Am I crazy for thinking 16 steps is enough steps?

[scootsity5]

I found a 4 gang, 16 position rotary grayhill switch with fancy gold plated contacts in my parts bin. It's been in there forever for the 'right project'. I'm currently working on a pre and have been debating building a stepped attenuator.

Am I crazy to think 16 positions is enough?

Has anyone built a 16 step attenuator?

How did you pick your resistor values? I was just going to plot a log curve, select 16 equally spaced points and get values, but if there's a better way, I'd love to know.

Thanks!

 

[trainbuffton]

16 positions is too few and a grayhill will be very small to solder to.

Id look for a Daven/shalco/shalcross switch.

I found a 24 position shorting switch on ebay with 10 gangs for $1

I removed 8 of the gangs and ended up with a nice attenuator which is in a preamp thats not finished.

I would hold out for a good shalco because youll get plenty of room to work.

Youll have 2 resistors per channel per step.
(You switch in a different top and bottom resistor for each position.)
That gets you a true h pattern ladder which is what you want. The h pattern presents the same load to the preceeding stage so it is operating into the same load regardless of the position of the switch. Shunt attenuators dont do that. They present a variable load to the preceeding stage which is not what ya want.

The goldpoints and other commercial units only switch one resistor out. (Shunt type) In order to do this right you need to build it yourself.

If i were you, id start with a cheap pot.
Figure out where max volume is for your system and record the high and low values to create the "top" of the attenuator.

Next find out where you listen most and consider putting a few closer grouped steps in the range where youll need more fine control.

Fwiw you can get a 48 position shalco but it'll probably cost you more than the $1 i paid.

 

[scootsity5]

16 positions is too few and a grayhill will be very small to solder to.

Id look for a Daven/shalco/shalcross switch.

I found a 24 position shorting switch on ebay with 10 gangs for $1

I removed 8 of the gangs and ended up with a nice attenuator which is in a preamp thats not finished.

I would hold out for a good shalco because youll get plenty of room to work.

Youll have 2 resistors per channel per step.
(You switch in a different top and bottom resistor for each position.)
That gets you a true h pattern ladder which is what you want. The h pattern presents the same load to the preceeding stage so it is operating into the same load regardless of the position of the switch. Shunt attenuators dont do that. They present a variable load to the preceeding stage which is not what ya want.

The goldpoints and other commercial units only switch one resistor out. (Shunt type) In order to do this right you need to build it yourself.

If i were you, id start with a cheap pot.
Figure out where max volume is for your system and record the high and low values to create the "top" of the attenuator.

Next find out where you listen most and consider putting a few closer grouped steps in the range where youll need more fine control.

Fwiw you can get a 48 position shalco but it'll probably cost you more than the $1 i paid.

I sorta figured I'd end up regretting 16 positions. I'll keep an eye out for one with more steps.

This grayhill has contacts spaced roughly as far apart as the leads on a through hole resistor, it may be a different part than you're thinking of.

I was going to keep the resistor across the input constant, and vary the output, but yes, I'd prefer to vary both (if I can find one with six decks).

Thanks!

 

[nerdorama]

You could consider using another 3 to 6 position switch as a coarse adjustment and the 16 position for the fine adjustment. If you set the 16 up for 1dB or some other fine increment, a 3 position coarse switch would yield 48 positions.

Here is article about this sort of scheme from TubeCAD Journal.

New and Improved TCJ Stepped Annuator

www.tubecad.com

 

[scootsity5]

You could consider using another 3 to 6 position switch as a coarse adjustment and the 16 position for the fine adjustment. If you set the 16 up for 1dB or some other fine increment, a 3 position coarse switch would yield 48 positions.

Here is article about this sort of scheme from TubeCAD Journal.

New and Improved TCJ Stepped Annuator

www.tubecad.com

I thought about that. It's an interesting idea.

 

[scootsity5]

I also did some thinking today and could absolutely build a voltage split on 4 decks by just soldering the ground side of the resistors together off the switch, they don't need to be switched.

