When is is a good idea to use a terminating resistor and when is it a bad idea?
I'm upgrading the output DC blocking caps on my Elektron Octatrack Sampler/Sequencer's single-ended line outputs. The stock output caps are 10uf, which places the RC HPF corner frequency nicely low when running into most line inputs. I want to use the best possible audio caps that I can fit inside the enclosure. With that being said there is the obvious trade-off between quality and capacitance given a certain maximum physical size, not to mention the fact that lower value caps will cost less.
If I opt to reduce the capacitance (and go with higher quality) I am wondering if it would be advisable "set" the corner frequency of the RC HPF by terminating the output to ground after the cap? Setting the terminating resistance would seem to give me predictable frequency specs--is this correct? Or wold I be setting the output impedance of the device higher as well, thereby raising concerns about parasitic cable capacitance, EMI, etc? Would it be better to terminate the far end of the cable? Or is it better still to just terminate at the input of the next device and preserve the low impedance for the length of the cable?
I presume that I will add a tiny bit of thermal noise from the use of the resistor. Is there another problem that I'm not seeing? Can anyone advise me in a general way about whether this would be a good idea? When is is a good idea to use a terminating resistor and when is it a bad idea?
Thanks a ton!
I'm upgrading the output DC blocking caps on my Elektron Octatrack Sampler/Sequencer's single-ended line outputs. The stock output caps are 10uf, which places the RC HPF corner frequency nicely low when running into most line inputs. I want to use the best possible audio caps that I can fit inside the enclosure. With that being said there is the obvious trade-off between quality and capacitance given a certain maximum physical size, not to mention the fact that lower value caps will cost less.
If I opt to reduce the capacitance (and go with higher quality) I am wondering if it would be advisable "set" the corner frequency of the RC HPF by terminating the output to ground after the cap? Setting the terminating resistance would seem to give me predictable frequency specs--is this correct? Or wold I be setting the output impedance of the device higher as well, thereby raising concerns about parasitic cable capacitance, EMI, etc? Would it be better to terminate the far end of the cable? Or is it better still to just terminate at the input of the next device and preserve the low impedance for the length of the cable?
I presume that I will add a tiny bit of thermal noise from the use of the resistor. Is there another problem that I'm not seeing? Can anyone advise me in a general way about whether this would be a good idea? When is is a good idea to use a terminating resistor and when is it a bad idea?
Thanks a ton!
It's a good idea to use a terminating resistor when the line is electrically long, which at audio frequencies means more than a few hundred metres. Do you plan to use such long lines?
The main purpose of a large-valued resistor to ground after a DC blocking capacitor is to ensure that the DC voltage will indeed go to 0. How large you can make the resistor depends on how large the capacitor is and how long you are prepared to wait for the voltage to settle.
Sometimes you see series resistors before or after the DC blocking capacitor. These may serve to keep the driving amplifier stable independent of the load impedance, or they may serve as a protection against short circuits.
Hi there MarcelvdG. Thanks for the reply. Did you read my post? I'd love to get answers to the questions that I asked--I think maybe you just read the first sentence? No, my situation doesn't have anything to do with driving long lines. It's about setting the RC HPF corner frequency. If you have some guidance, I'd be greatly appreciative.
Thanks!
Thanks!
My situation has something to do with this:
...although I'm looking to set the corner frequency for the RC HPF to optimize my selection of a blocking cap. The info is in my initial post.
...although I'm looking to set the corner frequency for the RC HPF to optimize my selection of a blocking cap. The info is in my initial post.
I read your initial post, but it wasn't clear to me what you meant. Hence my answer that deals with different reasons to use a resistor.
Anyway, if I understand you correctly, you are talking about a resistor to ground after a DC blocking capacitor. For some reason you want a well-defined cut-off frequency rather than as low a cut-off as possible. That means that you actually use the capacitor as a subsonic filter.
