Noob question about stepped attenuators

[ohshitgorillas]

I've been using a cheap DACT-type 21-step 100K attenuator I got from Hong Kong, and while the improvement in clarity over the stock pot is great, I feel like 21 steps is just not enough as I'm often forced to choose between too loud and too quiet. In the spirit of other Bottlehead amps that use dual attenuators, I'm wondering if I could just stick another 10K 21-step attenuator after the 100K for 441 steps of mega overkill super balanced volume control.

This is probably a dumb question and I feel like the answer is an obvious "yes that will be fine", but I have no formal electronics background and a lot of this stuff is a bit mystical to me... so I wanted to check before I pay for the thing and start drilling holes. Would 10K be enough or too much? A friend of mine has a vintage Alps 40-step 5Kohm attenuator that he told me I could have but I can measure the channel imbalance with my multimeter--whereas the Hong Kong DACT-type ones always measure dead-on.

Any input is appreciated.

[Paul Birkeland]

It may be a little easier than that.  Are you using all 21 steps, or just the first couple of steps?

[ohshitgorillas]

Depending on the music, compression, volume, phase of the moon, I use probably steps 5-12

[Paul Birkeland]

See FAQ #3 here:
http://bottlehead.com/smf/index.php?topic=4295.0

While you don't have the over-sensitivity problem, if you pad down the control a bit, you can spread out the 5-12 useful steps to something more like 2-18.

Since we don't have a ton of information about the attenuator itself, start with the 33K and 75K resistors on each channel, then let us know how it works out for you. 

-PB

[ohshitgorillas]

Thanks for linking me to that page. I was under the impression that the total impedance must be 100Kohm. That does seem much easier, although coarse/fine volume control is still tempting, less than $20 for the second pot. I need to mull this over a bit more...

http://www.ebay.com/itm/DACT-Type-21-Stepped-Attenuator-Potentiometer-100K-2A3-/201239685917

^This is the one I'm using, for reference. It's actually pretty great besides having too few steps. Small, easy to work with, sounds clearer than the stock pot. I needed to put in a new hole for the tab that holds it in, but otherwise it's a drop in.

[Paul Birkeland]

The 100K value isn't critical.  50K to 250K is an acceptable range.  Adding the 33K and 75K resistors keeps the impedance quite close to 100K.

-PB

[ohshitgorillas]

Sorry for bumping such an old thread. I'm planning to add a 3-way gain switch that will either add 0, 33k, or 50k ohm resistance. I just want to make sure I've got this right: What I want is a 4PDT switch, right? Is it best to split the signals into 3 and add resistors before the switch, after, or does it matter?

edit: holy shit these switches are expensive. what the hell? am I doing this right?

[Paul Birkeland]

You want either a 4P3T rotary switch, or a 2P3T toggle (not recommended).

I would suggest changing to a 50K pot, then adding 0, 50K, and 150K of resistance.  That will give you 0dB, -6dB, and -12dB.

[ohshitgorillas]

Nice, I would also have the finer control of 50k steps vs the 100k. Thanks very much! These switches look pretty complicated but I'm sure I'll figure it out.

edit: do I bridge the connectors on the switch with the resistors (or jumper)? Other than that I think I've got it figured out

[ohshitgorillas]

Quick question: I'm not sure I entirely understand how the 4P3T switch would be wired in. I understand the 3T part where I need 3 outputs for different gain settings, but wouldn't I only need two poles for the two channels?

[Paul Birkeland]

Quick question: I'm not sure I entirely understand how the 4P3T switch would be wired in. I understand the 3T part where I need 3 outputs for different gain settings, but wouldn't I only need two poles for the two channels?

Yeah, it looks like I didn't quite give you enough information on what I had in mind.  From a noise and loading perspective, it's nice to keep the input impedance relatively constant.  If you just use a single resistor per channel between the RCA jack center pin and the input of the volume pot, the input impedance will either be way too low on some settings, or way too high on some settings.  This is a bit part of the reason why the Crack FAQ mod has both the 75K resistors and the 33K resistors, the combination of both values dials the input impedance back to around 100K. 

What you'll want to do with the 3 different gain selections is to switch both the series and the parallel resistor.  In order to accomplish this, you'll need to either switch out the series resistor, or be able to add to it as you move up in switch positions, which is why a rotary switch works so nicely.  If you have a pot value in mind, and three values of attenuation, I can assist you a little with some calculations and sketches of how to make this work.  (I recommend -6dB, -12dB, and -24dB as your pre-attenuation values).

-PB

[ohshitgorillas]

Thanks for the help! Sketches would be great, since I don't really understand loading and am a little confused already, lol.

I was going to use a 50K stepped pot and 0, 50k, and 150k ohm to get 0, -6, and -12dB as discussed above although I have the feeling I might never use 0dB, so maybe -6, -12, -24dB would be better. I have both 50k and 100k stepped pots, but would probably prefer to use the 50k as it has smaller steps.

Edit: After thinking about it, I think I'd prefer -6, -12, and -24dB with a 50k pot

[Paul Birkeland]

OK, here is what you'd need for a 50K attenuator to get -6, -12, and -24dB of attenuation while keeping about 100K constant input impedance.  It looks like one 4P3T and ten resistors will do the job.  I put the switch letters and numbers on for the rotary switch that we use at Bottlehead.

-PB

[ohshitgorillas]

Unfortunately it seems that neither digikey or mouser actually stocks 44kohm resistors.... would it be too awkward to just use two 22k in series? Or would it be okay to use something like a 44.2k or a 43.5k?

[Paul Joppa]

Modern resistors (usually metal film) are inexpensive and readily available in +/-1%, but in the "golden age" resistors were typically 10%, sometimes 5% or 20%. The 44.2K and 43.5K values are close enough!