That is what I want to say to him Marcel. Of course you are true for shieldings and I will not permit to disagree with you as you are a valuable EE and I am not ( @wizeoldtech : the nope was more a joke in spite of the no I said to yoshimitsu that has really no clue...if we can not laugh anymore 😉 )
Though I am not sure the shielding is the main reason ? Is it not to split because of ruling ground loops between each others, making them the shortest with physical passive parts decoupling, rapid signal short trace with low , keep impedance and inductance constant, proofing high speed signal ground vs power supply return path and so on ? Of course with multi layer nothing impossible like adding a plain layer copper with just vias plunging, for shielding, but I remember member like Marce whom dayjob was to make complex HF pcb talking even of 18 mayers pcbs and dealing with lowest smd resistor case below 04xx size still. Shielding the main reason ?
Even with delta sigma dac or adc, 4 layers has some advantage over traces due to the clock speeds here, no ?
However for old pcm dac chip indeed, if well designed a 2 layers should be enough and a 4 layers no particular advantage but simplicity I assume (simplicity for the designer : you just not merge power and signal ground on a same layer...rapid and easy to design ?)
Well, way off topic, sorry for the op.
Though I am not sure the shielding is the main reason ? Is it not to split because of ruling ground loops between each others, making them the shortest with physical passive parts decoupling, rapid signal short trace with low , keep impedance and inductance constant, proofing high speed signal ground vs power supply return path and so on ? Of course with multi layer nothing impossible like adding a plain layer copper with just vias plunging, for shielding, but I remember member like Marce whom dayjob was to make complex HF pcb talking even of 18 mayers pcbs and dealing with lowest smd resistor case below 04xx size still. Shielding the main reason ?
Even with delta sigma dac or adc, 4 layers has some advantage over traces due to the clock speeds here, no ?
However for old pcm dac chip indeed, if well designed a 2 layers should be enough and a 4 layers no particular advantage but simplicity I assume (simplicity for the designer : you just not merge power and signal ground on a same layer...rapid and easy to design ?)
Well, way off topic, sorry for the op.
Here in EC, UN is watching closer what happen with all this mess on the far east of the pcb 😉
That was an over-simplification on my part. I just meant that electronic devices, in general, have gotten increasingly sophisticated (including smaller form factors), and having the possibility of making multi-layered PCB is one way for a designer to meet these standards. Some things are probably not doable without a multi-layered PCB.
Don't worry.
There was a very interesting thread in this forum from 2002 that discussed bypassing power supply capacitors in tube amps. I can't find it anymore tho 🙁
https://www.psaudio.com/pauls-posts/power-supply-madness/
Seems like it's not a myth at all to bypass power supply caps. PS Audio is rather thorough in their designs, and it's not something they advertise or anything (the bypassing). I found this article on the 3d google page. They wouldn't be doing it if didn't bring positive results.
Seems like it's not a myth at all to bypass power supply caps. PS Audio is rather thorough in their designs, and it's not something they advertise or anything (the bypassing). I found this article on the 3d google page. They wouldn't be doing it if didn't bring positive results.
If you knew all the things made that are unefficient...
Lookt at PSUD soft. Bypassing is not a refinment. Dimensioning the power supply caps is...
So you a signal tube, say an ecc88 twin triode for illustration, what would you do first ? Imagine a CRCRC...
Full diode bridge then 120 uf or 22 uF for the first cap ? Then the first R then second cap : 120 uF or 22 uF in relation to the first cap choice ? Then the third cap after the second R toward each triode, i.e. half tube : 125 uF or 22 uF ... x2 here as it is splitted in two B+ , B1+ and B2+. So in relation to the 9 possible choice, all will sound differently.
You see the bypassing choice is of a no importance, but the strategic choice of the main supply is.
So for the amp you questionned elswhere, your question should have been instead: are the main caps good dimenssioned ? (In spite of: should I bypass the a tual design because it is maybe not enough....)
Lookt at PSUD soft. Bypassing is not a refinment. Dimensioning the power supply caps is...
So you a signal tube, say an ecc88 twin triode for illustration, what would you do first ? Imagine a CRCRC...
Full diode bridge then 120 uf or 22 uF for the first cap ? Then the first R then second cap : 120 uF or 22 uF in relation to the first cap choice ? Then the third cap after the second R toward each triode, i.e. half tube : 125 uF or 22 uF ... x2 here as it is splitted in two B+ , B1+ and B2+. So in relation to the 9 possible choice, all will sound differently.
You see the bypassing choice is of a no importance, but the strategic choice of the main supply is.
So for the amp you questionned elswhere, your question should have been instead: are the main caps good dimenssioned ? (In spite of: should I bypass the a tual design because it is maybe not enough....)
I think you misunderstood the purpose of bypassing. You simply bypass to compensate for some of the drawbacks of using electrolytics - they struggle with higher frequencies.
But, yes, that is a good question. I always assumed that the designer of my amplifier knows what he's doing and that the size of power supply capacitors is carefully chosen based on the rest of the circuit and decreasing or increasing their value would be detrimental. Kinda like when cooking increasing or decreasing the amount of salt can be bad, more is not better it has to be just right.
But on the other hand, I totally agree that oversizing the power supply of any audio device is a good idea but I really don't know if justincreasing the capacitor size achieves that. And it's definetly something that ends up in a big debate again, some will say it's good, some say it doesn't matter/waste of money, some will say it makes things worse.
It's definetly something I'd like to explore further because someday I will replace the electrolytics in the PSU for film caps.
So, do you think increasing capacitor capacitance in the power supply does anything?
It would have been interesting if you used film caps instead of ceramics. Ceramics are not a good option.
You wrote earlier "You simply bypass to compensate for some of the drawbacks of using electrolytics - they struggle with higher frequencies."
If your circuit needs hf bypassing than put the bypass as close to the load as possible and use the proper capacitor for the job. Which is a (lossy) ceramic as it won't introduce oscillation as easily as a low ESR film cap.
If your circuit needs hf bypassing than put the bypass as close to the load as possible and use the proper capacitor for the job. Which is a (lossy) ceramic as it won't introduce oscillation as easily as a low ESR film cap.
If you have different ideas: soldering in 3 different values of film capacitors is trivial. Just spend the 5 dollars and do an objective evaluation. Get a 0.1uF, a 1uF and a 6uF Wima MKP (or similar), put them in one after the other and listen.
I didn't test film caps because one doesn't kick someone when he is down: film caps are known for their low losses, and relatively high ESL compared to group II ceramics which have intrinsic damping mechanisms and are physically smaller (does not apply to group I like COG, Nxyz, or Pxyz which are only available in low values anyway).
The high Q resulting from this combination is good in tuned circuits, but lethal for paralleling with larger caps.
The results would have looked similar, but the peaks and dips would have been more pronounced, by 10 or 20dB, and the frequencies would have been ~half a decade lower: it would have looked like a caricature, and I didn't want to leave the impression that the study was forced and artificial: I used the best possible practices, with suitable "candidates" and minimal parasitic inductances.
You can notice that in some (rare) cases (with solid tantalums amongst others), the combination result was slightly better than the large cap alone, which is a testimony to the neutrality and objectivity of the tests.
I now have no more access to sophisticated VNA's since I am retired, meaning I cannot show you graphs including film caps (but other members certainly can), but if you so wish, I can measure the minor parameters of any cap at home, and make a simulation based on these parameters: it will be as good as a VNA sweep, because in the material I showed, the agreement between the sim and the real measurement was near-perfect