Aiyima TPA3251 Modification Build Thread!

I'm starting this TPA3251 Modification Build thread so that everyone has a direct path for any and all modifications they wish to explore.

Like many of you Class D fans, I purchased the Aiyima TPA3251 A04 amplifier. After break-in and listening to it for awhile I realized that even though it sounds pretty darn good, it could potentially sound even better. So I then started to research what mods I thought it should get, after a lot of contemplation I came up with my build plan. Though my build plan exceeds the cost of my initial purchase price of ~$42 dollars, I believe I will end up with a Class D amp that sounds better than "many" higher dollar name brand amps. Plus this is simply a fun build, so I'm totally good with spending a few extra dollars in the quest to achieve magical sound.

For anyone that owns an Aiyima TPA3251 and is interested in modding it, please post up any & all mods that you've made or plan to make with pictures if possible. This way we can learn what works best with new innovative ideas. Also please state where you're buying your new parts from, this is very important because there are many fakes on the market. Almost* everything on my list was bought from either Mouser or Digi-Key USA, this way as a community we'll know that the sound performance we get is from a genuine part and not a fake.

My personal volume listening range is marked in Red and my Klipsch 8Ω speakers are very efficient at 95dB, thus the reason I chose the 10 uH inductors. If you like to play your music louder/harder Or you have less efficient speakers which requires you to push the amp harder, then you might want to consider a 7 uH inductor with the appropriate capacitor uF rating.

[0%----I-----25%----------50%---------75%----------100%] Of the usable volume range.

Texas Instruments LC Filter Design:
Page #17 (3.8 LC Filter Quick Selection Guide)
https://www.ti.com/lit/an/slaa701a/...A%2F%2Fwww.ti.com%2Ftool%2FLCFILTER-CALC-TOOL

8Ω Efficient Speakers Driven Light To Medium Hard = 10 uH Inductors & 0.47 uF Caps / 8 Ω Speakers Driven Medium Hard To Very Hard = 7 uH Inductors & 0.47** uF Caps

6Ω Efficient Speakers Driven Light To Medium Hard = 10 uH Inductors & 0.68 uF Caps / 6 Ω Speakers Driven Medium Hard To Very Hard = 7 uH Inductors & 0.47 uF Caps

4Ω Efficient Speakers Driven Light To Medium Hard = 10 uH Inductors & 1.0 uF Caps / 4 Ω Speakers Driven Medium Hard To Very Hard = 7 uH Inductors & 1.0 uF Caps

"From the results of the inductor study with the TPA3251 device, it is clear that the inductor plays a large role in the audio performance and total system power dissipation. Inductance value, DCR, linearity, and core loss factor into the total system performance and must be considered based on the design goals of a specific application. For the TPA32xx family of devices, a 7-µH inductor is a better value for performance-oriented applications due to improved linearity and generally improved distortion performance over higher inductances. However, for a more power-efficient system, 10-µH may be a better selection due to the reduced ripple current with a slight penalty on performance. In both cases, core loss must be considered as well as DCR. Although a 10-µH inductor may show improved power dissipation for low output power due to reduced ripple current, if it has high DCR, the losses may be greater at high currents than a 7-µH inductor with higher ripple current."

Humble Homemade HiFi Capacitor Test:

This is an excellent guide for choosing a capacitor.

Humble Homemade Hifi - Cap Test

Bypass Capacitors:

"A bypass capacitor is a capacitor that shorts AC signals to ground, so that any AC noise that may be present on a DC signal is removed, producing a much cleaner and pure DC signal. A bypass capacitor essentially bypasses AC noise that may be on a DC signal, filtering out the AC, so that a clean, pure DC signal goes through without any AC ripple."

What is a Bypass Capacitor?

"Roughly speaking", the formula for choosing a bypass capacitor is 1% of the total of the capacitor you want to bypass. If you own an oscilloscope, that would be the best way to choose a bypass cap. Having said this, most use a 0.1 uF OR 0.01 uF bypass cap.

