AIYIMA A70 Stereo Amplifier Review

[somebodyelse]

MLCC Bypass for Main Capacitors: Add a small (e.g., 1µF) Multi-Layer Ceramic Capacitor (MLCC) in parallel with each new Rubycon ZLJ capacitor. This is to filter out very high-frequency noise that the larger electrolytic caps are less effective at handling.

This is already present, close to the TPA3255 with minimal loop area. The datasheet emphasises that this layout is necessary for good performance, and I don't think we've seen an amp yet that doesn't follow that advice. When making changes (including to lower ESR main caps) take care to consider their combined resonant characteristics - if you don't then you could make things worse not better.

 

[Schlippwhip68]

The phillips 5532 cost 2.70 euro for one and sounds naturel and musical,the jrc 5532dd,opa 1656, BB5532 and more cheaper 5532 not so good or musical.

https://nl.aliexpress.com/item/1005003071748273.html?spm=a2g0o.order_list.order_list_main.15.684579d2cYyDQs&gatewayAdapt=glo2nld

I listened to the Philips NE5532 again today and compared them to a few others like the new golden choice the SX52B. I really like the Phillips NE5532, they have more rich textures than other versions of the same chip. Good depth to the soundstage and super clean vocal and good leading edges on notes. I actually prefer them in my Aiyima A70 more than the opa1656, SX52B, opa2134. I have a pair of Muses 01 ordered to try out but the Phillips are certainly impressive and I wouldn’t be surprised if I end up leaving them in the A70 even over the Muses 01.

The current set up is the Aiyima A70, Fiio K7bt and Q Acoustics 3020i. The Phillips are working really well in this set up.

 

[bramsat]

The original pililips NE 5532 is the best you ever heart...believe me!

 

[oyasai]

This is already present, close to the TPA3255 with minimal loop area. The datasheet emphasises that this layout is necessary for good performance, and I don't think we've seen an amp yet that doesn't follow that advice. When making changes (including to lower ESR main caps) take care to consider their combined resonant characteristics - if you don't then you could make things worse not better.

Thank you, somebodyelse, for your insightful feedback. I sincerely appreciate you taking the time to share your expertise.

You've raised an extremely important point regarding the resonant characteristics of the combined capacitor network. I completely agree that blindly swapping to lower ESR main caps without considering their interaction with the existing MLCC bypass capacitors near the TPA3255 could indeed make things worse. Your warning is well-taken and has prompted me to refine my approach.

Revised Plan: A Cautious, Sequential Implementation

Based on your advice, I will proceed only with Phase A for now, and even then, I will treat it as two distinct steps to isolate variables:

1. Step A1: Power Capacitor Swap. I will replace the stock main caps with the Rubycon ZLJ series. My primary motivation here is not just the lower ESR, but also to ensure the use of authentic, high-reliability components as a foundation, thus increasing my "peace of mind." I will perform listening tests after this step alone.
2. Step A2: Op-Amp Swap. After evaluating the capacitor swap, I will then proceed to replace the stock op-amps with the OPA1612.

Phase B, especially the addition of more bypass capacitors, is now officially on hold until I can fully analyze the results of Phase A. Thank you for preventing a potential misstep.

---

A Separate, but Related Question for the Community: Does Op-Amp Rolling Truly Matter in a PFFB Design?

This topic came up in my research, and I've seen @amirm's excellent video demonstrating that op-amp rolling has no measurable effect in a typical amplifier with pre-filter feedback, due to the corrective power of the feedback loop.

However, I have a specific hypothesis regarding PFFB architectures like the one in the TPA3255, and I would like to submit it for critique:

In a Post-Filter Feedback (PFFB) design, the main feedback loop is closed after the output LC filter. This means the input op-amp stage operates, to a significant degree, outside of this primary corrective loop. Its main role is to act as a high-impedance buffer and gain stage for the signal beforeit enters the TPA3255's modulator and power stage.

