I've seen on a bunch of schematics several capacitors wired in parallel to each other. This makes sense to me to increase the overall capacitance and energy reserve available to the amplifier. However, I have seen quite a few which would have for example several 10,000 uF caps, then a 100 nF cap along with them. What would this do? It's a speck compared to a mountain.
Long story short (the long story is sure to come out in this thread), the larger capacitors don't work as well (they don't have as low an ESR) at high frequencies as they do at lower ones, and the smaller cap makes sure there's a lower impedance across the cap at higher frequencies. These "higher frequencies" are usually above the audio range, but there's still a need for lower impedance to help with stability.
Modern electrolytic caps can have very low ESR. Paralleling small value film caps to a low ESR cap can lead to high frequency "ringing".
That`s the short story, no doubt the long story will follow when the more qualified members (than me) will chime in here.
That's 100% correct: you can see the deleterious effects of small parallel capacitors here:
http://www.diyaudio.com/forums/powe...lm-caps-electrolytic-caps-17.html#post2257381
The larger the electrolytic is, and the smaller the plastic or ceramic cap is, the worst the result is.
Absolutely nothing.
Like many other things you will see, do not expect a rational explanation. This goes from bypass capacitors to magic markers on CDs. I'm not kidding.
Tim
Just to give people an insight as to how it is done in industrial power supplies.
A piece of equipment I worked on last week 450VDC to 3 phase AC inverter, 100A per phase dI/dt 100A/us worst case.
Supply from the batteries was cable to a circuit breaker then a small air core filter choke. The choke fed a bank of decoupling capacitors (60000uF) these were connected by a sheet sandwich bus, top sheet positive bottom sheet negative with a 0.5mm mylar sheet separator. The sandwich bus which was 350mm wide fed the IGBTs directly, there were some box type decoupling capacitors directly connected between the rails at the IGBT end of the bus. This is how a load is fed where low inductance is a design goal. There were no film capacitors connected across the decoupling electrolytics. If the filter capacitors are not mounted on the same board as the load the stray inductance in the leads feeding the load may be more than the combined ESL of the filter electrolytics.
To put audio loads in perspective an amplifier producing 2000W of 20Khz sine wave into an 8 ohm load has a dI/dt of 1.9A/us so we are not exactly talking about demanding loads which require extreme bypassing measures.
As a counterpoint a 160Kw Inverter in my garage has the 3 phase rectifier coupled to the IGBT bridge with some copper straps and a fuse, filter capacitors are 12000uF and there are no film capacitors or sandwich bus used anywhere dI/dt is determined by the load which due to lead inductance is probably less than 20A/us
In PCB design ground planes for supply with distributed SMT capacitors rountinely handle dI/dt's over 1000A/us, the GSM telephony board lying against the screen as I type this is testimony to that. Even a modern CPU drawing 100A or so with a 3GHz "square wave" clock
will need some serious bypassing,
A piece of equipment I worked on last week 450VDC to 3 phase AC inverter, 100A per phase dI/dt 100A/us worst case.
Supply from the batteries was cable to a circuit breaker then a small air core filter choke. The choke fed a bank of decoupling capacitors (60000uF) these were connected by a sheet sandwich bus, top sheet positive bottom sheet negative with a 0.5mm mylar sheet separator. The sandwich bus which was 350mm wide fed the IGBTs directly, there were some box type decoupling capacitors directly connected between the rails at the IGBT end of the bus. This is how a load is fed where low inductance is a design goal. There were no film capacitors connected across the decoupling electrolytics. If the filter capacitors are not mounted on the same board as the load the stray inductance in the leads feeding the load may be more than the combined ESL of the filter electrolytics.
To put audio loads in perspective an amplifier producing 2000W of 20Khz sine wave into an 8 ohm load has a dI/dt of 1.9A/us so we are not exactly talking about demanding loads which require extreme bypassing measures.
As a counterpoint a 160Kw Inverter in my garage has the 3 phase rectifier coupled to the IGBT bridge with some copper straps and a fuse, filter capacitors are 12000uF and there are no film capacitors or sandwich bus used anywhere dI/dt is determined by the load which due to lead inductance is probably less than 20A/us
In PCB design ground planes for supply with distributed SMT capacitors rountinely handle dI/dt's over 1000A/us, the GSM telephony board lying against the screen as I type this is testimony to that. Even a modern CPU drawing 100A or so with a 3GHz "square wave" clock
will need some serious bypassing,
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.