Volume Control Resistance Stability Technical Measurements

Many audio amplifiers have the following gain configuration:


The potentiometer or stepped attenuator resistance stability can influence the system low level gain accuracy and residual noise.

A selection of four volume controls will be examined for resistance stability and compared to a stable laboratory reference standard resistor:
https://www.ietlabs.com/esi-sr1-calibration-resistor.html

Background information on resistors and measurements:
https://www.vishay.com/docs/28771/basics.pdf

Additional background information may be found in this IEC standard:
IEC 60195 Method of measurement of current noise generated in fixed resistors

Instruments used for resistance stability measurements:

Volume controls will be measured using 4 wire Kelvin probes.

In a Kelvin (4-wire) measurement two test points are connected to each side of the component measured. One point acts as the “force” that supplies the current and the second as the “sense” which is used to perform the voltage measurement. The place at which the force and sense paths meet is called a Kevin point.


The controls to be examined:
1) Laboratory Standard Resistor SR1 for reference stability
2) Goldpoint SA1X stepped attenuator
3) JBL NanoPatch
4) Alps RK27
5) Nobsound Little Bear MC102

Stability is measured with each control adjusted to 1000 ohms or as close as practical. Keysight 34465A DMM in 4 wire resistance mode sources 1 mA DC current which will be 1 Vdc across 1000 ohms. DMM is warmed up for 1 hour before measurements commence. Volume controls are allowed to stabilize as necessary with 1 mA current flowing through them. After stabilization 5 minutes of resistance data are collected at approximately 3 readings per second, NPLC set for 10. NPLC is Number of Power Line Cycles over which to integrate measurents.


Laboratory Reference Standard 1KΩ

Resistance varies ~ 800 to 1000 uΩ over short periods while drifting over a range of 1600 uΩ with an absolute accuracy of 1.0000048 KΩ

Goldpoint SA1X stepped attenuator:

Similar to the reference standard resistance varies ~ 800 to 1000 uΩ over short periods while drifting over a range of 1600 uΩ with an absolute accuracy of 1.00049 KΩ

JBL NanoPatch:

NanoPatch has greater resistance variance and drift, 20x greater than the SA1X stepped attenuator.

Alps RK27:

RK27 has short term variance approximately 10x greater than the stepped attenuator with drift similar to the NanoPatch.

Nobsound MC102:

MC102 behaves similar to the RK27 above in both short term variance and overall drift.



This composite graph was created by normalizing the first data point from each of the three potentiometer volume controls to the first Reference resistor reading. Normalization was not performed on the SA1X as it wasn’t necessary.

It is clear in this composite the SA1X stability is nearly identical to the Standard Reference SR1. All potentiometer based controls have similar behavior in short term and drift characteristics though the RK27 drifted in a positive direction while the NanoPatch and MC102 drifted in a negative direction.



Comparing Keysight 34461A and 34465A demonstrates similar capability to perform this particular set of measurements. This particular 34461A has slightly better absolute accuracy.

Stepped attenuators appeal to me for the ability to have repeatable, known volume levels which is very useful for estimating listening sound levels. They also appear to have better low level resistance stability.

For other measurements on passive attenuators:
https://www.superbestaudiofriends.o...uator-comparison-technical-measurements.7324/

https://www.superbestaudiofriends.o...ssive-attenuator-technical-measurements.7303/

 

Oh, you have no idea how important this is...................hints for the (near) future.

 

Lol, I think he probably does.

 

How about TKD ? Anyway you can get one of those in for comparison?

 

TKD 6R25 10K will arrive next week.
DACT 21 10K will arrive a week or two after the TKD.

 

Evil-me wants to try and convince the power hungry nutjobs* who run speaker amps for their headphones that they need to build their L-pads out of these for optimal sound quality...


*yes this is the pot calling the kettle black considering I used to run mine off monoblocks...

 

In the original post resistance stability of four volume controls were compared to a laboratory standard reference resistor. This post adds a “vintage” Penny & Giles RF11/2/CA obtained from ebay to the comparison.

See the first post for measurement methods employed and data collected.

Important Notes
The data below is NOT representative of a proper NOS P&G rotary fader.
Vintage and Brand Name do not automagically confer godlike audio quality upon the particular component.



The data from this specific P&G rotary fader should serve as a warning to the perils of acquiring used unobtainium.
There is a reason these components are available. They were likely removed from a functioning console as having become noisy.

Penny & Giles RF11 / 2 / CA test setup



Penny & Giles RF11 / 2 / CA adjusted for approximately 1K

After several days of measurements it became clear this potentiometer had issues. Exceptionally difficult to adjust for a specific resistance as it drifts for more than 35 minutes before settling after every adjustment. Every damn time the pot is turned. There are 2 gangs on this pot. The second gang was even worse with dead zones.



Once settled the stability observed varies 600 mΩ over short periods and 1.6Ω overall. This is orders of magnitude greater than the four controls originally tested.



Composite chart of the original four controls and lab reference resistor measured. Note y-axis is 0.005Ω per minor division.

P&G RF11 data overlaid on data from previous graph

y-axis is 0.1Ω per minor division. 20x greater than the previous graph.

This specific P&G rotary fader data overlaid on data from the previous graph. Definitely a defective potentiometer. The other four controls look very stable in comparison. Again, note this is not representative of a functioning P&G control. This highlights the problems with acquiring “vintage” components.

 

The sad part is this particular pot has probably been pulled and resold numerous times simply due to its branding.

 

Kind of like tubes...