Therefore, my hypothesis is that the op-amp's own self-noise and distortion are not as effectively canceled out by the PFFB loop compared to a traditional pre-filter feedback design where the op-amp is directly inside the loop.

If this is correct, then swapping to an op-amp with a demonstrably lower noise floor and lower distortion (like the OPA1612) should, theoretically, result in a small but tangible improvement in the final output's purity, as it is feeding a cleaner signal into the core of the amplifier before the main PFFB loop does its work.

Am I fundamentally misunderstanding the topology here? I'm eager to be proven wrong and learn from your collective expertise.

Thank you all again.

---

P.S. - An Immediate Real-World Result

I should add that my primary motivation for upgrading the power supply was not purely theoretical. With the stock power adapter, I was able to hear a faint but noticeable noise floor (a sort of "hiss" or "digital haze") from my full-range drivers when no music was playing and my ear was close to the speaker.

I can confirm that after switching to the lab-grade stabilized power supply (running at 36V), this idle noise has been completely eliminated. The background is now subjectively "black."

This initial result reinforces my belief that addressing the power supply quality is the most critical foundational step for this class of amplifier, before even considering other internal modifications.

 

[Bruce Morgen]

Thank you, somebodyelse, for your insightful feedback. I sincerely appreciate you taking the time to share your expertise.

You've raised an extremely important point regarding the resonant characteristics of the combined capacitor network. I completely agree that blindly swapping to lower ESR main caps without considering their interaction with the existing MLCC bypass capacitors near the TPA3255 could indeed make things worse. Your warning is well-taken and has prompted me to refine my approach.

Revised Plan: A Cautious, Sequential Implementation

Based on your advice, I will proceed only with Phase A for now, and even then, I will treat it as two distinct steps to isolate variables:

1. Step A1: Power Capacitor Swap. I will replace the stock main caps with the Rubycon ZLJ series. My primary motivation here is not just the lower ESR, but also to ensure the use of authentic, high-reliability components as a foundation, thus increasing my "peace of mind." I will perform listening tests after this step alone.
2. Step A2: Op-Amp Swap. After evaluating the capacitor swap, I will then proceed to replace the stock op-amps with the OPA1612.

Phase B, especially the addition of more bypass capacitors, is now officially on hold until I can fully analyze the results of Phase A. Thank you for preventing a potential misstep.

---

A Separate, but Related Question for the Community: Does Op-Amp Rolling Truly Matter in a PFFB Design?

This topic came up in my research, and I've seen @amirm's excellent video demonstrating that op-amp rolling has no measurable effect in a typical amplifier with pre-filter feedback, due to the corrective power of the feedback loop.

However, I have a specific hypothesis regarding PFFB architectures like the one in the TPA3255, and I would like to submit it for critique:

In a Post-Filter Feedback (PFFB) design, the main feedback loop is closed after the output LC filter. This means the input op-amp stage operates, to a significant degree, outside of this primary corrective loop. Its main role is to act as a high-impedance buffer and gain stage for the signal beforeit enters the TPA3255's modulator and power stage.

Therefore, my hypothesis is that the op-amp's own self-noise and distortion are not as effectively canceled out by the PFFB loop compared to a traditional pre-filter feedback design where the op-amp is directly inside the loop.

If this is correct, then swapping to an op-amp with a demonstrably lower noise floor and lower distortion (like the OPA1612) should, theoretically, result in a small but tangible improvement in the final output's purity, as it is feeding a cleaner signal into the core of the amplifier before the main PFFB loop does its work.

Am I fundamentally misunderstanding the topology here? I'm eager to be proven wrong and learn from your collective expertise.

Thank you all again.

---

P.S. - An Immediate Real-World Result

I should add that my primary motivation for upgrading the power supply was not purely theoretical. With the stock power adapter, I was able to hear a faint but noticeable noise floor (a sort of "hiss" or "digital haze") from my full-range drivers when no music was playing and my ear was close to the speaker.