I gather you are familiar with the equations for first-order filters, so they will then tell you precisely what resistance and what capacitance give you what roll-off (taking into account the input resistance of the input you are driving). I generally prefer to have no subsonic filtering in the signal path unless there is a good reason to expect subsonic signals, so I would go for the highest resistance that gives a reasonable settling time. Settling to within 1 % takes ln(100) RC ~= 4.6 RC.
A resistor to ground will only reduce the output impedance of the amplifier (albeit normally by a negligible amount), so the worries you have about cable driving make no sense to me.
Your cable is electrically short. Assuming that at least one side is terminated with a resistance greater than a few hundred ohms, you can regard the whole cable as a lumped capacitance to ground. It therefore makes no difference on which side you put the resistor. If the cable can get disconnected with the equipment powered up, it may be a good idea to put some resistance to ground on either side to always have a defined path to ground.
Is this the kind of information you were looking for?
Anyway, if I understand you correctly, you are talking about a resistor to ground after a DC blocking capacitor. For some reason you want a well-defined cut-off frequency rather than as low a cut-off as possible. That means that you actually use the capacitor as a subsonic filter.
I gather you are familiar with the equations for first-order filters, so they will then tell you precisely what resistance and what capacitance give you what roll-off (taking into account the input resistance of the input you are driving). I generally prefer to have no subsonic filtering in the signal path unless there is a good reason to expect subsonic signals, so I would go for the highest resistance that gives a reasonable settling time. Settling to within 1 % takes ln(100) RC ~= 4.6 RC.
A resistor to ground will only reduce the output impedance of the amplifier (albeit normally by a negligible amount), so the worries you have about cable driving make no sense to me.
Your cable is electrically short. Assuming that at least one side is terminated with a resistance greater than a few hundred ohms, you can regard the whole cable as a lumped capacitance to ground. It therefore makes no difference on which side you put the resistor. If the cable can get disconnected with the equipment powered up, it may be a good idea to put some resistance to ground on either side to always have a defined path to ground.
Is this the kind of information you were looking for?
Yes. I was wondering if I was missing anything. I guess from your response that I am not missing anything and that a terminating resistor will cause no harm. The reason I want to set the HPF freq is that I don't want it to ever be above 2hz (not to create a rumble filter, as you suggested). I would like to terminate the output at a higher resistance than I'm likely to find on the input of the following device. A 1Mohm termination, for instance, would allow me to use a 1uf cap and still pass full audio, even when driving a standard line input. Does my question make sense to you now? I'm wondering if cable capacitance will be a concern with a large terminating resistor, because the cable and the resistor will form another RC filter. it sounds like you're saying that it will not be a problem. Is that correct?
Thanks!
you need one (1) DC blocking capacitor between your audio modules.
You can place that capacitor at the output of the Source, or at the input of the Receiver.
But, don't use 2 Capacitors to block the DC.
Once you have decided where to locate it, you then add on a leakage grounding resistor, EITHER at the output of the Source, OR at the input of the Receiver.
Locating the DC blocker at the Source leaves one guessing at what the load might be and what the LF roll-off might end up at. That's where you are now.
I think it is better to locate the DC blocker at the Receiver where you know the LF you require and know the load that is being driven. Add on a 2M2 resistor at the input socket to dissipate any leakage. Add on a 22pF to 47pF (NPO ceramic in 603 or 805 SMD) across the input socket for a bit of RF attenuation BEFORE it gets onto the internal cabling.
1uF & 100k gives the same roll-off as 10uF and 10k @ 1.6Hz for f-3dB, or ~3Hz for F-1dB
You can choose any roll-off frequency YOU need/want and know it is fixed at that value for your Receiver.
If you have an extra DC blocker in your Source you can bypass that with an extra RCA output socket. Just label them as AC coupled and DC coupled.
You can place that capacitor at the output of the Source, or at the input of the Receiver.
But, don't use 2 Capacitors to block the DC.
Once you have decided where to locate it, you then add on a leakage grounding resistor, EITHER at the output of the Source, OR at the input of the Receiver.
Locating the DC blocker at the Source leaves one guessing at what the load might be and what the LF roll-off might end up at. That's where you are now.
