back to back coupling capacitors

[reddog]

I am wondering if "back to back" ( negatives tied together) caps follow the same rules as series ( positive tied to negative) caps in term of value and voltage..
Would two "back to back" 220uf 16v caps equal the value of 110uf 32v?
I want to replace my back to back output coupling caps with a single non polar and i am trying to figure out the corresponding replacement value..
Thanks for your thoughts..

[dhaen]

In my experience the value becomes imprecise. This is because various caps behave differently when reverse biased. Some are almost short circuit to the reverse voltage, while others behave almost correctly, thus reducing the overall capacitance value.

So in the case of your output coupling cap, the value could be anywhere between half and the whole value of one cap. I would choose the whole value, at a voltage rating just higher than one supply rail.

[AndrewT]

Hi,
try to estimate the high pass roll off frequency of your output capacitor. I think you should aim for about 10 times lower than the lowest frequency you want to pass. For an audio amplifier that equates to 1Hz to 2Hz for the -3db frequency.

Remember that the next stage may (should) also have a DC blocking capacitor and it will add together with this output cap to introduce a worse -3db cut off point.

The formula is Freq=1/2/Pi/R/C
so for a 10kohm input impedance and 1uF cap F(-3db) = 15.9Hz.
If the input stage has 2u2F as it's cap then the actual -3db is at about 23Hz. F(-1db) at about 46Hz and phase errors up to about 250Hz. there's why I recommend the 10 times figure.

Some research has suggested that the increase in distortion as the output frequency falls towards the F(-3db) point increases rapidly as the blocking voltage STARTS to rise, you might be able to measure this. This method tends to produce an even bigger output cap.

[rascal101]

Some research has suggested that the increase in distortion as the output frequency falls towards the F(-3db) point increases rapidly as the blocking voltage STARTS to rise, you might be able to measure this. This method tends to produce an even bigger output cap.

Just wanted to understand what you mean by distortion. Is this due to coupling capacitor esr having different values with respect to frequency and temp? Or the decrease in the impedance of the low pass filter cap at roll-off frequency?

Thanks.

[davidsrsb]

It's due to the ecaps having zero or even -ve bias. The only safe way to run ecaps back to back is too have a high value bleed resistor from the common junction to a power supply rail, so correctly biasing the caps. Trouble is that this can cause some big switch on thumps.

[clem_o]

Originally posted by reddog

I want to replace my back to back output coupling caps with a single non polar and i am trying to figure out the corresponding replacement value..
Thanks for your thoughts..

Its my understanding that nonpolar electrolytics are in fact built using back-to-back polar types. So, you may not really gain anything by doing so... (?)

Cheers!