Breeze TPA3255 XLR Balanced Amplifier TearDown : DIP8 Op Amps.

[dfuller]

Why not reply to two posts with two replies?

Because doing so clogs the thread with a bunch of replies from the same person back to back. It's considered good manners to keep it to one reply at a time.

 

[ICIETDIYEUR]

People hi

@ dfuller.

They are two reasons to want replacing the 220μF polymer with an electrochemical 470μF/25V (Cout of the LM2575HV regulator):

1> These are the recommendations in the datasheet of the regulator LM2575HV on page 13.

2> It is used to feed the two double OPAMPs and VDD of the LM2575HV regulator, which is not negligible !

I would also point out that it is necessary to improve the decoupling of the OPAMPs. (aka 'decoupling bases' for dual OP AMP).

For this I will use THIS that I receive this week (Tomorrow with my OPA1656 ?)

At the moment I use THIS

I preferred this solution because it does not require modification of the PCB or any special knowledge for the realization of the 'MOD' which leaves it within everyone's reach and allows the return to the original state without any difficulty.

I think that as things stand with this version 1.5 of the PCB, replacing the original OPAMPs with its 'decoupling bases' with OPA 1656 (FET) is enough to optimize the amp.

To go further, still without special knowledge, replacing the original power supply with the following one (Built-in PFC function) will bring you to the top by staying within reasonable budgets in relation to the starting price of the amplifier

It is available HERE

I forgot to mention that it is also necessary to change the four WIMA capacitors from 1μF to 10μF to decrease the cut-off frequency (and improve the phase of the signal at low frequencies) of the high-pass filter formed with the volume circuit at the input of the OPAMPS for reasons mentioned earlier in this thread.

A big lack at very low frequencies will be felt with WIMA capacitors of 1μF originally

Quick !!!

Replace them with 10μF

Regards

 

[SMen]

Because doing so clogs the thread with a bunch of replies from the same person back to back. It's considered good manners to keep it to one reply at a time.

All for good manners .

I would also point out that it is necessary to improve the decoupling of the OPAMPs. (aka 'decoupling bases' for dual OP AMP).

For this I will use THIS that I receive this week (Tomorrow with my OPA1656 ?)

View attachment 237983



I preferred this solution because it does not require modification of the PCB or any special knowledge for the realization of the 'MOD' which leaves it within everyone's reach and allows the return to the original state without any difficulty.

@ICIETDYIEUR could you explain more the decoupling bases and the ones you selected in the photo? How do they improve noise already 'in the signal' from the power supply, or just from the power supply? Perhaps this explains here.
It would be great to hear your thoughts on using the OPA1656s you are getting with and without these, over the next few days maybe, to give an idea of the improvement. Also the impact on the bass response - of this chip vs the WIMA capacitors - I have a big increase in bass response without any change to the capacitors.
Upgrade 'phases' are a very good idea

 

[ICIETDIYEUR]

Hello all

Before discussing the subject of 'decoupling bases', I think that some clarifications must be made on the choice of OPAMPS

I will try to explain it in a simple way using documents from Texas Instruments ->

The THD+N with BIPOLAR OPAMPS gets worse with increasing source resistance, whereas the FET OPAMPS amplifiers have negligible current noise to increase the total distortion.
In a low-noise, low-distortion application with a large source resistance, the FET OPAMPs would be a better choice

I hope this summary is helpful to you and that I wasn't too clumsy

Here is a link to the document used for this attempt at an explanation -> DOCUMENT

See you soon for the rest

Regards.

 

[ICIETDIYEUR]

Hello all

How wise you have been, a little bonus ->

Be careful though:
this is only OBJECTIVE technical data and not SUBJECTIVE qualitative musical assessment. In addition, there are still a large number of other references including DISCRETE OPAMPS (DEXA, SPARKOS LAB, BURSON, SONIC IMAGERY, AUDIO-GD,...)

The final choice is yours

Regards.

 

[SMen]

Hello all

How wise you have been, a little bonus ->

View attachment 238255

View attachment 238256

Be careful though:
this is only OBJECTIVE technical data and not SUBJECTIVE qualitative musical assessment. In addition, there are still a large number of other references including DISCRETE OPAMPS (DEXA, SPARKOS LAB, BURSON, SONIC IMAGERY, AUDIO-GD,...)

The final choice is yours

Regards.

This is really useful.

If you will forgive the pun, I was 'musing' over the MUSES8820 and MUSES8920 because the MUSES02 is way over budget for me.

Am I correct then that the MUSES8820 might be suitable in your low Ze table, and the MUSES8920 in your high Ze table?

Or ... what criteria to assign different op amps to each table? There is a lot of crossover regarding specfications.