I can confirm that after switching to the lab-grade stabilized power supply (running at 36V), this idle noise has been completely eliminated. The background is now subjectively "black."

This initial result reinforces my belief that addressing the power supply quality is the most critical foundational step for this class of amplifier, before even considering other internal modifications.

View attachment 476745

That's one of the secondary reasons I eschewed sealed "power brick" PSUs in favor of a single 10A industrial SMPS unit for the two A07s in my main system. That approach allows conveniently adding extra shunt capacitance (at the barrier strip DC output terminals of the SMPS) and avoids modding the amplifiers themselves, which not only adds more capacitance, but also lowers the aggregate ESR of the resulting de facto multi-component capacitor. I'm not sure if the effects of doing that actually improve amp performance, but I already had a few suitable capacitors and adding one of them in that way certainly didn't cause any harm.

 

[oyasai]

Thank you again for the ongoing discussion. The points raised have helped me structure my thought process more rigorously. I want to present the logical chain of reasoning that led to my current hypothesis, framed not as a definitive conclusion, but as a series of questions and deductions.

I'm approaching this as a thought experiment, and I invite you all to critique the steps in my logic.

---

Step 1: The Initial Observation - A Discrepancy Beyond Frequency Response

My starting point was the linked video (https://youtu.be/AN9OjM1L_BA?si=2kXgf4VIvKSYKo0I).

The key takeaway for me was not that the frequency response (SPL) was dramatically different. In fact, the A80 showed more output in the low bass region (a peak around 30-40Hz). This contradicts the simple notion that "expensive amps produce more bass."

Instead, the critical data point was the severe phase deviation in that same low-frequency region (10:50).

This leads to my first question: What physical mechanism could cause an amplifier to maintain (or even boost) SPL, yet fail to maintain phase coherence, specifically under heavy, low-frequency load?

---

Step 2: The Hypothesis - From "Power" to "Control"

This observation shifts the focus from "lack of power" to "lack of control." An uncontrolled woofer cone will overshoot and resonate, which could manifest as a peak in the SPL graph, but its timing (phase) will be completely detached from the input signal.

This leads to my second question: What part of an amplifier is primarily responsible for controlling a speaker driver, especially against its own back-EMF and impedance swings?

My hypothesis is that this is fundamentally a power supply issue. Not a lack of sustained power (W), but a lack of instantaneous current delivery and low impedance to effectively damp the driver's movement. A high-impedance power supply would struggle to sink the current generated by the driver's back-EMF, leading to a loss of control and, plausibly, the observed phase shifts.

---

Step 3: The Experiment - Testing the Power Supply Hypothesis

This is why my first and most critical modification was to change the power supply for my two A70s.

• My System Context: I am running a demanding speaker setup: dual-mono A70s driving a 2-way system. The low end is handled by four 8" Dayton DS215-8 drivers in sealed enclosures (stereo pair, two drivers per channel wired in parallel, creating a nominal 4-ohm load with complex impedance behavior).
• The Change: I replaced the stock power adapters with a single, high-current (36V/20A) lab-grade stabilized power supply.
• The Subjective Result: I must emphasize this is purely subjective, but after the power supply change, I ran a 20Hz sine wave test. Before, the sound was barely audible. After, it was clearly perceptible as a physical pressure wave. This suggested to me that the amp's ability to control the drivers at very low frequencies had fundamentally improved. This anecdotal observation, combined with the video's data, solidified my belief that the power supply is the primary bottleneck.

---

Step 4: The Refined Goal - Optimizing the Internal "Last Mile"

Now, I view my internal modification plan (the "Phase A/B" I posted earlier) not as a random parts swap, but as a logical extension of this power supply-centric hypothesis.

My reasoning, informed by the helpful feedback from Bruce regarding his similar setup, is as follows: A powerful external supply is crucial, but its effectiveness can be limited by the "last mile" of the power delivery chain inside the amplifier.

