Removing "pop" by adding isolation capacitors to amp

[stoopalini]

Those of us with amps that do not use isolation capacitors (or direct coupled amps) with a UMC-200 get an audible pop whenever the UMC-200's muting circuit is applied.  This can be simply pressing the mute button on the remote, or anytime the input signal switches formats, or anytime I FF or RW my DVR.  This get's very annoying after a while, so I set out to see if I could remedy it somehow.  I understand the muting strategy in the UMC-200 is very common in pre amps, and I'm not wanting to debate that end of things; I just want to see if there's a way to modify the amplifier to remove the pop without sacrificing the amps capability.

I'm running my front height speakers off of an Adcom GFA-545, and this is the only amp I have which produces this pop.  Apparently it's a fairly common complaint with the Adcoms when used with a source which mutes by shorting the output to ground.  From what I can tell, this causes a level shift on the input of the Adcom and produces the popping noise.

Well, I found some info on creating an isolation circuit which consists of wiring a capacitor in series with the positive leads of the RCA inputs, and then wiring a resistor from the RCA's positive lead to ground (to drain the cap).  This would need to be done on both RCA inputs of the GFA-545.  So to test the theory, I grabbed a 1uF cap and a couple of RCA couplers I had in my spare parts bin, and wired up an adapter to place the capacitor in line with the RCA center pin, and sure enough, the pop is gone.

So, my question is, which I'm hoping the Emotiva guys (or anyone with the required electrical knowledge) can help with, what value cap and resistor should I use to not reduce the freq response of the Adcom amp?  The 1uF cap did the trick for removing the pop, but I don't know if this value cap will reduce the frequency response of the amp at all.  I believe a 100k resistor from the center pin of the RCA to ground will be sufficient to drain the cap, but would love someone with more experience designing circuits to chime in.

Thanks for any help!

[Boomzilla]

My gut feeling, stoopalini, is that the pop isn't caused by the switching, but rather by a leaky capacitor in the signal path of the Adcom amplifier itself. The capacitor leaks DC into what should be an AC circuit, eventually lifting the entire circuit to a small voltage. When the muting circuit of the preamp engages, the DC is shunted to ground, thus the pop. The best way to fix this is to recap the Adcom. This isn't a cheap option, but it fixes the cause, not the symptom.

Yes, a capacitor coupled input box for the Adcom would also fix this. RC crossover calculators abound on the net. Use a 20K-ohm resistor and a crossover frequency of about 60,000 Hz. and you'll get a capacitance for the job. Use the capacitor in series with the signal, and the resistor to ground. Use good (polypropylene or better) capacitors.

[stoopalini]

Thanks for the reply. I called an authorized amplifier repair shop in Austin (same guys who repaired by JBL AVA7), and they said it would be extremely rare to have a GFA-545 which needed to be recapped. When I explained the issue, he said this was normal due to the design of the Adcom and the fact the pre was muting by shunting to ground? I don't know enough about amplifier/circuitry design to understand what is happening though ...

But, if I understand this filter correctly, basically I'm building either a low pass filter with a set point above the freq response of the amp (ie: 60khz like you recommend), or building a high pass filter with a set point below the freq response of the amp (ie: 10-20hz) .... is this right?

If so, I found this diagram out there, and according to it, the high pass filter is the one which has the capacitor in series and the resistor in parallel; while the low pass has the resistor in series with the capacitor in parallel



Assuming this is correct, if I used a value of 60khz like you recommend, and placed the cap in series, wouldn't I be setting up a high pass filter at 60khz, essentially muting the signal?

I played around with a calculator, and since I have some .1uf and some .047uf capacitors laying around, I used those values.

So according to this, I can use a 200k resistor in parallel and a .047uf capacitor in series, to create a high pass filter at 16.93hz, correct?



I hope I'm understanding this correctly.

[geebo]

Why would you want to remove pop? He's a pretty good guy IMHO.

[KeithL]

The pre mutes by shunting to ground. This means that, IF THERE IS DC ON THE INPUT OF THE AMP, you will hear a pop or click when the mute shunts it to ground. (There shouldn't be significant DC on the input of any piece of audio equipment unless it is having a problem; the most common reason for this is that the equipment is capacitor coupled at its input and that capacitor has gone leaky - often from age. However, there are other reasons, and some direct coupled inputs may have a tiny bit of DC on them for any of several reasons - including that the input component itself just has a little input bias. Even a tiny bit of DC may make an audible tick; as long as it's minimal, it's no cause for alarm.)

DC is "0 Hz", so to block DC you are indeed making a HIGH PASS filter which passes all audio frequencies but blocks lower frequencies (like DC). Simple R-C filters roll off somewhat gradually, and so will start affecting frequencies well above their calculated roll off frequency; they will also cause significant phase shift at frequencies well above it. For that reason, and since you are only worried about blocking 0 Hz, you should calculate your filter to be HIGH PASS with a roll off at one or two Hz (not 20 Hz) to keep side effects well away from the audio band.

