastly, the MOSFETS could be dead. In that case, that's usually caused as a consequence of another defect.

If the 6 caps have a voltage, then we need to continue on the secondary side. In that case, please clean U7, D1 and D2 and identify the parts. They might be linear voltage regulators that have failed.
Post results. Good luck and be careful with the primary side. If your house does not have an RCD (Jordfelsbrytare in Swedish), please consider leaving this to a repair shop with the proper equipment.

[Whales]

Half of that boards is a mains power supply.  It will be deadly to touch, I recommend wrapping it in lots of tape during your work so you can't accidentally have contact.

Running it without a load is probably OK but I'd chuck some 100 ohm resistors or so on the output anyway.  Resistors are better than speakers for some styles of debugging, the signals will look a lot less noisy.  You can use 8 or 4 ohm resistors but they'll either have to be high power or you'll need to make sure to keep the volume down.

Start with checking the power supply voltages.

It looks like the amplifier uses integrated ICs (metal chips on bottom of PCB).  If both failed at the same time then it's likely the fault is elsewhere.

[JHAnd]

Thanks a lot for your replies, I really appreciate the help!

Allright, here we go.
Let's take a look at that PCB. There's a primary side, which has line voltage on it. Then there's a switched power supply (SMPS), that creates the DC for the Class D amp.

  Commandment
  Rule number one of all repairs:
THOU SHALT CHECK VOLTAGES! (Imagine this spoken by a loud voice from the sky for dramatic effect).

You have no signal, so likely there's no power at all. Could be the SMPS. Let's check.
Two difficulties:
1: You don't have an isolating transformer (and no, your variac likely isn't one), this is dangerous.
2: The heat sink is not attached, so U8, MOS1 & MOS2 are not cooled during measurement.
You want to measure the voltage over the 6 electrolytic capacitors to check if the SMPS is working. You can switch the unit on briefly to avoid overheating the now uncooled parts. Make sure not to touch any spicy parts on the primary side. Best to cover it with copious amounts of an insulating material  (Kapton tape, iso tape, whatever you have at hand).

If there is no voltage across those caps, we need to work on the primary side. I didn't see any fuse at first glance. Likely broken parts are the bridge rectifier BR1 on the left, the controller U2. Not sure what U3 is. A MOSFET gate driver perhaps? Please read off the part number of U2 and U2. And lastly, the MOSFETS could be dead. In that case, that's usually caused as a consequence of another defect.

If the 6 caps have a voltage, then we need to continue on the secondary side. In that case, please clean U7, D1 and D2 and identify the parts. They might be linear voltage regulators that have failed.
Post results. Good luck and be careful with the primary side. If your house does not have an RCD (Jordfelsbrytare in Swedish), please consider leaving this to a repair shop with the proper equipment.

There is power, as the preamp output is fully functional. Also the display turns on, so I am able to adjust settings, EQ etc in the digital menu system... If the power supply was bad, wouldn't the unit be totally dead?

The preamp output and display are part of another board that attaches to the main board by way of a 26-pin connection (the black one), and are powered through this 26-pin connection.

Here are some measurements of the main board:

Bridge rectifier input: 230 VAC
Bridge rectifier output: 330 VDC

Two big caps on the primary side: 163VDC and 167VDC
(what are these for? Smoothing caps?)

6 electrolytic caps on the secondary side: 35VDC on each cap
(these are smoothing caps to get rid of ripple on the DC supply right?)


Here's what I read on the components:


U2: AVS10CB by ST electronics
This seems to be an Automatic mains selector 110/230V?
https://www.allelcoelec.com/datasheets.c3/AVS10CB.pdf

U3: S27952 530P C092S_ by manufacturer "International Rectifier"
https://pdf1.alldatasheet.com/datasheet-pdf/view/530872/IRF/IRS27952.html

U7: LM337SP GK0G2 V6 CHN S28 by ST electronics
https://www.allelcoelec.com/datasheets.86/LM337SP.pdf

D1 and D2: STTH1002CB GKOSS534 by ST electronics
https://www.digikey.lu/en/products/detail/stmicroelectronics/STTH1002CB/669389

What IC are U11 and U6?

They are TDA8954TH by NXP semiconductor.
https://octopart.com/datasheet/tda8954th%2Fn1-nxp+semiconductors-21124594

So these seem to be the actual Class D power amps ICs. So perhaps one chip is run in mono mode for the Sub speaker and the other chip is run in stereo mode for the left and right (or mono if in mono mode setting).

[Vicus]

The ES503 is still in production today and the TDA8954 is discontinued from August 2020 so 

Have you an oscilloscope? In a older design of dB they did use an indipendent clock source to sync the TDAs so if it is missing no sound. Also it happened that if one TDA is faulty it stop also his companion due to the fact the protection indication pin is tied togheder.