************************************************************************************************************************

My Modification Build List:

1) Op Amps, OPA1656 (2)
Elna RFS SILMIC II, 10 uF 25v Plus a Russian K42Y-2 PIO 10%, 0.1 250v as a "double" bypass cap combo on Op Amp pin 4 (Neg.) & pin 8 (Pos.)

2) Wurth Elektronik 7443631000, 10 uH Shielded Inductors mounted on their sides to save space. (4), Location = L2, L4, L6, L8

2a) Coilcraft VER2923-103KL, 10 uH Shielded Inductors mounted on their sides to save space. (4), Location = L2, L4, L6, L8
At this point I will only recommend the Wurth Shielded Inductors, the Coilcraft are to big and mounting them will be difficult.(See post #4)

3) Nichicon LKG Gold Tune, 3300 uF 50v, LKG1H332MESZBK*** (2), Location = C36, C37 (High Ripple Current at 2.8A)
Wima MKP 10, 0.1 uF 250v OR Russian K40Y-9 PIO 10%, 0.1 200v (2), Location = C36, C37 as a possible Bypass Cap.

4) Russian K42Y-2 0.47 uF 160v PIO 10%. (4), Location = C32, C33, C38, C39
Wima MKP 10 OR Rifa PHE426 0.01 uF 250v (4) Location = C32, C33, C38, C39 as a possible Bypass Cap.

4a) Russian KZK K78-34 0.47 uF 250v 5%. (4), Location = C32, C33, C38, C39
Wima MKP 10, 0.01 uF 250v OR Russian K40Y-9 PIO 10%, 0.01 200v, (4) Location = C32, C33, C38, C39 as a possible Bypass Cap.

4b) Wima MKP 10, 0.47 uF 250v, MKP1F034704J00JSSD, (4), Location = C32, C33, C38, C39
Russian K40Y-9 PIO 10%, 0.01 200v, (4) Location = C32, C33, C38, C39 as a possible Bypass Cap.

4c) Kemet (Rifa) PHE426 0.47 uF 250v, PHE426HB6470JR06, (4) Location = C32, C33, C38, C39
Russian K40Y-9 PIO 10%, 0.01 200v, (4) Location = C32, C33, C38, C39 as a possible Bypass Cap.

5) Elna RFS SILMIC II, 22 uF 25v, (4), Location = C5, C10, C15, C20
Russian K40Y-9 PIO 10%, 0.01 200v (4), Location = C5, C10, C15, C20 as a possible Bypass Cap.

6) Meanwell LRS-200-36, 36 Volt / 5.9 Amp Switching Power Supply. (1)

7) Nichicon LKG Gold Super Through, 2200 uF 50v, LKG1H222MESACK (1), Smoothing Capacitor for the positive & negative output terminals of the Meanwell LRS-200-36 PSU. Bypass capacitor choices will be the Wima MKP 10, 0.1 uF 250v and the Russian K40Y-9 PIO 10%, 0.01 200v.

8) Volume Potentiometer, I'm totally open to suggestions because the stock Aiyima could be better. Size of course is a major concern.

9) OFC 20 AWG (2x = 17 AWG) pure copper build wire and silver solder.

10) Everything gets plugged into my Panamax M4300-PM surge/power conditioner for cleaner power.



Rob43


* The K42Y-2, K78-34, & K40Y-9 capacitors come from Russia.

**This is a guess, I'd also consider trying a 0.39 uF capacitor. Possibly a Kemet (Rifa) PHE426 0.39 uF OR KZK K78-34 0.39 uF 250v in this location.

