"Quad-track potentiometer" for accurate attenuation?

[leigh]

I was just reading a review in Sound on Sound for the excellent-looking new monitor controller from Drawmer, called the MC2.1. In the review, they talk about the design used for the main volume knob, and about how they achieved tight tracking in the attentuation level between the stereo output channels.

Now, SOS is usually pretty technically accurate in their reviews, and they go to lengths that (sadly) few other publications bother with anymore, like checking manufacturer's specs using spectrum analyzers and whatnot. That said, I cannot make heads nor tails of their explanation of how this unit's attenuator works:

Drawmer have used a custom quad-track potentiometer configured with interleaved and paralleled connections. This clever approach minimises tracking variations extremely well, and I found that there was never more than 0.25dB difference between the two output channels...

It's really the "interleaved and paralleled" part I don't get. I could see taking two potentiometer tracks per channel, and averaging them to smooth out local variances in the amount of carbon deposited on a particular point on one of the tracks. You figure that's what they mean, or is there more to it, on the "custom pot" side of things?


PS: The review is here, although you'll need a subscription to read the whole thing: http://www.soundonsound.com/sos/apr14/articles/drawmer-mc21.htm

 

[JohnRoberts]

To understand you need to think about how pots are made and what causes tracking errors.

Pots are generally fabricated by screening a resistive ink onto a non-conductove substrate forming an arc for rotary pots and linear traces for slide pots. Voltages (or a voltage and ground) are connected to both ends of this resistive element. Then a conductive metal wiper makes contact with the resistive element along it's length to pick off a voltage.

Generally the bulk resistance or total resistance is not very precise (20% ?) but the precision of the curve, which us usually silk screened can be very repeatable. Finally for stereo dual-ganged controls we have mechanical coupling accuracy between the two sections.

The dual wiper design gets a little tricky, but you can make a pot with two independent screened resistive tracks side by side, and dual wiper that picks off the voltages from both separately. This way the bulk resistance will track better because it was literally screened and processed (cooked) identically on the same substrate. The single two part wiper tracks a lot better than mechanically coupled dual pots.

We can imagine a difference in a rotary pot between the inner and outer arc since the length of the outer arc will be longer. Perhaps they screened 4 interleaved traces and combined them such that the length difference between the two pairs was more alike.

I have seen and used special oddball pots, not exactly like this one, but similar tricks inside. I expect the tooling cost to make the oddball wiper the hardest part, Another consideration the 4 traces and extra spaces in the area that normally holds one pot trace, means the traces will be pretty narrow. The narrow traces create another issue related to the number of wiper fingers contacting each trace which will be less than standard, but 4 traces with two pairs in parallel might help that some.

There is no free lunch but this should deliver much improved section to section tracking with perhaps a slight increase in wiper bounce. If you take apart and look at a typical pot wiper you will see multiple small wiper fingers side by side. The concept with multiple wiper fingers is that there should always be one or more making good electrical contact with the resistive element while others are momentarily bouncing and not in contact. (I have experienced issues from too few fingers and too narrow resistive tracks in exotic pot designs).   

This is somewhat speculation on my part, but how I would do what it sounds like they are describing.

JR

NOTE: THis approach will only improve the section to section tracking, it will not affect the pot taper attenuation accuracy or anything else.  I recall back in the day Alps made some rotary controls that were literally stepped attenuators contacting a small array of screened(?) resistors to form a precision attenuator. Perhaps the resistors were early days SMD, but IIRC they were screened.

 

[thermionic]

Is it one of these that you're thinking of John: http://s646.photobucket.com/user/Justin-G-2009/media/LevelAttenuator-crop.jpg.html ?

The one shown in the link, after cleaning, fitted to the battleship Sony preamp it came as standard with, measured better than 0.1dB at just about anywhere on the dial. This is a random scope shot I took: http://s646.photobucket.com/user/Justin-G-2009/media/LineLevels-crop.jpg.html?sort=3&o=6

As much as it's a testament to Alps' manufacturing tolerances, it also says something about the Sony preamp's QC tolerances.

It saddens me immensely to think that there are virtually no stereo pots available today with this kind of channel matching. Even P+G doesn't come close. I've tried them all: Ko On, you name it. The Alps RK50 can be found in narrow values,  but it's about $700USD and I'm not sure if it has the stepped etchings  you see in the one seen in the Sony preamp.

I sold the Sony preamp to an owner of a high-end studio (he had a K-series and all manner of outboard). He wasn't happy at all...even though the Sony measured its exact figures in my analyser... He felt it sounded bland and one-dimensional... I'm sure it would've been a different story if they'd stuck a dirty great transformer on the output... From memory, the Sony had no caps directly in the path - it was complementary from in to out. It seems a distant memory to think that companies the size of Sony made cost-no-object uber-hifi....

 

[JohnRoberts]

While that is a different animal, it looks like it was laser trimmed to dial in the taper... Kool, excessive but kool.

If I cared that much about tracking I'd use a switch, or DIGITAL.... ;D ;D

JR

 

[leigh]

Cool info, thanks!

If I cared that much about tracking I'd use a switch, or DIGITAL...

Yeah, that's a good option! I'm currently using an all-passive monitor controller I built, centered around a Dantimax stepped attenuator relay board. That has very tight channel matching...

 

[topcat]

I saw this at the factory the last time I visited (They made 250 custom "N@Wal" noise at work limiters/headphone amps for my previous employer) and it is very clever...IIRC (always difficult) they use a 4 gang pot with layers 1 and 3 and 2 and 4 paralleled to get over the 'mis-tracking' between left and right particularly at low levels.....

The 'box' sounds very sweet and is well engineered.....

These guys are very very smart but fortunately don't take themselves too seriously...They just seem to want to make great sounding kit that is well engineered at an honest price.....they are also very nice people.

Sorry if it sounds like a sales plug but working with the team at Drawmer was the work highlight of last year.

tc