Yes and no. I didn't read the text too carefully. The text mentions 100 uF but the schematic has these caps labeled 100 nF. Sorry for this. In that case I'm with you.Kuei Yang Wang said:
Excellent visualization of the problem but this shows the theory but I wonder how the problem is in real life.
The PSRR isn't so high at high frequencys but still how much "bôs" do you want to reject or surpress?
Kuei Yang Wang said:
Do you mean that noise above 250kHz somehow intermodulates back to the 20Hz - 20kHz band (with what?) and become audible? And that the changes you found in the simulation somehow take care of that problem?
I rather wouldn't count the times that a supply oscillated on me. I learned the hard way too, as I'm sure you did. But that type of instability readily jumps at you on the scope. Instability of the chip itself at 300kHz ditto.
It is a great leap of faith to jump from what *could* be a *possibility* to concluding that these are actually present and that the snubber things cure them. This ice is too thin for me to skate on.
And please, don't take this personal, as Carlos seems to do. I merely want to point out that sometimes you need to step back and critically ask yourself whether you are still firmly rooted on Mother Earth. I think you are just about to start floating!
Jan Didden
PS And don't let your eagerness to show a certain person that he Don't Get it cloud your professional judgement...
Well put, Johannes. Besides my first question was a missunderstanding due to an error in the schematic I referred to. Nevertheless the arguments are rather thin, I'll agree Jan.janneman said:
One thing in the simulation I wonder about: You have a _very_ low series resistance from the transformer.... so what about the 1-10 ohms?
Konnichiwa,
I don't think it is news or anything extraordinary to state that with nonlinear circuits noise outside the band of interest can result in noise "folding back" into the band of interest. So I thin the "somehow" is not somehow but "it went thataway Sheriff".
A powersupply line that has a high impedance at certain frequencies will not elimiate noise at such frequencies, be it the result of exciting the mains transformers resonances or plain old RFI of all sorts of nature which is passed readily through by the torroidal mains transformers and is conducted on by the ill advised capacitors in parallel to the rectifiers which are intended to "slow down" the rectifiers reverse spike (why not use rectifiers free from minority carriers to start with?).
Adding 1 + 1 gives the potential at least for problems and indeed such problems actually CAN be measured, can they be heard? We have no proof beyond a reasonable doubt either way.
I have found NOTHING. I merely illustrated that placing the RC circuit in parallel with a bypassed Electrolytic capacitor having relatively high ESL causes the peaking in the supply impedance to much reduced and thus the filtering action is improved, by as much as 20db, based on the simulation.
It jumps out if on the testbench the required conditions to induce instability are fulfilled. If they are not the circuit will appear stable but will be marginally so once the load is changed to a real one, instead of an 8Ohm resistor.
As someone who has worked for a long time with Chip's that are "twitchy" I have had enough time to learn how marginally stable most looped feedback circuits really are, especially "high speed" ones and how dependent on a well controlled PSU impedance they are. If your LM6172 in the CD Player output stage sounds "glassy" it is near oscillation at RF and you need to work out where in the supply you have goofed.
For power amplifiers the LM3875 is pretty "fast" as well so layout and supply impedances can do all sort of interesting things.
The original gaincard/gainclone layout with proven low ESR & ESL capacitors in very close vicinity of the chip actually makes for a supply that at >> Audio frequencies is better behaved than an ill concieved Supply bith bigger value capacitors and "audiophile" bypassing senselessly thrown in. And yes, I remain convinced that the results are audible under certain conditions, both of marginal stability and RFI folding back into the audio range.
Sayonara
janneman said:
I don't think it is news or anything extraordinary to state that with nonlinear circuits noise outside the band of interest can result in noise "folding back" into the band of interest. So I thin the "somehow" is not somehow but "it went thataway Sheriff".
A powersupply line that has a high impedance at certain frequencies will not elimiate noise at such frequencies, be it the result of exciting the mains transformers resonances or plain old RFI of all sorts of nature which is passed readily through by the torroidal mains transformers and is conducted on by the ill advised capacitors in parallel to the rectifiers which are intended to "slow down" the rectifiers reverse spike (why not use rectifiers free from minority carriers to start with?).
Adding 1 + 1 gives the potential at least for problems and indeed such problems actually CAN be measured, can they be heard? We have no proof beyond a reasonable doubt either way.
janneman said:
I have found NOTHING. I merely illustrated that placing the RC circuit in parallel with a bypassed Electrolytic capacitor having relatively high ESL causes the peaking in the supply impedance to much reduced and thus the filtering action is improved, by as much as 20db, based on the simulation.
janneman said:
It jumps out if on the testbench the required conditions to induce instability are fulfilled. If they are not the circuit will appear stable but will be marginally so once the load is changed to a real one, instead of an 8Ohm resistor.
As someone who has worked for a long time with Chip's that are "twitchy" I have had enough time to learn how marginally stable most looped feedback circuits really are, especially "high speed" ones and how dependent on a well controlled PSU impedance they are. If your LM6172 in the CD Player output stage sounds "glassy" it is near oscillation at RF and you need to work out where in the supply you have goofed.
For power amplifiers the LM3875 is pretty "fast" as well so layout and supply impedances can do all sort of interesting things.
The original gaincard/gainclone layout with proven low ESR & ESL capacitors in very close vicinity of the chip actually makes for a supply that at >> Audio frequencies is better behaved than an ill concieved Supply bith bigger value capacitors and "audiophile" bypassing senselessly thrown in. And yes, I remain convinced that the results are audible under certain conditions, both of marginal stability and RFI folding back into the audio range.
Sayonara
peranders said:
Change the deam PSU on your amp for this one in 5 minutes and listen to the deam thing.
