I have a little headphone amp project going on that I'd like to try with a 10k-log pot as the volume control.
All I have are 50k-log and 100k-log pots.
I was thinking that if I take a 12k resistor and connect it from the top to bottom lugs of the pot (pins 1 and 3), that will put that 12k ohms in parallel with the 100k ohms of the pot, which should end up making the pot value 10.7k ohms (close enough).
However, I've read that the output impedance of the pot (used as a volume control) won't change if I do that. Huh??
I figure the wiper is moving on the carbon (or resistive plastic) track, and will always be in parallel with the Rparallel (12k) between the pot's pins 1 and 3. It should act pretty much exactly like a (nearly) 10k ohm pot.
No? Now what piece of elementary electronics am I missing?
😕
All I have are 50k-log and 100k-log pots.
I was thinking that if I take a 12k resistor and connect it from the top to bottom lugs of the pot (pins 1 and 3), that will put that 12k ohms in parallel with the 100k ohms of the pot, which should end up making the pot value 10.7k ohms (close enough).
However, I've read that the output impedance of the pot (used as a volume control) won't change if I do that. Huh??
I figure the wiper is moving on the carbon (or resistive plastic) track, and will always be in parallel with the Rparallel (12k) between the pot's pins 1 and 3. It should act pretty much exactly like a (nearly) 10k ohm pot.
No? Now what piece of elementary electronics am I missing?
😕
Looking at it from a linear track perspective:
At maximum setting, the resistance between wiper and terminal 3 would be that of 100k in parallel with 12k, which is 10.7k.
At the mid-point setting, the resistance would be 50k in parallel with 62k (50k in series with 12k), which is 27.7k.
At maximum setting, the resistance between wiper and terminal 3 would be that of 100k in parallel with 12k, which is 10.7k.
At the mid-point setting, the resistance would be 50k in parallel with 62k (50k in series with 12k), which is 27.7k.
Last edited:
It won't change the output impedance of the pot. All you are doing with the 12k is adding a load to the source component driving the pot.
To put a different spin on it... if the pot is fed from lets say an opamp output then the impedance driving the pot is close to zero... now the 12 k really does do nothing at all (apart from add more load for the opamp to drive).
To put a different spin on it... if the pot is fed from lets say an opamp output then the impedance driving the pot is close to zero... now the 12 k really does do nothing at all (apart from add more load for the opamp to drive).
Thanks for the replies.
Oh well. That is not what I was hoping to find out.
I need to find a 10k pot somewhere. The application is a headphone amp made from a 12GN7A frame-grid RF pentode wired triode, into an Edcor 8k:50 OPT (single-ended). I think the Miller capacitance is way up there, even though there is no spec for triode Cag in the datasheet. (I'll bet there's a way to infer that from the pentode specs, though.) My guess is that the Cmiller is going to be about 200pF, because triode mu is pretty high (roughly 35).
--
Oh well. That is not what I was hoping to find out.
I need to find a 10k pot somewhere. The application is a headphone amp made from a 12GN7A frame-grid RF pentode wired triode, into an Edcor 8k:50 OPT (single-ended). I think the Miller capacitance is way up there, even though there is no spec for triode Cag in the datasheet. (I'll bet there's a way to infer that from the pentode specs, though.) My guess is that the Cmiller is going to be about 200pF, because triode mu is pretty high (roughly 35).
--
This might be interesting and might even help:
I recall being told that a _linear_ taper pot of resistance R, driven by a source resistance much less than R, and loaded with a resistance R/5, will exhibit an approximate (over a certain region) log characteristic. Be sure to account for _all_ loads from wiper to ground.
One can always build a spreadsheet and calculate magnitude in dB. Then see over what range of locations on the pot the change in dB with pot travel is nearly linear.
Maybe someone has a refinement on the R/5 value?
Tom
I recall being told that a _linear_ taper pot of resistance R, driven by a source resistance much less than R, and loaded with a resistance R/5, will exhibit an approximate (over a certain region) log characteristic. Be sure to account for _all_ loads from wiper to ground.
One can always build a spreadsheet and calculate magnitude in dB. Then see over what range of locations on the pot the change in dB with pot travel is nearly linear.
Maybe someone has a refinement on the R/5 value?
Tom
A note on potentiometer output impedance: it is a function of the pot travel, as noted in an earlier post. If the top of the pot is driven by a low impedance, and the bottom of the pot is ground, the output impedance will be maximum at the electrical center of the pot, when the resistance above the wiper is equal to the resistance below (the "electrical center" of the pot). In a log taper pot, this corresponds to -6dB which should occur at about (if memory serves) about 10% of the way down from the top. So if the pot output in dB is linear with pot travel, all the way down (or fully counter-clockwise on a rotary pot) will correspond to -60dB. In practice, to obtain a completely "off" condition all the way down (which is - infinity dB) there is a departure from a pure log characteristic near the bottom. But the signal is more than 50dB down anyway, so who cares?
The maximum relative error in pot output voltage due to loading the output occurs at the electrical center of the pot. This is 50% mechanical travel for a linear pot, or about 10% down from the top for a log pot.
Got into a debate once with a professor about maximum error. Only later did we discover that he was thinking absolute error and I was thinking relative error.
About the project: If the headphone amplifier input impedance is say 50k ohm
then using a 100k ohm log taper pot will will have a maximum relative error of -3.5dB at 10% down from the top. So at 10% down from the top, the output will be -9.5 dB instead of -6dB.
Tom
The maximum relative error in pot output voltage due to loading the output occurs at the electrical center of the pot. This is 50% mechanical travel for a linear pot, or about 10% down from the top for a log pot.
Got into a debate once with a professor about maximum error. Only later did we discover that he was thinking absolute error and I was thinking relative error.
