I am building a small (small as I can make it) board for Balanced-to-Unbalanced (and vice-versa) conversion. Everything fits so far as I used 22uF 4mm electrolytic bypass capacitors on the incoming rails. I usually see 47uF caps used for this purpose, but those are 5mm caps. I would have to enlarge the board slightly and redo my KiCad layouts to fit 5mm caps. Will the 47uF to 22uF increase in bypass capacitance REALLY make any difference? My power supplies are superbly well-filtered to begin with so I kinda doubt that it would, but I am wondering in case my design gets used in other equipment that has less robust power supplies. Opinions?
If the rails are clean and this is all low power opamp type circuitry then no problem using 22uF. In fact even that is pretty large for a bypass cap. Everyone has their own ideas and there is no right and wrong.
1uF, 2.2uF or 4.7uF should also work just as well. They are not reservoir caps, they don't need to try and hold the rails up as current is drawn, they just provide a low impedance at higher frequencies.
1uF, 2.2uF or 4.7uF should also work just as well. They are not reservoir caps, they don't need to try and hold the rails up as current is drawn, they just provide a low impedance at higher frequencies.
I would go to 22uF. As the capacitance increases, the resonant frequency of the cap goes lower. This mean that over this frequency, the cap abandones its behaviour as cap and acts as an inductor.
Mainly for electrolytic units.
In any case add a ceramic 0.1uF in parallel to overlap its resonant frequency, and the 0.1 remains capacitive when 22 is inductive
Mainly for electrolytic units.
In any case add a ceramic 0.1uF in parallel to overlap its resonant frequency, and the 0.1 remains capacitive when 22 is inductive
100nF between + and - rails is usually enough if you put it straight on the Ic's pins.There are many professional circuits with very high psrr that need no bypass cap at all.Even the ne5534that you didn't like can work with 100nF depending on the circuit's gain and structure.
Thanks for the replies. The small capacitance addition is a good idea; I already have 100nF caps across the rails on every one of the ICs.
Be careful doing that, as the smaller capacitor can resonate with the larger capacitor's ESL, counterintuitively raising impedance. I would stick to a single capacitor unless I had a specific reason to do otherwise.
10-22uf should be great in addition to the small ceramic ones.
More will begin to degrade imo, add more inductance, which isn’t desirable.
As long as the electrolytic parts aren’t of the super low esr variety, everything should work well for any unwanted interaction.
More will begin to degrade imo, add more inductance, which isn’t desirable.
As long as the electrolytic parts aren’t of the super low esr variety, everything should work well for any unwanted interaction.
This is true, but lets suppose that the impedance looking backwards to the PSU is suffuciently low to reduce rhe Q of this resonance to negligible values. This is why in RF we use several values in parallel of capacitors of different kinds.
for digital but for analog .... ?! Is he decoupling a crystal of a dac ? For e ceramic class II in the analog domain is a no go... sonically, most of the time. prefer a radial mkp, acrylic smd, PEN, etc... all but ceramic class II if on the signl way.
Suppose the 0.1 was tight to the opamp but the 22uf was some length of track separation in the pcb layout. Would that help
It really depends on if you mean a "decoupling" cap or a "Bulk" cap.
The former is to remove HF noise by dumping AC to ground.
The later is to provide immediate current to the device/board under power. Power actually takes time to arrive. It might not seem like that's significant, but it is.
If your signals are line level in and out and accepting 10K or higher impedance there will be virtually no current for the your opamps to deliver. So the bulk cap can be pretty small. On my headphone amp I used a 400uF as the main power supply cap and 10uFs for the each opamp + a 100nF (or maybe a 10nF) as per the datasheet for decoupling. The 400uF is the bulk cap providing a near infinite amount of instantaneous current. Required because the amp has a 1 watt output.
I'm not sure about the "becoming" inductive being the issue, but large caps tend to pass lower frequencies. Small caps high frequencies. So there is a technique of "decading" decoupling/bypass caps. Things like a 1uF, 100nF, 10nF in parallel. It covers a wider range of noise frequencies.
Using >1uF as a decoupling cap isn't really a thing, the frequency response of large caps isn't going to filter much except maybe 50/50Hz ripple.
In digital electronics it's typical to place something like a 22uF and a 100nF across the power pins of all "powerful" chips like MCUs but that is to combat the square wave current surges.
The former is to remove HF noise by dumping AC to ground.
The later is to provide immediate current to the device/board under power. Power actually takes time to arrive. It might not seem like that's significant, but it is.
If your signals are line level in and out and accepting 10K or higher impedance there will be virtually no current for the your opamps to deliver. So the bulk cap can be pretty small. On my headphone amp I used a 400uF as the main power supply cap and 10uFs for the each opamp + a 100nF (or maybe a 10nF) as per the datasheet for decoupling. The 400uF is the bulk cap providing a near infinite amount of instantaneous current. Required because the amp has a 1 watt output.
I'm not sure about the "becoming" inductive being the issue, but large caps tend to pass lower frequencies. Small caps high frequencies. So there is a technique of "decading" decoupling/bypass caps. Things like a 1uF, 100nF, 10nF in parallel. It covers a wider range of noise frequencies.
Using >1uF as a decoupling cap isn't really a thing, the frequency response of large caps isn't going to filter much except maybe 50/50Hz ripple.
In digital electronics it's typical to place something like a 22uF and a 100nF across the power pins of all "powerful" chips like MCUs but that is to combat the square wave current surges.
The other place you will find "bypass" / "decoupling" caps is when they are used in a slightly different context.
Coupling / AC pass caps on amplifier input and line outputs that allow two devices to float a different DC offset but still exchange AC audio signals.
It's exactly the same premise. The only difference is, with an AC coupling cap it's the AC you want to transit the cap to the input of the amp, rather than dumping it to ground.
Consider that 220uF is more than enough to capture all the audio band, all the way down to about 25Hz. It says nothing about how it will handle 1Mhz noise though. That's where the smaller caps come in. If you put a 100nF as your AC coupling cap, you wouldn't get any bass.
Coupling / AC pass caps on amplifier input and line outputs that allow two devices to float a different DC offset but still exchange AC audio signals.
It's exactly the same premise. The only difference is, with an AC coupling cap it's the AC you want to transit the cap to the input of the amp, rather than dumping it to ground.
Consider that 220uF is more than enough to capture all the audio band, all the way down to about 25Hz. It says nothing about how it will handle 1Mhz noise though. That's where the smaller caps come in. If you put a 100nF as your AC coupling cap, you wouldn't get any bass.