You can measure it later and take your conclusions.
Listening is the ultimate test, or do you have a secret formula to explain everything with measurements?
Deam it, if I was as complicated as you, I would be quiet as a mule, listening to what people sell me, and making my system threoretically perfect but crappy in practice.
PS: too many theoretical orgasms around here, back to my hole.
Konnichiwa,
???? What on earth are you talking about? The simulation attemps nothing else than to illustrate the impedance of the capacitor combo's, nothing else is modelled.
The only reason the resistor is there is that P-Spice needs a device somewhere "to sense the current". Consider the resistance the internal resistance of your AC current Meter, as it's a pretty good meter the resistance is very low....
Sayonara
peranders said:
???? What on earth are you talking about? The simulation attemps nothing else than to illustrate the impedance of the capacitor combo's, nothing else is modelled.
The only reason the resistor is there is that P-Spice needs a device somewhere "to sense the current". Consider the resistance the internal resistance of your AC current Meter, as it's a pretty good meter the resistance is very low....
Sayonara
janneman said:
Please, we are talking about an impedance matching ZOBEL, not a snubber...but let me interject on the matter of diode snubbers and power supply noise...
I don't think it's "noise", as in Gaussian noise here -- in the diode circuit there is some oscillation, not random but periodic, caused by the diode capacitance, transformer inductance, leakage and interwinding capacitance. I have measured it from ~50kHz to 1.6 MHz -- it certainly isn't a lot of power -- if each diode has a different capacitance the ringing frequency of each diode will differ -- so if you have one resonating at 100kHz and the other at 101 kHz you should expect sidebands at 1 kHz and 201kHz, many, many dB down, however. That's from the textbook, what you see in real life is periodicity and randomness -- and some of this is crud on the transmission lines to boot!
Well, I posted this once before -- this was a quickie experiment I did with a small 80VA torroidal transformer, etc. -- When there aren't a lot of harmonics on the transmission line you can see the periodicity better.
An externally hosted image should be here but it was not working when we last tested it.
carlosfm said:
Change the deam PSU on your amp for this one in 5 minutes and listen to the deam thing.
You can measure it later and take your conclusions.
Listening is the ultimate test, or do you have a secret formula to explain everything with measurements?
Deam it, if I was as complicated as you, I would be quiet as a mule, listening to what people sell me, and making my system threoretically perfect but crappy in practice.
PS: too many theoretical orgasms around here, back to my hole.
Carlos,
Listening is the MOST UNRELIABLE and WORTHLESS test you, as the designer/developer/having a stake in the results can do, UNLESS your results are collaborated by independent controlled/blind tests. Your statement that you heard a difference is WORTHLESS for anyone except yourself.
Jan Didden
Kuei Yang Wang said:
Still thin ice! The fact that something is possible does by no means mean it is happening here and now!
Kuei Yang Wang said:
I'm with you 100% here. Looking at those caps in parallel of those diodes you can almost SEE the junk getting into the amplifier! (At least I Get It this time
)Kuei Yang Wang said:
Indeed. But I think many reading it picked up the implication that THEREFORE the amp sounds better with the snubbers.
Jan Didden
Hmm..
somebody's rather hung up on this subject, why not just deny the existance of this thread and live happily in the bliss of ignorance?
I see both "sides" logic here, but having actually implemented this thing and heard the audible differences, i'm with Carlos on this one.
may i just add that it will cost you under 4 euro to try it?
IT ISN'T HARDER THAN THAT!
somebody's rather hung up on this subject, why not just deny the existance of this thread and live happily in the bliss of ignorance?
I see both "sides" logic here, but having actually implemented this thing and heard the audible differences, i'm with Carlos on this one.
may i just add that it will cost you under 4 euro to try it?
IT ISN'T HARDER THAN THAT!
janneman said:
Jan, my oppinion is just my oppinion.
You can go and ask people that tested this PSU what they think about it.
And you can present me with 1000 different CD players with ultra-flat frequency response and minimal distortion and they all sound different.
Now you tell me which one to pick.
If I was to buy a system, would I only trust measurements, even if very complete, instead of listening?
I told you many times, and I don't want to talk about that anymore: I listen to the same thing on blind and on sighted tests.
But they are very unreliable, with most listeners.
I have many experience with blind and sighted tests.
It's up to you to believe me or not.
I don't care.
I listen to the music with my ears, I love music, I'm mad about it.
Are you sure you LISTEN?
demogorgon said:Hmm..
somebody's rather hung up on this subject, why not just deny the existance of this thread and live happily in the bliss of ignorance?
I see both "sides" logic here, but having actually implemented this thing and heard the audible differences, i'm with Carlos on this one.
may i just add that it will cost you under 4 euro to try it?
IT ISN'T HARDER THAN THAT!
I'm working on it
Jan Didden
Attachments
janneman said:I'm with you 100% here. Looking at those caps in parallel of those diodes you can almost SEE the junk getting into the amplifier! (At least I Get It this time
Surely you are not talking about my schematic, because I don't use caps across diodes.
P-A may explain why he uses them on his chip amp, and I advice people to remove'em.
carlosfm said:
Surely you are not talking about my schematic, because I don't use caps across diodes.
P-A may explain why he uses them on his chip amp, and I advice people to remove'em.
No no, no particular schematic, but often you see them, in an effort to short whatever the diode might generate when it switches off when the mains dips below the reservoir cap voltage and the transformer is disconnected from the amp for the next 80% of the mains cycle period
.Soft recovery diodes take care of this much better, and prevent the caps to be a conduit for mains-borne junk.
Jan Didden
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.