Breeze TPA3255 XLR Balanced Amplifier TearDown : DIP8 Op Amps.

[v1adpetrov2]

I received my copy of TPA3255XLR by mail!
The autopsy revealed that I also have an updated v1.5, just like ICIETDYIEUR.



Output ripple SMPS without connected amplifier board: near 340mV, which is bad enough.

Ripple on the connected amplifier board, which is already catastrophically bad.

We will figure out why the amplifier board adds such huge ripples to, to put it mildly, a low-quality power supply.
1. I removed the cooling radiator from the TPA3255 chip: half of the legs are covered with a huge amount of thermal paste.
This caused the excitation of the chip and the generation of high-frequency interference.
And cleaned the thermal paste and washed the chip with isopropyl alcohol.
At the same time I made two modifications:
- replaced the 30 kOhm resistor with 10 kOhm, which raised the frequency of the modulator from 450 kHz to 600 kHz.
- shunted a 220uF/63V electrolytic capacitor to power the chip with a 1nF/100V film capacitor to remove some of the high-frequency ripples.

The manufacturer again saved and install 1uF capacitors instead of WIMA 2.2uF film capacitors.
Added four 1uF capacitors to increase the total capacity to 2uF:

I carefully applied a new thermal paste to the TPA3255 chip and returned the cooling radiator to its place.
And started the main modification - the addition of a resistive-capacitive load to filter ripples and provide peak power surges to heavy speakers.
Many years ago, I developed a filter-booster "snowflake" for small-sized amplifiers with an external power supply and successfully used it in various projects.

According to the same principle, I added four 2200uF/63V electrolytic capacitors and one 0.015uF polypropylene capacitor to lower the ESR.
And I added two 0.1 Ohm/20W resistors to the plus and minus terminals between the power supply and the amplifier board.

Without resistors, the power supply unit, when turned on, tries to charge a capacitor bank with a capacity of almost 10,000 Muf, does not withstand peak values of current consumption and protection is triggered.
Plus, some of the high-frequency oscillations from SMPS are quenched on these resistors.

The variable resistor in the power supply easily allows you to change the output voltage from 32 V to 49 V.
Some users have noticed that when the power supply voltage levels are close to the maximum, TPA3251 and TPA3255 sound better than when lowered.
TPA3251 from 32 volts and above,
TPA3255 from 40 volts and above.
Although this is of course subjective.
For myself, I set the voltage to 42V, the benefit of the converter chip up to 12V LM2575HVS-12 works with an input voltage up to 60V.


Ripple on the connected amplifier board, after all modifications.

Vpp=180mV.
Much better already!
Linear power supplies with a toroidal transformer usually have output ripples of 100 - 120 mV.

A better power supply can be achieved if you replace the standard SMPS with another power supply with the PFC function.
For example, like this!

 

[ICIETDIYEUR]

I received my copy of TPA3255XLR by mail!
The autopsy revealed that I also have an updated v1.5, just like ICIETDYIEUR.
View attachment 237203 View attachment 237205 View attachment 237209
View attachment 237210 View attachment 237214 View attachment 237216

Output ripple SMPS without connected amplifier board: near 340mV, which is bad enough.
View attachment 237224

Ripple on the connected amplifier board, which is already catastrophically bad.
View attachment 237223

We will figure out why the amplifier board adds such huge ripples to, to put it mildly, a low-quality power supply.
1. I removed the cooling radiator from the TPA3255 chip: half of the legs are covered with a huge amount of thermal paste.
And cleaned the thermal paste and washed the chip with isopropyl alcohol.
At the same time I made two modifications:
- replaced the 30 kOhm resistor with 10 kOhm, which raised the frequency of the modulator from 450 kHz to 600 kHz.
- shunted a 220uF/63V electrolytic capacitor to power the chip with a 1nF/100V film capacitor to remove some of the high-frequency ripples.
View attachment 237226 View attachment 237227
The manufacturer again saved and install 1uF capacitors instead of WIMA 2.2uF film capacitors.
Added four 1uF capacitors to increase the total capacity to 2uF:
View attachment 237238 View attachment 237228
I carefully applied a new thermal paste to the TPA3255 chip and returned the cooling radiator to its place.
And he started the main modification - the addition of a resistive-capacitive load to filter ripples and provide peak power surges to heavy speakers.
Many years ago, I developed a filter-booster "snowflake" for small-sized amplifiers with an external power supply and successfully used it in various projects.
View attachment 237229 View attachment 237230
According to the same principle, I added four 2200uF/63V electrolytic capacitors and one 0.015uF polypropylene capacitor to lower the ESR.
And I added two 0.1 Ohm/20W resistors to the plus and minus terminals between the power supply and the amplifier board.
View attachment 237235
Without resistors, the power supply unit, when turned on, tries to charge a capacitor bank with a capacity of almost 10,000 Muf, does not withstand peak values of current consumption and protection is triggered.
Plus, some of the high-frequency oscillations from SMPS are quenched on these resistors.

