I have built a preamp with active volume control.
This feeds a modified class d amp.
For dc blocking in the signal path i use 22uF electro caps - Jamicon TM-series.
I didn't buy those because of any audio reputation they have, but because they are the cheapest and tiniest part my local store sells.
I have them at the input of the preamp and between the preamp and power amp.
I'm really surprised how good they sound - at least to my noob-ears.
I have also ordered a bunch of 10uF MKP caps.
Those things will be huge compared to the electros.
I will do the listening comparison anyway.
But i would like to hear some opinions from you more experienced guys.
Are there any measurements which (in theory) disqualify electros when film caps are available?
This feeds a modified class d amp.
For dc blocking in the signal path i use 22uF electro caps - Jamicon TM-series.
I didn't buy those because of any audio reputation they have, but because they are the cheapest and tiniest part my local store sells.
I have them at the input of the preamp and between the preamp and power amp.
I'm really surprised how good they sound - at least to my noob-ears.
I have also ordered a bunch of 10uF MKP caps.
Those things will be huge compared to the electros.
I will do the listening comparison anyway.
But i would like to hear some opinions from you more experienced guys.
Are there any measurements which (in theory) disqualify electros when film caps are available?
There are measurements and there are measurements.
The leakage current of a film and leakage vs temperature will be lower. Characteristics at hf (way beyond the audio band will differ.
You say you are surprised at how good they sound... its really just that they have "no sound" at all. The large film caps will be more prone to picking up electromagnetic interference simply because they present a large physical form.
This was fun at the time (the files are long since deleted)
http://www.diyaudio.com/forums/everything-else/250147-listening-test-part-1-passives.html
Look at the picture in post #23 😀
The leakage current of a film and leakage vs temperature will be lower. Characteristics at hf (way beyond the audio band will differ.
You say you are surprised at how good they sound... its really just that they have "no sound" at all. The large film caps will be more prone to picking up electromagnetic interference simply because they present a large physical form.
This was fun at the time (the files are long since deleted)
http://www.diyaudio.com/forums/everything-else/250147-listening-test-part-1-passives.html
Look at the picture in post #23 😀
...yes, that's what i think about:
I have my perf-board-design so tight with the 22uF electros - i struggle to give that up for the 11 times larger lead-spacing of the film caps.
Wow, the replies to the listening tesr are surprising.
It starts interesting, but i didn't read the whole article yet.
Will take the time to read the whole thing during next week (when i get tired of learning mathematics for university).
I scrolled to the end of it to read the conclusion.
There are some advices on how to use electros.
Then i asked myself (and now you guys) the question:
Am i doing that right?:
Mooly adviced me in another thread that using electros on the preamp should be done with the "+" polarity facing the preamp at the in- and output.
If i now omit the dc coupling caps that where originally mounted on class d amp board....do i go better with using two of the 22uF as one bipolar cap?
Always polarise your caps by making an actual measurement of voltage across them if possible. This is important where nominal values are near zero but natural offsets and so on could swing the result either way. I see no advantage in making a bi-polar cap for the majority off applications.
so the correct way of using one single electro cap between preamp and amp would be face its plus towards the preamp and the minus towards amp?
i do not understand theory of the 'direction' of the polarity in the signal path.
i do not understand theory of the 'direction' of the polarity in the signal path.
It's the capacitor's DC polarity, not the signal path. A polarized capacitor must have a positive DC voltage across it, in the way that the capacitor's case is marked.
For example, if there is a +10VDC at the output of the preamp, and about 0VDC at the amplifier input, then the + capacitor lead goes toward the preamp,
so that the voltage across the capacitor is positive, the same as the marking.
Some electrolytic capacitors are non-polarized, so then either connection is ok for them. Often these types have a + at both ends, to denote that they are non-polarized.
^As rayma says 🙂
(and when its not obvious which way around a cap should go then you measure the voltage across it and fit it accordingly. Suppose you had a preamp constructed using opamps (or discrete) and you wanted to use coupling caps at the output. One channel might have an offset voltage of +6 millivolts and the other -11 millivolts. You would fit one cap one way around and the other cap the opposite way such that each cap always had a DC voltage of the correct polarity across it with reference to the cap markings.)
Yes, to be more explicit, most electrolytic capacitors (and tantalum types as well) cannot tolerate significant reverse (negative) DC voltage.
They will "short" if reverse-connected to a DC voltage.
If there's a chance that the DC voltage could be either polarity, or vary in time, then I'd use a non-polarized (NP) type capacitor.
