• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Fosi Audio V3 op-amp rolling, has anyone tried it? Snake oil? Or are there actual differences?

I have a practical question about the Fosi V3 amplifier - is it technically correct to solder a wima mkp10 0.15uF in parallel with 63v/2200uF capacitors? Doesn't this increase distortion in class D amplifiers?
 
I have a practical question about the Fosi V3 amplifier - is it technically correct to solder a wima mkp10 0.15uF in parallel with 63v/2200uF capacitors? Doesn't this increase distortion in class D amplifiers?
It might cause some parasitics. It might cause physical damage. Or it might have no effect. It's a fashion tweak.
 
It might cause some parasitics. It might cause physical damage. Or it might have no effect. It's a fashion tweak.
I've already done this and there was no damage, but subjectively I liked the sound with Wima more - now I've returned to the factory version without them and the magic is gone!)
 
I've already done this and there was no damage, but subjectively I liked the sound with Wima more - now I've returned to the factory version without them and the magic is gone!)
Almost certainly placebo/sighted bias.
 
I've already done this and there was no damage, but subjectively I liked the sound with Wima more - now I've returned to the factory version without them and the magic is gone!)
Wima's of that value aren't as good for decoupling a power supply rail than some lower value ceramic caps around 10nF or compound using 1-10uF in parallel to it near the power pins of opamps or so which can be confirmed by RF measurements in the MHz to GHz range on the power-pins. It is pointless to decouple reservoir caps themselves because of the inductance of the traces going to the circuits.
That said, most likely the wima does absolutely nothing for technical performance nor will the measurements improve but knowing they are in there WILL do something (positive) in your brain and if that's your magic...by all means solder them in and once again enjoy the magic. I am all for that but won't do anything for me... That's how perceptual bias works. Enjoy it and or playing with more expensive caps others claim bring even more magic. As long as you believe it will work for you.
 
Last edited:
Wima's of that value aren't as good for decoupling a power supply rail than some lower value ceramic caps around 10nF or compound using 1-10uF in parallel to it near the power pins of opamps or so which can be confirmed by RF measurements in the MHz to GHz range on the power-pins. It is pointless to decouple reservoir caps themselves because of the inductance of the traces going to the circuits.
That said, most likely the wima does absolutely nothing for technical performance nor will the measurements improve but knowing they are in there WILL do something (positive) in your brain and if that's your magic...by all means solder them in and once again enjoy the magic. I am all for that but won't do anything for me... That's how perceptual bias works. Enjoy it and or playing with more expensive caps others claim bring even more magic. As long as you believe it will work for you.
They do something, because when comparing the sound with other class AB amplifiers, the "puff" when instantly switching with the selector button (between two amplifiers for one pair of speaker systems) is heard more clearly than in the version without Wima and at the same time Fosi goes into protection and switches off a little earlier than in the basic version.
 
They do something, because when comparing the sound with other class AB amplifiers, the "puff" when instantly switching with the selector button (between two amplifiers for one pair of speaker systems) is heard more clearly than in the version without Wima and at the same time Fosi goes into protection and switches off a little earlier than in the basic version.
If you do this without basic controls, what you have there is a brain trick having nothing to do with the actual sound.
 
If you do this without basic controls, what you have there is a brain trick having nothing to do with the actual sound.
of course, that's why I asked the question, which I'll repeat again - does this theoretically improve or worsen the sound in class D amplifiers?
 
It is not Class dependend...
It depends on the OP amps used. Some degrade audio, some don't have any measurable effect, others might be untested / undecided....
As the last instance the personal subjective impression might be the "real" decider. If that is physically "real" is another question...
 
Last edited:
of course, that's why I asked the question, which I'll repeat again - does this theoretically improve or worsen the sound in class D amplifiers?
It doesn't do anything consistent to the Class. The answer will be specific to the amplifier you are making the mod to (regardless of class). And no-one can answer it unless they measure the results, or do the detailed design calculations.

Even controlled listening can't answer the quesstion - it can only determine if there is an audible change. Not if that change is for the (objective) better or not.
 
of course, that's why I asked the question, which I'll repeat again - does this theoretically improve or worsen the sound in class D amplifiers?

The question is too generalized. Most likely it won't do diddly-squat. There's a small risk of it forming an unintentional resonance circuit, but you'd need to do measurements to verify whether or not it happens in this specific case. It also helps to make a schematic before you just start soldering in more components. If Fosi already bypass the electrolytics with SMD ceramics, then the Wimas absolutely won't do any good.

EDIT: @antcollinet beat me to it :D
 
The question is too generalized. Most likely it won't do diddly-squat. There's a small risk of it forming an unintentional resonance circuit, but you'd need to do measurements to verify whether or not it happens in this specific case. It also helps to make a schematic before you just start soldering in more components. If Fosi already bypass the electrolytics with SMD ceramics, then the Wimas absolutely won't do any good.

