Class AB input coupling cap question

[Bengali]

Hi,

I would appreciate your experience and thoughts on input capacitors in class AB amps. I have built few and noticed they usually require minimum 4.7uf-10uf cap. These kits I have built are designed with electrolytic capacitor in mind.

To replace these with nice film cap is very costly. My question is if you know your preamp has output coupling cap, is it still better to have the electrolytic input cap on the amp? I was told I may get turn on thump if no cap is used.

My dilema is this. The preamp I will use has nice mundorf output coupling cap.
The kit amp uses a panasonic fc 10uf 50v cap for input coupling. The panny electrolytic would defeat the sonics of the mundorf, would it not?

Thanks

 

[Mooly]

A single cap between preamp and power amp provides all the DC isolation needed. There is almost certainly going to be switch on/off thumps and noises. Thats easily fixed with a good speaker delay circuit that drops out quickly on power off and gives a repeatable and reliable switch on delay.

 

[indianajo]

None of my 4 amps, the preamp, or the mixer, thump at turn on or off. The 4 organs don't, with 16-35 W amps. The Peaveys have delay circuits, the Dynakits and Hammonds don't. So what is wrong with your amps?
I built a sine wave generator that was whanging the mixer op amps right to the rail due to the lack of output caps on the generator, the lack of input caps on the mm cartridge input of the mixer, and the different power supplies. So in some situations there have to be caps or transformers coupling things.
I replaced the tantalum 5uf caps at the inputs of my Dynakit ST120 amp, which pair of had popcorn noise from the day I bought them, with two 4.7 uf 50V Aerovox Gold ceramic caps. Much better. Was my first solid state amp repair, no internet support or schematic diagram then, I had no scope, or any idea where the popcorn noise was coming from, 1990 until 2010.
Aerovox golds are $5 each, not cut rate bargain parts. Started using Aerovox gold 10 uf 50V ceramic caps in the organs, sparkling highs and tight bass hits replacing the tired 38 years old electrolytic coupling caps electric kazoo effect. I think the 50V rating on a 1.6 V signal is swamping the reputed ceramic non-linearities, but what do I know? Watch out, the ceramic cap bogeyman will get ya!@

 

[Bengali]

Thanks guys. Good to know info you provided.

Nothing wrong with my amps. I was stating that it kinda makes no sense in putting down good money for high end film caps in a pre and then using electrolytic down the chain in the amp. That's all.

Never heard aerovox gold ceramic before. nos part? Will google it now.

Have a blessed 2013!!!

Cheers

 

[indianajo]

AVX 4.7 uf 50V X7R cap SE035C475KAR - AVX - CAPACITOR CERAMIC 4.7UF, 50V, | Newark
I managed to snag ten COG 10 uf caps somehow for $4 each, that is a better dielectric. I put those in the organ where Roederstein wound 63V film caps wouldn't fit.
With the AVX 4.7 UF 50V Z5U caps at the input of my ST120, it sounds the same as the CS800s amp which has no caps at all in the path. This is on SP2-XT speakers, which are the best I've ever owned but can be beat I suppose. The ST120 with djoffe bias set accessory sounds better than the stock ST70 tube amp with new e-caps, plate caps, output tubes and good B+ voltage (so 5AR4 tube is okay).

 

[JMFahey]

class AB amps. I have built few and noticed they usually require minimum 4.7uf-10uf cap.

Input cap values have absolutely no relationship to amp class.
Period.
Please read about RC time constants, there's your answer.

 

[Ian Finch]

Exactly. A large (10x the required RC constant) value Electrolytic is often fitted at the input to avoid the rising low frequency distortion due to an electrolytic - even high quality electrolytic. In this case, a small 1uF film cap is adequate for the role of a 4.7-10uf electrolytic. Much cheaper than a big one too!

 

[Barleywater]

Exactly. A large (10x the required RC constant) value Electrolytic is often fitted at the input to avoid the rising low frequency distortion due to an electrolytic - even high quality electrolytic. In this case, a small 1uF film cap is adequate for the role of a 4.7-10uf electrolytic. Much cheaper than a big one too!

Nonesense! RC constant determines cutoff. In cases where previous stage output capacitor and input capacitor are used, series connection reduces this, requiring large value capacitors for good low frequency performance without large phase shift in 20Hz-200Hz range.

 

[TANDBERGEREN]

Nonesense! RC constant determines cutoff. In cases where previous stage output capacitor and input capacitor are used, series connection reduces this, requiring large value capacitors for good low frequency performance without large phase shift in 20Hz-200Hz range.

Good call.

And to have this DC-barriere between especially separate components is actually a very good idea anyhow.

 

[Ian Finch]

Nonesense! RC constant determines cutoff. In cases where previous stage output capacitor and input capacitor are used, series connection reduces this, requiring large value capacitors for good low frequency performance without large phase shift in 20Hz-200Hz range.

Yes, RC constant determines cutoff as agreed already here. However, many designs appear to have cutoffs ~1Hz which is how necessary? See Self - Distortion 8: Capacitor distortion, in recent editions of "Audio power amplifier design Handbook". There are examples of distortion measured with both input and output caps. The recommended values required are in the order of 10 x the cutoff frequency with electrolytics to maintain lowest THD. In the past we often saw film caps paralleled with electrolytics in an attempt to improve this. It is not a single issue with one solution but a single film cap is usually better at maintaining low distortion at the input of a domestic amplifier.

 

[AndrewT]

Barley and Tand,
What RC time constant is required for passing the FULL AUDIO bandwidth of 20Hz to 20kHz?

 

[Barleywater]

In: Question on input dc blocking capacitors #32 I show simulation of simple 1st order high pass filter with -3dB point at 6Hz. Response is effectively flat >20Hz, but phase skewed by group delay, as exhibited in square waves depicted. This is accurate representation of how low frequency transients get re-ordered, sometimes quite audible.

This is another reason why 10x is often suggested in choosing cutoff frequencies.

Regards,

Andrew

 

[AndrewT]

Are you saying that 10times wider than required bandwidth is adequate, or are you saying that 6Hz (about 3times), or are you saying some other value for RC?

What about the HF end of the bandwidth?

 

[Barleywater]

10x, so as elsewhere: in the neighborhood of 2Hz and lower for cutoff frequency.