Any resistor value above about 15k is fine (you just want to avoid putting too much load on the source preamp). HOWEVER, note that the input impedance of the amplifier itself is in PARALLEL with the resistor you pick. (So, if you use a 200k resistor, but the input impedance of the amp itself is 20k, the resulting "R" for purposes of calculating the filter will be 20k in parallel with 200k - which comes out to around 17k; the formula for two resistors in parallel is (R1 * R2)/(R1 + R2).) If the resistor value is lower than you think, then the cutoff frequency will be higher - and you'll risk losing audible bass.  

May 12, 2014 8:46:41 GMT -5 stoopalini said:

Thanks for the reply. I called an authorized amplifier repair shop in Austin (same guys who repaired by JBL AVA7), and they said it would be extremely rare to have a GFA-545 which needed to be recapped. When I explained the issue, he said this was normal due to the design of the Adcom and the fact the pre was muting by shunting to ground? I don't know enough about amplifier/circuitry design to understand what is happening though ...

But, if I understand this filter correctly, basically I'm building either a low pass filter with a set point above the freq response of the amp (ie: 60khz like you recommend), or building a high pass filter with a set point below the freq response of the amp (ie: 10-20hz) .... is this right?

If so, I found this diagram out there, and according to it, the high pass filter is the one which has the capacitor in series and the resistor in parallel; while the low pass has the resistor in series with the capacitor in parallel



Assuming this is correct, if I used a value of 60khz like you recommend, and placed the cap in series, wouldn't I be setting up a high pass filter at 60khz, essentially muting the signal?

I played around with a calculator, and since I have some .1uf and some .047uf capacitors laying around, I used those values.

So according to this, I can use a 200k resistor in parallel and a .047uf capacitor in series, to create a high pass filter at 16.93hz, correct?



I hope I'm understanding this correctly.

[KeithL]

Right answer - and right topology - but wrong values....

Since you want to block DC but pass audio, you want a filter that is HIGH PASS at a few Hz (NOT kHz). Calculate the values for high pass at 2 Hz or so and you'll be fine.

May 12, 2014 7:36:29 GMT -5 Boomzilla said:

My gut feeling, stoopalini, is that the pop isn't caused by the switching, but rather by a leaky capacitor in the signal path of the Adcom amplifier itself. The capacitor leaks DC into what should be an AC circuit, eventually lifting the entire circuit to a small voltage. When the muting circuit of the preamp engages, the DC is shunted to ground, thus the pop. The best way to fix this is to recap the Adcom. This isn't a cheap option, but it fixes the cause, not the symptom.

Yes, a capacitor coupled input box for the Adcom would also fix this. RC crossover calculators abound on the net. Use a 20K-ohm resistor and a crossover frequency of about 60,000 Hz. and you'll get a capacitance for the job. Use the capacitor in series with the signal, and the resistor to ground. Use good (polypropylene or better) capacitors.

[Boomzilla]

You're right, Keith - It's been quite a few years since I calculated filters... To be effective the capacitor must be in the signal path - not to ground. Therefore, this will be a high-pass filter, not a low-pass.

[stoopalini]

Thanks guys, great stuff!

Since the Adcom is direct coupled, my next steps are to measure the impedance of the amplifier's inputs, then calculate the cap and resistor values for 1-2 Hz, and build 2 filters.

Since this amp is driving the front heights, which are NHT SuperOnes, impact to the low frequencies shouldn't be too big of a deal, considering they're crossed over at 90Hz by the UMC anyway.

I assume I want the "Time Constant" as small as possible? And this is achieved with a higher resistance value and a smaller capacitor value; but can I also assume the higher resistance value places a larger load on the amp so there is a balance to be achieved here?

Here is what I came up with:




What about capacitor and resistor "types" for this application?

Would this Dayton Audio DFFC-0.22 0.22uF 400V By-Pass Capacitor be suitable for the cap? It's a "film and foil" type capacitor.

As for the resistor, what wattage would be suitable? Most of the high ohm resistors seem to have a rating of 1w. Will this 300K Ohm 1W Flameproof Resistor be suitable?

Thanks again for your help, it's greatly appreciated!

[Boomzilla]

The wattage of signal is low. You won't need more than a ¼ to ½ watt resistor. The resistor's tolerance is more important than its wattage. If you can, buy +/- 1% resistors, although anything up to +/- 5% is probably OK. The voltage of the signal is also low, so you won't need high voltage capacitors either.

[stoopalini]

Ok, great ... thanks.

I just placed an order for a few of these Dayton Audio PMPC-0.22 0.22uF 250V Precision Audio Capacitors, some of these 300K Ohm 1W Flameproof Resistors, and a project box, RCA connectors, etc ...

For less than $20, I should be able to build the filter. Sure beats $200 for the amp repair shop to modify the Adcom

Thanks guys, I'll let you know how it goes. If it works well, with no drawbacks, this may help some others who are complaining of the popping when the audio signal changes formats.