[JHAnd]

The ES503 is still in production today and the TDA8954 is discontinued from August 2020 so 

Have you an oscilloscope? In a older design of dB they did use an indipendent clock source to sync the TDAs so if it is missing no sound. Also it happened that if one TDA is faulty it stop also his companion due to the fact the protection indication pin is tied togheder.

Interesting,

I do have an oscilloscope, (analog HAMEG 20MHz with digital storage) however I don't have access to it until tomorrow. Perhaps I can look at the input signal on pins 4/5 (channel 2) and 8/9 and (channel 1) and if there is a PWM output at pins 16 (channel 1) and 21 (channel 2).

If you can guide me on what to test, that would be great!

I'll attach the pinning information below:

[CaptDon]

Class D amplifiers are not harmed by open circuit / no load. For the most part the only amplifiers that ever suffered damage from no load were the tube and solid state amps using output transformers. With no load on the output the primary or secondary side could have spike voltages big enough to damage the transformer through arcing or insulation damage. Also excessive voltage spikes on the primary side of the transformer could kill solid state output transistors. Another rare failure with no load on a tube amp with tetrodes or pentodes as the power tubes is the plate voltage swinging excessively low causing the screen grids to draw excess current and become damaged leading to total tube destruction. This isn't so much a problem when the output tubes are wired in ultra-linear fashion since the screen voltage drops along with the plate voltage.

[Vicus]

Class D amplifiers are not harmed by open circuit / no load. For the most part the only amplifiers that ever suffered damage from no load were the tube and solid state amps using output transformers. With no load on the output the primary or secondary side could have spike voltages big enough to damage the transformer through arcing or insulation damage. Also excessive voltage spikes on the primary side of the transformer could kill solid state output transistors. Another rare failure with no load on a tube amp with tetrodes or pentodes as the power tubes is the plate voltage swinging excessively low causing the screen grids to draw excess current and become damaged leading to total tube destruction. This isn't so much a problem when the output tubes are wired in ultra-linear fashion since the screen voltage drops along with the plate voltage.

I agree but I some type of class D amp doesn't oscillate without a load.

[Vicus]

The ES503 is still in production today and the TDA8954 is discontinued from August 2020 so 

Have you an oscilloscope? In a older design of dB they did use an indipendent clock source to sync the TDAs so if it is missing no sound. Also it happened that if one TDA is faulty it stop also his companion due to the fact the protection indication pin is tied togheder.

Interesting,

I do have an oscilloscope, (analog HAMEG 20MHz with digital storage) however I don't have access to it until tomorrow. Perhaps I can look at the input signal on pins 4/5 (channel 2) and 8/9 and (channel 1) and if there is a PWM output at pins 16 (channel 1) and 21 (channel 2).

If you can guide me on what to test, that would be great!

I'll attach the pinning information below:

You could check if there is oscillation on the output pin. A straight line, sometimes with a low voltage dc offset without a load, means that the output stage is off.

[BreakingOhmsLaw]

Voltage on the caos seem about right. Check the voltage from the LM337 (see datasheet for pinout).

[JHAnd]

Thanks,

I will check the TDA8954 and the LM337 tomorrow...

one question;

The components that get hot and require a heatsink... you mentioned U8, MOS 1 and MOS 2... and there are others components too that have CPU paste on them and should be against the heatsink. If I set the output volume low will these components be cooler or are they always in need of cooling? Is there anything I can do to not fry stuff while I'm testing? Perhaps the only thing is to have short testing periods and then shut the unit off like @BreakingOhmsLaw mentioned.

As I mentioned in my first post I have two of these PA-systems at my disposal; a faulty one and a fully functional one.

I have been toying with the idea of comparing voltages between the two units. However I'm not so keen on frying any of the units and especially not the healthy one! One good unit is better then two faulty units!

Thanks again for helping me!

[DavidAlfa]

The won't get hot unless you put a heavy load on them, no problem.

[TroelsM]

Without knowing that specific amp-module, I would strongly advice to watch the temperature very closely if you apply power without heatsink. Yes, the switcher should not get warm with no(little) load, but any small mistake can cause power dissipation and with no heat sink you have very, very short time before the part is damaged.

Kind regards TroelsM

[JHAnd]

Voltage on the caos seem about right. Check the voltage from the LM337 (see datasheet for pinout).

Output: -12.4 VDC
Input: -35.9 VDC
Adjust: -11 VDC

You could check if there is oscillation on the output pin. A straight line, sometimes with a low voltage dc offset without a load, means that the output stage is off.

I fed a 1000hz sine into the XLR input of the device and tried probing the TDA8954TH. I did not know what voltage and frequency ranges to expect but here's what I got:

Scope ground was tied to earth ground on the device.

