summing large amounts of channels.

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
madswitcher said:
Hi Pucho,

If you can, get hold of a circuit diagram of an EMI TG console, or even the Users Manual.  The designer) used a transistor as a voltage to current converter for the channel mix output and a simple virtual earth amp which simply summed the currents.

Simple, no mix resistors and did the job effectively.

Regards

Mike
I would be interested in seeing that too...

In theory a single transistor (common base topology) with audio signal feeding through a resistor into the emitter could output a high compliance AC current from the collector, albeit +/- some class A current. This class A current (just like inside class A VCAs) must be large enough to absorb max signal swings and limits the noise floor performance. In a patent search decades ago I found a very old circuit the summed current outputs from vacuum tubes. 

The patent office denied my patent application back in the late 70s because the wet behind the ears patent examiner didn't understand the difference between summing "currents", and summing "current sources". He cited a text book description of a VE sum amp that literally says it is summing currents (argh). At the time I didn't have enough cash sitting around to educate the examiner while paying a lawyer's hourly rates. I briefly considered driving down to VA to educate the examiner in person, but I had better things to do with my cash and time back then.

For the record I never figured out how to make a discrete low noise, low distortion, bilateral current source. I did figure out I could make an active op amp synthesized bilateral current source with 10x the source impedance of simple resistive summer, delivering 20 dB less noise gain. 20 dB less noise, less distortion, and less phase shift was a win in my book.
=======

Apparently I am a bad communicator, lack technical authority, or both...  :-[  Back in 1980 when I shared my current source summing invention with the world through my article, many at the time did not believe me. I admit it is kind of fantastic to suggest a console sum amp structure with no N+1 multiplication of sum amp ein.

As I recall exactly one fellow consoles designer (Steve Dove... sorry to name drop) shared with me when I ran into him at a trade show later that year, that not only did he get it, he copied me and used it in a small broadcast console design (high noise gain in bipolar VE bus amps can suffer from rectification when exposed to high RF environments.) 

JR
 
madswitcher said:
Hi Pucho,

If you can, get hold of a circuit diagram of an EMI TG console, or even the Users Manual.  The designer) used a transistor as a voltage to current converter for the channel mix output and a simple virtual earth amp which simply summed the currents.

Simple, no mix resistors and did the job effectively.

Regards

Mike
The user manual describes the circuits quite well, but does not go into details, which, as you know, is where the devil is.
Can you post a schemo in the Documents folder?
We'd like to see how the designer ironed out the DC and headroom issues.
 
Unfortunately I cannot post the circuit as experience has shown that the copyright police are still quite active.  This is why I suggested/asked if someone else may have it in the public domain.

However, one can imagine such a common emitter amplifier acting as a current source with a 27k up to +Vcc and a 6k8 (down to -Vcc that gives an output current of proportional to the voltage input.

Before anyone jumps down my throat about such a circuit, it is perhaps good to consider the heritage of such a console and the quality of the music produced by it.  Even Mr. Neve admired the work of its designer - Mike Bachelor IIRC.

Mike
 
madswitcher said:
Unfortunately I cannot post the circuit as experience has shown that the copyright police are still quite active.  This is why I suggested/asked if someone else may have it in the public domain.

However, one can imagine such a common emitter amplifier acting as a current source with a 27k up to +Vcc and a 6k8 (down to -Vcc that gives an output current of proportional to the voltage input.

Before anyone jumps down my throat about such a circuit, it is perhaps good to consider the heritage of such a console and the quality of the music produced by it.  Even Mr. Neve admired the work of its designer - Mike Bachelor IIRC.

Mike
Copyright is only on the particular drawing. Feel free to redraw yourself, no harm in that.
 
madswitcher said:
its designer - Mike Bachelor IIRC.

Correct, Mike Batchelor.    Responsible for a great body of design work.

I've already lost a few friends by posting my fair share of EMI schems on this site through the years, it's certainly not worth any more personal grief for the sake of swinging dicks with Mr. Roberts in this thread so, up to someone else.

There never was a patent or copyright on any of this stuff though.
 
ruffrecords said:
The manual is available on the web here:

https://www.manualslib.com/manual/769744/Emi-Tg-12345-Mk-Ii.html

See section 12.4 3

Cheers

Ian
While I would much prefer an actual schematic they describe a plausible psuedo current source summing but... the transistor stage biased at 0.25% Vsupply  has a resistor load (27k?) so that defines the source impedance, much lower than a true current source.  N channels of that, even capacitor coupled will have source impedance equal to all those 27k s in parallel.