*** This capacitor is 50mm's tall and will require case modification to fit.
 

Attachments

  • thumbnail (2).jpg
    thumbnail (2).jpg
    167.9 KB · Views: 5,074
  • thumbnail_(20)_-_Edited_(2) - Edited.png
    thumbnail_(20)_-_Edited_(2) - Edited.png
    576 KB · Views: 5,167
  • thumbnail_(18)_-_Edited_(4).png
    thumbnail_(18)_-_Edited_(4).png
    717.5 KB · Views: 5,289
  • thumbnail (1).jpg
    thumbnail (1).jpg
    113.7 KB · Views: 5,134
  • basics1_-_Edited.png
    basics1_-_Edited.png
    24.3 KB · Views: 4,655
  • thumbnail_(27)_-_Edited.jpg
    thumbnail_(27)_-_Edited.jpg
    125.9 KB · Views: 2,109
  • Screenshot_2020-06-30_at_4.35.53_PM_-_Edited.png
    Screenshot_2020-06-30_at_4.35.53_PM_-_Edited.png
    108.3 KB · Views: 2,105
  • thumbnail_(23)_-_Edited_(1).jpg
    thumbnail_(23)_-_Edited_(1).jpg
    114.1 KB · Views: 2,079
  • thumbnail_(22)_-_Edited_(1).jpg
    thumbnail_(22)_-_Edited_(1).jpg
    106.7 KB · Views: 2,166
Last edited:
Single question Rob43
Finally which is the total cost of the final update ?

I share again the 360 Customs Inductors test (TPA325X) so it can help :

inductor_cmp.png



I also add this PSU from Meanwell that performs very well too :

https://www.mouser.fr/ProductDetail/MEAN-WELL/ELG-240-36A?qs=rJnG6vTwR7RLtfNoQ6SlQA==

An externally hosted image should be here but it was not working when we last tested it.


Another superb alternative for PSU (same used in the famous BOSC Gan modules) :

Murata PQC250-36

https://www.mouser.fr/ProductDetail/Murata-Power-Solutions/PQC250-36?qs=AQlKX63v8RvHt6DepqpADg==

An externally hosted image should be here but it was not working when we last tested it.
 
Last edited:
Moving with pleasure to this thread from the general thread… THANKS Rob !


What follows is often close to what Rob has already posted, so hopefully just confirming or infirming a similar tweaking idea for the community.


We have 2 of these amps, internaly indentical sounding indeed identical. We found the volume pot to be a weak point and, as we wanted a remote control and a single volume control for both amps (bi-amping is the goal as that sounded better) we decided to go for an external pot. We have yet to remove the existing tiny low quality pots and go for a straight signal path instead and tell the sonic difference, at this stage the baseline are the original amps with an additional Alps blue pot (not the best but well).

The amps are fed by a Meanwell SMPS, LRS-350-24. This adjustable unit allows to go up to 29V (from a soundpressure POV you don’t lose anything really) and has with 150mVpp a much lower ripple current / better specs than the next unit with 36V. With 350W it has enough power to drive 1 or 2 units.

The list of mods we intend to do on this set up are, in random order :

A- Fit a home made filter between SMPS and power amp(s), various possible configurations

B- Increase the PS capacity with external caps after our filter… just to give it a try given we will get some unused desoldered 3000uF internal PS caps ;-)

C- Replace the existing op amps with OPA1656 (I have many other “flavours” ready to go in case of final fine tuning, but that’s clearly our first choice)

D- Replace/discard the volume pot as we feel that’s a potential bottleneck

E- Install Polypropylene DC blocking caps (Panasonic ECW, 4.7uF) and get rid/bypass the relevant electrolytics at the entry of the op amps and between op amps stages (C4, C11, C16, C19)

F- Replace the electrolytics (C5, C10, C15, C20) at the output of the op amps stage with 1uF WIMA MKS quality caps, as that should do the trick with the (roughly) 24k power amp input impedance that follows

G- Bypass various smaller PS electrolytic caps with either Murata 0.1uF (C0G, had to go for 0.082uF in fact due to the 0.1uF being in back order) or 2.2uF Murata X7R quality caps, more details to follow as quite a few

H- Replace the 2 big electrolytic PS caps (C37, C38) with high quality low ESR ones (Nichicon UBY series, 3000uF/35V)

