cap coupled output combined with cap coupled input?


2009-02-28 1:24 pm
Most of my dacs have cap coupled outputs. Most of my amps have cap coupled inputs for various reasons (stability or input floats at 2.5volts). Now isnt it a bad thing to run 2 caps in series ( halving the value)? Or does the attenuator between the 2 make it so they dont fight?

Normally the signal path is:
comes out of the dac
goes into cap in series and a resistor to ground
goes to some rca wires
goes to a 50k attenuator
goes into cap in series and a resistor to ground
goes into amp

Is this scenario bad?
It depends on what is the other side of the capacitors. You need to keep DC away from the volume control, so that is why there are usually capacitors on both sides. Capacitors at the output and input of equipment also provide some protection from faults in other equipment.

If the stage immediately after the volume control has little or no input current (e.g. JFET, valve, or high-beta BJT) then you may be able to omit the capacitor after the volume control but I would not recommend it. You may get slightly more bass but you will probably not be able to hear it as most loudspeakers give up at LF before amplifiers do. On the other hand the extra bass may cause more intermodulation in the amplifier - sounds you can't hear can damage sounds you can hear.
Last edited:


2009-02-28 1:24 pm
my sources are amb gamma1 and gamma 2's. So the outputs are 22uf electrolytics bypassed with MKP.

The attenuators are 50k.

the inputs of the amps range from 20k at the lowest(41hz amp4), to a oddwatt el84 tube. It doesnt have input caps yet, but not complete. The last amp is a chipamp lm3886 kit. Im not sure what caps it uses. None of my amps are complete yet so i wanted to get this cap issue over with before i start completing the builds.
if you can guarantee that you will never plug in a source with an output offset and can never send a DC during fault condition, then your receiver does not need a DC blocking cap.

However, this means you can never experiment with equipment swapping, for yourself or friends.

I find it much more flexible to design in DC blocking to every receiver and to add a second input socket that bypasses the DC blocking cap.

If you do have series connected DC blockers then the lower quality one can be safely bypassed.

But, be careful with multiple receivers. The input conditions to the first stage amplifier are critical to the performance of the whole amplifier. If two or more power amps are paralleled from a common source and two or more have no DC blocking capacitors then the input offset currents will interact. This may interfere with stability, reduce sound quality or even cause damage.

My advice: if there is any doubt always DC block.
the outputs are 22uf electrolytics bypassed with MKP.

The attenuators are 50k.

the inputs of the amps range from 20k at the lowest
there is an incompatibility between Rin=22k and Rs of pot <=12k5 Do you plan to buffer the pot?

22uF//0.1uF & 50k give an RC = 1.1s. This will pass bass and sub-bass and the decade below sub-bass. No problem with bandwidth.
That introduces a problem. The power amplifier needs to be bandwidth limited at the input to prevent potentially destabilising input signals from interfering with correct amplifier performance.
You can reduce the high pass filter RC to somewhere between 0.1s and 20ms, but this must be chosen to be compatible with the characteristics of the power amplifier. This requires the input blocking cap value to be designed as part of the power amplifier. Your source manufacturer has no idea what your power amp needs as a high pass filter.

I see solutions:
Keep the gammas as is and fit DC blocking to all your power amps.
Bypass the Gamma filters and fit DC blocking to all your power amps.
Modify the Gamma filters to improve the quality of the signal, but do so in conjunction with the design of all your power amplifiers.
I think it is Rane that suggests a Panasonic bipolar 16V 100uF or 220uF paralleled with a film cap for the sources output. alternatively fit a 10uF film and foil alone as your DC blocker and adjust the receiver Rin to set the filter frequency.

Lots of options. Switch on your calculator.


2009-02-28 1:24 pm
Ya i was afraid that introducing a 50k pot on an amp with a 20k input was not going to work. Guess ill have to put in an opamp buffer or use only my external preamp with a buffer.

Back to caps. My main concern(reason for starting this thread) was that i didnt want my 22uf source caps rendered useless by the amp having a 1uf blocking cap in series with the 22uf. I might have to brush up on my electronics, but i thought 2 caps in series will act like the smallest cap, just like 2 resistors in parallel will act like the smallest resistor. However the circuit goes crc with a 47k resistor to ground at the gamma rca's end. Then when you introduce an attenuator i loose my mind set.

I can comprehend and calculate a dac with an output cap and shunt resistor (depop) going into a 50k load (attenuator). I can comprehend and calculate a cr filter for an amp being driven by a load impedance load. Its putting them together that throws me off.

My other main concern was making the caps to big. My rule of thumb was you only needed to be 1.5-2 octaves below your lowest note. And if you put to big of a cap in, it would affect the highs. But all the dacs have huge caps for the outputs and even you say to use 100-220uf. I take it that doesnt cause a massive turn on thump because of the resistor from the rca to ground? I have been warned not to use a big cap in the amp section from 2 amp build manuals.
The output from an item will typically have larger capacitors than the following input. This is because at the output you want to preserve the output impedance down to LF, because you don't know what is going to follow. At the input you know what follows because it is part of the same equipment, so the aim is to protect the circuit from DC and low LF.

So if you know you will be feeding a fairly high impedance then you can reduce the 22uF source cap - the opposite of what you were expecting?