I say it because I believe it to be true. There is a big difference between using low ESR caps and bypassing. One is sometimes useful. Reported big improvements may or may not be real; after spending time and money any change in the sound is reported as an improvement, and some exaggeration may occur.goodguys said:
Elyics have a bad press because originally, when they were made for tube circuits, series parasitic impedances were mostly unimportant.
When they became used extensively for transistor circuits, ESL and ESR mattered a lot more, and to use the existing caps, expedients had to be used, like bypassing with a paper, film or ceramic cap.
However, 50 years ago, manufacturers like Siemens, Philips and others did their homework and adapted their production to the current trends.
This means that the tricks usable in the sixties are not only useless, but in fact detrimental, because paralleling two low esr capacitors results in unwanted resonance, something that the "poor" caps of the pre-60's avoided thanks to their high esr.
A modern E-cap (that's post 1970) has an ESL ~equivalent to the length of wire separating its electrodes, and that is true even for "poor", generic types.
In fact, some boutique caps manage to do worse, but all of the normal production adheres to that principle.
Another myth worth dissipating: older caps are generally not worse than their modern counterparts: regarding esr, I have 1973 NOS Siemens caps capable of beating most of the current production, even though they have an axial construction.
They are of course much, much larger than modern ones, but they are also infinitely more robust.
It is probably one of the reasons why many recapping stories have an unhappy ending.
That said, E-caps, especially from dodgy manufacturers (most are now gone as a result of natural selection) can dry out and fail in various ways, but good brands like Nichicon, Siemens, etc. last forever when used within their limits
When they became used extensively for transistor circuits, ESL and ESR mattered a lot more, and to use the existing caps, expedients had to be used, like bypassing with a paper, film or ceramic cap.
However, 50 years ago, manufacturers like Siemens, Philips and others did their homework and adapted their production to the current trends.
This means that the tricks usable in the sixties are not only useless, but in fact detrimental, because paralleling two low esr capacitors results in unwanted resonance, something that the "poor" caps of the pre-60's avoided thanks to their high esr.
A modern E-cap (that's post 1970) has an ESL ~equivalent to the length of wire separating its electrodes, and that is true even for "poor", generic types.
In fact, some boutique caps manage to do worse, but all of the normal production adheres to that principle.
Another myth worth dissipating: older caps are generally not worse than their modern counterparts: regarding esr, I have 1973 NOS Siemens caps capable of beating most of the current production, even though they have an axial construction.
They are of course much, much larger than modern ones, but they are also infinitely more robust.
It is probably one of the reasons why many recapping stories have an unhappy ending.
That said, E-caps, especially from dodgy manufacturers (most are now gone as a result of natural selection) can dry out and fail in various ways, but good brands like Nichicon, Siemens, etc. last forever when used within their limits
Thanks for the replies, much appreciated.
I went ahead and soldered 2000uf panasonic electrolytics across the op amp power supply, also the driver transistors power supply and the output transistors power supply.
After listening the sound has become a little more solid, a touch more of everything, detail, seperation etc. I and the family can hear it without a struggle. It is not much but okay for a quick cheap mod. Overall i am disppointed i thought i would get more than that.
My speculation was that the 2000uf caps would charge up and act as a local power supply for the components to use, do you think i was wrong in this assumption and that the components really use the mains power supply. Perhaps the 2000uf caps are just serving the smooth the dc a little.
What do you think.
I went ahead and soldered 2000uf panasonic electrolytics across the op amp power supply, also the driver transistors power supply and the output transistors power supply.
After listening the sound has become a little more solid, a touch more of everything, detail, seperation etc. I and the family can hear it without a struggle. It is not much but okay for a quick cheap mod. Overall i am disppointed i thought i would get more than that.
My speculation was that the 2000uf caps would charge up and act as a local power supply for the components to use, do you think i was wrong in this assumption and that the components really use the mains power supply. Perhaps the 2000uf caps are just serving the smooth the dc a little.
What do you think.
McCormack did something similar in their DNA series of amplifiers, which avoided large tank capacitors in favour of smaller ones between supply and ground at each output device.
The self-resonance of smaller caps occurs at higher frequencies than for larger ones, part of the reason this technique is used for switching power supplies. The distributed configuration does want a hefty ground plane, otherwise one could get ground bounce from charging currents into the caps since rectifier conduction for capacitor-input filtering will be short pulses with high peak currents.
Whether the equivalent series resistance (ESR) of the small caps at the output devices is lower than the ESR of a tank with the wire leading to the output stage will be a function of the particular design. Equivalent series inductance would be reduced (see self-resonance above), which might have an effect if the power supply rejection ratio is moderate.
To answer your question, at a first approximation both sets of capacitors are running in parallel so you simply increased the supply capacitance. More thorough modeling would include ESR and ESL in the main & auxiliary capacitors, and also for the wire from main caps to output devices. One might think that would be negligible, but I ran across a power supply problem where a wide circuit board trace showed enough resistance to trigger a low voltage detector. We ended up widening the trace some more and using 2 ounce copper to get around that, so a class-AB amp driving a speaker with a funky impedance curve might well show a non-trivial voltage drop on a reasonable hunk of wire.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.