• The Rubycon ZLJ(2,200uF, 63V) Swap: This isn't about increasing capacitance. It's about lowering the ESR of the main reservoir caps. This is the final buffer that provides the instantaneous current the TPA3255 needs for a low-frequency transient. A lower ESR capacitor can deliver that current faster and with less voltage drop, thus improving the "control" we discussed in Step 2.
• The OPA1612 Swap: This is to ensure that the signal being fed into this now-robust power/amplification stage is as pure as possible, minimizing noise and distortion from the very start of the chain.

In essence, I am postulating that there is a strong correlation, if not a direct causation, between the stability/impedance of the entire power supply architecture and an amplifier's ability to maintain phase coherence at low frequencies. My modifications are an attempt to test this by reinforcing every link in that chain.

Does this thought process seem sound to you all? I'm open to the possibility that other factors are at play, but this seems to be the most logical explanation for the observed phenomena.

 

[somebodyelse]

A Separate, but Related Question for the Community: Does Op-Amp Rolling Truly Matter in a PFFB Design?

This topic came up in my research, and I've seen @amirm's excellent video demonstrating that op-amp rolling has no measurable effect in a typical amplifier with pre-filter feedback, due to the corrective power of the feedback loop.

However, I have a specific hypothesis regarding PFFB architectures like the one in the TPA3255, and I would like to submit it for critique:

In a Post-Filter Feedback (PFFB) design, the main feedback loop is closed after the output LC filter. This means the input op-amp stage operates, to a significant degree, outside of this primary corrective loop. Its main role is to act as a high-impedance buffer and gain stage for the signal beforeit enters the TPA3255's modulator and power stage.

Therefore, my hypothesis is that the op-amp's own self-noise and distortion are not as effectively canceled out by the PFFB loop compared to a traditional pre-filter feedback design where the op-amp is directly inside the loop.

If this is correct, then swapping to an op-amp with a demonstrably lower noise floor and lower distortion (like the OPA1612) should, theoretically, result in a small but tangible improvement in the final output's purity, as it is feeding a cleaner signal into the core of the amplifier before the main PFFB loop does its work.

Am I fundamentally misunderstanding the topology here? I'm eager to be proven wrong and learn from your collective expertise.

Thank you all again.

I think you're fundamentally misunderstanding it. With TI's datasheet implementations with and without PFFB there is no feedback to the opamp's input, just to the TPA3255's input. This seems to cover most of the PFFB implementations we see with the TI amps as they broadly match the datasheet performance. The opamp operates with its local feedback in both cases - there is no difference, so your hypothesis is based on a faulty premise. It is possible to send feedback to the opamp input, making a compound amp with performance comparable to the opamp rather than the power amp that follows it. This is how Neurochrome's Modulus amps outperform the LM3886 Done Right. It's been suggested this is how the best performing TPA3255 based amps work too - I've not seen their schematics. Wrapping the feedback round both amps makes it a bit more tricky though, so if you're looking to change the opamp in one of these then be sure you know how to test its stability afterwards.

 

[Schlippwhip68]

Hopefully this will offend no one. I stuck the Muses 8920 into the Aiyima A70 today and they really did outshine even the rather good Phillips NE5532. I fed it via my Fiio K7bt from the laptop using Apple Music and APO EQ. I am using the Q Acoustics 3020i for speakers and I wilI say the whole system gelled very well. The AKM chip based K7 seemed to match the Muses 8920 very, very well and coming through the Q acoustic at 6ohms the A70 did not break a sweat. What was provided was a wide soundstage with excellent centre and stereo imaging with quite a bit of texture across the board and more air than the Phillips allowed, more space in general. I tried the Burson v5d but only in one channel as I have only one of those at the moment but they also sounded so good especially the bass….i ordered the second.

