Reed Switch Audio Attenuator

A bit of background for this project. I'm getting to the point in my home studio where I want to have a dedicated monitor controller. Currently I'm using a cheap Yamaha MG166cx to route signals from my various devices to my Adam A5X studio monitors. My audio interface is a humble Echo Audiofire 2 that is also connected to the mixer. I can only imagine how many low quality components the signal goes through in that mixer. Needless to say when I directly connect the interface to the monitors and use software to control level I notice a definite improvement in sound quality. I'm also, slowly, working on a BIII build and when it's done there's no way I'm going to run it through that mixer.

Now I'm in the process of designing a custom monitor controller that brings together everything I need. In addition to transparent level control I'd like to have source selection, mute, dim and simple mixing functionality. I don't mean a summing mixer, more of to combine an aux signal with the main stereo signal and it's specifically for when I video game. ;) This controller will be a project thread in itself.

So to get to the point of this thread how do I control level in the most transparent way possible? I've looked at many of the options: stepped attenuators, LDR's, transformers, relay switched arrays, chips etc. Personally I don't want to deal with micro-controllers and I don't need remote control. What I want: ideally it would be a proper ladder attenuator, allow use of through-hole resistors, have enough positions to not be too course, be able to support balanced stereo with the possibility to expand it to balanced 5.1 in the future. Easy right? :rolleyes:

I decided to pursue using relays as the most promising avenue for development. My initial thought was to use a single gang rotary switch to manually control multiple gangs of relays. This approach is nice since one switch can control many gangs simultaneously and the actual contacts carrying audio (inside the relays) can be sealed away from the atmosphere. Of course the amount of relays gets stupid quickly. For a 36 position balanced stereo ladder attenuator you'd need 144x DPST relays. The ubiquitous Omcron G6 series are a few bucks each so this would end up costing quite a lot. Not too mention the gangs would be huge due to the physical size of the relays. For my application I don't mind the size, but the cost would be killer for balanced 5.1.

I've been thinking about this for a while now and have been making drawings of various possibilities and combinations. I started looking for smaller and cheaper relays which led me to reed relays. Better but still not quite what I was looking for. Then one night the moment came when I realized you could just use reed switches and a neo magnet to directly activate them.

For those that don't know a reed switch is basically a reed relay with no coil: Reed switch - Wikipedia, the free encyclopedia

The great thing is that they are cheap and very small. Unfortunately no one makes a DPST version so it gets more complicated doing a ladder style layout. While personally I've never heard of or seen a reed switch audio attenuator they are out there apparently: Zero Feedback SE Transformer Mosfet Audio Preamplifier-Line Driver
 
I started working on the first prototype last night after work. This one will just be a simple 24 position series attenuator to prove the concept. I just want to determine the proper spacing of the switches and what size magnet will give a make-before-break behavior. Unfortunately the reed switches I bought locally have super short leads so they are rather hard to work with. Here's where it's at now:











The left side of this will have the reed switches much closer together to see what happens. I'm going to wire up some LED's in a row to see how it all behaves as you rotate the magnet arm. Will have to order a variety pack of small neo magnets to play around with. Also some smaller reed switches would allow for a tighter layout. This is going to be an on going design thread so any input/feedback is welcome. Thanks for looking. :)
 
Neat design!
How do obtain a position stop - so that when the control is mounted vertically the magnet doesn't fall to the lowest, 6 clock, position?

Thanks. I was thinking the position stop would be a gang all on its own. That way when I make a 12 gang version for balanced 5.1 the single position stop gang would be closest to the knob to minimize twisting of the shaft. The position stop itself could simply be a thin metal plate with an arc of holes drilled to match the number and position of the reed switches. So for the prototype shown above it would be 24 holes, 12.5 degrees apart, starting at 233.75° and ending at 306.25°. The spring-loaded ball detent might be something like this ball plunger: Stainless Steel Ball Plungers | Ball Plunger The threaded body would allow for fine tuning of the desired spring pressure.
 

Thank you for the excellent link. Weiss really does make some nice stuff. It's helpful that they show so many pictures of the ATT202 internal construction on that page. Looks like they have the reed switches rather close together. That's great to see because I was worried they would have to be spaced 1/4" apart. Since I plan to build a 48 or 60 position attenuator this is good news.
 
FYI - You don't need 36 relays for a 36 position attenuator. It is possible to make 127 step using 7 relays. Arranged as 1/2dB per step gives a practical 0dB to -63dB attenuator. Think binary...

Sent from my Nexus 7 using Tapatalk HD

Are you referring to the R-2R resistor ladder topology? They do indeed save on relays, but they also have widely varying impedance. There's also quite a few components in the signal path at times. Plus they need a binary counter to operate the relays which means a micro-controller of some sort.

None of the above traits are desirable to me. My goal with this project is to build a true ladder attenuator with passive operation. If there are any tricks to doing that using less switches I'm open to input.
 
Are you referring to the R-2R resistor ladder topology? They do indeed save on relays, but they also have widely varying impedance.

Not necessarily R-2R but that kind of thing. Input resistance can be fixed vs. level setting. Output resistance varies still. But that will with you project as is, I think?

There's also quite a few components in the signal path at times. Plus they need a binary counter to operate the relays which means a micro-controller of some sort.

Yes, more components in path, it is true. For me that just meant finding sealed signal relays and high quality res. Binary counter, however, does not necessitate using a microcontroller.

None of the above traits are desirable to me. My goal with this project is to build a true ladder attenuator with passive operation. If there are any tricks to doing that using less switches I'm open to input.

I have no tricks for less switches and maintaining constant input resistance. Soz.

All the best with your project. I'm not on a mission to dissuade you. Hopefully my input will be useful/interesting/thought provoking for others visiting this thread.

Regards,
 
But they will be along soon to tell you how magnetics screw up

Not so much the magnetics, but the required lumped mass, and the cost.

Contact force of Reed switches is rather low, and they're prone to vibration.
So the attenuator has to be a bit bulky, which comes as a gratuity by using a stepper motor, the natural choice for accurate switching.
An Astrosyn type-14 mini stepper motor, as used by Weiss, easily adds $60 to the total cost (which is cheap, compared to what e.g. an Airpax stepper motor used to cost)

A Hamlin MDSR-4 reed switch now does a few dimes each, but before 2k they cost two times as much, >100 wholesale, without counting inflation.
(made a stepped attenuator similar to Weiss' in the late '80s, had to resort to the cheapest ones available, MDRR-4, still ~$60 just for the Reeds)

Very nice ones to look for is the Hamlin MDC series, mercury wetted.
An MDC-3 e.g. was over $5 each at 100 minimum in the early '90s, NOS ones now sometimes $50 for hundred.
(downside is the diameter, 0.1", requires an H33 magnet)