I- Bypass the 2 big PS caps (C37, C38) with quality 2x2.2uF Murata X7R caps each

J- Replace the 4 output filter coils (L2, L4, L6, L8) with quality coils (Coilcraft MA5172AE) having other values matching the LS impedance

K- Replace the 4 output filter caps (C32, C33, C38, C39) with quality caps (real Epsons, as I suspect the ones in the unit are fakes) having other values matching the LS impedance

L- Resolder various locations starting with the awful output connectors


M- We tried my trustfull big size passive RCL main filter in front of the Meanwell SMPS and that brought absolutely no change despite filtering in both directions (preserving other units from junks), so we moved on to cleaning the output of the SMPS (point A)


ALL FINDINGS AND RESULTS ARE / WILL BE IN OUR VERY HUMBLE OPINION, TO OUR EARS, AND NOT WILLING TO GET INVOLVED IN ANY DEBATE.


The sole purpose is to point the best VFM mods, rank them in magnitude if possible and describe the sonic benefits they give so you can spice your system accordingly to taste and budget.

All parts were purchased at Mouser's except the motorised Alps potentiometer. The total for Mouser parts is roughly 75E/unit - yeah, we did buy some extra parts for experimenting, that's part of our fun, but you won't have to of course :)


Claude
 
So, that's our play ground, pix courtesy of Gilles.

Yes, there are ways to put bigger PS caps into the unit having them lying sideways, yes it is probably possible to fit quality big shielded coils (same price as ours) sideways or upwards, fiddling a lot with their connections and hot glue in hope to close the lid and having them not moving around etc.

However, we felt we should go for 90% quality and 20% hassle rather than messing up, so at the end we found very acceptable coils (that are unlikely to saturate given bi amping and high efficiency speakers). We also felt it would be nice if all was a tight fit, but still a nice feet without to many connections lying around and putting beginners off. Last but not least we felt it would be nice to have all the mods reversible and not very visible, not that we want to sell on the units one day, but at least easily reversible so we can play without having to stick with too much proudness to our inefficient mods should it be the case :)

Therefore we went for similar sizes and locations (except the external 'additonal" master volume pot) and we went the trouble finding bypass caps that could fit... under the board while keeping of course the casing untouched. It is a choice, there are of course other options, even for us as all this comes in a bigger old amp housing anyway.

Oh, and Gilles had an excellent idea. If you are not confident when fitting parts under the board, you can still add a very very thin isolation sheet on the casing's underpart to avoid potential shorts. It is true I didn't bother, but it is indeed a nice idea as the casing is conductive despite the paint and all underneath additional bypass caps will be PS ones, so defo grounding them wouldn't be nice!

As a reminder from another thread, we will increment the mods step by step, having 2 absolutely identical (tech and sound) amps connected such a way that we can move very quickly from an amp to the other (not modded) to drive the Klipsch RFP8000F - or of course to move to the final bi-amping- enables easy direct comparisons.

Let's get started...
 

Attachments

  • IMG_20200724_193337.jpg
    IMG_20200724_193337.jpg
    692.8 KB · Views: 1,497
  • IMG_20200724_193100.jpg
    IMG_20200724_193100.jpg
    748.8 KB · Views: 1,364
  • IMG_20200724_192816.jpg
    IMG_20200724_192816.jpg
    704 KB · Views: 1,492
  • IMG_20200705_154940.jpg
    IMG_20200705_154940.jpg
    690.8 KB · Views: 1,800
  • Capa.pdf
    463.8 KB · Views: 381
Last edited:
First tweak, “A” in my list, is an external SMPS filter of mine, between the Meanwell SMPS and the amps.

To start with, the Meanwell SMPS we went for has 150mVpp ripple, whereas the 36V up the ladder model has +40% advertised junk. So our PS is not doing a bad job at all to start with and is IMHO probably very suitable for Class D with quick current needs, not to mention it has enough oomph with 350W, probably more than our ears could bear with the Klipsch regardless distance. My understanding is it includes already a small output filter. Although not a fan of SMPS myself, I was learned better by Papa and Mark’s SMPS filter, so we thought with some tuning and clever choice we might well compete with linear devices for such a Class D amp, linear PS that would anyway… not fit into the old amp’s casing we have chosen. Not to mention price and convenience.

