MultiChannel Volume Controls

[WJS]

I'm thinking about making a mutli-channel (5.1 or so) tube preamp but I was wondering about how to design the volume control sections. It?s easy with single or two channel designs, but there must be a better way than having a huge pot.
I know there are switches that are (many)Throw(>6)Pole but that would require a lot of resistors.

Any creative solutions to simultaneously control the volume of several tube channels evenly?


There are way's to do it that involve silicon... but I'd rather not go that route.
Thanks,
Bill

 

[adamasd]

You can use LDRs/LEDs or vactrols, but you have to match them, which can be a pain. Although if you went the LED/LDR route, you only need on LED to drive all the LDRs, so that simplifies things, no LED matching. There was a thread about how to get vactrols to have a more even response from part to part not to long ago, it would not get you perfect resaults, but it should make it so you do not have to match them as closely to get good resaults.

But how close do you really need to be? Even matched tubes do stay matched all that long. Some quick LDR or vactrol matching to get it close might be good enough.

adam

 

[WJS]

hmm... I've never tried that approach. I wonder how well it would work with 6+ LDR's though... the light from the led would have to be fairly evenly spread, and that spread be static if it was to be compensated some how.

Definatly something to think on.
Thanks

 

[adamasd]

Clear epoxy should work good to diffuse the light and spread it evenly. I would use a single bright LED and suspend it in the center a some sort of half sphere mold. A mellon baller seems like it might make a good mold. center the LED left to right and have the bottom of the LED level with the flat side of the mold. Then fill it with epoxy and let it dry. Maybe use something to vibrate the mold, get out any air bubbles that would screw with the light. Once dry get it out of the mold and arrange the 6 LDRs around the top like a halo. That should keep them all the same distence from the LED and at the same angle from the LED, which should get pretty consistent light to each LDR. Then just encase the whole assembly in something light tight and off you go.

The only problem is keeping the epoxy from sticking to the mold. I have had some luck doing things like this by useing nonstick sprays, but there is probably something better. Abit of experimentation should get some useable resaults. If you can find a mold made out of something very much non-porus, there are a few varieties of epoxy that will not hold to such things, like highly polished metal.

adam

 

[Michael Tibes]

[quote author="adamasd"]You can use LDRs/LEDs or vactrols, but you have to match them, which can be a pain. [/quote]

I doubt that this route will be successful. LDRs are infrared sensitive - which means they change value with temperature. I had a programmable guitar tubeamp 20 years ago in which it was tried to use LDRs as control elements. It sounded great, but the sound was never consistent for even 10 minutes.
There are not many ways to go which come to my mind if you don't want to involve silicon in the signal path:
- a potentiometer works well, is pretty expensive in a good quality and match and might get scratchy after a while (biggest downside for me),
- switches with resistors, perhaps even more expensive than a good pot http://www.dact.com/,
- maybe relay switched resitors like in a R/2R D/A converter, I believe http://electronics.dantimax.dk/Kits/index.html might use that. This looks promising to me, though it seems most complex at first look.

Michael

 

[Boswell]

I know there are switches that are (many)Throw(>6)Pole but that would require a lot of resistors.

It's no more resistors than individual channel switched gains - it's where they are situated that makes the difference.

Have you considered the kit switches where you buy wafers and shaft assemblies separately and make up your own? With careful mechanical design, you could keep the wafer with its resistors local to its channel and have a long shaft running right through all the channels.

There was a thread about how to get vactrols to have a more even response from part to part

Yes, that was me mentioning one of the commercial designs I did some time ago. That one was in relation to a compressor, and gave about 40dB of gain control. Having done many Vactrol designs, both single and matched multi-channel, I would be very cautious about using them for the application you mention.

 

[Michael Krusch]

How about using THAT2180 as VCA? You can use the audiopath of the Gyraf SSL, omit the sidechain and add a pot for the control voltage.

 

[TedF]

Yes, you could use VCAs, but the amount of electronics needed would make it expensive.
It is not too difficult to use LED/LDR combinations to make a decent fader.... If you use an LDR in series, and then another down to ground (a potential divider) and design a little bit of analogue to make one string of LEDs go down in brightness while another goes up, you can get reliable noise and distortion free fading over a range of at least 80dB.

If anyone's really serious about trying it, email me direct and I will let you have a circuit that works. :wink:

 

[Michael Krusch]

Optos are less expensive, but VCAs will give you better tracking. If you select optos for good tracking you have to buy a lot of optos. I think in the end VCAs will be the less expensive way.

 

[WJS]

I kinda like the idea of a VAC, but it requires a good bit more electronics, and parts (usually another pentode from the little bit of research i was able to do(if i'm wrong please tell me!)). The LED/LDR option, sounds interesting but also has the potential to be a big problem. As this is going to be the main preamp for my system, any slight deviations introduced here will be amplified by the rest of the system... how much to these change with temperature? I'm guessing this will be a factor because of the heat from the tubes.
I'm thinking about going on a resistor soldering run and make a giant stepped atten. but this has problems also, I need to match not only pairs of resistors, but 6...

i'm starting to look at things like this:
http://www.analog.com/en/prod/0,,761_797_AD5206%2C00.html

 

[chrissugar]

Hey Bill

Why overcomplicating the whole thing. Buy a six waffer ELMA switch with 24 steps and build yourself a stepped attenuator. It is not cheap but very precise on all channels if resistors are measured, and sonically there is no better solution. You will not regret it.

chrissugar

 

[Michael Krusch]

[quote author="WJS"]i'm starting to look at things like this:
http://www.analog.com/en/prod/0,,761_797_AD5206%2C00.html[/quote]

This one is controled by SPI, so you have to add a microcontroler for control.

 

[Boswell]

i'm starting to look at things like this:
http://www.analog.com/en/prod/0,,761_797_AD5206%2C00.html

I've used the AD5206. It's fine for instrumentation work, but has limitations for use in audio. Despite the 256 steps, it's still a linear device, so the steps are coarse in dB terms at high attenuation settings. There is also some digital breakthrough unless you are very careful with the layout. You really need to use optoisolators on the digital inputs.

 

[WJS]

I think i might just do a stepped atten. I think that might just be the easiest/best.

Its very nice to have access to such knoledgeable people.
thanks for you help!

 

[WJS]

this project got stuck under some work, but i've still been designing...


I was looking at the ELMA swiches, quiet a variety!

any recomendations?
I also need a distrib. that ships to the us.

it doesn't have to be elma. I just need at least 6 channels, and i guess 24 steps should do just fine.
resistors are fine, i live near an nte place that will sell me them.

thoughts?

 

[kiira]

I have a 5 deck 24 pole Greyhill I bet you could use... five channels right? You could build a shunted step attenuator on each deck for each channel.It's a high quality switch.

Kiira

 

[Rochey]

If you can tolerate generating an SPI chain, then I'd look at using the PGA2310 (www.ti.com)

SPI is just a clocked through serial "word" -- if you look at TI's MSP430 processors (which you can get the dev kits for $20) then we already have an example of serial SPI output.

There's also an example (in code) of using the onboard ADC to read a voltage value and generate a "word" from it that you can clock through to the PGA2310.

just a thought.

cheers

R

 

[shabtek]

try reading data sheet for vishay vcr2n series jfet...has sample applications; works simply as shunt resistor for low level audio and can handle hotter signals with simple feedback design. needs low level negative rail and pot or stepped switch to vary resistance.
they are cheap so you could find 5 that closely match for not too much dough.

could be part of simple compressor too

 

[emrr]

Second vote on the stepped passive approach...