I think it is better to locate the DC blocker at the Receiver where you know the LF you require and know the load that is being driven. Add on a 2M2 resistor at the input socket to dissipate any leakage. Add on a 22pF to 47pF (NPO ceramic in 603 or 805 SMD) across the input socket for a bit of RF attenuation BEFORE it gets onto the internal cabling.
1uF & 100k gives the same roll-off as 10uF and 10k @ 1.6Hz for f-3dB, or ~3Hz for F-1dB
You can choose any roll-off frequency YOU need/want and know it is fixed at that value for your Receiver.
If you have an extra DC blocker in your Source you can bypass that with an extra RCA output socket. Just label them as AC coupled and DC coupled.
Thanks a ton, Andrew. That's very helpful. So I gather that a terminating resistor attached to the source output is always of limited use since the load provided by the input and interceding cable is potentially variable. I suppose, if I wanted to be certain that I can achieve the absolutely highest bandwidth, I should build a custom cable that terminates at the far end, and to an extremely high value resistor.
This whole topic is still a bit confusing to me--and I seem to have also caused confusion in even asking these questions. This sort of inquiry seems pretty vital to me, especially in the context of electronic musical instrument design (unlike in the world of Hi-fi), where a particular instrument must be designed to be loaded by a wide range of device inputs--gigging musicians like myself have to deal with innumerable variables when we set up to perform in a different city each night and I would love to design a way to always achieve the best bandwidth with this (and other) devices, regardless of the input impedance of the venue mixing console, etc.
Thanks for the careful and detailed reply.
Cheers!
This whole topic is still a bit confusing to me--and I seem to have also caused confusion in even asking these questions. This sort of inquiry seems pretty vital to me, especially in the context of electronic musical instrument design (unlike in the world of Hi-fi), where a particular instrument must be designed to be loaded by a wide range of device inputs--gigging musicians like myself have to deal with innumerable variables when we set up to perform in a different city each night and I would love to design a way to always achieve the best bandwidth with this (and other) devices, regardless of the input impedance of the venue mixing console, etc.
Thanks for the careful and detailed reply.
Cheers!
OK, I'm starting to get the picture.
You have an electronic music instrument with unbalanced output that has to drive an electrically short cable and an unknown line input of some unknown other equipment. Under these conditions, you want to keep the -3 dB frequency below 2 Hz without using an excessively large capacitance.
In this case a resistor to ground isn't going to help. That is, a large resistor to ground will help to keep the output DC level at 0, but it does not prevent the cut-off frequency from going above 2 Hz if the input resistance of the unknown equipment happens to be small.
A large resistor in series with the input of the unknown equipment would help, but it would among other things also result in a large and unknown attenuation, in HF roll-off due to cable capacitance if it is placed on the driving side and in extra thermal noise. I wouldn't even try it if I were you.
Possible alternatives:
1. Using a large capacitor after all
2. DC servo loop
3. Feedback loop incorporating the output capacitor
4. If and only if you know for sure that the unknown equipment always has a capacitor at the input, which you probably don't: replacing the output capacitor with a short circuit
You have an electronic music instrument with unbalanced output that has to drive an electrically short cable and an unknown line input of some unknown other equipment. Under these conditions, you want to keep the -3 dB frequency below 2 Hz without using an excessively large capacitance.
In this case a resistor to ground isn't going to help. That is, a large resistor to ground will help to keep the output DC level at 0, but it does not prevent the cut-off frequency from going above 2 Hz if the input resistance of the unknown equipment happens to be small.
A large resistor in series with the input of the unknown equipment would help, but it would among other things also result in a large and unknown attenuation, in HF roll-off due to cable capacitance if it is placed on the driving side and in extra thermal noise. I wouldn't even try it if I were you.