The beauty of an external filter is it is easy to add and take off, plus you don’t need any skills once it is assembled. I am not overly confident posting in details what I did here, because:

- these 2 filters (can be one in front of each amp, or daisy chained) are prototypes, designed with rough rules of the thumb and no proper professional calculations, too time consuming and too many variables otherwise

- Lucky me, while thinking about RCRC and LRCR, Mark Johnson came up with a very clever SMPS filter for SMALL power needs. A very clever RLC filter I shamelessly took inspiration of, thanks again Mark

To understand what it is about I would say have a look at this very professional thread

PO89ZB , an inline DC filter for SMPS wall warts . Preamps, HPA, Korg NuTube, etc

My variation/prototype is for high power devices. It is half of this (double) RLC filter with instead an overall 0,05R damping resistor, a Bourns 1.5uF /5A coil damped in parallel by a 1k resistor, C being in my case a huge very low ESR 3000uF cap (Nichicon UBY series, 3000uF/35V). If you don’t follow all this, please have your own filter built by someone else, there is quite some energy at play!

The nice bit is it adds also PS capacity and it opens further doors for even more outboard PS caps… should the need arise once we have done all the internal PS tweaks we have in mind.

It is dimensioned for effortlessly 5A, cost less than 10$ to build and, depending on the C value, has a cut off frequency below 2kHz with a steep filtration curve. Should in real life (assembling quality = always less than simulation) dampen the SMPS junk by well over 50dB.

Anyway, only results matter, and these are:

- Wider soundstage +1

- Deeper soundstage +1, with also more precision in the instrument positioning

- Spatial airiness +2: as flying in a bigger room but without losing focus, enjoyable!

- Middle frequencies / voices +1: more realistic sounding, more presence/body

- Bass +1, more finely defined and better body in the lower bass

Overall perception is also a somewhat cleaner and more precise sound with a better register integration and better flow, somewhat more natural.

There is no downside, but due to what is perceived as a less distorted sound you may initially perceive a tad less energy / speed on impulses. In direct comparison it turned out energy/impulses were absolutely unchanged, but blur and extra harmonics could make some rock music sound that way. At the end taking the filters away was perceived as a clear negative so they stay.

As of the magnitude of the changes, we have to admit it is not huge: say perhaps +1.5 overall in our scale, close to changing a no name electrolytic DC coupling cap with a top electrolytic one. You hear it clearly, but you do need to listen carefully. In terms of VFM it is still OK, such a filter cost less than 10$ t build and is the icing on a good SMPS, but let’s be honest: it isn’t a world beater mod or a must do.

It will be interesting to see if the differences are quite small because the existing low ripple SMPS is already good performing and with this filter we are close to what can be achieved re external SMPS regardless the PS technology FOR THE AIYIMA (the sonic quality tend to make me believe so), or if it is because SMPS are crap and my filter also (I didn’t like SMPS but was taught a lesson recently). Or perhaps all this is not “that” relevant at that level given the PS rejection capability of the TI amp chip, helped also by the existing PS caps? Anyway, we have some other mods to perform re PS caps and if sensitivity is low whatever we try then it will probably mean it is more profitable tweaking elsewhere.

Enjoy music

Claude… and Gilles
 

Attachments

  • 20201101_132011.jpg
    20201101_132011.jpg
    799 KB · Views: 1,532
  • 20201101_132020.jpg
    20201101_132020.jpg
    789.8 KB · Views: 1,542
Moving on to tweak C, OPA1656. Those reading me in the past know I warn from blind op amps rolling-op amps are designed for a given application field and perform accordingly, even gain alter sonic quality, hence performing differently in various set ups. On the other hand, I was lucky enough to have free samples at product launch and was overwhelmed by its performance in various configurations, beating my old beloved AD825 biased in class A and many other more recent op amps. Needless to say I purchased (at Mourser’s only) many of these OPA1656 since and was never disappointed regardless the application, from RIAA amp to high impedance HP amp.