The power supply unit easily allows you to change the output voltage from 32 V to 49 V.
Some users have noticed that when the power supply voltage levels are close to the maximum, TPA3251 and TPA3255 sound better than when lowered.
TPA3251 from 32 volts and above,
TPA3255 from 40 volts and above.
Although this is of course subjective.
For myself, I set the voltage to 42V, the benefit of the converter chip up to 12V LM2575HVS-12 works with an input voltage up to 60V.
View attachment 237213

Ripple on the connected amplifier board, after all modifications.
View attachment 237225
Vpp=180mV.
Much better already!
Linear power supplies with a toroidal transformer usually have output ripples of 100 - 120 mV.

A better power supply can be achieved if you replace the standard SMPS with another power supply with the PFC function.
For example, like this!

Hello

Great job v1adpetrov2 !

Regarding the capacitor of 220μF/63V, wouldn't it be better to put a 470F as on the datasheet ?
Are the 1nF/100V to be paralleled enough ?

Regarding the 1μF WIMA, you added 1μF in parallel to arrive at 2μF: could we not rather replace them with 10μF in electrochemical as ELNA SILMIC II ?

I have not been able to determine if Mosfet or bipolar OPA would be best suited for this amplifier because I have not been able to make the diagram
Your advice ?

Otherwise, I understood the principle for your "snowflake" filter booster but I did not understand how to wire it exactly

Your link for a new SMPS power supply with the PFC function does not work (I live in France...), do you have a reference at 'ALI' ?

Thank you in advance for your answers.

See you soon

 

[v1adpetrov2]

Hello

Great job v1adpetrov2 !

Regarding the capacitor of 220μF/63V, wouldn't it be better to put a 470F as on the datasheet ?
Are the 1nF/100V to be paralleled enough ?

Regarding the 1μF WIMA, you added 1μF in parallel to arrive at 2μF: could we not rather replace them with 10μF in electrochemical as ELNA SILMIC II ?

I have not been able to determine if Mosfet or bipolar OPA would be best suited for this amplifier because I have not been able to make the diagram
Your advice ?

Otherwise, I understood the principle for your "snowflake" filter booster but I did not understand how to wire it exactly

Your link for a new SMPS power supply with the PFC function does not work (I live in France...), do you have a reference at 'ALI' ?

Thank you in advance for your answers.

See you soon

1. Links for a new SMPS power supply with the PFC function:
from Global site Ali
from USA site Ali

2. You can replace them with 10 uF in the electrochemical of ELNA SILMIC II, but for better sound quality, it is still recommended to use film capacitors.
Even with an input buffer resistance on 10 kOhm operational amplifiers, 2 uF is enough to pass frequencies from 7.957 Hz and higher.

3. We connect the "snowflake" to any place convenient for you, preferably as close as possible to the TPA3255 chip.


4. The 220μF/63V capacitor does not need to be touched. After all , we connected as many as 8800μF in parallel to it + 1880μF was installed by the manufacturer (4 * 470μF/50V) .
A 1nF/100V capacitor is needed only to filter high-frequency ripples.

5. The "native" operational amplifiers of the NJM5532DD sound with juicy deep low frequencies that lack attack. And against their background, the high frequencies are very muted.
There are fans of this sound and they do not need to change operational amplifiers.
I prefer the "monitor" sound with an aggressive attack.
So I replaced NJM5532DD with OPA1656.

The low frequencies have become less deep, but the bass drum attack has become just fantastic!
The high frequencies have become bright, but not tedious.
In general, I am satisfied with the result!

Yes, I completely forgot to point out that the amplifier is rocking two large vintage Optonica CP5000 speakers with 12" woofers.