Of course maybe I am imagining it wrong...  ::)

JR
 
JohnRoberts said:
Of course maybe I am imagining it wrong...  ::)

No, you're always king John.
This is but an example of using  voltage to current amplifiers fed into a virtual earth amplfier for audio mixing.  This was a 1966/7 design.  Not bad for the time IMHO. 
 
Winston O'Boogie said:
No, you're always king John.
This is but an example of using  voltage to current amplifiers fed into a virtual earth amplfier for audio mixing.  This was a 1966/7 design.  Not bad for the time IMHO.
I am not king, if I was I would demand to see a schematic.  8)

JR
 
JohnRoberts said:
While I would much prefer an actual schematic they describe a plausible psuedo current source summing but... the transistor stage biased at 0.25% Vsupply  has a resistor load (27k?) so that defines the source impedance, much lower than a true current source.  N channels of that, even capacitor coupled will have source impedance equal to all those 27k s in parallel.
That's my understanding too. It's supposed to load the bus at 27k/N. So a 24 channel would have a noise gain of a tad under 6dB. Certainly much better than the 28dB of a standard configuration.
One thing I can't figure out is where this 6.8k is connected and typically what the voltages are and how the signal is connected.
Actually, the manual says the max level is 0.4mArms, which in 6.8k corresponds to 2.7Vrms or +11dBu. This is to be put in perspective of the +/-20V supply rails, that would support an operating level of +22dBu.
Now it is probably not an issue since there is so little noise to amplify...?
What would be more of a concern is the linearity/distortion performance, both from the voltage-to-current converter and the VE summing stage with only one transistor providing gain.
 
JohnRoberts said:
I am not king, if I was I would demand to see a schematic JR

I have a copy.  Other folks who have worked on these desks will, no doubt, have copies.  The EMI archives at Hayes have copies. 
The most authoritative person on the desks these days would probably be Mr. Brian Gibson who's been maintaining them since he helped install the first one at Abbey Road in the late 1960's. 

But some people get tetchy  when EMI schematics are posted publicly, I've been there so I choose to sit this one out.

It isn't rocket science to design a high impedance current source though, given the knowledge we've gained in solid-state design since 1966,  so interested folks could take the ball and run with it.


 
abbey road d enfer said:
  .
Actually, the manual says the max level is 0.4mArms, which in 6.8k corresponds to 2.7Vrms or +11dBu. 

Just as an FYI, the reference level at EMI was lower than 1.23v.  Of course, John R. uses the rather quaint dBv   reference in his 1980 paper  ;)
I know you're more than capable of putting together a better voltage to current amplifier Abbey and I think you're correct in that a one transistor solution wouldn't be the best in terms of linearity etc.

P.S.  just reminded by a friend that the designer was Mike Batchelor with a 't.

This is what Brian Gibson relayed to another friend a few years ago:



"Last night I went along to an evening with Rupert Neve, organised by The APRS (Association of Professional Recording Studios) and hosted by Dave Harries.

Rupert gave an interesting talk on his life, with a question and answer session. At one point in the proceedings he was talking about other consoles and designs that were around, and EMI desks were mentioned (not by me I hasten to add) and Rupert came out with the comment "I was in awe of Mike Batchelor" - which I considered to be praise indeed.

I spoke to Rupert afterwards and asked him how well he knew Mike. He said he'd met him a few times at design seminars and had been very impressed with Mike's knowledge and design approach.

Rupert Neve says "I was in awe of Mike Batchelor" must go down as a quote to remember."
 



 
Winston O'Boogie said:
I know you're more than capable of putting together a better voltage to current amplifier Abbey and I think you're correct in that a one transistor solution wouldn't be the best in terms of linearity etc.
I toyed with simulation, trying to imagine what kind of THD performance to expect. It's not too difficult to cook a two-transistor summing amp with good performance, about 0.05%, and the one-transistor current source bottoms out at 0.1% for the specified 0.4mA rms. Not too bad.
 
Ha!  No, not too shabby at all.  Good job.
Imagine what could be done if only transistors didn't cost so much per, or if there was only a way to fabricate a bunch of transistors on one piece of silicon!  ;) 

If we were doing it with IC's and we have a nice transamplifier doing the mixing, what would you recommend for the current sources, something along the lines of the attached picture?


