OTB Mixer Using Current Summation

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
Post Reply
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

OTB Mixer Using Current Summation

Post by mediatechnology »

This thread has been renamed since it picks up JR's original thread on an OTB mixer using VCAs and current summation. I happen to be the guilty party who mentioned that Pease's article contained info which could potentially solve a VCA bus capacitance issue. The first couple of posts discuss Pease and then we get to the real project.

Wayne


The December 1, 2008 Electronic Design has an interesting article by Pease:

http://electronicdesign.com/Articles/Pr ... leID=20110

He says two interesting things:
However, I remembered a good circuit I cooked up 40 years ago in Fort Wayne, Ind. A customer was using a photoresistor to vary an amplifier’s gain, as in Figure 2. But when the Rf went up to 5M, the amplifier became much too slow, due to the 5-pF capacitance inherent in the photoresistor. What to do?

At first I was going to use an extra op amp to make a negative capacitor to cancel out the Cf. But then I figured out it might work well if I just connected an adjustable C' back to the positive input. We tried it and it worked fine! It cancelled out the Cf under all conditions and extended the BW by 10×.
Question for JR: Would the above trick (figure 2) be a means to offset the FB capacitor size in a VCA current summed mixer where the I-V converter was an op amp? See: viewtopic.php?f=6&t=183&start=12

(For our readers: Ignore the fact that Rfb is a photoresistor in Pease figure 2.) If the combined output C of 16 VCAs was 15 pF x 16 (240 pF) and the required Rfb was 20K, then Cfb would have to be >240 pF. That would put the corner frequency around 33 kHz. Could this trick be used to speed up the IV stage?

And then there's this "gem":
I’ve heard arguments that every audio amplifier should have a 7:1 or 10:1 safety factor between its actual slew rate and the biggest, fastest signal it will have to handle. I used this circuit to show that a 3:1 margin would probably not cause 0.01% distortion in that signal. I was all set to give the demo, when I discovered that nobody wanted to listen to the demo. They had made up their minds and didn’t want to listen!

Anyhow, a factor of 1.5:1 or 2:1 is probably not safe, but 3 should be plenty. Don’t waste a lot of money on ultra-fast amplifiers to get a safety factor of 7 or 10. Also, don’t worry too much about slew-rate symmetry. If an amplifier is fast enough in one direction, and faster in the other direction, that’s not a big deal. I mean, who the heck spends a lot of time LISTENING TO SQUARE WAVES?! (Don’t answer that question...) (If you wanted to be sure to get matched positive and negative slew rates, you could throw out P2 and use an op amp to invert the voltage at the wiper of P1 to put the same magnitude of voltage at the foot of R2.)
I remember once JR where you characterised slew rate limiting as "clipping in the time domain." That simple explanation stuck with me. I'm curious to see what kind of mail Bob gets where he sort of says the same thing.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: Pease - What's All This Adjustable Slew Rate Stuff Anyhow?

Post by JR. »

A capacitor across the feedback R can be compensated nicely by a similar C to the positive input (assuming Rs at both inputs), a C from the - input to ground is not the same animal. In fact a C to ground can generally be compensated for with a C across the feedback R. These Cs form a parallel feedback network with the resistive feedback. For a unity gain stable opamp that feedback C can equal or exceed the input C to ground. For decompensated opamps that aren't unity gain stable the feedback cap can't be more than 1/3 or whatever of the input C to meet stability constraints. (I recall using low pF feedback caps with decompensated opamps to get better stability than without, despite fact that too much C there would cause oscillation).

Yes slew limiting is like clipping, and just like clipping, headroom or safety margin depends on the mechanism or immediate slew limiting factor. Similar to how a solid state circuit can be clean in the amplitude domain until just before clipping, while some tube circuits will start distorting several dB before amplitude clipping, slew limiting is not necessarily a sharp threshold event but can be a gradual degradation to rate of change in response to input. So appropriate headroom depends on the specific circuitry (mostly LTP and compensation approach). I have pondered coming up with a new data sheet spec that plots rate of change vs input error voltage. The linearity of this relationship in the audio range of rate of change could be informative, but I fear would be misunderstood like all the simpler specs on data sheets.

Bob doesn't explore the use of rise time limiting in the front end of audio circuits to moot slew limiting. A properly established 1 pole LPF early in the audio path can guarantee never encountering slew limiting, for valid signals that don't exceed amplitude clipping.

-------

Back to my VCA based current source summing bus structure, the brute force solution is to add feedback C to swamp the input C, and while perhaps workable for a modest number of inputs, in a larger bus structure, the feedback C could get uncomfortably large. I still prefer my trick of incorporating the bus C into one of two poles forming a compound LPF.

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Pease - What's All This Adjustable Slew Rate Stuff Anyhow?

Post by mediatechnology »

Another compensation question for you John:

Some time ago I saw a schematic for a distibution amp and in it they had used a 5534 as a polarity inverter with a signal gain of -1. It was an internal stage without external loading (cable). IIRC the Rin and Rfb were 10K. What was different about it was they had tied a 4K7 from the inverting input to ground making the signal gain to ground "-2." This resistor, at first glance, appeared to have no purpose. Also absent was any compensation...