 

[v1adpetrov2]

Another modification to increase the life of the amplifier is properly organized passive cooling!
Although we are dealing with a Class D amplifier, but the radiator temperature when listening at medium volume is 45-48 degrees Celsius.
And if you decide to have a small party at home, the radiator will heat up to 60-65 degrees.
Therefore, the ventilation holes above the radiator and the power supply will never be superfluous!


Now, the new amplifier has become a worthy part of my home music system!

 

[SMen]

What you have posted is really interesting v1adpetrov2 ...
I was going to post a question about the power supply itself, and you have answered it ... but your post with your measurements has gone (the one that
ICIETDYIEUR replied to this morning, your one starting "I received my copy of TPA3255XLR by mail!") I received mine very recently and it is the same version

I was just just wondering if you could repost for the links. Not sure why I can't see it, for the ripple etc. OK ... your post has just re-appeared... no re-post necessary.

The only change I have made is to use the OPA1656 following daniboun's recommendation (thank you daniboun ).
I have been astonished really just by this one change!

It has been a 'slow burner' this thread since daniboun first posted about it the BRZ3255's potential ...

I had also been wondering about something like Allo's 'capacitance multiplier' going in after the power supply, but it appears this is only good to max 32v and 5A. Maybe that is similar to your 'snowflake' that I am unable to link to ... but that is a above my diy "paygrade" I have to confess... a higher rated allo CM would be ok though.


When you say "
A better power supply can be achieved if you replace the standard SMPS with another power supply with the PFC function"

Will the power supply you link to achieve the same as some / many of your modifications?

It looks as though you achieved your Vpp=180mV with the existing power supply. Sorry if I missunderstood.

​

Another modification to increase the life of the amplifier is properly organized passive cooling!
Although we are dealing with a Class D amplifier, but the radiator temperature when listening at medium volume is 45-48 degrees Celsius.
And if you decide to have a small party at home, the radiator will heat up to 60-65 degrees.
Therefore, the ventilation holes above the radiator and the power supply will never be superfluous!
View attachment 237374 View attachment 237375

Now, the new amplifier has become a worthy part of my home music system!
View attachment 237373

 

[v1adpetrov2]

When you say "
A better power supply can be achieved if you replace the standard SMPS with another power supply with the PFC function"

Will the power supply you link to achieve the same as some / many of your modifications?

It looks as though you achieved your Vpp=180mV with the existing power supply. Sorry if I missunderstood.

Yes, replacing the power supply with the PFC function (Power Factor Correction) really improves the noise characteristics of the amplifier.
After all, improved technical characteristics are the reason for using such SMPS in audio equipment.

Output ripple: less than 150mV
Total output power: 400W/4 Ohm; with PFC function, PF value greater than 0.98;
Output voltage adjustment range: 10%
Incoming line stability: plusmn; 0.5%
Load stability: plusmn; 1%

If you have small shelf speakers connected to the amplifier, then a simple replacement with SMPS with PFC will be enough.
But this power supply will not be able to provide peak power at high volume levels for a 12" woofer without increasing the capacitance of capacitors.
I hope I was able to answer your question...

 

[ICIETDIYEUR]

Good evening everyone

(it's night, I'm French: no one is perfect... LOL)

Thanks to SMen for joining the conversation that is becoming more and more interesting.

First of all, I wanted to thank daniboun again who made us known that amplifier on this forum

I live deepening on different subjects but not necessarily in the order of things presented: I recall that I am French and it is not always easy for me to express here correctly his ideas, you will excuse me I beg.

1> Great idea to have added vents for the amplifier, I had also thought about it just when I saw it in the photo.
Have you also created vents from below but with a smaller number of holes (less air passage surface) and more dispersed on the hood, which should create a VENTURI effect and thus make it even more effective ?

2> For the capacitor connected to pin 2 (VDD) of the TPA3255, I will replace the 220F and parallel a 100nF as indicated in the datasheet ->

3> Regarding the power supply, it is obvious that the original mounted one just does the job, we can of course increase its voltage. But the one you have given the link does not pass into the amp (too high = 45mm) without leaving the hood open or making significant enough modifications to try to fit it in. I found another one better and one that should fit well HERE

4> For the resistance of 30K to replace by a 10K (at 1%) connected to the pin 8 (FREQ ADJ), I do not think a good idea...
It does not only seem to condition or repel the 'AM BAND FREQUENCY' or its original value (30K) is called 'NOMINAL, MASTER MODE':
I hovered over the datasheet and it seems that its value acts on the 'OC Threshold' which basic with 30K corresponds to 12.9A, for 20K to 17A and therefore much more with 10K !
This also acts on the 'OVERLOAD PROTECTION COUNTER' (OLPC) which goes from the original 2.3ms to 1.7ms with a resistance of 10K.
Its values interact with each other as can be seen in the tables of the datasheet.
I believe that with this modification, the acceptable current value becomes too important and therefore goes a faster and especially too important heating of the TPA3255 on with the size of the case of this amp. But I may have been wrong and misinterpreted what I read.

