Well now- I have been working with Circlotrons for over 3 and a half decades and I can verify that this is true. The more AC you can read across the last capacitor in the power supply - the worse for the sound of the amplifier. This is simply solved by increasing the value of the cap; you have to have a value that indeed is a real bypass for a low impedance circuit. IOW the bigger you make it, the less effect it seems to have. If you can regulate, so much the better (although I have found that in power amplifiers, the failure of an output device can sometimes doom the regulator circuit, so for reliability I usually avoid them). In my book if a cap sounds better by *not* passing AC, then it fails the definition and function of a coupling cap.
Now with a true coupling cap, such is not the case; The better it passes AC the better it sounds and the bigger you make it the more coloration it imparts. Plus if it is so big that its timing constant exceeds that of the power supply, it will induce low frequency instability.
I think we need to make a difference here between passing AC currents and having an AC voltage across it.
It has been shown quite convincingly that the AC voltage component is proportional to the distortion a capacitor imparts. Of course, it can happen for many reasons, like the basic Xc resulting in a non-negligible voltage drop at said current, or the internal parasitics of the cap.
A coupling cap by definition has negligible AC voltage over it for all frequencies of interest, hence it 'couples' AC. In general currents are small in what we ordinarily think of as coupling caps, although in reality practically every cap not designed to give a specific RC constant, such as couplers, DC blockers, bypass caps and especially power supply caps are really all 'coupling caps'.
As far as power supply caps go, just increasing their value is a two-edged sword - on one side the AC component drops, but on the other, it becomes richer in high order harmonics as it decreases the conduction angle of rectifier diodes (this in itself has other repercussions on the diodes, cap and transformer which might turn dire when the cap value becomes too large). A better solution is forming a CRC filter. Even a fraction of an ohm in the resistor can drastically reduce peak current from the rectifier through the output cap, lowering not only the RMS AC over the cap, but also the high harmonic content, intermodulation of rectifier and load current, and finally reducing the chance of the internal ESL ringing with the capacitance at the start and end of rectifier commutation. On class AB amps the resistor can be increased quite a lot without incurring significant losses, due to the high peak to average ratio of supply current, so CRC filtering has an even better effect.
Member
Joined 2009
Paid Member
I think this is an important point. I marvel at tube amps that can afford to use choke loaded supplies because their dc current requirements are lower. So much nicer than a big capacitor after the rectifier.
You can use common mode chokes of course, because of the dc balance achieved. I've tried both CRC and CLC for SS amps.
Another approach: http://www.diyaudio.com/forums/digi...e-nos-dac-using-tda1541a-175.html#post2354992
My 0.02 here:
Just because signal current is supplied by a capacitor does not automatically mean that said capacitor's dynamic resistance appears as distortion of the signal.
Consider the example circuit attached. Q1 drain is supplied by a bridge rectifier and a single cheap-az electrolytic capacitor. It has a few ohms or even tens of ohms dynamic resistance at audio frequencies.
Q1 source-drain dynamic resistance is in the megohm range at audio frequencies.
Q1 transfer conductance is about 50 mA/volt which provides a dynamic source resistance wrt the input signal of about 20 ohms.
So the equivalent circuit is say 1 megohm (MOSFET s-d) in series with the generator, with an equivalent "load" of say 100 ohms (the capacitor) "reflected" back to a 20 ohm source resistance.
Not a lot to worry about, I would say, in this case.
If the active device is a triode or something with local feedback then the dynamic resistance to the capacitor becomes much smaller and we start seeing significant effects.
Cheers,
Michael
Just because signal current is supplied by a capacitor does not automatically mean that said capacitor's dynamic resistance appears as distortion of the signal.
Consider the example circuit attached. Q1 drain is supplied by a bridge rectifier and a single cheap-az electrolytic capacitor. It has a few ohms or even tens of ohms dynamic resistance at audio frequencies.
Q1 source-drain dynamic resistance is in the megohm range at audio frequencies.
Q1 transfer conductance is about 50 mA/volt which provides a dynamic source resistance wrt the input signal of about 20 ohms.
So the equivalent circuit is say 1 megohm (MOSFET s-d) in series with the generator, with an equivalent "load" of say 100 ohms (the capacitor) "reflected" back to a 20 ohm source resistance.
Not a lot to worry about, I would say, in this case.
If the active device is a triode or something with local feedback then the dynamic resistance to the capacitor becomes much smaller and we start seeing significant effects.
Cheers,
Michael
Attachments
Last edited:
True. The feedback ("miller") capacitance is a big component of the megohm dynamic resistance that isolates the 20 ohm source from the ~100 ohm "load" made by the dynamic resistance of the electrolytic cap. So the audio current imposed on the cap is indeed equal to the audio current at the stage's output. But the back-reflected dynamic resistance of the cap is attenuated by that megohm (or so) dynamic resisitance of the MOSFET (in this example).
What I'm saying is that the PSRR of something like my example is very high, and that self-same PSRR also functions to isolate the potentially nonlinear dynamic resistance of the power supply output filter from the audio "signal path". In the case of my example at least.
Cheers,
Michael
PS Would the effect of the back-reflected nonlinear dynamic resistance of the power supply filter be measurable as an added component in the output spectrum of the amplifier?
Last edited:
Member
Joined 2009
Paid Member
I tried this in Google Translate but it didn't help
Try Wiki. The Bestseller is "Tube Sound"
....and constantly discharged from Output power Tube and Load(Speaker), both conected in series.......
Best regards
To my opinion this is signal current, flowing from the cap.
That doesnt neccesary mean that this cant be an exceptional amplifier...
Yes,Amp will continue to `sing` for a several seconds even with Main Tranformer turned off until this smoothing Capacitors not become fully discharged.
