Transforming TPA3255 into ABD modulation.

[JayGilb]

Filter PCB. 80x80mm. From JLCPCB costs $4 for 5 pieces.

It’s so nice to be able to get pcb boards at such a low cost. I still have bottles by of etching solution and resist tape at home from many years ago.

 

[Hayk]

For your first order, you have $2 voucher that makes less than $5 with shipment 10-12days.

Tomorrow I will test a neutral tube interface replacing the opamps. Once tuned, then I will proceed to add 2nd harmonic in phase and reverse phase to move the staging.

 

[Hayk]

I just found this on AliExpress:
US $31.11 | ZK-3002 Pure Rear Level Digital Amplifier Board Stereo 300Wx2 TPA3255 Bridged Mono 600W AMP Audio Music Power Amplifier Board

https://a.aliexpress.com/_c4laBEXf

Nowadays, one can by for $14 free shipment a decent 3255 board, much better facture then mine, (got it free with 600w smps for $12.87).

 

[Hayk]

These are the simulation distortions for neutral and forward sounding interfaces at maximum input level. What makes the second, sound forward, is the 2nd harmonic phase -180°. It amplifies more the positive coming signal than the negative. At 2w it will have 0.1%THD. For the back staged needs the phase to be zero as is the case for the neutral, but 0.008% THD, at full power, it doesn't make any effect.
I will try to make a simple back staged, as many call it warm sounding, to be low cost. It requires an input transformer, but good and cheap ones cost over $20 a pair.

 

[KSTR]

Very interesting topic, subscribed.
It is such a simple and elegant concept one has to wonder why nobody though about it before. EMI concerns are valid but can't be the only point.

 

[Hayk]

I am preparing severa,l simple, low cost, single to differential converters to try them comparatively. For now, I don't have a single circuit switchable at low cost. Below is a 2 transistor with required harmonic phases for slight warm sound. It is the same circuit as above but transistor replacing the tube. I will design its complementary, it should give still different spectrum.
@KSTR. Welcome to the thread. What do you mean by EMI concerns?
The only EMI problem I have, is a Samsung refregerator of the neighborhood, that pollutes my measurements. The filter has a 1mH CM choke to eliminates all the 450khz, shorted by 470nF capacitors to avoid the speakers wires as antenna, radiate the carrier frequency.

 

[KL....]

Very interesting topic, subscribed.
It is such a simple and elegant concept one has to wonder why nobody thought about it before. EMI concerns are valid but can't be the only point.

Thank you, @KSTR, it has been thought about via the Axign AX5689 Controller chip & PFFB (a different approach), hasn't it, although what @Hayk is presenting is quite Novel and Simple/Elegant, as you have mentioned. Note that the AX5689 can output PWM@4 or 5Mhz (at least for GaN, as mentioned by @GaNMaster) which means that the 1mH CM choke (as mentioned by @Hayk) could be much smaller, couldn't it?

@Hayk to ask, will this Architecture/Topology work with PFFB? If yes, then potentially/technically, it could be even more Effective/Efficient, couldn't it? Can it handle varying Dead Time?

 

[Roland68]

Thank you, @KSTR, it has been thought about via the Axign AX5689 Controller chip & PFFB (a different approach), hasn't it, although what @Hayk is presenting is quite Novel and Simple/Elegant, as you have mentioned. Note that the AX5689 can output PWM@4 or 5Mhz (at least for GaN, as mentioned by @GaNMaster) which means that the 1mH CM choke (as mentioned by @Hayk) could be much smaller, couldn't it?

@Hayk to ask, will this Architecture/Topology work with PFFB? If yes, then potentially/technically, it could be even more Effective/Efficient, couldn't it? Can it handle varying Dead Time?

At MPS, this is called PFF (post-filter feedback). The feedback point is located after the output filter.
The difference is that this signal is digitized and then processed in the digital loop filter, which is much more precise. This is called a fully digital feedback loop design. Additionally, the feedback loop is programmable.

According to MPS, long dead times or low-power stages are not a problem due to the feedback. Power supply interference and noise are also significantly suppressed by this type of feedback.

