Dear friends!
I am designing a headphone amplifier with a built-in DAC (PCM2704C). The datasheet recommends using 1uF ceramic decoupling capacitor on every power pin.
1) Would it be better to replace 1uF ceramic capacitors with film ones?
2) Would it be better if I replace each such capacitor with a pair connected in parallel: 0.1uF (ceramic X7R) and 100uF (electrolytic or tantalum)?
3) Does a tantalum capacitor have an advantage over an electrolytic in this case?
I am designing a headphone amplifier with a built-in DAC (PCM2704C). The datasheet recommends using 1uF ceramic decoupling capacitor on every power pin.
1) Would it be better to replace 1uF ceramic capacitors with film ones?
2) Would it be better if I replace each such capacitor with a pair connected in parallel: 0.1uF (ceramic X7R) and 100uF (electrolytic or tantalum)?
3) Does a tantalum capacitor have an advantage over an electrolytic in this case?
Last edited:
1: no
2: no
3: no
To qualify it a little, use a good quality X7R smd as close to the pin as you can. Using a through hole film will likely put the decoupling further away from the dac which is not what you want. In the other case, you could get a 100n ceramic close enough but the 100µ would be further away. Making your decoupling in the lower frequencies worse. Having the 1µ ceramic close by and a couple of lower value eclo's around can help.
And just keep away from tantalum.
2: no
3: no
To qualify it a little, use a good quality X7R smd as close to the pin as you can. Using a through hole film will likely put the decoupling further away from the dac which is not what you want. In the other case, you could get a 100n ceramic close enough but the 100µ would be further away. Making your decoupling in the lower frequencies worse. Having the 1µ ceramic close by and a couple of lower value eclo's around can help.
And just keep away from tantalum.
Hi. 1uf is just for decoupling agains oscillations and for circuit stability, but so small capacity will not help agains voltage drops because of voltage drop on tiny tracks or current peaks or just fluctuations. I would place also 10uf ceramic near and event bigger , maybe electrolytic , too. Short saying, decouple at all frequencies. Tantalum capacitors are known for their capacity , and not lowest esr, some voltage regulators oscillate with low esr capacitors at output, so tantalum then is the choice.
Thank you!
So, ceramic 1uF as close, as possible. And 10uF ceramic also very close. And a little further electrolityc 47uF? 100uF? 470uF?
So, ceramic 1uF as close, as possible. And 10uF ceramic also very close. And a little further electrolityc 47uF? 100uF? 470uF?
Capacity depends on what exactly are you decoupling, it's reference or supply voltage of some part of chip. Digital part or analog part, and how distant is voltage regulator or voltage source. For digital supply voltage is less critical as for analog. Any noise on analog supply you may hear as noise on audio output, whatever line level or headphone output, for digital you may have absolutly no difference, till ripple begins disturb digital part operation. I would chose 100uf maybe, but check where supply is coming from, and do source allows such loading.
1) That would be a classic mistake - the requirement is ultra-low inductance in such a high-speed decoupling capacitor. Small SMT ceramic caps within a few mm of the chip's pins are the norm here. Specifically an MLCC ceramic cap is mandated (any 1uF chip capacitor will be MLCC)
2) That can help reduce inductance, so yes, so long as they are not too far from the pin, but I'd stick to datasheet recommendations for preference.
3) Any kind of electrolytic won't work for fast decoupling, but instead provides bulk decoupling to reduce supply rail variation at lower frequencies.
Just follow the guidance in the application section of the datasheet...
You can think of it like this: high-speed decoupling prevents device malfunction, bulk decoupling improves performance.
High speed is in the nanosecond regime incidentally.
At these speeds a few cm of PCB trace has significant inductive impedance - it looks more like a resistor than a wire, this is why the cap needs to be right on the pins.