[richardrc]

Why are you using a resistor?
Most audio input stages I have seen are already ac coupled so as to eliminate the DC component.

[stoopalini]

From what I gather it's because the Adcom amp is direct coupled; which is the reason this issue exists with this amp. Without the resistor, apparently the cap wouldn't discharge upon power down.

[stoopalini]

Although now that I look at the schematic for this amp, maybe I am mistaken?

I don't know enough to read this correctly, but does R603 placement mean I don't need another resistor in the filter I am building?

[richardrc]

That schematic shows a DC coupled amp. The capacitor in series with the input will be fine. The resistor will make little difference.

[stoopalini]

Thanks for that ... so what does this mean in terms of choosing the correct capacitor value to ensure I won't be losing any of the frequency response?

The calculator I was using showed a 300kOhm resistor with a 0.22uf capacitor would produce a high pass filter at 2.41hz.
The schematic shows a 1kOhm resistor in series, followed by a 22kOhm resistor in parallel, followed by a 330pF capacitor in parallel.

Would I use the existing 22kOhm resistor as the R value of the high pass filter equation, then choose the appropriate capacitor to dial in the desired Hz?

[richardrc]

22k would be be a good enough approximation. A value if 2u2 or 2u7 would probably fit the bill. Remember that this filter will introduce a 90° phase shift.

[stoopalini]

Hmm, I didn't realize this would add a phase shift into the audio signal. I assume the UMC-200's time delay won't be adequate to compensate for the shift considering different frequencies have different lengths, so is there a better filter design I should be looking at to filter out the DC without adding in a phase shift, or one which compensates by adding another 90* reverse shift?

[srrndhound]

May 14, 2014 7:01:11 GMT -5 stoopalini said:

Hmm, I didn't realize this would add a phase shift into the audio signal. I assume the UMC-200's time delay won't be adequate to compensate for the shift considering different frequencies have different lengths, so is there a better filter design I should be looking at to filter out the DC without adding in a phase shift, or one which compensates by adding another 90* reverse shift?

Don't worry about the phase shift. It is essentially nonexistent in the audible band, as this filter's cutoff is 3 Hz (2.2uF, 22k). Red herring.

[stoopalini]

May 15, 2014 13:06:17 GMT -5 srrndhound said:

Don't worry about the phase shift. It is essentially nonexistent in the audible band, as this filter's cutoff is 3 Hz (2.2uF, 22k). Red herring.

Ok, does that mean the phase shift only occurs within the vicinity of the cutoff value? 

I actually ordered .22uf capacitors when I thought I would be using a 300k resistor, so I don't have any 2.2uf ones.  I could order 2.2 ones, but with the 22kOhm resistance value and a .22uf capacitor, the cutoff value is 32.88Hz; and since these channels are crossed over at 90Hz by the UMC anyway, I imgaing it won't matter.

Thanks again guys, I love little projects like this where I can learn a bit more about these things

[KeithL]

22k is the value you would use (and that 22k resistor also means that you don't need that 330k one.)

The phase shift is significant at the cutoff frequency, and decreases as you get further away. Technically, the phase shift will never go to zero, but as long as you keep the cutoff frequency a factor of ten away from the audio band, it will be a few degrees at 20 Hz at most - which will be entirely inaudible. (Of course, the reason why that amp was designed as DC coupled is to prevent that phase shift, so you are compromising the design.) The capacitor should be non-polar (so film or something equivalent), and anything over 20v or so should be fine - although 50v or over would be better (see below). You're not going to be using the 330k resistor but, if you were, a 1/10 watt one would be just fine.

The bigger question is:
Why is that DC there?

A DC coupled amplifier should have very little DC on its input (a few millivolts is normal). If that's really all that's there, then it won't hurt anything and you're wasting your time fussing with a filter. If there's a lot more there, then there are parts beginning to fail on the amplifier - and it really should get looked at. Your filter WILL protect the UMC if the amp fails (especially if you use a capacitor that's at least 50v), but it won't do anything to prevent the amplifier from failing eventually when/if the fault gets worse. (And, in case you were wondering, if the amp fails altogether it may well destroy itself beyond repair, and could take your speakers with it. Without the filter, it could also destroy your UMC.)

May 15, 2014 15:46:43 GMT -5 stoopalini said:

May 15, 2014 13:06:17 GMT -5 srrndhound said:

Don't worry about the phase shift. It is essentially nonexistent in the audible band, as this filter's cutoff is 3 Hz (2.2uF, 22k). Red herring.

Ok, does that mean the phase shift only occurs within the vicinity of the cutoff value? 

I actually ordered .22uf capacitors when I thought I would be using a 300k resistor, so I don't have any 2.2uf ones.  I could order 2.2 ones, but with the 22kOhm resistance value and a .22uf capacitor, the cutoff value is 32.88Hz; and since these channels are crossed over at 90Hz by the UMC anyway, I imgaing it won't matter.

Thanks again guys, I love little projects like this where I can learn a bit more about these things