Pro Audio Design Forum

I'm trying to replace some obscure pots with switches. However, some switches are so large that I have to reside to relays. My goal is to minimize the relay count. Now, with linear pots it's quite simple to use 4 relays and get 16 even steps feeding binary numbers into relays coils. I can even use a binary-coded switches for that. I'm stuck with log/antilog potentiometers that I can't get to work in the same manner.

I've looked up r-2r log ladders, but they are made as attenuators (mostly L-pads), and this is not suitable since I don't have the 3-d connection (ground). I need a variable resistor, not attenuator.

Are there any solutions to this problem?

So you're looking for a "two terminal" (rheostat) log/antilog taper "potentiometer?"

Not sure it will help but have a look at this "Resistors Provide Nonlinear Pot Tapers," Mark Rumreich EDN November 13 1986. viewtopic.php?f=12&t=866#p10260

You might be able to switch both R and R' in the above to reduce the number of relays.

I realize that you cannot use a three terminal device but for our readers interested in R2R ladder networks this is an interesting resource: "Logarithmic Attenuator Calculator" http://www.eijndhoven.net/jos/attenuato ... index.html

This is most helpful! Thanks! Rheostat is what I need. Those taper "bending" resistors are very handy. Will see how it works with linear switching pot.

The attenuator site you mentioned is also a very interesting resource. Unfortunately it doesn't suite my particular need.

I've explored the tapering option and it's not that easy. The problem is: I need the "20" multiplier for the correct taper. When I select the appropriate values for R and R' (2020k and 105k), the minimum resistance (for 16 steps, equally divided for the linear pot) is 58K which is too much. I need a span of 500R - 100K.
Gonna tinker with series/parallel circuits and scaling resistors some more...

No luck with relays. It takes too many of them to cover any useful range of the log pot.

I started to look in the direction of JFET switches which led me to one interestion TI app note with the schematic of video switcher. I played with it in the sim and it looks fine, I don't get any drastic increase in THD. But this is sim. Real life may be quite different. What do you think of this solution?
The plan is to switch only one resistor in the circuit by JFET at a time thus minimizing the number of JFETS the signal passes through. Any thoughts?

EDIT: It just occured to me that maybe DG412 will do the job...

This seems like an unusual amount of work to replace a few pots..

I am not a fan of slugging pots but that is one way to trick linear pots into non-linear transfer functions.

JFET switches are not going to be very linear especially for low Z.

I have done some work to make electronic switches very linear (by embedding them inside the NF loop), but that was several decades ago, and not practical for a simple pot replacement.

JR

This is an EQ project, so replacing a pot with a rotary switch is a very useful option. Especially if this pot is a dual rev log beast. My concern is tracking between the decks (which in rev log may be horrible). Slugging is an option, but it will require a higher resistance value pot, in my case around 2Meg.