Input pins (4,5,8,9)
Voltage: 10mV/Div
Time: 2 μs/Div

Scope reading:



PWM output pins (16, 21):
Voltage: 10mV/Div
Time: 0.1s/Div

Scope reading:





Two more zoom levels of the PWM output:

[DavidAlfa]

1KHz input, then why are you probing the input at 2uS/div? That's probably noise.

For 1KHz, set to 500us/div or so.

[JHAnd]

1KHz input, then why are you probing the input at 2uS/div? That's probably noise.

For 1KHz, set to 500us/div or so.

This was the only oscillation I could find on the input so if it is noise then there is no signal reaching the power amp inputs.

[DavidAlfa]

10mV is probably just noise, inject a 1Vpp 1KHz signal and trace it from the input!
It could be a bad power rail, check VDD voltages of all ICs.
Signal ground must be taken from the secondary side.

[Vicus]

Consider that 2 channel have an high pass e 2, in bridge have a low pass. Probably at 150Hz ish.

[JHAnd]

10mV is probably just noise, inject a 1Vpp 1KHz signal and trace it from the input!
It could be a bad power rail, check VDD voltages of all ICs.
Signal ground must be taken from the secondary side.

Tried tracing from the input today but found it quite difficult... I can follow the signal from the input XLR pins past through some resistors and caps and I can see the trace in the board and then confirm with the probe that the signal is still there... but then the trace disappears and I can't see where the connection goes... perhaps it goes to the other side of the board? Or can there be traces that are not visible in a middle layer?

I also tried probing the 26 pin connector that connects the preamp board to the main board. I did this for both the faulty unit and the working unit and tried to compare them... but I was quite nervous about firing up the working one without a heat sink. Since this unit is working I could also hear the sine in the speaker, but all of a sudden the signal disappeared... perhaps some overheating protection got activated?

Anyway I saw that on the working unit there was a 1Khz sine on one one of the pins on the 26-pin connector, but the voltage was only 40mVpp compared to the 1Vpp that I was feeding at the input. Seemed strange that it was lower in gain going from the preamp to the power amp board then from the input to the preamp.

Anyway on the faulty unit I got some kind of distorted signal at the same pin, not a clear sine like on the working unit.

I've also started looking for the data sheets for the IC's on the preamp board:

U1: Vishay/Siliconix DG412DY
U14: ST electronics 33078
U3: ST electronics
U7: ST electronics
U10: LM2594 PCCD

There is another side of the preamp board too but I will have to unscrew it to get a peak at it and I do not have access to the board again until Monday.

[Jeff eelcr]

Nakamichi T-100 and Beginner do not go together, nice piece of equipment to learn audio with, I still want one.
Some output IC's will continue to heat up until they are destroyed always check for overheating without a heatsink
you can use some type of heatsink that will not be in your way.
Jeff

[JHAnd]

Nakamichi T-100 and Beginner do not go together, nice piece of equipment to learn audio with, I still want one.
Some output IC's will continue to heat up until they are destroyed always check for overheating without a heatsink
you can use some type of heatsink that will not be in your way.
Jeff

thanks for the input Jeff,

about the T-100... I'm just using it as an sig gen because it is what is available at work.


There are eight chips that should touch the heatsink on the back of the board. I could probably use 3-4 heatsinks to get them all covered.

Any ideas on what material I could use as a heatsink, and how big they would have to be? And how I can monitor heat on all these chips at the same time while making measurements?

[Jeff eelcr]

If it burns your finger after a few seconds/minuite, I use aluminium bar stock, drill holes where needed and clamp or screw.
Its reuseable.
Jeff

[Vicus]

TDA8954 can't stay more than a few secondo without heatsink when they work. Pay attention to isolate what need to be isolated and pay attention to what you touch on the primary side. My 2 cents is to measure with the board mounted in the factory heatsink. By simply measuring residual oscillation at around 400kHz on the output you can check if there is a problem in the output stage. A residual oscillation scould be visible especially without a load.

Did you try to swap the preamp board from the working unit to the faulty and vice versa?

[JHAnd]

TDA8954 can't stay more than a few secondo without heatsink when they work. Pay attention to isolate what need to be isolated and pay attention to what you touch on the primary side. My 2 cents is to measure with the board mounted in the factory heatsink. By simply measuring residual oscillation at around 400kHz on the output you can check if there is a problem in the output stage. A residual oscillation scould be visible especially without a load.

Did you try to swap the preamp board from the working unit to the faulty and vice versa?

I thought about inserting the other preamp board, but not sure if it is safe because they are different revisions and have slightly different layouts...plus if there was something seriously wrong with the main board perhaps I'd fry the working preamp board? Or am I being too careful?

Compared the output of the faulty one and the working one looking for a 400khz oscillation... difference is the working one is indeed outputting a 400khz oscillation but the faulty one not.

[Vicus]

So the problem is in the output section.

Just curiosity, what's difference between boards?