So, I can still wire it like a pot, it just will only have 16 steps. Which, if I then added a 2db pad on a second switch, while spacing the 16 step switch at 4db, it would basically allow 0-48db attenuation in 2db steps. I'm not sure I actually want to do that, but it's actually a workable idea.

 

[rothwellaudio]

You could consider using another 3 to 6 position switch as a coarse adjustment and the 16 position for the fine adjustment. If you set the 16 up for 1dB or some other fine increment, a 3 position coarse switch would yield 48 positions.

Here is article about this sort of scheme from TubeCAD Journal.

New and Improved TCJ Stepped Annuator

www.tubecad.com

I was doing that 30 years ago. Yes, it's very effective. The simplest way to expand the usability of the existing 16 way switch would be to add one (dual gang) on-off toggle switch to knock down the entire signal by 20dB to change the range over which the 16 way operates. Alternatively, wire the toggle for a 2dB drop so it "fills in" the coarser steps of the 16 way.
Ideally, you want a range on the volume control of about 60dB. Or course, the gain of the preamp you're making will be a big factor. Too much gain will exacerbate the problems of a 16 way attenuator. Give it as little gain as you can get away with and you'll be ok. You shouldn't need any more than 10dB.

 

[scootsity5]

I was doing that 30 years ago. Yes, it's very effective. The simplest way to expand the usability of the existing 16 way switch would be to add one (dual gang) on-off toggle switch to knock down the entire signal by 20dB to change the range over which the 16 way operates. Alternatively, wire the toggle for a 2dB drop so it "fills in" the coarser steps of the 16 way.
Ideally, you want a range on the volume control of about 60dB. Or course, the gain of the preamp you're making will be a big factor. Too much gain will exacerbate the problems of a 16 way attenuator. Give it as little gain as you can get away with and you'll be ok. You shouldn't need any more than 10dB.

No gain on the pre, it's just a relay-based selector switch (except on the phono, there's a phono pre).

I just went through the math, based on available precision resistors at digikey, and I actually think 16 steps may suffice, with the following attenuation:
54, 32, 28, 26, 23, 20, 17, 16, 14, 12, 10, 8, 6, 4, 2, 0

I'm going to build it and give it a try.

 

[scootsity5]

Actually, if I parallel a 5m with a 50k, I can get the highest attenuation down to 40dB. Given that I also built in a mute switch, I think that should suffice.

 

[rothwellaudio]

I just went through the math, based on available precision resistors at digikey, and I actually think 16 steps may suffice, with the following attenuation:
54, 32, 28, 26, 23, 20, 17, 16, 14, 12, 10, 8, 6, 4, 2, 0

1% resistors will give you 0.1dB channel balance, which is far superior to any pot. You don't need better tolerance than that.
The addition of one switched resistor will turn your existing 16 steps into effectively 31 steps, which is usually considered sufficient for a stepped attenuator.

The stepped attenuators which I used to make had 132 steps plus full mute, the channel balance was 0.1dB, the attenuation went down to -67dB and the step sizes were less than 1dB. That was done by using a 12 way coarse control and a 12 way fine control.

 

[scootsity5]

1% resistors will give you 0.1dB channel balance, which is far superior to any pot. You don't need better tolerance than that.
The addition of one switched resistor will turn your existing 16 steps into effectively 31 steps, which is usually considered sufficient for a stepped attenuator.

The stepped attenuators which I used to make had 132 steps plus full mute, the channel balance was 0.1dB, the attenuation went down to -67dB and the step sizes were less than 1dB. That was done by using a 12 way coarse control and a 12 way fine control.

I think I'm going to do this in the future, but for now I'm going with a TKD pot. The price was right, and their ladder pots seem like a pretty solid compromise between a standard pot and a stepped attenuator. Mostly, I prefer this particular amp to just have a single knob.

You're right about resistors, I made too much of higher precision, fancy resistors. I'll spec it out with regular 1% and see how cheap it could be.

Maybe I'll get/build both and see which I prefer.

Thanks!