Shunt volume controll (scaling the shunt resistors)

I have calculated some on a shunt volume controll. Where there is a fixed resistance lets says 10k (R1) in series with the signal. And then there are 8 relays which are controlled be an AD converter or an optical encoder which each will tie one resistor to ground (R21-R28).
<img src="">
However I have some trouble getting even steps (true logaritmic). So I'm wondering if anyone has got any idea, how to scale to shunt resistors.

An other idea I saw on, is to use six voltage dividers in series. Where the voltage dividers are 48dB, 24dB, 12dB, 6dB, 3dB and 1,5dB. This way I can get very precise 1,5dB steps from -94,5dB to 0dB (64steps). But the disadvantage is that there are more resistors in series with the signal.

What do you think?


[Edited by Freddie on 08-28-2001 at 12:13 PM]


abs right: 2^8 is 256.

I'll take a step size of 0,5 dB as 256*0,5 is 128dB and this in theory at least, would be enough to turn down the noise of a jet-engine at 1 m to nearly dead quiet...

I'll propably end up with an Excell 5.0 spreadsheet where you can input step size, value of R1 etc and it returns the values of R21 etc


The first figures out of the spreadsheet give some reason to rethink it all.

When using in total 256 steps of 0,5 dB you actually will be setting the volume at 75% or more in normal day-to-day use. That is around the -30dB point in this case...

That feels not right...


[Edited by dutch diy on 08-28-2001 at 05:12 PM]
Re: scaling

Hi Freddie,

more figures turn up from the spreadsheet and it looks even worse.

I was designing this model 'backwards' that means I first do basic calculations and formulas to get the needed equivalent for R21 - R28 and than concentrate on the formulas for individual values for R21 etc.

Part one has been completed and behold:
Given 10k R1 and 256 steps of 0,5dB (where I've assigned step 256 (all resistors used)to be the 0% turn, lowest volume.)
until you reach 45% turn the equavalent of R21-R28 is to be below 0,001 Ohm which is not ever feasable.

So, I quit for now.
I suggest you use less resistors and use a spreadsheet which
produces the actual dB value for a given set of values for R21 - R26.

Let me know if you need help on that part.



2001-06-01 4:53 pm
Just because you have 256 (or 64) possible steps doesn't mean you have to use them all. You need some decoding logic anyway to turn on the right combination, maybe a mapping circuit that maps each of, say, 20 switch positions to one of the 256 codes. Having 256 codes to choose from makes it pretty certain you can get the right 20.
Just a thought.
Re: scaling


First results (values in Ohms, based on R1 = 10k and input stage assumed to have infinit resistance):

R21 24240, R22 10047, R23 3354, R24 673, R25 39.9, R26 0.16

NOTE: These values only 'match' the ideal sequence at certain points: -1.5; -3; -6; -12; -24 and -48dB and produce 'less' than desired reduction in all between points

Because R26 is magnitudes smaller than any other resistor putting those in parallel doesn't reduce V<sub>out</sub> any further than -48 dB.

I have a RTF-document avaliable which contains more info (and all the calculated values of the R<sub>shunt</sub>) if you want, but I need your e-mail adress.


[Edited by dutch diy on 08-30-2001 at 05:00 PM]