To put it short, to our ears our self-made boards seem to perform very well in all areas without any real downside. Bass, mids, treble, flow, music integration, 3D soundstage, airiness, transparency, details, richness of tones, delicacy… the list is very long and magnitude in progress is each time big. I recommended several months ago to replace the existing op amp (probably a fake 5532) and to give the OPA1656 a go in that Class D amp… and the rest is really past history now, so popular and straightforward that mod has apparently become since (whereas I hadn’t tried it for myself not owning that Class D amp myself, so time for me to catch up!). Of course, depending on your hifi system and taste, you could still prefer a darker or more bass heavy sound (it is VG in the bass, but not bass biased whereas more common Op amps as the one fitted tend to be, mainly because other registers don’t shine) etc. and there are (we have…) other op amps flavours, but as of me I prefer to use what I consider today’s best, IMHO competitive with discrete stages, and adjust flavours elsewhere should I still want it to. What is lost at op amp level is lost forever, whereas I prefer adjusting elsewhere at a final stage using a transparent and neutral addictive op amp as baseline to evaluate progress. Should that not be enough, then I have anyway a flavour box with compatible op amps with me J

So being late to the party, I am only listing this mod here because most of you went for it (and if not already this is a must do and possibly one of the best VFM mod for this amp) and because Gilles and I use a point system to rank mods based on their perceived magnitude. I thought it would be good for you to know how many points we give to this mod so you can evaluate other mods based on this popular benchmark. In short all the above applied, making it an excellent upgrade worth to our ears +5 points overall. In my experience this is kind of lifting this Class D amp at very least 1 full class up the ladder if you would follow traditional product lines from specialised amp manufacturers. Call it lifting a heavy bottleneck. Needless to say both amps are now fitted with the OPA1656 modules and that will be the new baseline for us.

Enjoy music

Claude… and Gilles
 

Attachments

  • 20201101_131925.jpg
    20201101_131925.jpg
    868.1 KB · Views: 1,527
I worked on the Op Amp bypass caps late tonight but must have been more sleepy then I realized.

I unfortunately mixed up pins 4 & 8 and soldered things in backwards... :(

Luckily I caught it in time, I'll fix things up tomorrow.


Rob43
 

Attachments

  • thumbnail_(21)_-_Edited_(3).jpg
    thumbnail_(21)_-_Edited_(3).jpg
    86.1 KB · Views: 1,476
  • thumbnail_(22)_-_Edited_(2).jpg
    thumbnail_(22)_-_Edited_(2).jpg
    99.6 KB · Views: 1,413
  • thumbnail_(1)_-_Edited_(4).png
    thumbnail_(1)_-_Edited_(4).png
    961 KB · Views: 616
Hi Daniboun,

Nice to meet you "here"...

Thanks for posting the various PS tips, always welcome, especialy as that unit doesn't come with one!

While it is impossible to predict sonic qualities on specs only, the first one you mentioned has the merit to be ready to go so easy to use for people not liking to fiddle with casing and connections!

Just as a side note, I went some trouble finding a "on the paper" suitable SMPS. The one I chose is oversized in terms of overall power -you could pick the same unit with less power if driving just 1 amp- but in either case in direct comparison of Meanwell's own spec sheets, the one you mentioned that comes ready to go in a casing has according to Meanwell 67% more ripple and noise than mine.

That ripple and noise figure is quite a critical criteria for PS, and as both units come from the same manufacturer it is probably not far fetched to believe evaluation was done following the same process. In short, that SMPS performs OK to good on the paper, but following Meanwell they have better products in store.

I would say if my filter worked a bit on my SMPS, then it should even more on that one, but then given the choice why not picking up a better spec'd SMPS at Meanwell directly? Having said that, other than PS ripple and noise it is indeed a nice quality SMPS from Meanwell, no worries.... and it is ready to go.