 

[Schollaudio]

Are these amps still available in the US? I can't find one, please share a link.

 

[ICIETDIYEUR]

Hello everyone

@SMen:

You have determined in which table to place the two OPAMPS you are talking about

I made them appear on the tables for more readability ->

I put on a blue background the important criteria in my opinion to remember for a 'preselection'.
They are only technical and do not refer to a possible musical quality, it is important to specify it I think.
Regarding its criteria, they can be discussed with other members which is always interesting

For the allocation of OPAMPS by array, it is simply related to the technology use by them: BIPOLAR or FET (CMOS->MOSFET).

@ Schollaudio

Thank you for joining this thread

I can't tell you if they are available in the United States but a simple search on eBay should give you the answers

Regards.

 

[SMen]

I put on a blue background the important criteria in my opinion to remember for a 'preselection'.
They are only technical and do not refer to a possible musical quality, it is important to specify it I think.
Regarding its criteria, they can be discussed with other members which is always interesting

For the allocation of OPAMPS by array, it is simply related to the technology use by them: BIPOLAR or FET (CMOS->MOSFET).

Thank you for updating your tables ... this fascinating thread continues!

I have a question (actually a couple):

When you say:

"For the allocation of OP AMPS by array, it is simply related to the technology use by them: BIPOLAR or FET (CMOS->MOSFET)"

I am understanding that the array you mean is relating to the three chips in the DA10 - so looking at your tables you have 2 AD797(x2) in the I/V stage with 1 OPA1656 as the Buffer, together making up the "buffering output high quality DAC", so AD797(x2)x2 -> OPA1656 ?

I think that the DA10 could well justify a new thread so I don't want to divert, but I refer to it because that is how I interpreted what you have written here about the "array".

I have more questions, but I think I will hold them back and not all at once! Plenty still to read about ...

 

[ICIETDIYEUR]

Exact SMen

You have understood everything for the OPAMPS of the DAC Rod Rain Audio DA10

As a second choice instead of the AD 797, we can choose two other references (in my opinion):
- the LME 49990 which is very very good but more difficult to find for sale with very high technical characteristics.
- the OPA 2211 which has been specially developed for the current outputs (electrical current) of current DAC chips as indicated at the bottom of the table.

FYI, the AD 797 is the first choice of OPAMP advised by ESS

Afterwards, we do as we want because nothing is ever fixed with the progress made on the components.

To summarize on the DAC, the two AD 797 give the signal to the XLR outputs then the OPA 1656 use what is produced by the two AD 797 to give in turn the signal to the RCA outputs.
In fact, a user who only uses XLR outputs might not put the OPA 1656, which has the effect of relieving the power provided by the regulators who will therefore work in better conditions for 'the rest'

Regards.

 

[ICIETDIYEUR]

People hi

I received my 'decoupling bases' that I talked about earlier in this post.

Here are just a few pictures to present them to you ->

1 - Box received (x2)

2 - View above

3 - View below

4 - General view 1

5 - General view 2

6 - Climb with OPA 1656

7 - My old ones 'decoupling bases'

8 - In the amplifier

And that's it

( As a reminder, you can find them -> HERE <- )

Regards.

 

[SMen]

That's it?! - It would be great if you could give an opinion on listening?
This looks like a fantastic single and retro fittable upgrade, thank you
I wonder if these also reduce ripple noise?

 

[ICIETDIYEUR]

Then.

Question 'sound improvement' my opinion will only be subjective because I do not have measuring instruments adapted to validate it...

However, the 'message' seems to have gained clarity on the big sets, the 'attacks' also seem faster to me

Otherwise, no 'magic' but simply by looking at the 'logic' of the assembly used, OPAMPS can only work in better conditions, that's for sure

And as you have pointed out so well, it is a very simple 'upgrade' just as much as the 'step back'

Regards.

 

[ICIETDIYEUR]

People hi

I had talked about replacing the link capacitors located near the volume potentiometer.
Here's a quick reminder ->

The purpose of this replacement is to lower the low cut-off frequency below 1 Hertz but especially to correct the phase shift at low frequencies as can be seen on the violet curve (result of the 'MOD').

I take this opportunity to make you see that the impedance that results from this assembly is 50 kOhms: it is the input impedance (Ze) that will 'load' the OPAMPS that follow.
We are in the presence of a high input impedance and we have seen that it was preferable to use FET type OPAMPS in this case, hence the choice of OPA 1656

I had the choice between different capacitors of 10μF for this replacement ->

I chose WIMA (polypropylene) because they are better suited for this type of circuit (linking), they are however less simple to set up than the others because they are too big and it will still be necessary to set them up

As much as it is very simple to remove the existing capacitors as it will be more complicated to put the new ones.
Indeed, on these Chinese cards the soldering is 'not terrible' but above all it tends to stay in the holes provided for the 'legs' of the components.
Forget about the desoldering pump or the ground braid: it does not work on this type of board and you risk damaging it

So, i will try to explain my 'technique' with some photos ->

In the first photo, I use resistors whose 'legs' are the same diameter as those of the capacitors I unsoldered and push them about 5 millimeters (which will protrude on the other side) into the hole by heating with my soldering iron.