The Burson’s bass through the A70 was a serious step up from everything I have tried so far, even the Muse 8920. Same power supply used throughout, an 36v 6a power brick. The bass notes from the Burson is where that little Australian made chip shines. Every note was clean, clear and tuneful. For Dub fans, Reggae etc I would highly recommend this chip. I played Solomon Jabby and his latest album today and was immediately caught up in those bass notes. The rest of the frequency’s I have not studied as of yet but there was nothing there to really cry over in regards to shortcomings straight off the bat, with a little more time I could probably pick out a few pros and cons with the rest of the sound but is that the case with everything. I only spent a few minutes with the Burson in because only one channel was possible but the second was ordered only after that few minutes.

For now, the mighty Phillips NE5532 has been superseded by the Muse 8920 for the wider soundstage mostly, slightly better bass response and a less boxy sound with more air. I personally find the 8920 specifically in the A70 and working with the AKM chip based Fiio K7bt one of the most musical combinations I have heard in sometime especially in the price bracket, it’s a very engaging sound that has now been on in my room for over 4 hours without myself getting bored, trust that is saying something these days. The 8920 sits very well in the A70 sonically speaking, I am not so clear on the overall compatibility with the A70’s circuitry, sadly I am afraid I cannot entertain my friends here with that kind of expertise but at least so far there has been no issues at to speak of toward any detrimental affect on the amp with this op amp.

I know this is more the audiophile side of the topic but nevertheless the end result is what the engineer is aiming towards after all.

In regard to power supply I used a standard older Aiyima 36v (huge) the new newer models are much smaller. I would agree that with the amps as good these now are it maybe worth it for the manufacturers looking into building better power bricks especially @ 36v 6a which imo is the more balanced option for the TI3255 chip.

Cheers!

 

[Roland68]

Hopefully this will offend no one. I stuck the Muses 8920 into the Aiyima A70 today and they really did outshine even the rather good Phillips NE5532. I fed it via my Fiio K7bt from the laptop using Apple Music and APO EQ. I am using the Q Acoustics 3020i for speakers and I wilI say the whole system gelled very well. The AKM chip based K7 seemed to match the Muses 8920 very, very well and coming through the Q acoustic at 6ohms the A70 did not break a sweat. What was provided was a wide soundstage with excellent centre and stereo imaging with quite a bit of texture across the board and more air than the Phillips allowed, more space in general. I tried the Burson v5d but only in one channel as I have only one of those at the moment but they also sounded so good especially the bass….i ordered the second.

The Burson’s bass through the A70 was a serious step up from everything I have tried so far, even the Muse 8920. Same power supply used throughout, an 36v 6a power brick. The bass notes from the Burson is where that little Australian made chip shines. Every note was clean, clear and tuneful. For Dub fans, Reggae etc I would highly recommend this chip. I played Solomon Jabby and his latest album today and was immediately caught up in those bass notes. The rest of the frequency’s I have not studied as of yet but there was nothing there to really cry over in regards to shortcomings straight off the bat, with a little more time I could probably pick out a few pros and cons with the rest of the sound but is that the case with everything. I only spent a few minutes with the Burson in because only one channel was possible but the second was ordered only after that few minutes.

For now, the mighty Phillips NE5532 has been superseded by the Muse 8920 for the wider soundstage mostly, slightly better bass response and a less boxy sound with more air. I personally find the 8920 specifically in the A70 and working with the AKM chip based Fiio K7bt one of the most musical combinations I have heard in sometime especially in the price bracket, it’s a very engaging sound that has now been on in my room for over 4 hours without myself getting bored, trust that is saying something these days. The 8920 sits very well in the A70 sonically speaking, I am not so clear on the overall compatibility with the A70’s circuitry, sadly I am afraid I cannot entertain my friends here with that kind of expertise but at least so far there has been no issues at to speak of toward any detrimental affect on the amp with this op amp.

I know this is more the audiophile side of the topic but nevertheless the end result is what the engineer is aiming towards after all.

