While replacing the electrolytic caps (10/43uF and 25V) in my in/out audio signal path of my audio interface, is it worthwhile to also bypass with ceramics? Like a 0.1uF 50V X7R? on each? Or is this primarily used for PSU caps?
It may cause a 'tanking' resonance which depending on the circuit may upset things and cause some oscillation.
Stefan
Stefan
Don't think bypass. This implies placing a low esr capacitor across an electrolytic to allow high frequency to pass more easily.
Think supply rail decoupling.
This is placing medium esr capacitors with ultra low inductance at the point of current change. This is AT the POWER Pins of the current changing device.
Decoupling is just as important in the digital side as in the analogue side. The digital chips change current very quickly. It's this CHANGE in current that must be decoupled at the power pins, to help prevent the supply rail to other devices being starved of clean power.
Think supply rail decoupling.
This is placing medium esr capacitors with ultra low inductance at the point of current change. This is AT the POWER Pins of the current changing device.
Decoupling is just as important in the digital side as in the analogue side. The digital chips change current very quickly. It's this CHANGE in current that must be decoupled at the power pins, to help prevent the supply rail to other devices being starved of clean power.
Last edited:
Ah never noticed it was coupling caps. Have the amps got coupling caps? If so you could just do away with them altogether. Normally gives a nice boost
Coupling caps are there for a reason.
'Bypassing' coupling caps is unnecessary. If it sounds different with a 'bypass' then there is something wrong with the design or the bypass arrangement.
'Bypassing' coupling caps is unnecessary. If it sounds different with a 'bypass' then there is something wrong with the design or the bypass arrangement.
OK, I'm confused, and bear with me because I just learned about all this stuff last weekend 😱 Here's the PSU of my audio interface:
Based on stuff I read I had already planned to bypass C12 and C18 with 0.1uF ceramic caps. C12 and C18 are my coupling caps for the 15V rails, right? So, to bypass those or not?
But there are also smaller 10uF electrolytic caps in my audio in/out signal path (not shown in the schematic), not really sure what they do but I think they maybe filter the signal before the op amps, and I was wondering about bypassing these with ceramics too. I had read in a few places that electrolytics suck for audio chains and that bypassing with small ceramics helps the sound, by allowing higher frequencies, as you mention.
Am I way off track?
If you don't know what the caps do why do you think they might need bypassing?
Don't believe everything you read, especially if you read it about audio on the internet.
Don't believe everything you read, especially if you read it about audio on the internet.
Care to make this a learning experience, or no?
I've read in several places from several "audio guru" types on the internet that you should bypass electrolytic caps with small ceramics, and I've also read the opposite. There seem to be two camps but I don't have the background to discern between the two. Can you or someone else please elaborate and help me clarify so I can take the best course of action?
All capacitors have some series inductance and resistance. This means that as frequency rises their impedance falls but not to zero. After the minimum the impedance rises again with frequency, but now it is inductive not capacitive. The frequency of minimum impedance can range from some kHz (large electrolytic) to hundreds of MHz (small ceramic cap with very short leads). In any case, the capacitor is still a small impedance for some frequency range either side of the minimum. Nothing awful happens if you go some way beyond the minimum!
I suspect the 'audio gurus' have read that impedance can rise with frequency, have been alarmed by this, and wrongly concluded that the solution is to add a parallel small cap as a so-called bypass. So for a start they are solving something which is not usually a problem. Secondly, their 'solution' can actually raise the net impedance above the minimum as the small extra cap is now in parallel with the inductance of the big cap. Fortunately, the big cap may be a bit lossy so the parallel resonance is well damped so little harm is done. If adding a 'bypass' results in a real change in sound (i.e. a change you can hear with your ears - no peeking!) then it is likely that this is caused by the increase in impedance above the minimum - so you are achieving exactly the opposite of what you wanted.
Coupling capacitors are often small electrolytics which are OK over the whole audio range and some way above it. No need for bypassing.
