3e TPA3255 amplifier PCBs - loud thump/splat noise on power-off (SE mode)

[JVN01]

Hi there, I'm having serious problems with my new 3e TPA3255 amplifier PCBs. With the boards modified for SE mode per the 3e datasheet I get a huge thump/splat on power down. The noise is a large DC thump which is big enough to damage any small driver (tweeter etc) connected. This is with no input connected.

Has anyone else had this problem?

• If I remove the In(-) link to GND the noise disappears, but I assume then the SE mode will not be correctly implemented.

• OPAMPs added for SE mode mod are genuine TI OPA1656 (same as input OPAMPs already on the PCB)

• I experimented using the shutdown/standby function (PS-CTRL > GND) which puts the amplifier into standby and this cures the noise on power down, however if I connect PS-CTRL > GND while amplifier is powered up, I also get a loud splat, so this isn't a good solution..

• My PCBs are Rev C-Aug-2024, Mono PBTL and Stereo BTL, both boards have same problem.

Having spent almost $600 on 4 PCBs I am very disappointed in this issue...

@3eaudio any advice gratefully received...

 

[charlielaub]

Please provide more details what you are doing when you say "power down". I assume this means you are turning off the power supply?

Are you leaving the inputs floating for these tests? If so, you should instead connect (ground) the input and ground together. This better mimics the conditions when you have upstream gear connected but with no signal (e.g. off). Also, you probably should have a load connected to the outputs anytime the amp is powered up.

 

[JVN01]

Please provide more details what you are doing when you say "power down". I assume this means you are turning off the power supply?

Are you leaving the inputs floating for these tests? If so, you should instead connect (ground) the input and ground together. This better mimics the conditions when you have upstream gear connected but with no signal (e.g. off). Also, you probably should have a load connected to the outputs anytime the amp is powered up.

Hi, just to confirm:

• Yes, "Power down" = turning off the AC power input to the SMPS power supply (Meanwell 350); this is representative of real-world operation ie how the system behaves when AC power is intentionally (or unintentionally) removed from the system.

• I do have a load connected to the outputs at all times (old speaker drivers that are tough and I am not concerned to damage).

The loud turn-off pops are present with the input floating ie nothing connected to in[+], and with input connected (RCA connection to MiniDSP Flex8 RCA outputs, regardless of whether MiniDSP is powered or unpowered).

It just seems a really poor implementation of the TPA3255, considering these PCBs are held to be 'state of the art' in terms of ready-made amplifier PCBs, it seems crazy that 3e would release a product with such a serious shortcoming.

My concern is whether 3e properly tested the PCB in SE (single-ended) mode, since the turn-off splat problem is not there when the In[-] input is not connected to GND...

 

[charlielaub]

If you made a mistake changing the board over to single ended, this could be the source of the problem. I would double check for any mistakes (op-amps installed backwards, forgot to remove or add resistors when making the conversion, etc.). I only have the rev 1.1 manual on hand, but it says:

• Add 1 op-amp for each channel (NE5532,LM4562,OPA1612,OPA1656,etc,SOIC-8 footprint).
• Remove 2 resistors (R17 and R21 in the schematic in my revision)
• add back one of the removed SMD resistors to R123 position

Did you do all of these steps correctly?

If all looks good, you just need to find a solution. If the PS_CTRL does not make a pop on power up when NOT connected to GND, but fixes the pop problem when connected to GND before power down, you just need to build a circuit with relays or whatever to make that sequence happen.

I am surprised to hear about the loud pop, and have not seen this reported before, so it is possible there is a mistake, or that the PS is interacting with the amp board on power down in a bad way, etc. You need to confirm the root cause is before blaming 3e-audio...

 

[JVN01]

If you made a mistake changing the board over to single ended, this could be the source of the problem. I would double check for any mistakes (op-amps installed backwards, forgot to remove or add resistors when making the conversion, etc.). I only have the rev 1.1 manual on hand, but it says:

• Add 1 op-amp for each channel (NE5532,LM4562,OPA1612,OPA1656,etc,SOIC-8 footprint).
• Remove 2 resistors (R17 and R21 in the schematic in my revision)
• add back one of the removed SMD resistors to R123 position

Did you do all of these steps correctly?

If all looks good, you just need to find a solution. If the PS_CTRL does not make a pop on power up when NOT connected to GND, but fixes the pop problem when connected to GND before power down, you just need to build a circuit with relays or whatever to make that sequence happen.

