Silver-mica I am told is manufactured in India but are virtually not available in the local market and if available, are very expensive. Farnell and RS Components stock almost all values but are very highly priced.
Hence, I am thinking of switching over to polypropylene type.
Has anyone compared the two types, for Cdom duties and what sonic signature does each bring to the table?
I for one have compared polystyrene and silver-mica and found the latter to be more precise but a bit sharper whereas the former is a tad bit softer in comparison but also thinner in the HF region.
Hence, I am thinking of switching over to polypropylene type.
Has anyone compared the two types, for Cdom duties and what sonic signature does each bring to the table?
I for one have compared polystyrene and silver-mica and found the latter to be more precise but a bit sharper whereas the former is a tad bit softer in comparison but also thinner in the HF region.
it is true that messing arround with this part will alter the sonic signature of the amplifier ... for better or worst i cannot really tell but for sure it will be changed
now from personal expirience going from one quality to the other means that you might also change the value with out problems
machines that had 100pf ceramic when this was changed to silver mica or styroflex the value easilly went down to 68 or82 pf without any oscilation possible to be found ....
i found also tha ceramic multileyer type sounded nice but since can be found in low voltage ratings i didnt want to risk there ...
isnt it the same issue with more simple things ???? like simple and not that sensitive matters like the input capacitors ...some go with MKT some with electrolytics some with bipolar and some with combinations .... be sure that each and every one of them will have diferent sonic signature ...
remains a very interesting issue to talk about ...i ll stay arround
regards sakis
now from personal expirience going from one quality to the other means that you might also change the value with out problems
machines that had 100pf ceramic when this was changed to silver mica or styroflex the value easilly went down to 68 or82 pf without any oscilation possible to be found ....
i found also tha ceramic multileyer type sounded nice but since can be found in low voltage ratings i didnt want to risk there ...
isnt it the same issue with more simple things ???? like simple and not that sensitive matters like the input capacitors ...some go with MKT some with electrolytics some with bipolar and some with combinations .... be sure that each and every one of them will have diferent sonic signature ...
remains a very interesting issue to talk about ...i ll stay arround
regards sakis
Its the highish dielectric absorption (DA) of the mica that gives it a dampening effect in that position so it can go lower in value. Valve experts preferred mica in step feedback compensation networks for that reason. It holds the loop ''softer'' transiently. Someone's got to see what happens with values on the oscope first, have the same shape square waves and then try subjectively judge.
ok then here is the simple steps ....
D.U.T is the P3A the most simple amplifier that uses 100PF and at square wave with 100pf square starts to get round at about 30khz very nice i think ...
change this with silver mica example 82 pf and sqaure wave still gets round arround 30khz or so ...
the question is if this ...oscope reading can tell you anything about the sonic signature ...
then again other question will be :what else is there to measure in order to evaluate option A from option B ???
to make this even easier you may simply trnasfer those above questions to the input or filter capacitor that exista in a more simple stage of the amp such is the input ...
thank you sakis
D.U.T is the P3A the most simple amplifier that uses 100PF and at square wave with 100pf square starts to get round at about 30khz very nice i think ...
change this with silver mica example 82 pf and sqaure wave still gets round arround 30khz or so ...
the question is if this ...oscope reading can tell you anything about the sonic signature ...
then again other question will be :what else is there to measure in order to evaluate option A from option B ???
to make this even easier you may simply trnasfer those above questions to the input or filter capacitor that exista in a more simple stage of the amp such is the input ...
thank you sakis
Nothing else to match with the scope, else you have an HP analyzer and you can see the closed loop shape. The point is, as you have done, to have same behavior before listening to it because same value with a previous ceramic can over damp if with mica. Try PPS too some day if you chance on some.
Wima FKP should also be nice for Cdome dis are the only wan,s the,t Wima self states to be OK for Audio Use.🙂
Plotting loop stability (gain vs phase shift) in LTSpice,
I've found to my surprise that CDOM usually wants a
series resistance? Else creating a Q peak that makes
phase shift far worse (almost guaranteed oscillation)
right before rolling under unity. Else having to abuse
such large value CDOM that is unconditionally stable,
but also distorts the resulting waveform to a triangle.
