Why a so high value for input cap?

[jeepy]

Hi everybody, I am just wondering why a 250uF! is used as the input cap in the original H. Kardon Citation 12 amplifier. Is there any way to lower this value with ad hoc modifications in the input stage? Many thanks
, JP

[lineup]

For 33k input impedance you can use 1uF or 2.2uF filmcap. Polyester or propylene.
I do not know why 250uF

[jeepy]

Originally Posted by lineup

For 33k input impedance you can use 1uF or 2.2uF filmcap. Polyester or propylene.
I do not know why 250uF

Thanks! much better than a big electro.!

[pauljguy]

It would appear that the designer was trying to match DC and AC components on both sides of the differential amplifier. By placing the 250uf on the input side, it balances out the time constant (roughly) from the other side (assuming input is from a low source impedance).
My guess is that by using a more normal input capacitor (a few uf's) you may get worse thumps during turn-on/off or low frequency artifacts during overload and non-symetrical behaviour of the output circuit. Matching DC and low frequency circuits on either side of the diff-amp will help stabilize the bias. If I make up an amp, I find it a good habit to match AC and DC properties of either side of the differential pair. Sometimes though, you end up with weird circuit components like the above case.
If you can tolerate some subsonic bias wandering, I think you can safely use the lower value input cap.

Since there should virtually no audio voltage drop across the original 250 uf cap, it should not affect the sound, even if non-linear. The existing circuit also incorporates the 250 uf on the other side - the two should more or less balance each other's imperfections out. Electrolytics have very poor value tolerance, so the balancing won't be all that great.
The 250 uf's probably don't see much DC voltage, so their polarization could be a problem. The originals may have been chosen with that in mind. A general purpose electrolytic may not behave as it should in this circuit.
The other 250 uf on the feedback side sets the low frequency cutoff - you don't want to mess with that value. It will also affect the sound (if non-linear) more than the input capacitor. Worst case scenario occurs at .3Hz (Fc where Xc=1800 ohms). Any non-linearity from the feedback side 250 uf would be reduced by roughly 100 (40 db) once you got up to 30 Hz. Electrolytics aren't all that bad.... I've checked quite a few, even ones that are 30-40 years old. High value ceramic caps are a very different story....
Replacing the feedback side cap with film types is worthwhile if the values aren't too large. 250 uf is a pain to replace in film type.
I would be more concerned for the feedback side cap, as it will have a greater influence on the sound. I hope I haven't opened a can of worms...

-Paul

[CBS240]

Originally Posted by lineup

For 33k input impedance you can use 1uF or 2.2uF filmcap. Polyester or propylene.
I do not know why 250uF

Xc for 1uf at 20Hz is 8K, or 24% of 33K. 2.2uf is 11%. I would probably go with a larger cap for this circuit.

20uf is 1.2%,................

20uf Polyproylene film caps???


As for me I no longer use such big input caps, I added an input buffer to my IPS and bumped the input Z up to ~500K so 800nf-1uf is plenty.

[Nelson Pass]

Highly likely Harmon had a big box of those caps. BTW, the Citation 12
is for all intents and purposes the amp from the original RCA transistor
manual.

[ostripper]

I have same 33K input imp. on all my amps. I won't go below 4.7uf for the reason CBS mentioned.

MKS2B044701K00KSSD WIMA Polyester Film Capacitors
VERY good input cap for $2.22 .

OS

[CBS240]

This is the input stage buffer I use, it is not inside any global loop, but it is quite linear. FJZ594 is a tiny J-fet intended for mic inputs, similar to a 4117 in Gm. Very little input capacitance, but quite linear as a follower if used at very low currents. 2SC6026 from Toshiba specializes in very linear gain, measured at about 300.


Now if I can just find a source for some 800nf - 1uf Polystyrene film caps....

[pauljguy]

Originally Posted by CBS240

Now if I can just find a source for some 800nf - 1uf Polystyrene film caps....

Why polystyrene? Your circuit looks like it isn't particularily fussy for capacitor types. You have low AC and DC voltage, your circuit is nice and linear, leakage current isn't a big issue , and ESR, "soakage", tempco are not critical here. The only audio thing that might be of concern is the voltage coefficient of capacitance (non-linearity), but your AC voltages across the cap aren't all that high.

I've tried to measure distortion across film capacitors and even under worst case conditions where capacitor reactance is same as load resistor, the distortion is below what I can measure (-110db). That plus the size, cost, poor availability, and vulnerability to soldering makes polystyrene a poor choice. Now if you were using it in a long time interval integrator with short circuit reset, polystyrene is a decent choice.

Even if you are paranoid about non-linearity, you can always increase the value of capacitor, and push the distortion products into subsonics. With higher capacitance, there is a smaller AC voltage across the cap to generate non-linearity. That voltage across the cap (and distortion) decrease with increasing frequency above the cutoff frequency of a highpass filter. That's why electrolytics are often used. Problem with THEM, is that they deteriorate in performance after many years. Where possible I try to avoid electrolytics, especially where non-linearity might affect audio signals. Some electrolytics (tantalum or aluminum) seem OK, some are kinda poor for linearity. Make the value large enough and you can effectively sweep the distortion "under the rug". Hence the large value of electrolytic in the circuit from the original post.

Paul

[DevilsDance]

What kind of nonlinearity are you talking about? Have ever tried to measure THAT nonlinearity on your acoustic system (drivers)? =) I bet more that 70% of people is using mp3 like a source of their music. Think about it.