Decoupling capacitors may be larger electrolytics and here there could be a problem of impedance rise at higher audio frequencies (although not always). In such cases it is important that the bypass is placed right at the circuit and not just across the big cap. Then the resistance of the wiring helps damp any parallel resonance. In some cases the real problem is not the big cap but the length of the wires to it.
To understand the problem and decide which camp to believe you need to learn about the following issues:
- first order passive CR filters
- coupling and decoupling caps - what they do (they are first order CR filters, coupling caps form part of high pass filters while decoupling caps form part of low pass filters)
- parallel and series LC resonance, and Q
I suspect the 'audio gurus' have read that impedance can rise with frequency, have been alarmed by this, and wrongly concluded that the solution is to add a parallel small cap as a so-called bypass. So for a start they are solving something which is not usually a problem. Secondly, their 'solution' can actually raise the net impedance above the minimum as the small extra cap is now in parallel with the inductance of the big cap. Fortunately, the big cap may be a bit lossy so the parallel resonance is well damped so little harm is done. If adding a 'bypass' results in a real change in sound (i.e. a change you can hear with your ears - no peeking!) then it is likely that this is caused by the increase in impedance above the minimum - so you are achieving exactly the opposite of what you wanted.
Coupling capacitors are often small electrolytics which are OK over the whole audio range and some way above it. No need for bypassing.
Decoupling capacitors may be larger electrolytics and here there could be a problem of impedance rise at higher audio frequencies (although not always). In such cases it is important that the bypass is placed right at the circuit and not just across the big cap. Then the resistance of the wiring helps damp any parallel resonance. In some cases the real problem is not the big cap but the length of the wires to it.
To understand the problem and decide which camp to believe you need to learn about the following issues:
- first order passive CR filters
- coupling and decoupling caps - what they do (they are first order CR filters, coupling caps form part of high pass filters while decoupling caps form part of low pass filters)
- parallel and series LC resonance, and Q
Thank you! I see that I have some learning issues, but for now, am I understanding the basics correctly?
Decoupling caps along with the RC network in the PSU help filter and shape the DC current, which drives the amps. The digital audio output from those amps is AC, which is HP filtered by coupling caps to remove any DC signal.
So if I placed 0.1uF ceramic cap flush with the underside of the PCB between the leads of my decoupling caps (C18, C12, C60, and C22, right?), would this be beneficial?
And no need to bypass the coupling caps.
Really appreciate the help!
Decoupling caps along with the RC network in the PSU help filter and shape the DC current, which drives the amps. The digital audio output from those amps is AC, which is HP filtered by coupling caps to remove any DC signal.
So if I placed 0.1uF ceramic cap flush with the underside of the PCB between the leads of my decoupling caps (C18, C12, C60, and C22, right?), would this be beneficial?
And no need to bypass the coupling caps.
Really appreciate the help!
looks like you have two different duties mixed up.
A Signal coupling capacitor passes a signal from one stage to the next.
It is usually there to block DC in case the DC voltages either side are different.
A signal coupling capacitor if of the wrong type or the wrong size can interfere with the wanted signal. and in the extreme becomes a filter or adds distortion. Size and type sort this potential problem.
A supply rail decoupling capacitor supplies short term current to a device that suddenly changes it's current demand. Again size and type and especially location/length of connections determine whether the capacitor is an effective decoupler.
I don't see any sch. Your link is corrupted.
A Signal coupling capacitor passes a signal from one stage to the next.
It is usually there to block DC in case the DC voltages either side are different.
A signal coupling capacitor if of the wrong type or the wrong size can interfere with the wanted signal. and in the extreme becomes a filter or adds distortion. Size and type sort this potential problem.
A supply rail decoupling capacitor supplies short term current to a device that suddenly changes it's current demand. Again size and type and especially location/length of connections determine whether the capacitor is an effective decoupler.
I don't see any sch. Your link is corrupted.