About the project: If the headphone amplifier input impedance is say 50k ohm
then using a 100k ohm log taper pot will will have a maximum relative error of -3.5dB at 10% down from the top. So at 10% down from the top, the output will be -9.5 dB instead of -6dB.
Tom
Thanks everybody for the replies.
carlthess40: Thank you for the kind offer, but that's OK. I need to purchase a few ganged 10k-log pots anyway.
tom31415926: In this particular project, the problem isn't so much deviation from a proper log taper or error between channels; it's bandwidth. The input (parallel) capacitance of the triode I'm using is very high, probably 200pF, maybe higher including strays. I fitted a 1k ohm gate stopper resistor, which is in series with the signal and adds to the 25k ohm output impedance of the 100k pot (at -6dB position), So that's 26k ohms in series with the input grid, in parallel with 200pF, which means a low-pass rolloff of (theoretically) -3dB at 30.6kHz. It's probably not that bad at the locations in the pot rotation that I'm using, but let's say it's 15k + 1k ohms, so Rseries = 16k and Cparallel = 200pF. That's still a LPF F3 of 50kHz. Still way too low.
I have one of those decent mini-stepped attenuators from eBay in 50k ohms. Let's say its max Zout is 1/4 of the total R, so 12.5k. Add 1k to that and we get 13.5k, for F3 of 59kHz, worst case. Let's say the Zout where I'm likely to be using the 50k pot will be 7k, so add the 1k for the grid stopper and Rseries = 8k. Now the F3 is 99.5kHz and we're in a better place.
So I'm installing that 50k-A stepped attenuator, and I'll see if that makes an audible difference in high frequency playback. Right now, with the 100k pot, response at 20kHz is not visually down by much on the 'scope (just visible), but the amp sounds 'dark' (four different headphones tried).
--
Theoretically, the worst case Zout of a 10k pot would be 2.5k ohms. Add 1k to that for 3.5k ohms. 3.5k ohms and 200pF gives LPF F3 of 227kHz, which is pretty much ideal. So I'd want a 10k pot in this case, or maybe a 25k pot.
--
carlthess40: Thank you for the kind offer, but that's OK. I need to purchase a few ganged 10k-log pots anyway.
tom31415926: In this particular project, the problem isn't so much deviation from a proper log taper or error between channels; it's bandwidth. The input (parallel) capacitance of the triode I'm using is very high, probably 200pF, maybe higher including strays. I fitted a 1k ohm gate stopper resistor, which is in series with the signal and adds to the 25k ohm output impedance of the 100k pot (at -6dB position), So that's 26k ohms in series with the input grid, in parallel with 200pF, which means a low-pass rolloff of (theoretically) -3dB at 30.6kHz. It's probably not that bad at the locations in the pot rotation that I'm using, but let's say it's 15k + 1k ohms, so Rseries = 16k and Cparallel = 200pF. That's still a LPF F3 of 50kHz. Still way too low.
I have one of those decent mini-stepped attenuators from eBay in 50k ohms. Let's say its max Zout is 1/4 of the total R, so 12.5k. Add 1k to that and we get 13.5k, for F3 of 59kHz, worst case. Let's say the Zout where I'm likely to be using the 50k pot will be 7k, so add the 1k for the grid stopper and Rseries = 8k. Now the F3 is 99.5kHz and we're in a better place.
So I'm installing that 50k-A stepped attenuator, and I'll see if that makes an audible difference in high frequency playback. Right now, with the 100k pot, response at 20kHz is not visually down by much on the 'scope (just visible), but the amp sounds 'dark' (four different headphones tried).
--
Theoretically, the worst case Zout of a 10k pot would be 2.5k ohms. Add 1k to that for 3.5k ohms. 3.5k ohms and 200pF gives LPF F3 of 227kHz, which is pretty much ideal. So I'd want a 10k pot in this case, or maybe a 25k pot.
--
Last edited:
Short answer, you can´t.
There arte various kludges floating around, and they are that: poorly performing kludges, all missing on some important point.
So either get the or, as suggested above, use what you have and pray.
There arte various kludges floating around, and they are that: poorly performing kludges, all missing on some important point.
So either get the or, as suggested above, use what you have and pray.
> My guess is that the Cmiller is going to be about 200pF
Then response at half level will be -1dB at 16kHz. Not perfect, but plenty good to try the idea and get you through until pots come.
Then response at half level will be -1dB at 16kHz. Not perfect, but plenty good to try the idea and get you through until pots come.
I fixed it.
For some reason, the pot I had in there (one of those Radio Shack 100k stereo pots with the loudness tap) wasn't passing high frequency signal well. It measures fine, though. A mystery.
I put a 50k ohm stepped attenuator in and all's good now. Nice, crisp highs into the cans now. Lots of 'air' and 'detail' now... 🙂
I figure if the theoretical -1dB point for 100k pot is 16kHz, 50k pot should give -1dB at 32kHz. That's good enough for me.
I also put in a Zobel network for Sennheiser HD650, 390R and 0.01uF in parallel with the OPT secondary of each channel. I made that switchable. I can't hear the difference between that in/out of circuit.
--
For some reason, the pot I had in there (one of those Radio Shack 100k stereo pots with the loudness tap) wasn't passing high frequency signal well. It measures fine, though. A mystery.
I put a 50k ohm stepped attenuator in and all's good now. Nice, crisp highs into the cans now. Lots of 'air' and 'detail' now... 🙂
I figure if the theoretical -1dB point for 100k pot is 16kHz, 50k pot should give -1dB at 32kHz. That's good enough for me.
I also put in a Zobel network for Sennheiser HD650, 390R and 0.01uF in parallel with the OPT secondary of each channel. I made that switchable. I can't hear the difference between that in/out of circuit.
--
Last edited: