I have a collection of solid-state amplifiers built from amplifier board kits from China. Some of them are chip amps (e.g., LM3886) and some of them use discrete components. One thing I noticed is that they all use a polar electrolytic capacitor bypassed with a small film cap (e.g., 0.1uF) as the DC-blocking capacitor in the feedback loop to get unity gain at DC.
This "practice" does not make sense to me, as the quality of the DC-blocking feedback cap should be as important as the DC-blocking input cap of the amp. As shown in the articles by Cyril Bateman, polar electrolytic caps are poor choices for signal coupling, due to their rather high distortion levels. Moreover, adding a small film shunt cap is a Band-Aid approach that does not really work, because it reduces the distortions by a negligible amount. See, p. 13 of https://linearaudio.nl/sites/linear...0 to 100uF caps and 100 Hz measurements_0.pdf
Some of the Chinese kit amps managed to provide space in their tiny PCBs to fit a film input capacitor of a few uF for the purported sake of better sound quality, but all of them seem to have forgotten about the DC-blocking feedback capacitors. Perhaps this is because the capacitance (ranging from several 10's of uF to hundreds of uF) of the DC-blocking feedback cap is always much larger than that of the input cap, and it is simply impractical in terms of cost and size to add a big fat film cap of 10's or even hundreds of uF to the PCB.
Cyril Bateman's article linked above shows that bi-polar caps are much better than polar caps in terms of distortion. So, last week I replaced the DC-blocking feedback caps (220uF) in my LME49810 amp with bipolar caps. The change in sound of the amp is not subtle to my ears. The sound seems to have become cleaner and, for lack of a better word, "shinier." I think it is better. Of course, I realize that the perceived change could be the result of confirmation bias.
Bateman's article also shows that "double bipolar" is even better, approaching the performance of a PET film cap. I have not tried that, however, as there is no room on the amp boards for two serially connected 470uF bi-polar caps (to have an effective capacitance of 235uF). I also checked my LJM MX50SE amp boards, and found it hopeless to try to replace the feedback cap (470uF) on an MX50SE board with even a single bi-polar cap.
I wonder whether replacing the polar DC-blocking feedback with a single bi-polar cap is pretty much as good as can be done in practice, and whether going "double-bipolar" or even (theoretically, I guess) a giant film capacitor would reach a point of diminishing return.
For the experts of amp design on this forum, do you think the quality of the DC-blocking feedback cap really matters? If so, what is the best practical way to minimize the distortions of such a cap?
This "practice" does not make sense to me, as the quality of the DC-blocking feedback cap should be as important as the DC-blocking input cap of the amp. As shown in the articles by Cyril Bateman, polar electrolytic caps are poor choices for signal coupling, due to their rather high distortion levels. Moreover, adding a small film shunt cap is a Band-Aid approach that does not really work, because it reduces the distortions by a negligible amount. See, p. 13 of https://linearaudio.nl/sites/linear...0 to 100uF caps and 100 Hz measurements_0.pdf
Some of the Chinese kit amps managed to provide space in their tiny PCBs to fit a film input capacitor of a few uF for the purported sake of better sound quality, but all of them seem to have forgotten about the DC-blocking feedback capacitors. Perhaps this is because the capacitance (ranging from several 10's of uF to hundreds of uF) of the DC-blocking feedback cap is always much larger than that of the input cap, and it is simply impractical in terms of cost and size to add a big fat film cap of 10's or even hundreds of uF to the PCB.
Cyril Bateman's article linked above shows that bi-polar caps are much better than polar caps in terms of distortion. So, last week I replaced the DC-blocking feedback caps (220uF) in my LME49810 amp with bipolar caps. The change in sound of the amp is not subtle to my ears. The sound seems to have become cleaner and, for lack of a better word, "shinier." I think it is better. Of course, I realize that the perceived change could be the result of confirmation bias.
Bateman's article also shows that "double bipolar" is even better, approaching the performance of a PET film cap. I have not tried that, however, as there is no room on the amp boards for two serially connected 470uF bi-polar caps (to have an effective capacitance of 235uF). I also checked my LJM MX50SE amp boards, and found it hopeless to try to replace the feedback cap (470uF) on an MX50SE board with even a single bi-polar cap.I wonder whether replacing the polar DC-blocking feedback with a single bi-polar cap is pretty much as good as can be done in practice, and whether going "double-bipolar" or even (theoretically, I guess) a giant film capacitor would reach a point of diminishing return.
For the experts of amp design on this forum, do you think the quality of the DC-blocking feedback cap really matters? If so, what is the best practical way to minimize the distortions of such a cap?
Last edited:
Definitely, use at least a good quality bipolar electrolytic there. Being part of the nfb loop, it is very important, at least as much as the input coupling capacitor. Try to set the time constant to 2Hz.
Last edited:
I wonder if the DC-blocking feedback capacitor is what prevents those cheap solid state amps from obtaining good performance in terms of THD. As Cyril Bateman's article shows, a polar cap can have harmonic peaks in the -90dB range. I wonder how good the amp's THD level could be when such a cap is in the feedback loop. In contrast, the Benchmark AHB2 has a measured THD level under -120dB. Maybe the removal of the DC-blocking polar cap from the feedback loop is necessary for making an amp with low TDH. Doing so, however, would sacrifice the unity gain at DC. Seems to be a dilemma.