I am surprised to hear about the loud pop, and have not seen this reported before, so it is possible there is a mistake, or that the PS is interacting with the amp board on power down in a bad way, etc. You need to confirm the root cause is before blaming 3e-audio...

View attachment 472848

Thanks Charlie, yes I have done all the mods correctly as outlined in the manual (I have 30+ years experience in electronic industry in audio/broadcast/power electronics etc so I am confident the mods are correct).

I realise I can come up with a bodge/workaround as you suggest (which is why I already explored the behaviour with the PS-CTRL function), however it seems odd that my PCBs have this problem. My guess is that maybe not many people are using the board in SE mode.

As I'm using exactly the same opamp for my SE mod (OPA1656) as the existing input opamp on the boards I'd expect the mod to work as intended.

Hoping for some feedback from @3eaudio soon!

 

[charlielaub]

Ah, OK. Thanks for posting the bit showing that this is not your first gig modding an amp, etc! Ya never know on these DIY forums what the level of the user might be...

Anyway, you are right it is a bit odd that this is happening. Did you try contacting 3e-audio for technical support directly through their website? There is an email address listed:

Contact Us | 3e Audio

www.3e-audio.com

 

[howard416]

You changed over all 4 without testing the first one?

Also, this isn't relevant now, but you could've used an SE source on balanced config with just a semi-modified cable.

 

[DVDdoug]

Also, this isn't relevant now, but you could've used an SE source on balanced config with just a semi-modified cable.

With most amplifiers yes, but not in this case. If you look at the modification circuit, amplifier "B" wouldn't be used. OUT-N would be operating as a virtual ground. You'd get half the output voltage for 1/4 of the power.

 

[howard416]

Sorry, rather than presuming I probably should've asked for the reason why.

 

[JVN01]

You changed over all 4 without testing the first one?

Also, this isn't relevant now, but you could've used an SE source on balanced config with just a semi-modified cable.

Well yes, since I had assumed the circuit modification advocated by the designer/manufacturer would work correctly first time, so I ordered the set of PCBs in full anticipation that they would need to be converted to SE mode before I could use them with my new MiniDSP Flex8...

As per @DVDdoug comment above, driving the amplifier boards simply with a modified 'pseudo SE' input cable would fail to drive the -ve side of the BTL output pair, resulting in reduced output.

 

[JVN01]

Ah, OK. Thanks for posting the bit showing that this is not your first gig modding an amp, etc! Ya never know on these DIY forums what the level of the user might be...

Anyway, you are right it is a bit odd that this is happening. Did you try contacting 3e-audio for technical support directly through their website? There is an email address listed:

Contact Us | 3e Audio

www.3e-audio.com

I have since contacted 3e, they have suggested adding a pair of 100uf capacitors to the PCB (unpopulated by default), have not had time to try it yet as it will require full disassembly of the build. I'm guessing it's psu decoupling caps for the opamps which may hold up the opamp rails for long enough for the TPA3255 to complete shutdown before the opamps crap out... hoping to try this weekend

 

[JVN01]

Update on this problem - adding the 100uf caps does not solve the problem. Exploring the board with a multimeter (without a schematic) suggests the opamps are running single rail off a 36V supply (GND common to audio GND on the board) with a pair of virtual mid-rail 18V DC (VCC/2) nets used for opamp biasing. The extra 100uf caps appear to just sit across the VCC/2 nets, adding some extra buffer capacitance to those voltages.

The input and output pins of the input opamps (providing the SE-DIFF conversion on the board to drive the TPA3255 inputs) sit (as expected) at around +18V. In a conventional opamp inverting buffer the + input would be connected directly to VCC/2; on these PCBs it seems to be connected to VCC/2 via a 10k resistor (to bias the input), and then it goes via the original (now redundant) 10uf input coupling capacitor to the external "-" input which according to the PCB legend should also be connected to GND for SE operation.. there also appears to be a parallel RC filter (10K/100pf or similar) between the external "-" input and GND (see sketch).

Looking at the equivalent schematic then connecting the external "-" input to GND means the 10uf input coupling cap will need to charge/discharge via the 10K bias resistor. Depending on the timing of the rise/fall of the opamp DC rails then this might explain the huge thumps during power down of the overall board..

I'm considering removing the 10uf coupling cap and connecting U1B directly to VCC/2 but I'm not keen to start pulling components from the PCB..

In the meantime I've ordered a couple of optomos relays to assemble a power on/off mute circuit to run off the incoming 230v AC...

Any other suggestions welcome!