DC blocking capacitors audibility.

[Ricardus]

Dumb question time: what's the audio/hi-fi application for DC blocking caps in the signal path? I've heard of - and used - DC blocking caps in the AC power chain (e.g. DC blocker between a component's power cord and the AC mains plug) to prevent mechanical toroidal transformer hum. But in the signal path?

Any time you feel you might need to block DC. But in high voltage tube gear you need to block all of that high voltage DC to get the lower voltage audio signal. For example in a preamp the preamp tubes are adding some gain, but you need to block the DC on their output to run the signal through the tone controls and out at line level to go into the inputs of a power amp or whatever component is next in the chain.

 

[Philbo King]

Dumb question time: what's the audio/hi-fi application for DC blocking caps in the signal path? I've heard of - and used - DC blocking caps in the AC power chain (e.g. DC blocker between a component's power cord and the AC mains plug) to prevent mechanical toroidal transformer hum. But in the signal path?

Amplifiers *usually* have the active components (transistors, tube and such) operating between one power rail and ground. Opamps can be an exception. It can also be done with discrete circuits, but at great trouble and expense.

Often there is a weak DC voltage at the input of the circuit (typically called bias), and a much higher voltage at the output of the circuit. A blocking capacitor (also called coupling capacitor) allows feeding the AC (signal) output of one circuit (which might be at dozens to hundreds of volts DC) to the input of the next stage, which might need a DC bias level of negative 30 VDC (for tubes) to plus 10 VDC for transistors.

An alternative and much more expensive way to do this for audio is with transformers.

It's never a great idea to feed any speaker DC... It kills them, either slowly or violently. Caps and transformers are the most common way to avoid this.

 

[Roland68]

Dumb question time: what's the audio/hi-fi application for DC blocking caps in the signal path? I've heard of - and used - DC blocking caps in the AC power chain (e.g. DC blocker between a component's power cord and the AC mains plug) to prevent mechanical toroidal transformer hum. But in the signal path?

In many of the devices tested for ASR (DACs, HPAs, pre-amps, integrated amplifiers, etc.) you will find electrolytic capacitors in the signal path to avoid DC.
Eg Topping L30, AIYIMA A07, SMSL A300, etc just to name a few.
The coils of loudspeakers and HP do not tolerate large DC components.

 

[Sebba]

Dumb question time: what's the audio/hi-fi application for DC blocking caps in the signal path? I've heard of - and used - DC blocking caps in the AC power chain (e.g. DC blocker between a component's power cord and the AC mains plug) to prevent mechanical toroidal transformer hum. But in the signal path?

Theoretically to prevent DC offset from preamps, dacs and also some recordings change the "zero position" of the woofer. But this can be done in tle NFB loop or DC servo and the rest of the chain can be DC coupled. My audio system has no capacitors in in the signal path., and I also use large capacitors in the AC line because toroidal transformers are very sensitive to even small DC components.

 

[Roland68]

Was this testing done doing curated blinded listening tests at matched volume levels? If not, the claims are suspect.

Do you mean that when switching between the capacitors in the signal path, levels have to be adjusted?

Excerpt from the TI document.
"Of course for any single or double blind test to be valid, output levels of the units under test should be matched as closely as possible. 0.1dB difference in output levels at a single location for a single listener is recommended, but 1dB level matching has proved to be sufficient."

 

[egellings]

Do you mean that when switching between the capacitors in the signal path, levels have to be adjusted?

Excerpt from the TI document.
"Of course for any single or double blind test to be valid, output levels of the units under test should be matched as closely as possible. 0.1dB difference in output levels at a single location for a single listener is recommended, but 1dB level matching has proved to be sufficient."

Nope, the caps don't change the sound level so long as the RC effects don't introduce an unintended roll-off at frequency extremes. Just be sure that after you switch your test caps, that you check that you have not accidentally modified the volume level, e.g. accidentally bumped the volume control, maybe?

 

[MaxwellsEq]

There are some other reasons why you might not want DC at some point in a circuit. If there is DC going through a volume control potentiometer, there will be wooshing noises as the volume is changed. DC through a potentiometer may shorten its usable life and if there is dirt on the track, there will be loud bangs.

 

[SIY]

TI is in the business of selling opamps. If they want to sell them to fashion audio customers, they need to talk fashion audio. Of course, there might be some quirk in the input performance of that particular opamp, but they may have chosen not to approach it that way.

 

[Sined]

Dumb question time: what's the audio/hi-fi application for DC blocking caps in the signal path? I've heard of - and used - DC blocking caps in the AC power chain (e.g. DC blocker between a component's power cord and the AC mains plug) to prevent mechanical toroidal transformer hum. But in the signal path?

An other application is for active speakers where your tweeter is directly connected to the amplifier without any crossover in between. This ensure no DC will get to the tweeter should there be an amplifier failure or more simply, from amplifier's turn on/off thump (should there is) or any other unwanted low frequency noise. The rule of thumb is to select a tweeter that will operate at least two octaves below the selected active crossover frequency. According to my old ears, there's no audio quality reduction but I feel much comfortable that my $$$ tweeters lives in a safer environment!