The parallel combo of Rin was ~ 3K2. Adding this extra resistor makes the noise gain about 4. Thinking about it I decided that they did this to raise the noise gain of the stage to >3. Doing this eliminates the need for overcompensation of the 5534 and it would appear to maintin slew rate. I had never seen this done before.

Have you ever seen it done with a 5534? I don't think it's particularly novel - just different.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: Pease - What's All This Adjustable Slew Rate Stuff Anyhow?

Post by JR. »

mediatechnology wrote:Another compensation question for you John:

Some time ago I saw a schematic for a distibution amp and in it they had used a 5534 as a polarity inverter with a signal gain of -1. It was an internal stage without external loading (cable). IIRC the Rin and Rfb were 10K. What was different about it was they had tied a 4K7 from the inverting input to ground making the signal gain to ground "-2." This resistor, at first glance, appeared to have no purpose. Also absent was any compensation...

The parallel combo of Rin was ~ 3K2. Adding this extra resistor makes the noise gain about 4. Thinking about it I decided that they did this to raise the noise gain of the stage to >3. Doing this eliminates the need for overcompensation of the 5534 and it would appear to maintin slew rate. I had never seen this done before.

Have you ever seen it done with a 5534? I don't think it's particularly novel - just different.
That's a pretty common technique for non-unity gain stable opamps, and a refinement of that is to couple that R to ground through a small capacitor so it doesn't increase the noise at audio frequency or DC offset, while still attenuating the feedback at very high frequency to provide stability.

A similar consideration in consoles with highly de-compensated discrete bus summing amps, is to back ground the bus feeds so the noise gain remains adequate for stability despite assignment status. Again this back grounding is only needed at very high frequency so can be cap coupled.

Another obscure trick I've seen with 5534 (IIRC) is to use a very very small feedback cap to cancel pin/package capacitance to ground, despite the prohibition against unity gain feedback connections. The cap ratio needs to still meet 3x or so (1/3 C to ground) criteria for stability.

But as you know I have a better plan for high performance summing. :D

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Pease - What's All This Adjustable Slew Rate Stuff Anyhow?

Post by mediatechnology »

Thanks John for the post and info.
But as you know I have a better plan for high performance summing.
Yes I do. I'm looking forward to that. We all are.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: Pease - What's All This Adjustable Slew Rate Stuff Anyhow?

Post by JR. »

I am reluctant to undertake a significant analog design project when in theory digital signals can be combined with no degradation, in fact after combining digital signals you get too much good information and some must be discarded.

The analog console company I worked with on the meter project have a VCA console so maybe I should look at modifying that? In my spare time.

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Pease - What's All This Adjustable Slew Rate Stuff Anyhow?

Post by mediatechnology »

There was a PAR article about DSP vs. Analog summation about a year ago. The guy that wrote it - and I'm paraphrasing a 4 page article - basically said that DSP-based mixes had a very narrow "fader" sweet spot. His opinion was, after doing the same mixdown both ways, was that analog solutions just sounded better. I'll see if PAR's site has a way for me to link to it. First I need to find it....
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: Pease - What's All This Adjustable Slew Rate Stuff Anyhow?

Post by JR. »

Thanks, I'd be very interested in finding some real there-there.. With all the smoke and arm waving I don't doubt there are problems with lesser digital hardware/software.

If there's a real need for such a product, and not just a perceived need, I would gladly rise to the occasion. While the correct response may be to improve the digital hardware. :lol:

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Pease - What's All This Adjustable Slew Rate Stuff Anyhow?

Post by mediatechnology »

I'll dig for it.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: ITB vs OTB

Post by JR. »

OK, I wasted a couple hours researching this and I have come up with a few themes.

#1 There does appear to be some breakdown in summing performance when many digital stems are combined in a lower cost digital environment. This was evidenced by comparisons between a large real time mix, and later combination of several smaller sub mixes.

#2 The comparison is not strictly apples to apples. The ITB vs OTB is made using OTB dynamics, EQ, etc. One pretty rigorous study, parsed out deviations from original in the OTB paths. Ironically these measurements were made with precision digital nulling.

#3 The preference for mixing OTB by big dog recordists, is for the improved ergonomic interface of fader and knobs, vs, mouse and menus.

#4- the DIY craze in simple OTB summing boxes may actually have merit, "If cheaper than" a digital system that doesn't corrupt mixes.

So I'm back where I started. I seriously question a real need for analog summing buses in a properly executed digital environment.

Please somebody tell me where I'm wrong..? Making an above average summing bus is one of the tricks in my kit bag, collecting spider webs. There is probably some demand for a simple OTB kit, but I am not enthusiastic about revisiting the kit business, after walking away from one over two decades ago. Ironically back then I kicked around the idea of a modular mixer kit,,, A bunch of different mixer circuit blocks that could be mixed and matched. Of course the economics today of competing with simple (manufactured) mixers that cost barely more than lunch money, make that kit idea even less viable than it ever was.

JR
Cancel the "cancel culture", do not support mob hatred.
Post Reply