renjy651 is a very knowledgeable man. He has helped me find caps and the right caps for my needs reg Mark Levinson he is the man period.
We all hear want a better audio experience and while we bicker we still learn.
Please consider renjy651 for what ever caps you need a real pro and helps me all the time find what I need.
We all hear want a better audio experience and while we bicker we still learn.
Please consider renjy651 for what ever caps you need a real pro and helps me all the time find what I need.
ALRAINBOW OP Contacted me 4 days ago about complete over haul all 28 power supply caps and 150 small caps replacement for No.33. Any way the Power supply 24 filter caps he ordered from mouser will not fit the reason It is 1/2 inch short and it is a cheap U36D 85C non stocked caps there tech support is a joke. The original power supply caps was United Chemicons most expensive 36da 95C caps not stocked any where in the world. The regulator caps Sprague 1900uF/150V part# 36dx192f150ac2a will be replaced with Cornell Dubilier 1900uF/250V best spec. I have these. Mouser recommended kemet 1800uF Red China made junk caps.
This is a gargantuan task even Nelson pass said in a diyaudio forum with out ML factory help he cannot repair a broken 33 amp.
I have the service manual for this amp more like a US defense computer.
I have ordered all the PSU caps and all the small caps for 33. All the parts will be worlds best. we will let you know the results when the big PSU caps arrive.
Thanks
Abe
This is a gargantuan task even Nelson pass said in a diyaudio forum with out ML factory help he cannot repair a broken 33 amp.
I have the service manual for this amp more like a US defense computer.
I have ordered all the PSU caps and all the small caps for 33. All the parts will be worlds best. we will let you know the results when the big PSU caps arrive.
Thanks
Abe
You can simply measure a cap for resp. capacitance, esr and Fres.
See image below, upper part for a low cost 4700uF cap, lower part for a similar cap from the same batch, now with 47uF and 100nF in parallel.
At Fres, behaviour is much more controlled.
In this way you can also compare old caps to new caps and compare difference in performance between different makes.
And keep in mind that a lower ESR is not always better, because the ESR may help to damp oscillations.
Hans
P.S. Also note that none of both caps have the specified 4700uf, but just 4000uF.
Attachments
Last edited:
That's not what I'm saying.
What I suggested earlier is to compare the old cap versus a new one before you start renewing all of them.
If there is hardly any difference, I think it's not to be expected that sound will be affected.
I don't want to go into the discussion that one maker sounds better than another one, just talking about the necessity of changing caps.
Each cap from a different manufacturer has a different measurable signature, but how this translates into sound differences is black magic.
I'm just warning against changing the engine of your Ferrari when the original is still working for the full 100%.
Hans
Oh, ok then
Do you need a network analyzer to get these measurements?
The OP stated renewing old caps with identical but new was audible. This may mean what your post indicates, but can also mean the new caps were just not broken in. A clarification from ALRAINBOW would be interesting.
Some simple sort of, yes, but not one that goes into the multi GHz region.Oh, ok then
Do you need a network analyzer to get these measurements? .
Hans
Oh, ok then
This may mean what your post indicates, but can also mean the new caps were just not broken in. A clarification from ALRAINBOW would be interesting.
If capacitors in electronic equipment needs to be "broken in" for hours before it works to spec, then it is very poorly designed indeed. Does your cellphone which works in the Megahertz regions with thousands of capacitors need to be broken in for hours before it can be used to make a call or function stably? Does your laptop need to be broken in for hours before it works perfectly? So why should a audio device which works on the miserably low 20-20000hz, have to have its capacitors broken in for hours or days? What people mistake for "capacitor burn in", is really our brain adjusting over days to a new sound character from a new system. It's not the system that is changing over time, but our brain adjusting. I do admit that there may be a role for burning in electro-mechanical devices like speakers or headphones, but NOT certainly for capacitors. Any respectable audio engineering society will laugh at the concept of capacitor burning in.