 

[milosz]

I did a blind A/B listening test (of so-so experimental quality) using a quality nonpolar electrolytic vs a quality (but not insane) metalized polypropylene as a coupling capacitor on the output of two different DIY headphone amps with capacitive outputs - one tubed and one solid state. I had a switch set up to select between the electrolytic and the film caps - each amp was set up with a nonpolar electrolytic output coupling capacitor and a film cap with a switch that select either . The switch was hidden from the listeners view and was switched by the experimenter as the listener was listening to Sennheiser HD800 'phones. Capacitors were selected to be of the same value within 2%- as determined by measuring the capacitor before it was soldered in.

Statistically, every listener could tell the difference between electrolytic and film at over 87% accuracy on average; some got 100% correct. Preference was 100% for the film cap.

Film cap was Audyn. NP electrolytic was Nichicon MUSE.

This is not the same as capacitors in a speaker crossover- nowhere near it. But it showed to my satisfaction that listeners can hear the difference between these particular capacitors used as output coupling caps. I infer that "film caps sound better than electrolytics" from this little study, which is really just my opinion, my bias. The experiment actually had too small a sample of listeners and too small a sample of capacitors to warrant my inference, yet I think my bias here has some accuracy when it comes to output coupling capacitors. It **MIGHT** also extend to interstage coupling caps, but really you'd need to do a listening study to confirm or debunk that.

FYI Frequency response of the headphones didn't change with electrolytic vs. film as far as I could measure using a calibrated mic and a home-made headphone coupler, and measuring only above about 250 Hz (I couldn't get good repeatability below 250 Hz with my home-made coupler.)

I don't think that what the listeners were hearing in this little experiment was related to frequency response. Listeners commented that the one switch position seemed to sound slightly distorted, and this turned out to be the NP electrolytic capacitor.

I lack the instrumentation to measure distortion, so I don't know if there is distortion being added or what.

 

[solderdude]

Experiment would need to be repeated under more controlled conditions and with measurements other a DIY headphone measurement fixture.
It is perfectly possible to measure FR and distortion at the output of an amp with an actual load.

A coupling cap would likely give the biggest differences at the lowest frequencies which you were not able to measure accurately so remains a question mark.

What value was the cap ?

A fun experiment, but not conclusive for others.

 

[milosz]

100 uF

Experiment would need to be repeated under more controlled conditions and with measurements other a DIY headphone measurement fixture.
It is perfectly possible to measure FR and distortion at the output of an amp with an actual load.

A coupling cap would likely give the biggest differences at the lowest frequencies which you were not able to measure accurately so remains a question mark.

What value was the cap ?

A fun experiment, but not conclusive for others.

What value was the cap ? 100 uF, as called for in the circuit design.

The experiment was a blind listening test. When you say "need to be repeated under more controlled conditions" what exactly are these more controlled conditions and why would the listening test "need" to be repeated under these conditions? I think the conditions my test was conducted under were quite well controlled. The only thing that I think would increase it's validity would be to use a larger sample of listeners. I don't really have access to such a pool of volunteers, the best I could do was conduct the test with some friends, some of whom are professional musicians and some of whom are pro audio / video media producer-engineers.

I attempted the FR measurement just for fun, not part of the experiment. Yes you can measure distortion from an amp, and etc and so forth, but that is not a listening test. My intent was to determine IF people could hear the difference between two types of capacitors, not learn what it was that their hearing was picking up on. The statistics I collected clearly show that they can, at least for the sample of people that participated. As to exactly what it is that they are hearing, I won't speculate.

 

[solderdude]

5Hz -3dB on HD800 should be low enough for high impedance headphones.

 

[nutzandvoltz]

In Texas Instruments' snaa031a paper I've come across the following text

........In listening tests at TI's sound room evaluating different circuit components used in the LM4702 demo
amplifier, there was one part whose negative effect on audible signal quality was undeniable. A DC
blocking capacitor on the input of the LM4702 degraded sound quality. In multiple listening tests, with
different participants and at various locations around the country, the negative effects of even the best film
and foil polystyrene DC blocking input capacitors in the audio signal path was confirmed.
It is therefore recommended that DC blocking capacitors not be used in the signal path for mid to high-end audio
equipment.........

Unfortunately, no measurements provided. I wonder what might be the reason for those diferences ?
In my experience, if the DC blocking cap is large enough (RC<1Hz) to ensure low impedance in the audio band, (very low voltage drop) the capacitors should be "translucent"

Capacitors is the worst AC coupling device, at best, yes I can see your term translucent is quite accurate, because they are not transparent.
I abandoned the DIY audio forum because of their attitude when I said that. If they wanted the science behind it, by someone writing a book on the subject, they should have shut up and let me say my peace.
Since this is a different forum, I imagine I have much of a mature audience. So I could go in detail if needed.