No. That is about the worst possible place to put a 'bypass'. As I said, put the 'bypass' (actually, it is extra HF decoupling) where it is needed. If the electrolytic is already right at the circuit then no extra bypass is needed.grantos said:
The embedded link won't work for me so
https://www.gearslutz.com/board/att...ommend-booster-caps-psu-delta1010psuschem.png
Well, these decoupling caps, particularly c12 and c18, do have longish (1cm or so) leads to allow them to be pushed out of the way by another pcb... I think this is is what you mean when you say they are potentially not "right at the circuit"
https://www.gearslutz.com/board/att...ommend-booster-caps-psu-delta1010psuschem.png
Well, these decoupling caps, particularly c12 and c18, do have longish (1cm or so) leads to allow them to be pushed out of the way by another pcb... I think this is is what you mean when you say they are potentially not "right at the circuit"
C12 and C18 are reservoir caps, not decoupling caps. What does the regulator datasheet advise? 1cm is not "longish" unless you are feeding fast logic or RF circuits. Forget bypasses.
OK I have just one more question for you, if you don't mind:
The coupling caps in the signal path are 47uF/25V. They're 5mm diameter, 2mm lead spacing.
Ideally I'd like to replace them with Silmic ii's but am wondering if it's worth the trouble. There are 47uF/10V caps of the right dimensions, but the voltage is below the originals. In this application, isn't the capacitance more critical than the voltage, which is pretty low in the signal chain? How could I find the actual voltage requirement without having a schematic?
If those are out, I could add a larger cap of the right specs but larger (8mm diameter, 3.5mm spacing), and because of the grouping some would have longer leads, but nothing crazy. Should be fine here right?
The coupling caps in the signal path are 47uF/25V. They're 5mm diameter, 2mm lead spacing.
Ideally I'd like to replace them with Silmic ii's but am wondering if it's worth the trouble. There are 47uF/10V caps of the right dimensions, but the voltage is below the originals. In this application, isn't the capacitance more critical than the voltage, which is pretty low in the signal chain? How could I find the actual voltage requirement without having a schematic?
If those are out, I could add a larger cap of the right specs but larger (8mm diameter, 3.5mm spacing), and because of the grouping some would have longer leads, but nothing crazy. Should be fine here right?
For coupling caps the minimum voltage rating should be equal to or greater than the highest possible supply rail voltage for the relevant part of the circuit. Note that the voltage rating is for DC voltage, so where they come in the signal chain is completely irrelevant. As always, the voltage rating is much more critical than the capacitance; wrong capacitance for a coupling cap and you lose bass or gain infrasonics (not too important), while too small voltage rating means you fry something in the amp.
The quick answer is that you need a schematic, or a voltmeter and the ability to work without a schematic. Best to leave things as they are, otherwise your 'upgrade' may become a significant downgrade.
The quick answer is that you need a schematic, or a voltmeter and the ability to work without a schematic. Best to leave things as they are, otherwise your 'upgrade' may become a significant downgrade.
OK thanks, it looks like I'd be under voltage requirement with 10V.
I'll go with the larger caps assuming the little extra lead length shouldn't matter if they're feeding op amps.
I'll go with the larger caps assuming the little extra lead length shouldn't matter if they're feeding op amps.
Lead length rarely matters in audio coupling caps, whatever they are feeding. Physical size may matter, as larger caps can pick up more hum and interference or increase unwanted capacitve coupling between stages. Hence 'upgrades' (which are unlikely to make an improvement anyway) can degrade sound.
Got it. You know how it goes... I first fixed the PSU and got overly excited about the possibilities, which lead to an obsession about having the "best" possible upgrades, which led me to be branded by Elna after some cool dudes said they're cool (even though I'm positive I'd never hear the difference, because e.g. I can barely detect a difference between amp tubes), all making me willing to compromise the original design for some peace of mind. All of which is really just an elaborate means of procrastinating the stuff I actually I have to do, which has absolutely nothing to do with audio. Cheers. 
- Status
- Not open for further replies.