Possible alternatives:
1. Using a large capacitor after all
2. DC servo loop
3. Feedback loop incorporating the output capacitor
4. If and only if you know for sure that the unknown equipment always has a capacitor at the input, which you probably don't: replacing the output capacitor with a short circuit
Thanks! I figured there must have been something about my initial wording that was obfuscating my aim. Okay--so there isn't such a simple way to achieve my goal. I will certainly look into the other more drastic options that you've outlined and assess them for feasibility. Thanks for spending so much time with this and continuing to assist!
Thanks a ton, Marcel!
Thanks a ton, Marcel!
There is supposed to be a 100 kohm resistor from the output to ground, apparently I forgot to draw it.
The reason why it may be unnecessarily complicated: when you use an op-amp with a low offset and near-zero bias current, you can probably get rid off C32, C10, C12 and R18 and turn it into a simple unity-gain buffer without causing any problems, even if the equipment you are driving is DC coupled. C5 and R15 block the offset or bias voltage of whatever is driving the buffer, so you only see the buffer's own offset. As long as that offset is, say, at least 100 times smaller than the nominal line level, chances are it will not trigger any DC protection circuits, not overheat any voice coils and not cause any significant increase of second-order distortion of the loudspeakers.
The reason why it may be unnecessarily complicated: when you use an op-amp with a low offset and near-zero bias current, you can probably get rid off C32, C10, C12 and R18 and turn it into a simple unity-gain buffer without causing any problems, even if the equipment you are driving is DC coupled. C5 and R15 block the offset or bias voltage of whatever is driving the buffer, so you only see the buffer's own offset. As long as that offset is, say, at least 100 times smaller than the nominal line level, chances are it will not trigger any DC protection circuits, not overheat any voice coils and not cause any significant increase of second-order distortion of the loudspeakers.
You get answers related to driving long audio lines because you *ask* about them by using the buzz word "terminating resistor".
Problem is , you are simply using the wrong term.
What you are actually talking about is a DC reference to ground resistor after a coupling capacitor, a very different concept.
It must NOT be the impedance defining the LF cutoff frequency, leave that to the input impedance of the audio device (power amp? ... preamp input? .... ) you are driving.
IF it were so, it means its value is way too low and it´s unduly loading your audio source; its real function is only to let the output side of the capacitor reach 0V a short time after audio source is turned on.
And don´t worry about its contribution to noise: at the output signal level is presumed high and it´s being driven by a source impedance way lower than its own value, so any noise contribution will be minor to begin with and swamped from driver out lower impedance.
As to using "high quality caps" there, ..... define "high quality" caps first.
Unless current ones are leaky or are dry/terminally ill with huge ESR or internal crimped terminals are failing, there is no advantage (read as "no sound change") by using one type or another, including 1000X more expensive ones, as long as you use same capacitance ones.
And although in theory having the ground reference resistor is the same whether it´s at the source output or the driven device input .... as long as both are permanently connected, it´s good practice using it the first way, so source element can be connected to different driven ones without undue pops and thumps, since its output will *already* be 0 DC volts.
Problem is , you are simply using the wrong term.
What you are actually talking about is a DC reference to ground resistor after a coupling capacitor, a very different concept.
It must NOT be the impedance defining the LF cutoff frequency, leave that to the input impedance of the audio device (power amp? ... preamp input? .... ) you are driving.
IF it were so, it means its value is way too low and it´s unduly loading your audio source; its real function is only to let the output side of the capacitor reach 0V a short time after audio source is turned on.
And don´t worry about its contribution to noise: at the output signal level is presumed high and it´s being driven by a source impedance way lower than its own value, so any noise contribution will be minor to begin with and swamped from driver out lower impedance.
As to using "high quality caps" there, ..... define "high quality" caps first.
Unless current ones are leaky or are dry/terminally ill with huge ESR or internal crimped terminals are failing, there is no advantage (read as "no sound change") by using one type or another, including 1000X more expensive ones, as long as you use same capacitance ones.
And although in theory having the ground reference resistor is the same whether it´s at the source output or the driven device input .... as long as both are permanently connected, it´s good practice using it the first way, so source element can be connected to different driven ones without undue pops and thumps, since its output will *already* be 0 DC volts.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.