 

This is in category of what lay audio people think dark magic is.
So simple and yet so little known mechanism.
What use of 16 bit microdynamic resolution when attenuator buldozes over this.
Well done investigation, ser.

Followup investigation idea - drift in sub second scale.

 

Since we're looking at stepped pots, what about IC volume chips? I believe that's how Jan Meier handles it in his amps. Granted, that adds a whole layer of complexity on top since they are active components, though I'm not entirely certain if some of them separate the signal path from the rest.

 

In the original post resistance stability of four volume controls were compared to a laboratory standard reference resistor. This post adds a TKD 2CP-601 100K potentiometer to the comparison.

See the first post for measurement methods employed and data collected.

Important Notes
This is an interim report. This was intended to be a 10K rather than a 100K TKD 2CP-601 potentiometer. Measurements were collected with the pot set for 10K out of 100K and scaling the results by a factor of 10 which may or may not be representative of a 10K pot set for 1K out of 10K.

In a week or two I will receive the following which will be added to the comparison:
TKD 2CP-601 10K
TKD 2CP-2511 10K
DACT 21 10K

4 wire measurements typically involve commercial test leads similar to those shown in the following picture:

When making sensitive, low level measurements, influences from the environment may introduce unwanted contamination to the data collected.

4-wire Kelvin method along with 10 NPLC is very robust at minimizing unwanted contamination. However, movement near the cables can still introduce contamination. Typically I must sit very still during a data collection, or delay start it and leave the room.

I decided to create a custom 4-wire test leads using Canare L-4E6S starquad cable, hopefully to further minimize issues such as cable movement contamination.

DMM end of custom test leads. The shield is connected to earth ground as an additional guard.


DUT end of custom test leads. In this case the DUT is my SR1 laboratory standard reference 1K resistor.


Commercial 4-wire Kelvin test leads connected to the SR1 for residual noise testing.
The cable was touched three times at
T = 12, 26 and 42 seconds. Observe the contamination introduced to the resistance stability measurement.


Using the custom starquad 4-wire cable with guard, the cable was touched again 3 times at T = 12, 26 and 42 seconds.
Observe no contamination introduced.

Details matter when making low level, real time measurments.

The following measurements were made with the custom starquad test leads abbreviated C4 hereafter.

Goldpoint SA1 compared to SR1 laboratory reference.
Performance is substantially similar over a 60 second period.


Alps RK27 potentiometer data added. Note y-axis has changed to accommodate the increased variance of the RK27 pot.

TKD 2CP-601 100K pot adjusted to 10K data:

This is the original data (not scaled) with pot adjusted to 10K out of 100K.

There was a problem with this data collection. More noise than expected is observed, approximately 1 Ω of variance. I will post an update when the issue has been resolved.



Scaling the data by a factor of 10 and normalizing to 1000 Ω.

If a linear scaling of the data is valid, then the TKD pot looks very promising as appearing to have lower resistance variance, better stability, than the RK27 pot.

There was a problem with a cut and paste operation between two spread sheets in which the formulae were pasted, not the values. If one looks carefully the data purported to be from the 2CP-601 is actually that of the RK27 scaled downward by a factor of 10.

This chart too will be updated when both of the above issues have been resolved.

When the 10K versions arrive, they will be added to this examination of passive volume controls.

 

Awesome!

Not surprised by the superior measurement of the TKD. Conductive plastic is a superior material to carbon film in just about every technical metric.

 

This is great data, thank you for putting these pots through such rigors and sharing the results! I've found it exceptionally difficult to measure long-term drift; awesome to see that you've been successful.

I'm curious if you considered either applying a very low frequency (say 0.01Hz or 0.1Hz), or looking at the data as an amplitude spectral density plot. My thinking is the stability of the resistance is effectively the stability of the voltage drop across the DUT under a 1mA stimulus. Removing the very low frequency information will help differentiate issues like self-heating of the smaller pots from other phenomena.

The idea with the amplitude spectral density plots is that you could then easily compare the noise density to the theoretical Johnson noise of the resistor, and get an understanding of what kind of excess noise we're dealing with-- broadband, 1/f, popcorn noise from intermittent wiper contact, thermocouple effects within the pot itself, something else? It's also possible to make this kind of measurement without applying a DC bias, so self-heating effects are reduced.

I attached an amplitude spectral density of a 20k RK27. My rig is not capable of sufficiently low noise measurement <5Hz or >1kHz as it's optimized for measuring low impedance noise sources. It does suggest that the noise has a 1/f spectrum.

Here are the APX settings. The lower trace is the noise floor of my measurement preamp.



And comparing that to the theoretical resistor noise:



Here's the computed excess noise. I subtracted the theoretical noise for a 20k pot from the measured value. The noise increases at lower frequencies, but the slope is flatter than typical for 1/f noise. The peaks all seem to be harmonics of 60Hz.

 

TKD 2CP-601 & 2CP-2511 series pots are currently 25% off at Parts Connexion (as are the equivalent ALPS):

https://partsconnexion.com/potentiometers/?sort=alphadesc

 

Crap. Going to spend cash lol. Thanks

 

Corrections necessary. Please re-read post 12 above

I am very sorry for any misrepresentations this may have given.
Updates will occur when issues have been resolved.

 

Engineers (the ones with imagination and capacity to invent) and spreadsheets do not get along all the time.

 

Especially tired ones that should have gone to bed instead of pushing through to see and share results.

 

In the first week of October, @atomicbob , I should have a Spectral Super Fader and another Penny&Giles pot to lend to the investigation. Will send to you as soon as I have them. Thanks for this work!