What attenuator impedance do I shoot for?
Dec 12, 2007 at 5:51 AM Thread Starter Post #1 of 12

cerbie

1000+ Head-Fier
Joined
Mar 12, 2005
Posts
1,219
Likes
12
I'm going to putting an attenuator, integrated into a cable, between my DAC (this is the future, not what's in my sig) and amp. I always have to have volume levels super low (this is true for any PC source I've used), and figure this will be a good way to slap it down and get some room to play...and be fun to do.

But, what kind of overall impedance should I be looking at using? I guess I want the lowest I can get away with, but how do I figure out what that should be?
 
Dec 12, 2007 at 8:28 AM Post #2 of 12
Cerbie, im afraid i can't help with the impedance, but i am very interested as to how you make this cable. Could you take some step by step pics?

Cheers
 
Dec 12, 2007 at 12:04 PM Post #3 of 12
Quote:

Originally Posted by J.D.N /img/forum/go_quote.gif
Cerbie, im afraid i can't help with the impedance, but i am very interested as to how you make this cable. Could you take some step by step pics?

Cheers



I'm interested in how I'll make the cable, too. I really won't know until I have the final parts in front of me. I'd like to fit it all inside the cable's shield, and solder the shielding nicely together at the Y...but I'll just have to see. The test unit (oh, I'll post pics of that!) will be my first IC at all, so I'm quite unsure about many things, and several of my parts choices are experimental, but the topic question is all I haven't gotten at least one good answer to by searching.
 
Dec 12, 2007 at 12:32 PM Post #4 of 12
Due to the amp's high input impedance (usually anywhere from 10K to 1M ohms), you'd have to use a very high value series resistor in order to achieve any significant attenuation. What is high? It'd have to be the same value as the amp's input impedance in order to attenuate by half (-6dB). Since the input impedance vary from amp to amp, a single inline resistor is really not a good plan if you want predictable amount of attenuation. Not to mention that such a high value resistor will likely introduce noise because a form of resistor noise is proportional to its resistance.

What's a better plan is to use two resistors per channel to form a voltage divider. For example, a 10K resistor inline with the signal, followed by another 10K resistor shunted to ground. This will also give -6dB attenuation and doesn't change too much unless the amp's input impedance is low enough to be in the same ballpark value as the resistors. Obviously, this wil be harder to do in a cable, but if you have fat RCA plugs you could probably squeeze tiny miniature resistors in there.
 
Dec 12, 2007 at 2:32 PM Post #5 of 12
A better solution might be to reduce the gain of the output stage of the DAC... sounds like it it putting out a really hot output. If you aren't currently using a DAC but coming out of the soundcard into the amp, once a DAC is put into the equation, it may no longer be an issue?
 
Dec 13, 2007 at 2:07 AM Post #6 of 12
Quote:

Originally Posted by Pars /img/forum/go_quote.gif
A better solution might be to reduce the gain of the output stage of the DAC... sounds like it it putting out a really hot output. If you aren't currently using a DAC but coming out of the soundcard into the amp, once a DAC is put into the equation, it may no longer be an issue?


Possible. It's been a problem with every PC sound card I've used (I've tried several older ones now that I switched to Linux and have a quiet mobo+video card), so I figured I'd get prepared for it to be a problem again.

amb: a divider is what I was thinking--just a basic L-pad per channel. But, I see many different resistance values used when I search here, diyaudio, just Googling it, etc., but no real explanation of why a given impedance is used.

I can see going into hundreds to thousands of Ohms to keep from drawing too much current on the attenuator, but beyond that, am not sure why person X may be using 500 and 1000 Ohms for -6dB, and person Y using 2200 and 4700--what does 4x Ohms do that's good or bad?

Aside: compared to a new pound of solder, Star Quad and Switchcraft connectors are downright cheap...
 
Dec 13, 2007 at 3:37 AM Post #7 of 12
The appropriate voltage divider resistor values depend on what type of load the preceding stage was designed to drive. After a DAC (or other line level outputs), the load should be high impedance, usually 10K ohms or higher. On the other hand, a headphone amp output could usually handle loads down to tens of ohms or lower, and a speaker amp output could handle (and in some cases, require) ~8 ohm load.
 
Dec 13, 2007 at 4:33 AM Post #8 of 12
OK, the tens of thousands of Ohms are dealing with replacing potentiometer-based volume control on the signal, and the others are probably going to some kind of load. Got it.

Would it be correct to say the 10k+ Ohms is a ballpark figure? That what is important is to be high R so that the opamp(s) don't start distorting from being too loaded here and there?
 
Dec 13, 2007 at 6:20 AM Post #9 of 12
You could go into the depth to measure the output impedance of your source, and input impedance of your amp, and create a divider network to achieve the desired attenuation while maintaining efficiency for both parts, but that's 2nd to 3rd year engineering work.

I would personally use a potentiometer to create a divider to find what ratio sounds good. 10k should be an OK starting point in general for most things I've encountered.
 
Dec 13, 2007 at 5:10 PM Post #10 of 12
For attenuation, a 10k pot is a good standard for most situations. A little better all-around compromise (nicer to the DAC) would be 20 or 25K, but then you really have to make sure you use good short cables (3' to no ill effect). 50K would push it for any passive.

With a 25k pot as a passive attenuator, if you hear a small bit of noise/buzz/radio station at half way with no signal (inputs shunted), most likely you're better off using a 10K.
 
Dec 14, 2007 at 12:20 AM Post #11 of 12
I think that gets me set up enough to experiment.

I don't think I'll go as far as perfectly matching it, especially if the input impedance would be important. If I start thinking like that, next thing you know, I'll be buying $12 resistors!
biggrin.gif
(on that note, I've pretty much settled on Panasonic EROS2P)

Cable length is likely to be 1.5', but less if I can manage it (it will basically depend on exactly how flexible the mini Star Quad is, and if I can do any easy cable management).
 
Dec 20, 2007 at 6:34 PM Post #12 of 12
Could someone possibly explain how to make an in-line attenuator? I've looked around on the internet but all i can find is stepped attenuator tutorials.

Id like to make one so it matches all my other cables (sad and anal i know but ho hum).
 

Users who are viewing this thread

Back
Top