Edit: the pic comes from "Impedance & Admittance Transformations using Operational Amplifiers"  published 1964, starts on page one and two then continues on page seven:    http://www.philbrickarchive.org/1964-1_v12_no1_the_lightning_empiricist.htm
 

Attachments

  •  Voltage to Current.png
    Voltage to Current.png
    52.7 KB · Views: 59
Winston O'Boogie said:
If we were doing it with IC's and we have a nice transamplifier doing the mixing, what would you recommend for the current sources, something along the lines of the attached picture?
From the beginning, it was my idea, using a modified Howland pump.
The circuit in your picture is too unstable. The Howland pump adds a current-sensing resistor between the positive and negative feedback paths. For good output Z, this resistor must be compensated for in the negative path. The modified Howland pump allows using identical resistors by adding a voltage-follower in the PFB path.
In order to achieve the targetted goal of making the output Z at least 10x the transconductance, the resistors must be matched better than 0.1%, so that conjures laser-trimmed monolithic to me.
There are probably other possibilities but I would think they have more or less the same sensitivity to component tolerance.
 
abbey road d enfer said:
From the beginning, it was my idea, using a modified Howland pump.
The circuit in your picture is too unstable. The Howland pump adds a current-sensing resistor between the positive and negative feedback paths. For good output Z, this resistor must be compensated for in the negative path. The modified Howland pump allows using identical resistors by adding a voltage-follower in the PFB path.
In order to achieve the targetted goal of making the output Z at least 10x the transconductance, the resistors must be matched better than 0.1%, so that conjures laser-trimmed monolithic to me.
There are probably other possibilities but I would think they have more or less the same sensitivity to component tolerance.


OK got it, thanks.  I'll look up the modified Howland, it's been a while,  but I see what you mean now. 
Too bad I don't have a big desk to modify, I'd give it a go.  Matching resistors to .1% wouldn't bother me  :)



 
abbey road d enfer said:
From the beginning, it was my idea, using a modified Howland pump.
The circuit in your picture is too unstable. The Howland pump adds a current-sensing resistor between the positive and negative feedback paths. For good output Z, this resistor must be compensated for in the negative path. The modified Howland pump allows using identical resistors by adding a voltage-follower in the PFB path.
In order to achieve the targetted goal of making the output Z at least 10x the transconductance, the resistors must be matched better than 0.1%, so that conjures laser-trimmed monolithic to me.
There are probably other possibilities but I would think they have more or less the same sensitivity to component tolerance.
I'm sure I have shared this before but back last century I used the improved howland current pump, that I first saw in a national applications manual back in the 70s but with impractical values for audio use (they used 1M resistors trying to get highest possible output impedance). For audio using more sensible resistor values (tens of K) deliver reasonable audio path noise floors while still delivering elevated output impedance to reduce noise gain. I confirmed that this practical synthesized current source delivered a 10x reduction in noise gain feeding a VE bus. The noise from these current sources sums incoherently.

---
For today's TMI, early this century I discussed with a senior engineer at THAT corp the possibility of using a custom metallization layer*** on one of their popular balanced interface ICs to repurpose it as an improved Howland current pump using their already existing precision resistors on that chip. THAT corp was unwilling to make an IC without their comprehensive (static/over-voltage) protection circuitry on "every" I/O pin. That IC silicon substrate only had 4 I/O protection wells, and my application required 5 pins of I/O. A custom metallization layer was almost practical (tens of $k). Full custom silicon was just not realistic for my hypothetical application.

JR

*** ICs are fabricated with subcomponents that get connected together by the final metallization layer, kind of like how traces on a PCB connect discrete components together...  ICs are like an upside down PCB but without the board. 



 

Attachments

  • imp_howland.pdf
    80.7 KB · Views: 49
Here is the actual synthesized current source from a 30 year old schematic I used in the AMR production series consoles.

Since there were 72 of these current source blocks in the 24 bus console (3 in each monitor section strip) I had to cut a few corners. (The market for big split consoles was already evaporating before my eyes thanks to Mackie's 8 bus success).

For a cost is not object design I would substitute lower noise modern JFET input op amps, tighter tolerance than 1% resistors, and more significantly buffer the inputs to the improved Howland. The varying source impedance from the pan pot position imbalanced and degraded the current sources output impedance. The fully loaded bus noise floor was lowest with all pan pots hard panned left or hard right, of course any music signal requiring other than a hard pan would mask the bus noise floor.   

I could make a better analog sum but so what? The market for this is microscopic IMO.  But here is the road map for any who want to go down that road.

JR
 

Attachments

  • amr_Isource.pdf
    451.2 KB · Views: 109
Back
Top