This is essentially two polarized electrolytic capacitors connected in series, with both negative terminals connected together,
and the two positive terminals used as the connections to the rest of the circuit.
Last edited:
Is time a factor for a polarised electrolytic?
If the electro has zero DC bias and is passing an audio (AC) signal.
Does the "time" that the signal is passing in the wrong direction matter?
How short a time period is of no consequence? 1ms, 10ms, 100ms 1s ?
How long a time period could be considered a danger/damaging to the electro? 100ms, 1s, 10s, 100s ?
Is the time related to the reverse voltage?
Is a smaller reverse voltage for a long time somehow equivalent to a higher reverse voltage for a short period?
Could Time*Voltage help define what is critical to non damage?
If the electro has zero DC bias and is passing an audio (AC) signal.
Does the "time" that the signal is passing in the wrong direction matter?
How short a time period is of no consequence? 1ms, 10ms, 100ms 1s ?
How long a time period could be considered a danger/damaging to the electro? 100ms, 1s, 10s, 100s ?
Is the time related to the reverse voltage?
Is a smaller reverse voltage for a long time somehow equivalent to a higher reverse voltage for a short period?
Could Time*Voltage help define what is critical to non damage?
Last edited:
No it doesn't matter as long as the caps reactance (its value) has been chosen correctly such that no voltage differential appears across it due to "slow moving" AC signals.
We know from a few manufacturers that polarised electros can tolerate some reverse voltage.
I have seen figures of 1V max and 1.4V max
What happens if those guide values are exceeded for a short time and a longer time?
I have seen figures of 1V max and 1.4V max
What happens if those guide values are exceeded for a short time and a longer time?
A zero DC bias AND an AC signal necessarily applies a reverse voltage across the capacitor for short (alternate half wave) periods.
It doesn't Andrew.
As the AC voltage goes up and down, each end of the cap does the same. Result, no voltage difference of any polarity across the cap.
Only if the cap is incorrectly chosen in terms of its value relative to the expected signal would you see a voltage appear across it. Think back to Doug Selfs precision preamp and all the interstage coupling electrolytics for a practical example. No significant DC of incorrect polarity can ever appear across them because they are scaled correctly.
Good questions. I would think that much of the reason for reverse polarity damage is from overheating from too much current.
I have seen current-limited reverse connected caps survive ok (not on purpose, though).
The voltage across an electrolytic does affect it's value. I'd always use a non-polarized MKP at the junction of the pre-power amps.
Nice to have someone with a different opinion in this thread.
You would have the MKP because of its non-polarity or because of other reasons (,too)?
What would be a sufficient measurement instruction to decide the polarity of electros?
(I assume it is not as simple as powering up all devices, not wire the signal path and measure potential differences (voltages) between outputs and inputs.)
Step by step details would be very nice.
Electrolytics are happiest with some polarising voltage across them. They may change their value a little if left with none, or become a little leaky. However, this should be of little consequence if the correct value is chosen.
Three options:
1. Use an electrolytic and don't worry too much about it. A shift of value won't matter too much, as an electrolytic should never be used as the setter of the main LF rolloff in the system. A little leakage won't matter either as it is fed from nominally zero volts.
2. Use a non-polar capacitor and accept that it will either be smaller in value or much larger in size.
3. Use two electrolytics with a DC bias resistor, in series for AC and in parallel for DC.
Three options:
1. Use an electrolytic and don't worry too much about it. A shift of value won't matter too much, as an electrolytic should never be used as the setter of the main LF rolloff in the system. A little leakage won't matter either as it is fed from nominally zero volts.
2. Use a non-polar capacitor and accept that it will either be smaller in value or much larger in size.
3. Use two electrolytics with a DC bias resistor, in series for AC and in parallel for DC.
And just to throw a spanner in the works 😉
Ever measured the voltage across electros in a circuit that has been powered down for a few minutes, or hours, or days ? All kinds of things happen at the few tens of millivolt level as rails collapse unequally and semiconductors drop out of conduction.
Ever measured the voltage across electros in a circuit that has been powered down for a few minutes, or hours, or days ? All kinds of things happen at the few tens of millivolt level as rails collapse unequally and semiconductors drop out of conduction.
Yes, that may happen, of course.
But I'd call it a "non problem" 😱
If polarity is forward, the small leak inherent in electros will discharge it.
If reverse, the same, only faster.
No big deal.
But I'd call it a "non problem" 😱
If polarity is forward, the small leak inherent in electros will discharge it.
If reverse, the same, only faster.
No big deal.
- Status
- Not open for further replies.