I am not entire clear about what you mean by "not directly in the signal path." Considering the amp in the simplified picture of an opamp. The opamp has not just one, but two input signals. One input signal is the signal to be amplified, and the second input is the feedback signal. The feedback current, which is the thing that generates the feedback signal, flows through the voltage divider formed by the feedback resistor and the resistor to ground (through the DC blocking cap), and "directly" through the DC blocking cap to ground, and the feedback voltage signal is sampled as the output of the voltage divider. So, to me, the DC-blocking capacitor is "directly" in the path of the feedback current and directly affects the feedback signal. I am not trying to play a game of semantics, but this seems to me to mean that the cap is directly in the signal path of the feedback signal. Is there something wrong with this view?
Last edited:
Of course, in fact the nfb components directly control the amplifier performance.
This is the entire premise of the nfb loop.
Lets be sensible here, the signal path is the path (or paths) carrying the signal. This is not hard to understand. The supply rails are not the signal path, a constant current source is not the signal path (but a current mirror might be).
The feedback path is definitely a signal path, and you'll realize that the signal can go round a looped path if you think a bit.
Things get more complicated sometimes, such as the splitting of the signal in two in a class-B output section.
And as for the DC blocking cap in the feedback its important that this is linear enough - which means having only small voltage across it, which means using a high value cap, perhaps 1000µF kind of range, so that the roll-off frequency it defines is way below the audio band (1Hz or lower). Then the signal voltages across it are measured in mV and µV, and any distortion is very small and very insignificant compared to the larger voltages across the feedback resistors.
Capacitor distortion in electrolytics is very real and the only defence is to keep the ac voltages small.
Last edited:
Certainly. You can inject a signal into any node in the amplifier circuit,
and get some sort of output.
Last edited:
I just use two normal electrolytics back to back.
The problem with that capacitor is making sure it is the right way around.
For some amps it needs to have +ve to feedback resistor.
In some quasi's or output stages with drivers it needs to be the other way around as the feedback signal has a negative value. If this value gets to about -1.5 volts the normal electrolytic becomes a short and the DC bias is amplified many times causing problems.
Wow, what a cantankerous thread.
What surprises me the most in the OP is "bypassed with a small film cap." Considering the reputation of much of Chinese devices, I'm surprised they put that cap on. Maybe they were just blindly following "good" schematics and don't know why it's there and/or never thought of Muntzing. The cap in parallel probably doesn't make (much of a) difference in casual listening.
But yes, modern power amplifiers are essentially op-amps where the negative input affects the output just as much as the positive, with the only difference being opposite polarity (or as it is too often called, "inverted phase").
Electrolytics (the usual polar ones) are indeed less than optimal for passing audio, and running one at essentially zero volts DC across it is about the worst way to operate it. There are solid-state amps from a few decades ago which no longer reproduce bass because the electrolytic capacitors in this application (in series with the feedback divider resistors) dried out and capacitance went way down. These wind up in landfills, or if they're really lucky, in thrift stores and bought by someone who knows what the problem is and how to fix it.
One way to fix the no-voltage-across-the-cap problem is to use two electrolytics of twice the value in series with their negatives connected together. The positive leads go to the two original locations in the circuit. Use a high value resistor (maybe 1 meg) from the negative connection to the negative rail of the power supply, and make sure the capacitors' voltage rating is more than the power supply rail. This will "bias" or charge up the capacitors during normal operation, put them in a (more nearly) linear range, and increase their lifetime by keeping them formed.
What surprises me the most in the OP is "bypassed with a small film cap." Considering the reputation of much of Chinese devices, I'm surprised they put that cap on. Maybe they were just blindly following "good" schematics and don't know why it's there and/or never thought of Muntzing. The cap in parallel probably doesn't make (much of a) difference in casual listening.
But yes, modern power amplifiers are essentially op-amps where the negative input affects the output just as much as the positive, with the only difference being opposite polarity (or as it is too often called, "inverted phase").
Electrolytics (the usual polar ones) are indeed less than optimal for passing audio, and running one at essentially zero volts DC across it is about the worst way to operate it. There are solid-state amps from a few decades ago which no longer reproduce bass because the electrolytic capacitors in this application (in series with the feedback divider resistors) dried out and capacitance went way down. These wind up in landfills, or if they're really lucky, in thrift stores and bought by someone who knows what the problem is and how to fix it.
One way to fix the no-voltage-across-the-cap problem is to use two electrolytics of twice the value in series with their negatives connected together. The positive leads go to the two original locations in the circuit. Use a high value resistor (maybe 1 meg) from the negative connection to the negative rail of the power supply, and make sure the capacitors' voltage rating is more than the power supply rail. This will "bias" or charge up the capacitors during normal operation, put them in a (more nearly) linear range, and increase their lifetime by keeping them formed.
It is what it is, ie the NFB shunt network has great influence on amp sound and this includes the associated resistors also. My advice is to experiment with makes and models of electro caps as suggested already and also to experiment with bypass cap types. There are all kinds of electros nowadays including solid polymer types, RS Components etc catalogs are your friend. Polypropylene bypass caps will 'dilute' Al Electro sound and give a 'blend' sound to the amp. For bypass caps Polystyrene caps are a 'must' try and bring a 'clarity' to HF and overall sound, ceramics are poison to the sound ime, Silver Mica would be worth a listen.
Amplifier input stage typically has series input resistor, series input capacitor, shunt resistor and shunt low value cap to suppress RF ingress and ime these component types are sonically important also and signal input stage functionally 'mirrors' the NFB 'input' stage. By juggling signal input and NFB point resistor and capacitor types you can 'voice' your amp how you like and over a surprisingly wide range. The suggestion of supplying DC bias to series connected midpoint is a good suggestion and will have another set of influences and 'tunings' to explore. So in answer to your question, imho capacitor type is critical in this circuit location and is partly what differentiates amplifiers. You have a lifetime of R and C permutations to explore, be confidant in your ears have fun learning and experimenting with 'tone'.
Dan.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.