In regard to power supply I used a standard older Aiyima 36v (huge) the new newer models are much smaller. I would agree that with the amps as good these now are it maybe worth it for the manufacturers looking into building better power bricks especially @ 36v 6a which imo is the more balanced option for the TI3255 chip.

Cheers!

Unless you test two identical devices with two different op amps, completely blinded and at exactly the same volume, your statement is unfortunately worthless.
The probability that you won't be able to tell which op amp is in which device is at least 99%.

For precisely this reason, when making changes to a device, I always work with a second, unaltered device as a reference. This has saved me a lot of unnecessary time. I know enough professional tuners who have sometimes wasted days and weeks on a perceived "improvement" that, in direct comparison, actually turned out to be a deterioration.

Likewise, many people were no longer able to distinguish between previously significant sound differences in a direct and blind comparison. The excuses for this would fill a book.
This is the experience of over 30 years.

 

[Schlippwhip68]

Unless you test two identical devices with two different op amps, completely blinded and at exactly the same volume, your statement is unfortunately worthless.
The probability that you won't be able to tell which op amp is in which device is at least 99%.

For precisely this reason, when making changes to a device, I always work with a second, unaltered device as a reference. This has saved me a lot of unnecessary time. I know enough professional tuners who have sometimes wasted days and weeks on a perceived "improvement" that, in direct comparison, actually turned out to be a deterioration.

Likewise, many people were no longer able to distinguish between previously significant sound differences in a direct and blind comparison. The excuses for this would fill a book.
This is the experience of over 30 years.

Well my friend at least people still have the freedom to read our posts here and make their own choices so that is a liberty that we can all enjoy.

 

[Roland68]

Well my friend at least people still have the freedom to read our posts here and make their own choices so that is a liberty that we can all enjoy.

Unfortunately, you're one of those people who didn't understand the "don't kill the messenger" thing.
It's not about freedom, but about unverified sonic statements that lack any metrological or physical basis. This sort of thing belongs more in "audiophile" forums.

I'm not even saying there aren't any differences, but from hundreds of blinded tests, I know that these differences, if they exist, are so small that even very experienced people, whether technically savvy or audiophiles, have a hard time hearing a difference blindly, assuming there is one.

Get a second A70 and try it out, with someone else switching the settings. But you probably won't like the result.

 

[Schlippwhip68]

I am a lot of things in peoples minds my friend and questionable, probably.

In regard to ‘small differences’ even those are deal breakers for a lot of people like myself. Call me fussy.

Thanks for the suggestion to further certifying a particular outcome either way but the outcome heard tonight with a single A70 needs no further investigation. Some of the differences were clearly heard were small as suggested, subtle yes but also very distinctive and large enough to warrant an immediate choice, an expression commonly used towards this particular outcome today is ‘night and day’ as other differences were and are very obvious.

Context of hardware should also be considered. Little attention is paid to a car with square wheels meaning that the modular approach to the system build has proven to be successful to one’s own desire which may find common appreciation with others if they should so experiment.

I would certainly recommend this set up to anyone looking for budget components and excellent sound quality.

But alas yes, no one intends to shoot the messenger and I agree my post would probably be more suited on an Audiophile forum.

God bless.

 

[Dave TR3A]

A striking case study showing Beranek’s Law.

 

[oyasai]

I think you're fundamentally misunderstanding it. With TI's datasheet implementations with and without PFFB there is no feedback to the opamp's input, just to the TPA3255's input. This seems to cover most of the PFFB implementations we see with the TI amps as they broadly match the datasheet performance. The opamp operates with its local feedback in both cases - there is no difference, so your hypothesis is based on a faulty premise. It is possible to send feedback to the opamp input, making a compound amp with performance comparable to the opamp rather than the power amp that follows it. This is how Neurochrome's Modulus amps outperform the LM3886 Done Right. It's been suggested this is how the best performing TPA3255 based amps work too - I've not seen their schematics. Wrapping the feedback round both amps makes it a bit more tricky though, so if you're looking to change the opamp in one of these then be sure you know how to test its stability afterwards.