EDIT: @antcollinet beat me to it :D
That's right, but we all know that Fosi doesn't have the best schematic solution for the TPA 3255 - there are already better measured amplifiers and most likely even better ones are in the pipeline. There is no need to convince me of the usefulness of bypassing electrolytic capacitors, in Sansui and Accuphase amplifiers (and not only) this has been used for a long time, and in top models it is even multi-stage. I am interested in using this trick specifically in class D due to the specifics of their operation, which differs from traditional amplification classes.
 
I am interested in using this trick specifically in class D due to the specifics of their operation, which differs from traditional amplification classes.
And what we are pointing out, is there is nothing in those specifics of Class D, that means "this trick"* applies more or less to class D than it does to any other class.

It will always be design specific - starting with the question "does the design already use a higher frequency capacitor in that position"


** and it is not really a trick. Just a well known and often used part of capacitive decoupling.
 
I am interested in using this trick specifically in class D due to the specifics of their operation, which differs from traditional amplification classes.

I assume you're talking about improving the output filter?

Again, doing these modifications is a shot in the dark without measurements for verification. You'd might be at less risk of f#cking things up on this particular model, since the filter isn't included in the feedback loop, but it's still a futile exercise. Doing verification with sighted listening tests is at a way too high risk of leading to false conclusions.

First step would be to show measurements of the stock filter giving a clear indication of it being inadequate. There's no sense in modifying things if they don't do anything wrong.
 
That's right, but we all know that Fosi doesn't have the best schematic solution for the TPA 3255 - there are already better measured amplifiers and most likely even better ones are in the pipeline. There is no need to convince me of the usefulness of bypassing electrolytic capacitors, in Sansui and Accuphase amplifiers (and not only) this has been used for a long time, and in top models it is even multi-stage. I am interested in using this trick specifically in class D due to the specifics of their operation, which differs from traditional amplification classes.
As others have explained, this is not FOSI specific, or Class D specific. Using a bypass capacitor on the main electrolytic supply bank is probably not beneficial, because you need to know the layout and length of the rails which come after. You may make things unstable.

The only way to do this is with test equipment that can measure up to RF, to ensure you are not creating an unintended resonant circuit. You should also measure amplitude response, THD, IMD and noise before and after to determine any benefits or disadvantages.

BUT, in general, randomly soldering components into a circuit without measurements is more likely to degrade it than improve it. You may prefer the degradation, if course.
 
There is no need to convince me of the usefulness of bypassing electrolytic capacitors, in Sansui and Accuphase amplifiers (and not only) this has been used for a long time, and in top models it is even multi-stage.
But does it actually do anything there or it it for show? Where I've encountered it, it was the latter.

Where bypassing is useful (rare in audio with modern electrolytics), it is done at the actual circuit, e.g., directly at the power pins of the opamp, not across the electrolytic itself.
 
That's right, but we all know that Fosi doesn't have the best schematic solution for the TPA 3255 - there are already better measured amplifiers and most likely even better ones are in the pipeline. There is no need to convince me of the usefulness of bypassing electrolytic capacitors, in Sansui and Accuphase amplifiers (and not only) this has been used for a long time, and in top models it is even multi-stage. I am interested in using this trick specifically in class D due to the specifics of their operation, which differs from traditional amplification classes.
In class A, AB, G and H there is no fast switching involved and usually one can get by with smoothing caps and the occasional decoupling near the power and/or voltage amplification stage to prevent instability.

For switching amps just having large electrolytics isn't enough because these may (PCB) design wise too far away from the power-switching devices. Even 10cm distance can be problematic due to the inductance of the traces/wires. Certainly if the device also has to comply to EMC/FCC rules.
A small capacitor parallel on those reservoir caps won't do anything up to say a MHz or so but can help in the 10Mz-10GHz range at the point where the switching power devices are located.
Small capacitors need to be charged/discharged quickly and with sharp rising/falling edges requiring high (narrow=HF) frequencies far, far above the switching frequency.
This requires local decoupling and (for RF) further removed caps like reservoir caps the inductance simply is too high so decoupling away from the switching components that draw the current peaks decoupling there will do nothing beneficial for the amp itself.
For that reason the decoupling needs to be done near the switching devices directly to a groundplane with as short as possible wires.
Compound is often needed using 10uF and 100nF or so (or something smaller) and need to be small in size. Larger sized caps (think WIMA) have too high ESR at GHz range.
Where you do see those used is at the output filter because of the large voltage swing across it where CML are poor performers.
At a constant voltage (power rail decoupling) the CML ones perform better than WIMA.

It is the 'audiophool myth' world that made the WIMA big in analog amps because they supposedly 'sound' better than other PP caps.

So while you may hear improvements and magic decoupling reservoir caps it won't do much (if anything) for the decoupling of the output devices as that has to be (and probably even is for cheaper designs) done locally near the output devices.

You need to measure in the GHz range to show these effects and is mostly about emission of RF that has to meet requirements and the output stage being fed from a low impedance voltage rail exactly at the pins of the powerIC/switching devices. Not at the reservoir caps.
 
Last edited:
Back
Top Bottom