5> When it comes to WIMA, the value you get even with your MOD is too small. Rather than explaining it badly, here are some pictures ->

6> Finally, speaking of OPAMPS: I also made the same choice as you (OPA 1656).
Indeed, if we base ourselves on the diagram of principle, the OPAMPS will see at their input an impedance of almost 10K.
The choice of OPAMP FET is much better suited for this circuit than bipolar people who prefer a low impedance (less than or equal to 1K generally).
Here is a small table that I am made for the choice of OPAMPS which will also serve me for my DAC ->

Regards.

 

[SMen]

Yes, replacing the power supply with the PFC function (Power Factor Correction) really improves the noise characteristics of the amplifier.
After all, improved technical characteristics are the reason for using such SMPS in audio equipment.

If you have small shelf speakers connected to the amplifier, then a simple replacement with SMPS with PFC will be enough.
But this power supply will not be able to provide peak power at high volume levels for a 12" woofer without increasing the capacitance of capacitors.
I hope I was able to answer your question...

Yes thank you, I think so, just so long as the specs for that power supply are as described. I have been looking up PFC function to understand more!

Your power supply information and ripple graph appeared just as I was going to ask the question!

The BRZHiFi stock power supply looks to be the same as the one used in the Aiyima A08.

Regarding speakers, surely it may be that larger woofers might be easier to drive than small, inefficient speakers? My standmounts are 90 dB.
I am using speakers from an older Triangle range where the largest floorstanders are 92dB efficient. (Titus 202, Aliote 202, Celius 202)

Thank you for your help and quick reply, and excellent analysis - you are taking this thread forward in an interesting way

 

[SMen]

Good evening everyone

(it's night, I'm French: no one is perfect... LOL)

Thanks to SMen for joining the conversation that is becoming more and more interesting.

First of all, I wanted to thank daniboun again who made us known that amplifier on this forum

I live deepening on different subjects but not necessarily in the order of things presented: I recall that I am French and it is not always easy for me to express here correctly his ideas, you will

Thank you ICIETDYIEUR

You have answered another one of my questions before I ask - input impedance - I was reading earlier about OPA1656 vs OPA1612, and bipolar vs FET. It is great to have some measurements behind the OP amp selection. Great table.

Psychic contributors here now . Great point about the power supply. Bravo ces français

 

[v1adpetrov2]

2> For the capacitor connected to pin 2 (VDD) of the TPA3255, I will replace the 220F and parallel a 100nF as indicated in the datasheet ->

View attachment 237423

You have confused the purpose of the capacitor.
The one that I shunted with a 1nF / 100 V film is located on the right on a typical scheme (instead of two pieces of 470uF) and the main PVDD power is supplied to it from 18V to 51V.

For 12V powering the modulator chip and OpAmp-s, a 220uF/16V capacitor is installed on the board near the step-down converter.
And another 10uF SMD capacitor near the chip.

 

[ICIETDIYEUR]

OK

It took you a while to answer... LOL LOL LOL I tease you

PS: Would you have managed to make the diagram of this amp ?
It would really be very useful to better understand the questions around this amp and you look well equipped especially to have excellent knowledge

Otherwise, for the newest power supply I confirm its compatibility:

ORIGIN = CDXS438-36 -> W=81,5mm / D=127mm/H=38mm
MEANWELL = EPP-400-36 -> W=76,2mm / D=127mm/H=35mm

But 100 euros...

But largely above quality level

 

[v1adpetrov2]

to ICIETDYIEUR:

 

[ICIETDIYEUR]

So they are all in parallel with different values and materials for a better response in frequencies and also in relation to time if I understand ?
Can you tell if the original 220μF value is better suited than with a 470F?

 

[v1adpetrov2]

So they are all in parallel with different values and materials for a better response in frequencies and also in relation to time if I understand ?
Can you tell if the original 220μF value is better suited than with a 470F?

