Digital pots using relays

[loudthud]

Wow, hard to believe this thread has been dead since 2009. Almost exactly 2 years. http://music-electronics-forum.com/t1015/ I thought I would start a new thread focusing on pots using relays. Refer to the old thread to view my posts on how to make a pot with relays and binary weighted resistors.

There was an article in the September 2011 issue of Audio Xpress where the author (Robert Nance Dee) used relays and binary weighted resistors to make a volume pot for a stereo system. He used 0.1% resistors, an optical encoder, and a micro-controller to control the relays. He says that there is no sense of stepping as the volume control is increased but he must not use the control near the ground end as you might do in a guitar amp with the tone controls. He uses 6 relays for 64 steps.

[R.G.]

There was an article in the September 2011 issue of Audio Xpress where the author (Robert Nance Dee) used relays and binary weighted resistors to make a volume pot for a stereo system. He used 0.1% resistors, an optical encoder, and a micro-controller to control the relays. He says that there is no sense of stepping as the volume control is increased but he must not use the control near the ground end as you might do in a guitar amp with the tone controls. He uses 6 relays for 64 steps.

I messed with this for a while. The use of relays sidesteps the issues of audio quality through a semiconductor switch. But if you can live with semiconductor switches, you might as well use a digital pot. Some of these are just resistor strings with a switch to the wiper contact on each tap. You can get them up to about 256 taps.

I remember an even earlier digital volume control that used JFETs for switches instead of relays, probably in the precursor name to Audio Xpress. That article pointed out the issues of relay timing, where if all the relays don't switch at the same millisecond, you will sometimes get a millisecond of detour through 100% volume. I think that was the rationale for JFETs instead of relays.

My conclusion on this is that the stepping motorized pot is still the best automation process if you demand audio through a pot and not semiconductor switches. But it's only my personal opinion. I wonder how the cost of six relays per pot compares to a 180 step stepper motor.

[jrfrond]

This is the basis of the design in some Digidesign/Avid products, usually for preamp gain control.

[Gregg]

Wow, hard to believe this thread has been dead since 2009. Almost exactly 2 years. LDRs as pots?

This is the thread I started couple of years ago. Since then I managed to build a complete programmable tube preamp with digital FX (based on AL3201) housed in a 1U rack case. I have 4 types of sounds - clean, crunch, drive and lead. Also 4 different "voicings" which in fact are a Tube screamer with fixed gain at zero and tone in the middle. What I do is switch different value caps in the 4k7/47nF circuit to get different frequencies.
Unfortunately I couldn't find time to "polish" it entirely but it was intended more as an experimental project and learning tool than as a complete finished device but it's working without problems so far.
Actually I used the LDR pot only as Gain pot. After the CF (loaded with a fixed TS) the signal is attenuated and further I'm using digital pots. I'll have to find some time and record couple of samples soon.
Recently I was thinking a lot about using relays for pots. There are SIL realys which are very compact in size and can be lined together to build a pot. I'm concerned however about vibrations because the next thing I'm planning to build is a complete head. I was wondering how the vibrations of a head sitting on top of a cabinet will affect the relays? Any comments on this would be appreciated.

[loudthud]

In the above referenced thread I posted a schematic of an idea for the binary/relay pot in post 139 and some attenuation numbers in post 146.

There is a problem as RG noted of the relay timing. One way to possibly get around the timing issue is to arrange the relays so that when all are released, the pot arm is connected to the low or ground side. So as a volume control, if you want to change positions you first release all the relays, then energize the ones for the new attenuation factor. That way it acts as a mute. But there is still a problem with relays when they are between energized and released. The end to end resistance will increase during that time.

Greg: Glad to hear you got something put together with the whole programmable pot idea. Looking forward to seeing and hearing the results.

[R.G.]

I think most if not all relays have specifications on what level of vibrational acceleration they can stand when operating. It's a big step from there to how much the relay shakes from air-born and structure-born vibrations. I suspect it's only an issue at the resonant frequency of the armature inside the relay...

Which we don't know.

The price of lots of relays is still an issue. Unless you're getting them surplus and of unknown parentage and provenance, they're about US$2.75 and up apiece. Assuming seven bits is enough (128 steps) and you can get that with seven relays, then one pot costs $19.25. I suspect 0.1% resistors cost something too.

[Gregg]

There is a problem as RG noted of the relay timing. One way to possibly get around the timing issue is to arrange the relays so that when all are released, the pot arm is connected to the low or ground side. So as a volume control, if you want to change positions you first release all the relays, then energize the ones for the new attenuation factor. That way it acts as a mute. But there is still a problem with relays when they are between energized and released. The end to end resistance will increase during that time.

I'm using a mute circuit all the time because I'm mostly into high gain stuff and if you have 3 channels you simply can't get without one.
The latch/release time of a SIL relay is around 1-2 msec which I think wouldn't be a problem if your mute circuit is 10-20msec.
I was thinking about 10 to 15 relays per "pot" which makes at least 100 relays for the whole amp. The prices of those can get less than a dollar the piece if you order directly from some chinese manufacturer.

[R.G.]

I'm using a mute circuit all the time because I'm mostly into high gain stuff and if you have 3 channels you simply can't get without one.
The latch/release time of a SIL relay is around 1-2 msec which I think wouldn't be a problem if your mute circuit is 10-20msec.

That should work.

I was thinking about 10 to 15 relays per "pot" which makes at least 100 relays for the whole amp. The prices of those can get less than a dollar the piece if you order directly from some chinese manufacturer.

Hey, if your budget can hold it, go for it.