Anyways, if series resistance is desirable (and I believe
it is). Then quality nuances of this particular cap become
almost irrelevant, because you are going to deliberately
sandbag the Q anyways...
The objective to gradually and with lowest phase shift,
take loop gain down, slightly below unity. Then other
poles (including parasitic ones) with higher corners can
roll off as aggressively as they like.
But again, this is just based upon LTSpice observation.
I have not compared to measurements of loop phase in
real circuits for sanity check.
I've found to my surprise that CDOM usually wants a
series resistance? Else creating a Q peak that makes
phase shift far worse (almost guaranteed oscillation)
right before rolling under unity. Else having to abuse
such large value CDOM that is unconditionally stable,
but also distorts the resulting waveform to a triangle.
Anyways, if series resistance is desirable (and I believe
it is). Then quality nuances of this particular cap become
almost irrelevant, because you are going to deliberately
sandbag the Q anyways...
The objective to gradually and with lowest phase shift,
take loop gain down, slightly below unity. Then other
poles (including parasitic ones) with higher corners can
roll off as aggressively as they like.
But again, this is just based upon LTSpice observation.
I have not compared to measurements of loop phase in
real circuits for sanity check.
Last edited:
Dissipation factor is loss and provides damping. Dielectric absorption is different and it's sonic effects are debated. Note that dissipation factor is not terribly frequency sensitive. A silver mica cap (or polypropylene) might have a dissipation factor of 0.0003 and it might not change much over the audio range. ESR, however, is related to dissipation factor by D/omegaCs, so it's very different than inserting a resistor in series with a cap. Silver mica and polypropylene are both very low DF caps, but the mica has more DA than one would expect from its otherwise excellent performance. Very low DF caps can be trouble in some places unless you have some series resistance as described in the above post. That's why tantalums are good for certain bypass tasks- they have a moderate and consistant DF, though I'd never use one for audio. IMO, properly applied, there shouldn't be any huge difference between silver-mica and polypropylene, and polypropylene is closer to a theoretically perfect cap. Let your ears be your guide.
What JCX is saying is that you stop the rolloff when another cap starts to come in, and this gives you less excess phase shift. The 'zero' is the opposite of a 'pole'. The 'pole' is the dominant rolloff made by the capacitor you are concerned with.
Indeed it is often the case that some resistance in series with Cdom helps stability. The reason for this is that it introduces a left-half-plane zero that can help cancel parasitic poles at higher frequencies. In theory, by the way, the Miller compensated stage without a resistor in series with the cap can actually introduce a right-half-plane zero in the open-loop transfer function, which is a bad thing.
I always make provision for a resistor in series with Cdom, but bear in mind that you should use it sparingly, else it can reduce gain margin. I recommend never placing the zero that it creates with Cdom any lower in frequency than twice the gain crossover frequency.
I have been fortunate to have a good supply of silvered Mica capacitors over the years, however I believe that COG (NPO) ceramics rated at a fairly high voltage can be very good, although I have never tried them.
Cheers,
Bob
I see soakage as a form of time lag in this application. I should have used the word 'delaying' not 'damping' my bad. It would be seen from the POV of second harmonic in a coupling situation.
Why is that? I always thought ceramics are bad in audio applications.
With kind regards,
Bas
It all depends on the dielectric used. C0G (NP0) are extremely stable with respect to temperature and voltage.
Never use an X7R, X5R or Y5V for this application.
For some LDOs and boost/buck regulators, I've actually included a series resistance option with the output filter capacitance which I've used ceramics for in case the circuit doesn't like the low ESRs of ceramic caps. I've had some bad experiences with tantalums -in their higher voltage ranges, say at 25Vdc or more, solid tantalum capacitors, particularly surface mount versions, actually need to be 'reformed' to some extent to meet their rated voltage after being subjected to the heat of wave or reflow soldering, or else may be liable to fail destructively at voltages substantially below their rated voltage, possibly even causing a pcb fire if enough current is supplied. Basically, solid tantalums are fairly crappy capacitors by today's standards, and don't sound very good, either.
Last edited:
- Status
- Not open for further replies.