I like the "brain adjusting" part best. Priceless, despite the anger issues. The question was not really directed at you but it's always nice to have another opinion for the sake of diversity
All you have to do is to just show a single published peer reviewed scientific paper supporting your "capacitor burning in" theory? Surely one of the thousands of eminent scientists and researchers must have published a paper validating your so obvious theory? Oh, i see you cant find any published paper, so maybe you should publish your findings instead as this could be a noble prize worthy discovery by you. In reality, the only people who benefit from your unscientific hocus pocus about capacitor burningin etc are capacitor sellers who are making huge profits by selling expensive capacitors to unsuspecting people who are willing to part with their hard earned money.
The choice of ESR in capacitors for power supply is determined by the characteristics of the power supply, ie: is it a switching power supply where very high peak currents are presents (low ESR caps are needed), or a conventional linear supply where the peak currents are not high.
If your power supply is the traditional conventional supply then just go with any capacitor for the transformer bulk decoupling which has sufficient peak current, voltage and temperature rating to support your amp's demands. A rough guide would be 1000uf per 10 watts of output power, per rail. If you have a 100 watt amp, use 10,000uf decoupling per supply rail.
For decoupling on the amplifier PCB, i use 1000uf standard electrolytic capacitors per rail. All other capacitor types are dictated by the electronic manufacturers recommendations, for example if a IC manufacturer says use MLCC caps, i stick with them.
I do not use esoteric capacitors for decoupling and my amplifiers measure well and sound great.
The only instance where i choose specific capacitors are when it comes to the capacitors in the audio signal path. For these i use C0G, or NP0 to improve linearity. If using non surface mount caps i use silver mica caps for compensations capacitors and non-polar electrolytic for the negative feedback cap and input caps.
These are standard design decisions which any rational circuit designer would use and which lead to very low distortion and excellent linearity.
Its important to note that amplifiers used in our homes are never driven anywhere near the maximum, if you have a decent speaker with 90db sensitivity, that means if your amplifier outputs 1 watt, your speaker is going to output 90db sound pressure levels. 90db is the equivalent of a power mower and you are likely to get hearing damage if exposed for long enough. So you may design a 100 watt amp, but it will hardly ever be called to deliver continuous 100 watt outputs.
Short answer: Theoretically yes, it would help to reduce heating of the cap , but practically not really at the low frequencies used. Just go with large electrolytic cap which has sufficient voltage and peak ripple current ratings for your amp.
Long answer:
ESR is just what it says, resistance in series with your capacitor.
Power-supply capacitors smooth ripple on DC power supplied from AC sources. Consider the following scenario (I have exaggerated the values to illustrate the point). When the AC source is low frequency (50 Hz, 60 Hz, etc) the capacitor values required for smoothing are large (1000's of uf) and hence physically large too, and could tolerate high ESR (ie: 1 ohm for a 1A supply with a 1000 uF filter capacitor). That's because a one-amp ripple current only created one watt of heat, and a large 1000uF capacitor can shed that heat.
When switchmode supplies went up to 100 kHz, and a suitable ideal-capacitor value for the higher frequency smoothing was (again for 1A output) a much smaller 3.3 uF. 1 ohm ESR means that same 1W would be dumped into a capacitor the size of a pea. It'd fail, because it's too small to dissipate one watt of heat. Thus it is necessary for the small 3.3uf capacitor to have a very low ESR to prevent over heating.
In summary the requirement for "low-ESR" capacitors normally arises in output filters of switching power supplies, where the frequency is relatively high (100's of kHz to MHz), so as to avoid overheating and premature failure of the capacitors.
Another important fact is that a good audio amplifier circuit has PSRR, or power supply ripple rejection built in (sometimes as much as -120db). This itself helps to reduce the effect of power supply ripple on amplifier performance. Which is why i recommend to go with a generic electrolytic capacitor for power supply smoothing with sufficient ripple current, and voltage rating as required by your amplifier.
I personally beleive that its the amplifier PCB decoupling which is more important. For the amplifier PCB decoupling i use a 1000uf generic electrolytic in parallel with a 4.7uf OS-CON (low esr solid polymer caps) and a 0.1uf MLCC SMD capacitor.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.