Impedance of Send and Return in the Loop

Rock_Zorro

I’m trying to find out what the output and input impedances might typically be for Send and Return in an effects loop. 

All I can find are references to Send connections usually being low impedance. Would 1k Send and 100k Return be somewhere close? Any actual specs?

What I want to do is put a voltage divider in the loop to attenuate the signal by -6dB making the master volume more sensitive at low playing volume.

Ta.

ICBM

It depends how well-designed the loop is. What amp is it?

In a *good* amp, yes - 1K send and 100K (or even higher) return would be expected. In a very few (eg Marshal Jubilee) the send impedance could be a lot worse than that.

But if you're just aiming to use a straightforward voltage divider, a 100K Log pot should work for almost all loops.

Rock_Zorro

Engl Gigmaster 15 combo. A fairly decent amp I’d say.

I was thinking of using a pair of 10k resistors worked into a short cable as I don’t need adjustment - just a fixed amount of signal attenuation so I have a greater working range on the master volume before it gets too loud.

Do you think this would work ICBM?

ICBM

I would expect that amp to have an IC-driven loop, yes - so low-impedance send/high-impedance return, which is good.

Two 10Ks will work but will give you only a 2:1 reduction (3dB 6dB), which is unlikely to be that useful. A 'Log' reduction - 10:1 rather than 2:1 - would probably work better. 100K (input to output) and 10K (output to ground) would do that easily. Actually 11:1 but it won't matter!

(Edit: doh. 2:1 on voltage is 6dB, not 3!)

Rock_Zorro

Oh, I must have my sums wrong. 

I thought 10k (input to output) and 10k (output to ground) would halve the input voltage (-6dB) which is the amount of attenuation I’m looking for. I need to do more reading on this!

I’ll try the 100k and 10k as you suggest, it’s only pennies to experiment. Thanks a lot for your input.

ICBM

Yes, you're right. Sorry, not thinking straight .

ecc83

Two to one is indeed a 6dB loss. 20log 2. Dunno where IC's head was!

But 10+10k will give you a bit more than 6 dB because Z out is never zero and Z in not infinity. Best way IMHO is to do what IC suggested, use a pot to find the loss you want then measure the two resistances so set and duplicate them with fixed Rs.

NB Always put attenuators at the SINK end not the source otherwise you could get HF cable loss, the dreaded "Tone Suck" as you gitiists would have it!.....And, LABEL THE FEKKIN' CABLE!

Of course SOME amp makers (cough!) give you decently low OPZ and high IPZ and a +4dBu/neg 10 option to boot!

Now, Maplin do a Lucky Bag of resistors and in it you will find some very wee ones and some weird values. Could be just the ticket!

Dave.

ICBM

ecc83 said:

Two to one is indeed a 6dB loss. 20log 2. Dunno where IC's head was!

Elsewhere .

I had just been talking to someone else about power attenuation for speaker ratings and forgot we were talking about voltage here… doh.

Rock_Zorro

Experimenting and measuring with a pot is a good idea - and the point about Z out and Z in not being ideal is understood. There’s a handy calculator which allows source and load impedance to be included in calculations to check.

http://www.sengpielaudio.com/calculator-voltagedivider.htm

Looking further into this I realise I’ve also made a mistake. I assumed a halving of voltage equated to a halving of perceived loudness, it doesn’t. To achieve this needs attenuation closer to -10dB I believe.

ecc83

Well it is not really a mistake because the "classical" figure of an INTENSITY change of 10 to 1 for a perceived halving or doubling of loudness was arrived at (years ago!) using AFAIK steady tone and lekky guitar is hardly that!

One thing I would say to anyone playing with and interested in sound, recording etc is, buy a cheap SPL meter. If you cannot run to one with switched curves get a C weighting jobby. Can be had well under £20.

Dave.