Regarding the Murata one, it looks indeed also very professional. Have you tried it? Specs are just that and when trying to compare different manufacturers it is a bit difficult as they not necessarly use the same measurement process or referential.

In the case of the Murata, I hope the built-in safety doesn't trigger when starting up the amps with additional PS caps in our applications. Perhaps their C-variant is a better bet?

Further, their advertised "ripple and noise value" is very high, up to 360mVpp if I translate their unusual percentage value right. That doesn't mean it is always so high (again, how do they measure... eventhough people should diverge too much on that simple criteria), but the lack of other indication is not a great sign (my quick look didn't allow me to find their operational frequency either, although some graphs could help deduce it roughly). Odd coming from such a good company - for comparison, the unit we use has on this very important criteria (less is better) advertises up to 150mVpp. Advertised values only of course, but that means the Murata could have potentialy 2.4 times more junk in terms of PS than our Meanwell SMPS... not a small difference on the paper and not very encouraging for such an expensive device I must say, until someone measures of course both units. Let's not forget they are both not designed to drive audio amps, so might be customised differently... To mitigate my words though, Murata does have a real life figure ripple and noise measurement on the spec sheet, but it is for their 12V model only. For whatever it is worth, in their specific conditions, that measured 75mVpp... a much more encouraging value indeed.

I hope these comments help and by no mean do I intend to put various PS proposals off, on the contrary! I just went for a given unit after looking at specs and never looked back: with more care and budget there must be other options to try!

Enjoy music

Claude
 
Hi Rob!

That looks great, well done!

Not easy to fiddle with such a limited space inside the unit : bravo.

It is also an interesting new path, it will be great to read how it performs.

We decided to go a much easier route, having off board caps at unit entry and then bypassing completely the existing caps (except the ones at power chip entry).

Your choice is great because thanks to your small mouting board you will be able to carry out many different quick caps comparisons / listening sessions without damaging anything: great idea I must say!

Read you soon

Claude
 
Last edited:
Very interesting.
Im having this Allo card between my mean well power supply.
The ripple is less than 3 mv.
The Capacitance multiplier is functioning like it has 1000 000 uf in noise rejection.

You can buy it at Audiophonics and directly from Allo. The price is about 17 dollars.
The Capacitance multiplier can take 30 volt and it can give 5 A . The ripple is less than 5 mv at 3 A output.
https://www.allo.com/marketing/brochures/allo-capacitance-multiplier-brochure.pdf

You loose some power, meaning 30 V becomes 29 V, or 24 V becomes 23 V .

How does it sound ?
Well, Its a mix of better and for worse. When you listen to good classical recordings with the Capacitance multiplier its like moving back the listening position in the concert hall about 5 meters. That can turn off the exitement somewhat whit rock music.

And by the way - the ne5532 thats inside the Aiyima a04 tpa3251 is NOT a fake one, as the community of faktiskt.io has discovered in sweden. A professional has been investigating the ne5532 and its a real one. The measurements below compairs two of the Aiyima a04 ne5532 with one lm4562.
But ...
The sound gets clearly better with for example a lm4562 and Im sure an opa 1656 is sounding better.
 

Attachments

  • FD28CC37-C40C-4C49-B552-ECF15F733331.jpeg
    FD28CC37-C40C-4C49-B552-ECF15F733331.jpeg
    78.9 KB · Views: 778
  • 618F7CF7-DC07-44F6-B0B8-FFAD80886D0F.jpeg
    618F7CF7-DC07-44F6-B0B8-FFAD80886D0F.jpeg
    72.1 KB · Views: 719
Last edited:
Yep, I read about the cap multiplier and even a report somewhere: was an option indeed, but I expected mixed results from this technology (for various reasons,) so decided to go another... unproven path, GLOL!

It could perhaps be fun if you want to explore that path aswell to compare to your cap multiplier vs my filter... or even perhaps once you have them why not trying a line like Meanwell / my filter / cap multiplier / and finaly amp PS input? That could perhaps be the winning combo and avoid mitigating results.