In a particular case, I do not see the need to replace this capacitor. The circuit designer specifically installed a Low-ESR electrolytic capacitor as close as possible to the TPA3255 chip to ensure peak power consumption spikes.
But if you really want to replace it, do it…

 

[SMen]

OK

It took you a while to answer... LOL LOL LOL I tease you



PS: Would you have managed to make the diagram of this amp ?
It would really be very useful to better understand the questions around this amp and you look well equipped especially to have excellent knowledge

Otherwise, for the newest power supply I confirm its compatibility:

ORIGIN = CDXS438-36 -> W=81,5mm / D=127mm/H=38mm
MEANWELL = EPP-400-36 -> W=76,2mm / D=127mm/H=35mm

But 100 euros...

But largely above quality level

Jut observing that the power supply that came with daniboun's original model is different to the latest version ours. Wondering how it measured in comparison.
Aiyima have discontinued the A08 which I believe also used the same power supply as the previous model.
The price of the upgrade is making me somewhat philosophical ... whatever happened to that super low noise powered Allo 3255 ...
Dear Allo...

 

[ICIETDIYEUR]

@ v1adpetrov2

Hello and thank you for your reply

I made a principle implementation for the MOD power supply of the amplifier.
The 8 capacitors would be in a plastic box with the dimensions of these to allow their replacement later if necessary.
What do you think?

I wanted to come back to the replacement of the 1% resistance connected to the pin 8 of the TPA3255.

In your opinion: what is the exact purpose of this modification?

Regards.

 

[v1adpetrov2]

I wanted to come back to the replacement of the 1% resistance connected to the pin 8 of the TPA3255.

In your opinion: what is the exact purpose of this modification?

About increasing the frequency of the modulator to 600kHz:
It is important to understand that the higher the pulse frequency of the master generator, the more detailed the sound of the amplifier will be. For example, if this frequency is 300 kHz (which is about 15 times higher than the highest audio frequency), then a Class D amplifier can already be attributed to Hi-Fi. The dynamic range of its sound and the signal-to-noise ratio also depend on the frequency of the master oscillator - the higher it is, the better. However, the disadvantage of increasing this frequency is that with its growth, the Class D amplifier will become less efficient, i.e. more energy will go into heat.

Another important note is to increase the power supply voltage of the amplifier board from 36 V to 42-44V.
The step-down converter to 12V LM2575HVS can indeed work with an input voltage up to 60V, but the Schottky diode 1N5819 is designed only for 40V.

Some users increased the supply voltage to 46-48 V, not taking into account this feature.
The result was a breakdown of the diode and a complete burnout of all the chips on the amplifier board.
Therefore, there is a table with a list of recommended diodes at different voltages in the datasheet for the LM2575HV.
!!! Replace 1N5819 with the one needed for increased voltage!!!

You have found a good place for an additional capacitor bank!
But the thickness of the wires to the power connection point should be 0.6 - 0.8 mm.sq.
And you need to install another 0.015uF polypropylene capacitor in this place.

I have added these changes to your schema:

 

[ICIETDIYEUR]

SO.
What you say about this 600KHz frequency is what I thought.
I would not make this change because I do not want more warmth.

Ditto for the increase to 42volts and the Schottky diode: it is imperative that you specify it in your post a little higher to avoid readers to make a big mistake !

Thank you for your advice and your opinion regarding my additional capacitor bank.
You will see that I have however integrated the polypro capacitor as shown in my photo ->

Regards

 

[SMen]

SO.
What you say about this 600KHz frequency is what I thought.
I would not make this change because I do not want more warmth.

Ditto for the increase to 42volts and the Schottky diode: it is imperative that you specify it in your post a little higher to avoid readers to make a big mistake !

Thank you for your advice and your opinion regarding my additional capacitor bank.
You will see that I have however integrated the polypro capacitor as shown in my photo ->

View attachment 237699
Regards

I think this picture still shows the 42v modification ... I was thinking also about 'the small print' modifications being lost in the details and maybe being a number of steps too far ... certainly for me! I am wondering if this explains why mine runs cool at the standard 36v!

I think I am needing a reset ;-) ... though following with interest.

 

[v1adpetrov2]

...I am wondering if this explains why mine runs cool at the standard 36v!

According to the datasheet, the TPA3255 will also work at 18 V supply voltage.
Another thing is that at 18V and at 50V the output ripples of power supply will be almost the same. So at 18V, the signal-to-noise ratio at the output will be worse than at 50V.
That's all the math...