 

[KSTR]

What do you mean by EMI concerns?

Any common-mode high frequency content is not filtered without any direct caps to GND. Notably when above the inductors self-resonant frequency.

 

[Hayk]

If you read the TI document of PFFB, it says, the purpose is to use low cost output filters. 8 years ago, I bought original Sagami inductors for $5 each. Today, On Ali, 2 pcs Sagami is $9.23 and the replacement, 10pcs for $7,89.
The PFFB given by the TI document, decreases the nonlinearty by 5db only.
In AD mod, each branch of the bridge is independent, so you can feedback individually. In BD mod, the output is differential, you need to convert it to single and feedback to the input before the opamps, hello stability. In ABD mod, I use 4th order filter, the phase shift above audio range is too high, it becomes too difficult to stabilize.
So, better to use bigger inductors.
Once the dead time issue is solved, next trouble, is the analog to 1 bit converter. The hysteresis of the comparator, and the linearity of the triangle generator becomes much more important than the nonlinearity of the output filters. To decrease such non linearity it requires local extra feedback. As the analog inputs are inverting opamps, it is very easy to increase the audio range feedback by 20db but keeping the noise gain the same. The neutral circuit in #24 does it by supplying the input in high impedance source. It needs 2 capacitors 100pf-330pf grounding the inputs, to keep the noise gain the same.

 

[Roland68]

If you read the TI document of PFFB, it says, the purpose is to use low cost output filters. 8 years ago, I bought original Sagami inductors for $5 each. Today, On Ali, 2 pcs Sagami is $9.23 and the replacement, 10pcs for $7,89.
The PFFB given by the TI document, decreases the nonlinearty by 5db only.
In AD mod, each branch of the bridge is independent, so you can feedback individually. In BD mod, the output is differential, you need to convert it to single and feedback to the input before the opamps, hello stability. In ABD mod, I use 4th order filter, the phase shift above audio range is too high, it becomes too difficult to stabilize.
So, better to use bigger inductors.
Once the dead time issue is solved, next trouble, is the analog to 1 bit converter. The hysteresis of the comparator, and the linearity of the triangle generator becomes much more important than the nonlinearity of the output filters. To decrease such non linearity it requires local extra feedback. As the analog inputs are inverting opamps, it is very easy to increase the audio range feedback by 20db but keeping the noise gain the same. The neutral circuit in #24 does it by supplying the input in high impedance source. It needs 2 capacitors 100pf-330pf grounding the inputs, to keep the noise gain the same.

For coils, I could recommend the CPD1495 series from CODACA. They're relatively inexpensive even in small quantities and offer some of the best price/performance ratio I know of for Class D amps.
Digikey CPD1495
Unfortunately, Digikey does not list all values.

The Axign AX5689 controller is superior to most solutions in terms of post-filter feedback and a few other aspects, but it's also a different approach and a fully digital concept.
You can't really compare this to the TPA325X.

 

[voodooless]

If you read the TI document of PFFB, it says, the purpose is to use low cost output filters.

That’s not the whole story. It says that you can do that if distortion isn’t as important to you. The primary reason however, next to lowering distortion significantly when using proper filters, is to remove most of the load dependence of the amp.

 

[Hayk]

 

[Hayk]

This 4 transistor circuit can function with 12v, hence it replaces physically the opamp to convert single to differential with adjustable staging. The when R7 is 0 it is backstage and front with 330k. Look No2 phase shift on the tables.
I will start building the neutral tube and this circuit today.

 

[KL....]

This 4 transistor circuit can function with 12v, hence it replaces physically the opamp to convert single to differential with adjustable staging. The when R7 is 0 it is backstage and front with 330k. Look No2 phase shift on the tables.
I will start building the neutral tube and this circuit today.

Thank you, @Hayk, but I can only see 2 transistors in the circuit? What is the output impedance?

 

[Hayk]

This is the circuit I already mounted it. It doesn't have yet the high impedance output to see first if the staging does really work.
The Russian tube 6N23p looks to be under embargo, Chinese 6j1 x2 can replace.