As of me, we still have 3 tweaks pending "in this field":
- the pending one, adding bypass caps to the existing PS caps in the unit
- replacing the existing big PS caps with hopefully better ones
- adding shamelessly capacity to the PS (but won't be as radical as you did)

As of the 5532, it is not a bad amp and it measures well in most areas, is safe, but the world has indeed moved on since a few decades re op amps.

BTW, a lot of parts "COULD" be fakes inside the unit (I will tell once taken out), but :
- fake doesn't necessarly mean bad performing for the purpose!
- at the end of the day, fake or not, the complete assembly works pretty well and sounds good, so either real, or fakes doing their job, or fakes and real well matched together with the righ receipe for a nice sound! Who cares...

I view my task more as finding potential bottlenecks that are quick fixes to make a good unit even better. Read mainly research areas around op amps, potentiometer, PS, output filter etc.

Fascinating (borrowed from M. Spock)

Claude
PS: re Aiyima A07 no clue, never saw one, as of the filter I see no reason it shouldn't, as of op amps you really need to know how that unit works, schematics preferably and investigate etc.
 
ClaudeG: I will try to build the filter you have made, the capacitance multiplier was not entirely positive sound-wise. The tpa 3251 has a PSRR of 60 dB. Texas instruments are recommending only 1000 uF for every halfbridge in the amplifier. Maybe the quality is of greater importance, than the size of the capacitor ?

Switching to lm4562 was, for me at least, an upgrade in every aspect of the sound. More dynamic, more clear sound. I will try to investigate further with opa 1642 that I will receive in a couple of days. A guy at the swedish forum are reporting that the sound is enhanced further with that OP-amp. I will report back here, later.
 
Last edited:
Don't expect miracles from the small filter, but then cost is low for an experiment.

Cap multipliers and mitigated results... Many uF for little money, if it would be that easy, we would all do it, wouldn't we? And with huge values it would blow the PS at start up LOL.

I have these cap multiplier as intermediate filtering in a Class A RIAA amp inside a very old H-K integrated amp, where it is wiseky used as low power and constant demand, so nothing wrong with it per se. Prob is you still need some capacity before the Allo cap multiplier to make sure you have enough oomph available, but that is often not too bad as included into a good SMPS at its output. But you also need some capacity at the output of the cap multiplier to make sure it does provide current in case of a surge (and not just to fill the small waves of a nearly constant demand) or when encoutering... big variations due to the fed devices requiring lot of current very quickly. In that matter (read class AB or D amps), it is far from 1000 000uF in terms of ESR and reserve, a lot of people prefer usualy caps banks with just 40 000uF total capacity against it.

These cap multipliers seem though IMHO very good when it comes to (not too high) power applications where the current needed is quite constant and demand is low (and variations also). Read Class A, low power and such.

Where it does act as advertised (equivalent lot of uF) is in fact IMHO for ripple reduction with constant load / demand, when there is dynamic demand with big surge they stand no chance against real caps IMHO. Class D chips seem to pump quite hard and quickly as I understand it... hence me going another route.

So yes, in opposition to usual amps, Class D amps might benefit from "hard" PS, hence SMPS and possibly, given some acceptable values, capacity figures being less important than say an excellent ESR across the entire required PS frequency band by the fed device. That's more the route I went in my mind. Speak not trying to increase C a lot, but chose caps with ESR & Co wisely. So as you said, probably "quality" rather than quantity.

Re op-amps, give OPA1656 a chance. It is not expensive. There are op amps with slightly better specs, but this op amp was designed with audio in mind for stages like in this amp and it manages then to be both transparent and musical. For other application in some intermediate stages on DACs etc., extremly neutral op amps can be better of course, but here it shines IMHO... Note I can also understand some people prefering even slightly coloured op amps or tubes in front of Class D to fight some of its coldness (in some cases)... then OPA1656 might not be the best choice, just a very good one.

Have fun!

Claude
 
Last edited: