Buffering, Opamps and Panpots

[franklinh]

So I've been piecing together a mixer (not passive), both on paper and in real life, and a stumbling block for me is buffering, specifically at the junction between the fader and the panpot. I understand that we need something there to prevent the panpot reacting with the fader, like a simple buffer - from there I get a little off track.

I've been told you can use something like a 5534 or any other opamp "as a buffer" in this situation. I've seen schematics of simpler ideas using discrete components like the emitter follower and others using a pair of transistors. Most of these don't include the use of an IC or DOA.

My design is using gliss faders (600 ohm printed on outside) and dual 10k linear panpots, which then go onto electrodyne ACN's (Active Combining Networks with summing resistors inclusive, don't know the value of them)

So the question is, does a (unity?) buffer like the discrete idea mentioned above need gain behind it to drive those 10k panpots? Do so many designs use an opamp set for gain here not only for more swing on the fader but to "push" through the insertion loss to the summing buss? Would there be one there at unity for the same reason? Am I thinking too hard?

I realize that if I understood impedance better I wouldn't be asking - my searching didn't net the exact answer. My concerns are partially finincial and partially trying to use as few components as possible. I might not want the extra gain at the fader if it means less components in the signal path.

As always, thanks!

Frank

 

[PRR]

> faders (600 ohm printed on outside) and dual 10k linear panpots

You should be able to drive 10K pan-nets directly from 600Ω faders no-problem.

 

[franklinh]

Thanks PRR! For this particular project that's the best news I've heard all day. Then there's that old saying "teach a man to fish..."

let's asume for the sake of argument that the respective values of the fader and panpot are such that a buffer IS necessary (would that have to be closer in value?). Could someone school me on why I would want an opamp or a discrete buffer, etc (as mentioned above)??

 

[NewYorkDave]

You really need to know the input impedance of the ACN in order to choose the component values in your panpot circuit. If you just take a guess and turn out to be wrong, the taper of the panpot will be all out of whack.

The easiest way to measure the input impedance is with the series resistance method. Connect an AC millivoltmeter across the input. Connect a sinewave to the input in series with a high-value pot wired as a rheostat. (The value of this pot should be quite a bit higher than the expected input impedance of the device; in this case, 1M should do). Measure the signal amplitude with the pot at zero. Then slowly turn up the pot until the signal "on the other side" drops by one-half (6dB). Remove the pot and measure its resistance; this number is equivalent to the input impedance of the device.

Speaking of panpots, here's one I designed that has a better taper and more constant input impedance than the usual panpot that's made from a dual-linear pot. The penalty is that it has 6dB minimum insertion loss, but usually that can be made up elsewhere.

http://electronicdave.myhosting.net/miscimages/panpot.gif

In your case, "RL" is simply the input impedance of your ACN.

 

[franklinh]

Thanks NYD!
Hmm.. I'm unfarmiliar with what wiring a pot like a rheostat is - input to wiper? I'll definitely do this measurement as it would be excellent to know the input impedance of those ACNs. Function generator kit should be arriving at my house soon.

I'm still unsure about the need for an opamp/opamp with buffer/just a buffer between the panpot and fader question (see first/third post). Can anyone set me straight on that one?

As always, thanks for the help.

Frank

 

[NewYorkDave]

I had a vague recollection that the Audio Cyclopedia had a blurb about the ACN, and I was right. Here it is:

728kB GIF

Input impedance is quoted as 10K. So this is looking pretty convenient for you since you want to use 10K dual linears for the panpots. Matter of fact, if you wanted to use the panpot I posted earlier, you'd use the "example" values given on the schematic. You don't need "RL" because that's already part of the ACN.

The input impedance of "my" panpot is about 50% of the value of the dual pot, so it's around 5K. That's a high enough impedance that you really wouldn't need to put a buffer after the 600-ohm fader--it's close enough to the "at least 10:1" rule. But if you plan to reduce the load impedance beyond that by adding aux send pots, etc. then you should consider a buffer after the fader--or you can do it old-school and use a resistive splitting band, which of course entails a signal loss penalty.

 

[franklinh]

That's even more good news. Thank you for the tidbit about the electrodyne ACN - not easily found!! I'll give that pan circuit a good hard look too.

The adding of the aux/echo send issue points me back to the opamp vs just discrete buffer circuit question though, sort of. Let's say it's a post fader aux send, whatever the first element of that is now in parallel with the panpot(s), which lowers the total impedance that the fader will see - ruining the 1:10 impedance thing - which is why we would use a buffer. Right? So why (except extra gain, cost or ease of construction) would a designer opt to use an entire opamp there as a buffer instead of a few discrete components arranged as a buffer? Is it a coin flip? can a buffer have gain?

I did some reading in the trusty Yam4ha sound reinforcement handbook about impedance, and it's starting to make sense. Thanks for hanging with this thread!

Frank

 

[pstamler]

Does anyone use the Orban panpot circuit anymore? It only needs a single, linear pot:

Assuming active summing, you lose about 10.6dB of level with the pan turned all the way in one direction.

EDIT: I tried some ASCII art to show the diagram, but it didn't work. I'll get a diagram up when I wake up.

Peace,
Paul

 

[NewYorkDave]

Sure, the Orban panpot is still alive and well. Here's one of many examples:

http://www.wavefront.mcmail.com/orban.gif

If you design it according to the formulae in Orban's AES paper, its adherence to constant-power law is very good and it has a fairly constant input impedance. But compared to dual-pot types, it has higher minimum insertion loss and (depending on the circuit values) can have less complete attenuation at "zero." The maximum attenuation improves when the series resistances are higher but then the minimum insertion loss increases as well.

Here's the info on Orban's paper, which is good reading and is available as a download from AES.

Notes on Panpots
JAES Volume 19 Number 11 pp. 935-939; December 1971
Author: Orban, Robert

When I was doing initial design on my mixer (which I still haven't actually built :roll I considered the Orban, the "British" type dual-linear-wth-slugging-resistors, and the ganged log/reverse-log with tailored loading (as seen in several models of Soundcraft, Mackie, etc.).

The Orban's insertion loss was too high for my application, the "British" type had an input impedance that was low and varied over a wide range and the last type required a special part, which is something I wanted to avoid. A ganged stepped bridged-T or ladder type as found in pre-1970 consoles was not feasible due to the high cost of switches with enough poles and positions. So I ended up cooking up my own variation on the dual-lin pot type that met my own needs very well with a fairly low minimum insertion loss.

Here's a chart showing the attenuation in each channel of my panpot per 5 % of rotation:
10kB GIF

The adherence to constant power law is pretty good, within 0.1dB throughout the range. Of course, that's with an ideal dual linear pot. Performance with real-world pots will depend on the tolerance and tracking. But for what it's worth, I measured it with a couple of cheap, off-the-shelf Alpha dual pots and it was still very close to the performance shown in the chart.

Frank: to answer your other question, since a channel fader is usually run at around 10dB of loss, the purpose of the fader buffer is normally to give about 10dB of gain in addition to buffering the fader from the panpots, aux sends, etc. And no, it doesn't have to be an IC; it can be any amp with a high enough input impedance, that can deliver a clean 10dB of noninverting gain with the capability to drive the following circuitry with adequate headroom. It's just that many designers find ICs convenient, and that's why you usually see them used in this duty.

Now I have a question for you: Is Tacoland on Broadway (near the hospital) still open? I used to eat there all the time when I worked nearby.

 

[franklinh]

NYD:

tacoland - hmm... not my part of town - I'll check when I get home.

Thanks for the answer on the fader buffer - so as long as i'm not running at 10db loss on my fader, the gain could be omitted and something like an emitter follower could be employed. purple audio has an interesting solution.

Thanks everyone! I will endeavor to post how things turn out to resolve the thread for posterity.

Frank