Once past the initial steps of will it work, you get into some of the practicalities. Consider in your implementation:
- how will you mount the relays? all on one controls mega-board, or one per "pot" distributed?
- how will you power the coils? super low power relays run 25-50ma per unit, sometimes more, depending on the coil voltage; that's 2.5 to 5 A of coil current to get from somewhere.
- how will you run the control and power lines to the coils to keep the amp quiet?

As I said, I thought about this before; I'd love to see pics of your implementation when you get it done. Go for it.

[Gregg]

Once past the initial steps of will it work, you get into some of the practicalities. Consider in your implementation:
- how will you mount the relays? all on one controls mega-board, or one per "pot" distributed?
- how will you power the coils? super low power relays run 25-50ma per unit, sometimes more, depending on the coil voltage; that's 2.5 to 5 A of coil current to get from somewhere.
- how will you run the control and power lines to the coils to keep the amp quiet?

The Gain pot will be as close to tubes as possible and will be a separate one. I was thinking of adding it as a second "floor" on top of the main PCB.
The tone stack pots will be a separate PCB close to the CF.
All relays will be powered from their own regulated power supply. It's not a problem to order from the local transformer shop a PT with the windings I need. The 12V SIL realys I have in mind draw ~12mA each which is not much even if you have 100 of them.
I'm going to be building a new amp, not modding an existing one so everything will be taken care of (I hope) including the transformers.

As I said, I thought about this before; I'd love to see pics of your implementation when you get it done. Go for it.

Thanks for the encouragement! Considering how long it took the previous projects I hope to move this one forward faster. I'll definitely have some questions when I start so I'll let you know.

[Steve Conner]

The DGxxx type analog switches are available with 2 ohms on-resistance now, which eases the traditional worries about them causing distortion. They can be used in current mode to lower the distortion even further. To me they should now compete with relays in audiophile kit.

I recently did an experiment using back to back power MOSFETs as an AC switch to replace the speaker relay in a hi-fi amp. The extra THD from this device was maybe 0.005% at 100w into 4 ohms. RG originally suggested putting them in the amp's power rails, but I was a bit more adventurous.

My experience with relays is that I installed some Hi-G sealed relays for channel switching in a combo amp. They withstand it fine. I salvaged them from an old TV camera control panel, they must be getting on for 20 years old. I think they must be no more vibration sensitive than tubes, which is faint praise mind you.

How do you make a log pot out of binary weighted resistors, other than making a linear one with more steps than you need, and using a log lookup table?

[R.G.]

The DGxxx type analog switches are available with 2 ohms on-resistance now, which eases the traditional worries about them causing distortion. They can be used in current mode to lower the distortion even further. To me they should now compete with relays in audiophile kit.

I do like the DG series. What signal voltage can they support? I've seen some tube amps switch 30-70V of signal, so I worry a bit about that, but the DG series does great in their range.

I recently did an experiment using back to back power MOSFETs as an AC switch to replace the speaker relay in a hi-fi amp. The extra THD from this device was maybe 0.005% at 100w into 4 ohms. RG originally suggested putting them in the amp's power rails, but I was a bit more adventurous.

I always wanted to do that experiment. Thanks for plowing into it! I'd say that 0.005% is fine for guitar amps. Audiophiles would complain that their 5% distortion SE tube amps were too dirtied up by 0.005% distortion to be listenable, but - well, they're audiophiles. The power rails were a place where I figured even tweekos would have a hard time arguing.

How do you make a log pot out of binary weighted resistors, other than making a linear one with more steps than you need, and using a log lookup table?

The original article did equation solutions for resistors for getting 1db steps from one to ?ten?, which turned out to be something like ten or twelve resistors, then added other switched attenuators for 10db and 100db which were separately switched in, as I remember. It's been a long time. Analog Devices actually did a db-step D-A converter from a laser trimmed network on a chip. I don't know if you can still get them.

[Gregg]

I do like the DG series. What signal voltage can they support?

I don't know about the DG series but Maxim', MAX4800 and Supertex's HV2201, HV2601 are suitable for various combinations of high voltage supplies, like +40V/-160V, +100V/-100V, and +160V/-40V but they don't come very cheap.

[Steve Conner]

I think those things had rather high capacitances for tube amp work.

The switches I was thinking of were the DG9421 and DG9422. They can only handle +/-5V. There are other analog switches with low on-resistance that cover the +/-18V range.

There is a very poor choice of analog switches suitable for tube amps. I'd suggest redesigning the circuit so the switching can operate at low voltage, but once you stray away from the classic tube amp topologies, it's a slippery slope, and before you know it you're playing guitar through a DSP amp simulator that you wrote yourself, and your tube amp building friends cross the street when they see you coming.

One issue with relays is reliability. A circuit with 100 relays in it will give 20 times more relay trouble than a circuit with 5 relays. For that reason I'd not settle for anything less than hermetically sealed relays with gold contacts, which will bump the price up somewhat.

[exclamationmark]

but once you stray away from the classic tube amp topologies, it's a slippery slope, and before you know it you're playing guitar through a DSP amp simulator that you wrote yourself, and your tube amp building friends cross the street when they see you coming.

This pretty much describes me... except I don't have any friends

I've always wondered about chip makers implementing an actual ohmic resistive element in combination with several back to back MOSFET's or any other semiconductor switching topologies (sort of like a linear rheostat... except with selectable resistance taps). I'd pay good money for one of those!

[Steve Conner]

Err, I think that's the "digital pots" we are talking about. Available from several chip makers, but restricted to low voltages like 30v total.