@somebodyelse, thank you once again. Your last post was exceptionally insightful and genuinely helpful. You were absolutely correct to point out the fundamental misunderstanding in my initial hypothesis.

I think you're fundamentally misunderstanding it. With TI's datasheet implementations with and without PFFB there is no feedback to the opamp's input, just to the TPA3255's input. ... The opamp operates with its local feedback in both cases - there is no difference, so your hypothesis is based on a faulty premise.

This was the critical piece of information I was missing. Your explanation sent me back to the TI application notes (specifically, SLAA788A for the TPA3255), and now I see it clearly. The PFFB loop is, as you said, entirely contained within the TPA3255's own ecosystem, feeding back to the chip's input, not the preceding op-amp stage. My premise was indeed faulty, and I appreciate the correction.

However, this correction does not invalidate my modification plan; paradoxically, it reinforces it from a different logical standpoint.

Here is my revised and now more accurate understanding of the signal chain in my system:

1. miniDSP SHD (DAC Stage): Creates the initial analog signal.
2. Op-Amp Stage (A70 Input): Functions as an independent buffer/pre-gain stage. It operates under its own local feedback. Crucially, any noise or distortion introduced at this stage is outside the TPA3255's PFFB loop.
3. TPA3255 Stage (Modulator + Power Stage + PFFB): Receives the signal from the op-amp. The PFFB loop works brilliantly to correct for non-linearities in the power stage and, most importantly, the output LC filter. But it can only correct the signal it receives. It cannot correct for flaws already present in that signal.

Therefore, my conclusion remains the same, but with a more robust foundation:

To achieve the highest possible fidelity, the signal fed into the TPA3255 must be as pure as possible. Since the op-amp stage is not "cleaned up" by the PFFB loop, its own self-noise and distortion directly impact the final sound quality.

This confirms that my planned upgrades are correctly targeted:

• Op-Amp Swap (OPA1612): This is a direct measure to improve the quality of the signal before it enters the TPA3255, minimizing the noise and distortion that the PFFB loop cannot address.
• Power Capacitor Swap (Rubycon ZLJ): This is to optimize the "power reservoir" directly serving the TPA3255 power stage, ensuring its ability to deliver instantaneous current without compromise, thus maximizing the effectiveness of the PFFB's control over the speaker load.

Thank you for helping me to refine my understanding. The discussion has solidified my confidence that this modification path, focusing on these two specific areas while leaving the core PFFB architecture untouched, is indeed the correct one for my goals.

---

In short, the correction clarified the limits of PFFB — and precisely because of those limits, op-amp and power supply upgrades remain the most effective path.

---

Documentation
TPA324x and TPA325x Post-Filter Feedback
Op Amp Measurements

 

[somebodyelse]

If that's what you want to do it's your choice - do it properly with controlled listening tests and it will be a learning experience. However even with PFFB the dominant source of noise and distortion in the A70 is the TPA3255, so changing opamp will have no appreciable effect. The measurements for 5532 vs Sparkos in a V3 Mono demonstrate this - it's not the exact same amp, but it is another decently performing TPA3255 with PFFB and socketed opamps. There's a whole other thread dedicated to this so I won't go into it further.

As a side note, unless you do some measurement before and after the mod, or record the output before and after so you can compare it directly, you are likely to perceive your change as an improvement due to confirmation bias. I say this as someone who's done similar mods in the past. Level matched recordings at the output will let you do blind ABX testing to see whether there really is an audible difference. You should be aware of the precautions you need to take when measuring or recording at the amp outputs - they're all floating at half the supply voltage so will cause problems or damage for the inexperienced or unwary. Don't try it unless you really know what you're doing.

 

[magreen]

As a side note, unless you do some measurement before and after the mod, or record the output before and after so you can compare it directly, you are likely to perceive your change as an improvement due to confirmation bias.