The reason why would be the input sensitivity is high enough to pick up interference from the capacitor plate as it turns itself into an antenna. Most chip amps that have this issue would always land a small value cap in parallel with a high ohm resistor to ground on each signal in out to snub this RF. Sometimes, putting a 1-3 pF cap across signal + and signal - is all you really need in a fully balanced circuit to snub the RF from coupling caps, because the common mode rejection effect the circuit would impose, would take care of the rest.

But capacitors induce distortion within the audio range as well. Especially if they are blocking voltages several times larger than the signal. This is caused by the magnetic field effects of its construction imposing changes in the circuit. A prime example of this would be using a WHIMA MKP cap that has excessive fields and a horrible coupling device in tube amps compared to a Solen or Jupiter non-metallized wound film and foil type.

Differently constructed caps have different distortion profiles and when you select one for coupling use you use the type and size in the circuit that is less sensitive to add its noise profile to the signal.

 

[nutzandvoltz]

Dumb question time: what's the audio/hi-fi application for DC blocking caps in the signal path? I've heard of - and used - DC blocking caps in the AC power chain (e.g. DC blocker between a component's power cord and the AC mains plug) to prevent mechanical toroidal transformer hum. But in the signal path?

Well its a different application you are trying to compare. In your transformer example, it is used as a current limiting device, like what is done on motors. In this circuit, its blocking the bias voltages from the chip to the input. But this one can be used in dc coupling if you set up the input for that. Which is better. Transformer coupling might be a little complex as you would have to set up the circuit so no dc current is flowing through the secondary. But you gain electrical isolation between devices.

 

[nutzandvoltz]

Do you realize that most of the important and used ICs come from these manufacturers?

Roland68 doesn't realize that author came from the pro audio world, and will share from time to time insights to design.

 

[nutzandvoltz]

Btw, their example circuit is for unbalanced input and probably wouldn't need the global negative feedback if the input was redone to a balanced input. The 200 uf cap is used intentionally in unbalanced mode to set a lower impedance point for the voltage divider that is formed between the terminated signal - input and the 68K feedback resistor.

 

[Cbdb2]

Do you realize that most of the important and used ICs come from these manufacturers? Without these manufacturers, most of the devices and measurements in this forum would not even exist, and above all not the performance that today's devices achieve.

I don't want to question the development performance of the developers at the audio device manufacturers, but most of them are based on the work and circuit proposals of TI and the other manufacturers in this industry area.

So if these manufacturers with their tens of millions of hours of measuring experience don't have the experience to talk about sound-relevant influences, who does?
This is documentation that is not intended for you at all, but for developers and manufacturers. It is extremely kind of TI to make these documents available to everyone, not every manufacturer does that.

I personally find your post very pretentious, but that's just my personal opinion and your post only reflects your opinion.
However, such statements have meant that TI and other manufacturers no longer share certain findings that can be described as scientific, taking into account the acquisition. The bottom line is that this is a loss for every end user (audio).


Addition:
I don't know of any statements from these manufacturers about the tonal quality of the component or OPAmps, but only about the tonal effect of the circuit. So pure information and development tips for the developers.
Very reprehensible.

Its TI, wheres there measurements to explain what they "heard"?

 

[LTig]

100 uF

What value was the cap ? 100 uF, as called for in the circuit design.

The experiment was a blind listening test. When you say "need to be repeated under more controlled conditions" what exactly are these more controlled conditions and why would the listening test "need" to be repeated under these conditions? I think the conditions my test was conducted under were quite well controlled. The only thing that I think would increase it's validity would be to use a larger sample of listeners. I don't really have access to such a pool of volunteers, the best I could do was conduct the test with some friends, some of whom are professional musicians and some of whom are pro audio / video media producer-engineers.

I attempted the FR measurement just for fun, not part of the experiment. Yes you can measure distortion from an amp, and etc and so forth, but that is not a listening test. My intent was to determine IF people could hear the difference between two types of capacitors, not learn what it was that their hearing was picking up on. The statistics I collected clearly show that they can, at least for the sample of people that participated. As to exactly what it is that they are hearing, I won't speculate.

Next time try bigger caps. Douglas Self published distortion measurements of el- and film caps, and while the film caps were good the elcaps had significant distortion rising with voltage across them. To get rid of distortion one needs to select them so large that the voltage across them is not higher than 80 mv at 20 Hz.

 

[nutzandvoltz]

As to exactly what it is that they are hearing, I won't speculate.

Harmonics mostly.
And some people in studio world have a few gizmos out there that its just a tube line stage that you change coupling capacitors and emitter resistors for the different harmonics each combination produces.

you see, not only the component would have a certain harmonic distortion profile (because its a form of in-harmonic distortion IHD) the circuit would have to be sensitive to adding these harmonics in the circuit.