Then I cut the 'leg' letting the maximum length protrude on top of the card (second photo).

On the third photo (below the card), I grab the tip of the 'leg' that protrudes with a clamp while heating it with my soldering iron and I pull it back gently until I pull it entirely: the hole is unclogged ready to receive one of the new 'legs' of my new capacitors

Preparation of new capacitors:

As you can see in the photos, I made sure to be able to shift the capacitors but so that the legs are aligned as is the case for the passage holes of the board.
I put heat shrink sheath to isolate the legs and prevent any short circuit. All that remains is to set them up and weld them ->

There you go !

It wasn't complicated

My amplifier almost 'finished' (two or three more things to see...).

You can see my 'DIY' cables that I use...

I was going to forget to give you my listening impressions following this change.
The high-bass/low midrange have gained in clarity/definition of quite little it is true but this is audible
For your information when I make changes, I call my neighbor who is a musician (guitarist) and we play the same piece of music many times to soak up the smallest details.
After the modification we both noticed an improvement during the passages on bass guitar and percussion.
This confirms that by listening, we validated what had been technically planned during my 'explanations' at the beginning of the post

 

[Roland68]

Hi everyone

@v1adpetrov2:

What do you think about replacing the 220μF polymer with an electrochemical 470μF ? ->

View attachment 237902

I also redid suggestions for more readable capacitor bank locations:

-MOD 1

View attachment 237904

Total = 10 000 µF

-MOD 2

​

View attachment 237906

Total = 10 000 µF

-MOD 3 (20 'HP' for 18 inches speakers )


View attachment 237908

Total = 20 000 µF

What do you think ?

What would be your choice ?

Regards

You should generally not add large capacitances to a SMPS unless you are familiar enough with SMPSs to really understand them and check the changes with an oscilioscope. Then you'll understand why that's a bad idea.
Capacitors for the amplifier should only ever be connected to one point, very close to the power transistors or the amplifier chip. Leads of more than 1-2cm (including connecting legs) are absolutely to be avoided.
And please forget everything you know or read about amplifiers with linear power supplies, it doesn't work with switching power supplies.

The same also applies to the LM2575. It's not as easy to use as an LM317. This voltage regulator must be correctly integrated and both the component quality and the component values must be adapted for use. Apparently someone designed the circuit who understands this and knows that the components are matched for the current loading since they are also used as a filter for this switching regulator. The 220uF is already on the upper limit for this purpose, changing it would make it worse.

 

[ICIETDIYEUR]

Hello Roland68

Your intervention is relevant and you are absolutely right: this is why I did not do it, as you can see on my last photo where the original 220μF capacitor is still in place

Kind regards.

 

[v1adpetrov2]

...Apparently someone designed the circuit who understands this and knows that the components are matched for the current loading since they are also used as a filter for this switching regulator. The 220uF is already on the upper limit for this purpose, changing it would make it worse.

Apparently, someone has developed a scheme for the LM2575 that includes the same type of components that are used in other parts of the general scheme.
And this was done solely to reduce the production cost of the amplifier board.
And the filter there turned out to be disgusting - 52 kHz modulation frequencies just form the basis of the noise on the power supply.

The ingenious developer installed a 1μF capacitor at the input of the LM2575, while the datasheet indicates the minimum capacity of the input capacitor = 47μF.
The load current (of 12V supply) from two operational amplifiers and a TPA3255 chip = 0.170A. Then the inductance should be = 2200μH (in extreme cases - 1500μH), but the inductance of 47μH is set into board.
And the output capacitor should be at least = 470μF.

You can make the calculation yourself, the methodology is described on pages 13-15 of the datasheet.

 

[ICIETDIYEUR]

Hello v1adpetrov2

You have right.

So, you mean that ? ->

Is this correct ?

Regards.

 

[v1adpetrov2]

Is this correct ?

Yes, of course!
Exactly.
And, if you remember, a Schottky diode at 3 A 60 V

 

[ICIETDIYEUR]

Yes in order to switch the main power supply to 42 Volts

I added it to the photo ->

Otherwise I have these capacitors which I think could be suitable for 'Cin' (100µF/50V) and 'Cout' (470µF/25V) ->

So I seem to be missing the diode and the self that I don't have yet

Have you already made these changes on your side ?

Regards.