Hey, don’t underestimate just how good confirmation bias sounds. It’s real. And it sounds great!
It’s like the placebo effect. People say, oh that’s just a placebo it’s not real. The placebo effect is very real. And it works!

 

[Roland68]

I am a lot of things in peoples minds my friend and questionable, probably.

In regard to ‘small differences’ even those are deal breakers for a lot of people like myself. Call me fussy.

Thanks for the suggestion to further certifying a particular outcome either way but the outcome heard tonight with a single A70 needs no further investigation. Some of the differences were clearly heard were small as suggested, subtle yes but also very distinctive and large enough to warrant an immediate choice, an expression commonly used towards this particular outcome today is ‘night and day’ as other differences were and are very obvious.

Context of hardware should also be considered. Little attention is paid to a car with square wheels meaning that the modular approach to the system build has proven to be successful to one’s own desire which may find common appreciation with others if they should so experiment.

I would certainly recommend this set up to anyone looking for budget components and excellent sound quality.

But alas yes, no one intends to shoot the messenger and I agree my post would probably be more suited on an Audiophile forum.

God bless.

Differences like night and day? Really? When swapping op amps?

Technical basics aside, have you ever wondered why people keep popping up on forums, YouTube, etc., who hear such significant differences but can't provide any measurements or other evidence?
When you then ask them to do a blind test, these people either back out or have to admit that they can't hear any difference in a blind test.
As I said, the excuses for this could fill books, and the list of users who promised proof and blind tests and couldn't deliver is endless.

In the same way, you could easily record and compare the signal, as several people have already done here in the forum, but there are no such recordings with audible differences.

How much money could someone, or even the manufacturers, make if there were even one consistent proof of such differences?

I always recommend conducting your own blind tests; afterward, the issue is always resolved, even if some people quickly fall for the next snake oil trick.

Does Op-amp Rolling Work?
Op-amp Rolling Using Sparkos on Fosi V3 Mono
Does Op-amp Rolling Work? [Video]
Amplifier Op-amp Rolling Part 2 [Video]

 

[somebodyelse]

Hey, don’t underestimate just how good confirmation bias sounds. It’s real. And it sounds great!
It’s like the placebo effect. People say, oh that’s just a placebo it’s not real. The placebo effect is very real. And it works!

I'm fully aware, both in theory and from experience. There have been studies published on both with robust experimental method and solid data.

I've done similar mods in the past without good listening controls and 'heard' clear differences. If you keep doing it then sooner or later you'll make the revelatory mistake - you'll hear the difference, then find you forgot to change the thing you thought you had. Nothing had actually changed, but you still 'heard' a difference. Others have had a similar thing in the recording studio setting - you make a change, hear the improvement, then find the change was on a muted channel or something that wasn't patched in.

 

[DavidM1]

Unless you test two identical devices with two different op amps, completely blinded and at exactly the same volume, your statement is unfortunately worthless.

Such a simple message but it has as much impact as a bug hitting a windscreen.

Why do folks resist the obvious?

 

[Roland68]

Such a simple message but it has as much impact as a bug hitting a windscreen.

Why do folks resist the obvious?

The same reason why charlatans and miracle healers are still so successful in medicine and make money from it.
People want to believe in something, the more inexplicable and mystical, the more they do it.

It's not ineffective either, because more and more people are gaining access to ASR and objective tests. There are also more and more people who do their own blind comparisons and then discover that the differences are only marginal or not audible at all.
Likewise, there are many people who have discovered this for themselves and only then discovered ASR.
So it's not entirely hopeless.

I had to get used to this method of working with two devices 25 years ago, when I was still working with people who heard a difference like "night and day" every time a component was replaced. In a blind test, the same people couldn't hear even the slightest difference, or, even better, couldn't even tell if anything had been replaced in either device.
For me, it made such work much easier and also shed a lot of light on the darkness. But most importantly, it saved me hundreds, if not thousands, of hours because you could immediately hear the difference when you switched.