But throgh this capacitors dont flow any AC or DC current.They have double function in same time.1)To smooth 100Hz DC pulsating current from Bridge Rectifier.2)To properly suply Output Power stage special under high Peak DC current demands,since this capacitors are Big DC current Reservoirs,much bigger than Main Transformer/Bridge Rectifier actually is.
OTL Circlotron(Balanced Bridge) Tube or SS Amp it is exceptional good sounding Amps topology,but thats not include each Circlotron design.
(`Signal path,`hmmm
,maybe Signal Loop......)Best Regards
Last edited:
Yes,Amp will continue to `sing` for a several seconds even with Main Tranformer turned off until this smoothing Capacitors not become fully discharged.
But throgh this capacitors dont flow any AC or DC current.They have double function in same time.1)To smooth 100Hz DC pulsating current from Bridge Rectifier.2)To properly suply Output Power stage special under high Peak DC current demands,since this capacitors are Big DC current Reservoirs,much bigger than Main Transformer/Bridge Rectifier actually is.
OTL Circlotron(Balanced Bridge) Tube or SS Amp it is exceptional good sounding Amps topology,but thats not include each Circlotron design.
(`Signal path,`hmmm
Best Regards
The smoothing of 100hz is by short circuiting the 100hz AC ripple on top of your DC to ground! this is AC short circuited!!!!! A capacitor can not do anything else; it wont let dc flow through it. even the initial charging is considered ac; because its voltage changes! after that, as the voltage on the capacitor stays the same, it does nothing. But connect an amplifier to its leads. The ac signal current goes through it to ground.
This is not theoretical, but easyly proofed. Put a resister in series with it, so all current going through the capacitor must go through that resister as well. then measure the voltage on the resister with a scope.
You will see your audio signal and ripple mixed!
The only amplifiers I know who avoid this are constant current amps; the current drawn from the supply is really constant, so there isnt AC involved.
Signal LOOP, YES!!! that what were talking about... current comes from supply and goes back to supply. it gets modulated in approx. form of audiosignal.
I imagine it always like this:
there comes current from the supply, it divides and split and goes through al the parts and components, and at ground it comes together from all the components and parts and adds up to exactly the same amount of electrons as left the supply. problem is, capacitors and coils will delay some parts of the add up current...
Last edited:
Very good point.
If You look(and think) on (Tube) Audio Circuit in that way, next thing to considere is `Star Ground` Return/Start curent loops Point,whatever we use `point to point` or PCB layout.
Star Ground Point basicly is good and nonreplacable for use to avoid Negative interaction of weak input/driver stage curent(signal) loops with stong current(signal) loops of Output Power Stage/PSU unit .Thats is the basic of good Audio aparatures engenering Design.Any one who dont use this Design aproach made big mistake and poor sounding Amps/Preams or ... .
About Elko Capacitors Again:
Just finished nr#153798 succesfull TV set repairing in my life,Allways the same:`Dry` Electrolitic Capacitors is 99% of the reason which made malfunction in this TV Set Aparatures.
Conclusion:Elkos dont last forever,and to chose right quailty one for each Amps stage is not easy job at all.
Best Regards to All
Last edited:
Very good point.
If You look(and think) on (Tube) Audio Circuit in that way, next thing to considere is `Star Ground` Return/Start curent loops Point,whatever we use `point to point` or PCB layout.
Star Ground Point basicly is good and nonreplacable for use to avoid Negative interaction of weak input/driver stage curent(signal) loops with stong current(signal) loops of Output Power Stage/PSU unit .Thats is the basic of good Audio aparatures engenering Design.Any one who dont use this Design aproach made big mistake and poor sounding Amps/Preams or ... .
About Elko Capacitors Again:
Just finished nr#153798 succesfull TV set repairing in my life,Allways the same:`Dry` Electrolitic Capacitors is 99% of the reason which made malfunction in this TV Set Aparatures.
Conclusion:Elkos dont last forever,and to chose right quailty one for each Amps stage is not easy job at all.
Best Regards to All
Yes!
The grounding is a very important aspect of your design. I always want the currents from charging powersupply caps to go back to the powersupply as soon as possible. I only connect the latest cap its ground to the star ground!
That way the current charging the caps (pulses) never flow through the star ground...
Ill quit threadhijacking now
Acros this serial conected resistor with PSU Capacitor You can actually measure only charge/discharge double direction current flow of this PSU Elkos, but Not the the current flow across this PSU Elkos,which dont exist.
This ` inside`DC current flow is measurable only when Capacitors` leak`or is internaly shorted.
Best Regards
NO NO NO!!Acros this serial conected resistor with PSU Capacitor You can actually measure only charge/discharge double direction current flow of this PSU Elkos, but Not the the current flow across this PSU Elkos,which dont exist.
This ` inside`DC current flow is measurable only when Capacitors` leak`or is internaly shorted.
Best Regards
The charge/ discharge is also single direction! But its AC on top of dc. And that AC gets shorted!
as a matter of fact, there is no "discharge" cycle; thats your signal...
Please try it in real; you'll see it!
Greetings, Paul
Last edited:
NO NO NO!!
The charge/ discharge is also single direction! But its AC on top of dc. And that AC gets shorted!
as a matter of fact, there is no "discharge" cycle; thats your signal...
Please try it in real; you'll see it!
Greetings, Paul
Really dont want to Hijack this great thread started by respected DIY Member Audiodesign.
Just my last quick answer for Pauldune:
Capacitors can only to smooth 100Hz DC pulsating current but Not to `short `any DC pulsating current komponents.Try a litle scope probe,You can see thats always remain some 100Hz pulsating rest across capacitors whatever which capacitors value is incorporated.
If capacitors realy `shorts` any AC component on the top of DC current, on the scope screen will be perfect Flat line but usualy is Not.
Best Regards