 

[KL....]

This is the circuit I already mounted it. It doesn't have yet the high impedance output to see first if the staging does really work.
The Russian tube 6N23p looks to be under embargo, Chinese 6j1 x2 can replace.

Thank you, @Hayk, 150K input impedance, the top Transistor pair are bootstraped to act as a CCS, is that correct?

 

[Hayk]

The Japanese, in mid 60's, wanted to know why the Marantz 8b sounded similar to single ended amps and passed through spectrum analyzer to find out it had similar even harmonics dominant. Since diffused by Jean Hiraga the recipe, it became popular idea that to have warm sounding amp you need to have such spectrum.
These days, deaf designers convince noobs that his amp measures 0.002% THD but dominant even harmonics, so it is a certainty that sounds warm.
I simulated below, the 2nd harmonic of 10% with +90° phase and -90°.
What does effect is not the hearing of the harmonics that makes the difference, but in time domain the transients get reversed. If the positive going input pulse, as piano notes, the amp amplifies less than should, than the sound becomes less expressive and more comfortable for long term listening. With reverse phase, it becomes more vivid and more exciting rhythmic.

 

[Hayk]

I mounted the circuit after testing on bench. I will try it out tomorrow.

 

[KL....]

The Japanese, in mid 60's, wanted to know why the Marantz 8b sounded similar to single ended amps and passed through spectrum analyzer to find out it had similar even harmonics dominant. Since diffused by Jean Hiraga the recipe, it became popular idea that to have warm sounding amp you need to have such spectrum.
These days, deaf designers convince noobs that his amp measures 0.002% THD but dominant even harmonics, so it is a certainty that sounds warm.
I simulated below, the 2nd harmonic of 10% with +90° phase and -90°.
What does effect is not the hearing of the harmonics that makes the difference, but in time domain the transients get reversed. If the positive going input pulse, as piano notes, the amp amplifies less than should, than the sound becomes less expressive and more comfortable for long term listening. With reverse phase, it becomes more vivid and more exciting rhythmic.

Thank you, @Hayk, Yes, Phase/Time Domain Distortion/s is (very) relevant and must always be considered, mustn't it, especially with reference/relevance to the word/s 'Neutral' and/or 'Transparent'?

Distortion causes Phase/Time Domain Shift/Distortion

• Non-Linear Processing: Most audio gear (EQs, compressors, amplifiers, digital filters like FIR filters) are non-linear and alter different frequencies' timing differently, creating phase shifts.
• EQ (Equalization): Boosting or cutting frequencies with steep curves (especially digital EQs) shifts the phase, altering the waveform's shape and sound.
• Clipping/Overdrive: Hard clipping (a severe form of amplitude distortion) generates high-frequency harmonics, which shifts the phase of these new frequency components relative to the original signal, causing a harsh, metallic sound.
• Tape Recording: The physical process of recording to tape introduces phase shifts.
• This video explains how phase distortion affects the sound, The Hardware Hub.... What is Phase Distortion and how does it affect Speakers

Noise causes Phase/Time Domain Shift/Distortion

• Adding Jitter/Timing Errors: Random noise adds unpredictable fluctuations (jitter) to the signal's timing, effectively scrambling phase information, especially when mixed with the original signal.
• Interference: When noise interferes with the desired signal (especially at low levels or in multi-mic setups), it creates complex phase cancellations and reinforcements, muddying the sound.
• This video demonstrates the concept of phase and phase shift, Sound & Frequencys Explained by Wickiemedia

Effects on Sound

• Phase Distortion: Makes audio sound "phasey," thin, hollow, or blurry, reducing definition, especially on transients (like drum hits) or when tracks are mixed.
• Phase Cancellation: When signals are out of phase (180 degrees), they can cancel each other out, leading to a loss of bass or reduced volume.

For Consideration.... Phase/Time Domain Shift and Noise/Distortion are deeply linked where it's hard to have one without the other in complex audio systems, as processing that changes Amplitude (Distortion) almost always affects Timing (Phase)