Virtual earth summing with transformer

[ruffrecords]

With the risk of repeating myself, the circuit is undoubtedly VE, although the poor implementation (inadequate DCR of the windings) makes the gain sensitive to bus loading. The low impedance of the summing amp is not due to the shunt resistor, but to current NFB.

I think it is not that simple. I think it is voltage summing with an inverting op amp as gain make up.

Cheers

Ian

 

[Newmarket]

Passive summing is not as much of a compromise as people surmise. The noise gain of a virtual earth summer**** will be similar to the makeup gain of a passive summer for same number of inputs  so six of one half dozen other. ding ding ding... the secret sauce for passive summing is that people use their favorite "colorful" makeup gain stage, so it adds a specific sound signature having little to do with the summing technique.

JR

**** Back in the day i substituted current sources for resistors in a VE summer and reduced that noise gain significantly but Yawn.... digital summing is arbitrarily perfect so why bother.  :

Fair points. And I'll agree about not associating either approach with any particular 'magic sound' 

Now the 'Current Source' approach you mention is very interesting. I have thought along those lines previously but have never got beyond the "handwaving stage". I recall discussing it when I did audio stuff for money but there was never much enthusiasm for it IIRC.
I'm thinking it would take 'proper' development and I don't really have the facilities for that now 

I've tried finding your 1980 article online with no success.
And the link to the Peavey AMR manual on your website is a cul-de-sac.
Any help on those would be greatly appreciated.

 

[madswitcher]

Now the 'Current Source' approach you mention is very interesting. I have thought along those lines previously but have never got beyond the "handwaving stage". I recall discussing it when I did audio stuff for money but there was never much enthusiasm for it IIRC.
I'm thinking it would take 'proper' development and I don't really have the facilities for that now 

I've tried finding your 1980 article online with no success.
And the link to the Peavey AMR manual on your website is a cul-de-sac.
Any help on those would be greatly appreciated.

If you have a look at the post I made above, you should see a pdf of a circuit doing this.  I haven't seen John's article, and in fact would very much like to, but I get the impression that he has been bitten by this subject for a long time.

The circuit I posted is from the EMI TG12345 mixing console Mk2 which is the famous type used by EMI in their Abbey Road (and other) studios.  I will own up and say that I am a fan of these due to their  architecture and sound. Think Kate Bush, Beatles Abbey Road and Dark Side of the Moon, so it has a good heritage.

Happy soldering and listening,

Mike

 

[Newmarket]

If you have a look at the post I made above, you should see a pdf of a circuit doing this.  I haven't seen John's article, and in fact would very much like to, but I get the impression that he has been bitten by this subject for a long time.

The circuit I posted is from the EMI TG12345 mixing console Mk2 which is the famous type used by EMI in their Abbey Road (and other) studios.  I will own up and say that I am a fan of these due to their  architecture and sound. Think Kate Bush, Beatles Abbey Road and Dark Side of the Moon, so it has a good heritage.

Happy soldering and listening,

Mike

Yes. I saw that but noted John's comment too:
"OK not exactly a current source. Operating class A means not very quiet either. "
so it seems he was referring to something a bit different.
I do appreciate that he's been 'bitten' as you say - but I'm not trying to shift the discussion in that direction. Just looking for the info.
Thanks.

 

[abbey road d enfer]

I think it is not that simple. I think it is voltage summing with an inverting op amp as gain make up.

Cheers

Ian

OK, does the opamp sense voltage or current?

 

[JohnRoberts]

Yes. I saw that but noted John's comment too:
"OK not exactly a current source. Operating class A means not very quiet either. "
so it seems he was referring to something a bit different.
I do appreciate that he's been 'bitten' as you say - but I'm not trying to shift the discussion in that direction. Just looking for the info.
Thanks.

I am not "bitten", whatever that means. The current source summing approach worked and I used in a few consoles since the 80s. It mainly makes sense for larger numbers of inputs because current sources cost more than resistors , and I never did like throwing money on the ground for no reason. The last big console had well over 100 stems feeding the L/R bus.

Here is a working link to my old article (I still need to fix the spelling of bus).  http://www.johnhroberts.com/des_art_1.pdf

That design describes current source summing in general but did not literally show how...

I have a copy of the AMR console manual but it is a large file (16M) and I haven't been able to upload it to my website (yet even though my host claims 20M capability  :-[ ).

Here is a clip of one current source (below).

This design is some 25 years old, and I had an improved version finished and working in my lab back then, but never released the improved version to production because we were getting killed in the marketplace by the inexpensive Mackie 8bus that was selling like hot cakes. Mackie ran more ads in one month than we did over the entire life of my console.

The major improvement I made (back then) was was to switch the current sources off the L/R bus when not in use. There was a small but measurable noise floor contribution from 72 of these puppies always connected (switching them off when not in use completely resolved that).

If i wanted to revisit this today I would use a more modern (lower noise) Bifet op amp than TL07x.  I would also buffer the input feeding the current source, since the impedance of the level pot, degrades the output impedance of the current source that depends on the precise matching of all 5 resistors.

I really really really wish there was some need for quality analog summing today (I have the superior technology gathering dust) but this has been eclipsed by digital summing that actually increases quality (resolution) from combining more stems, rather than degrading like analog does from combining more inputs.

I have looked at the passive summing fad and my conclusion is that people like the warmth from their less than linear make up gain stage. There is no real magic or "there" there IMO.

Again I apologize for this veer...  Current source summing was a thing decades ago... now moot IMO.

JR

 

[JohnRoberts]

Hi John,
Dust off that summing design, if you get a chance.   I'm sure someone would like to try it out, sometime.
Adam

You could always buy an old AMR console .....    ;D ;D ;D (but don't expect free service calls from me. )

I do not see much attraction for a summing bus that sounds like nothing at all....  : 

JR 

 

[Newmarket]

I am not "bitten", whatever that means. The current source summing approach worked and I used in a few consoles since the 80s. It mainly makes sense for larger numbers of inputs because current sources cost more than resistors , and I never did like throwing money on the ground for no reason. The last big console had well over 100 stems feeding the L/R bus.

Here is a working link to my old article (I still need to fix the spelling of bus).  http://www.johnhroberts.com/des_art_1.pdf

That design describes current source summing in general but did not literally show how...

I have a copy of the AMR console manual but it is a large file (16M) and I haven't been able to upload it to my website (yet even though my host claims 20M capability  :-[ ).

Here is a clip of one current source (below).

This design is some 25 years old, and I had an improved version finished and working in my lab back then, but never released the improved version to production because we were getting killed in the marketplace by the inexpensive Mackie 8bus that was selling like hot cakes. Mackie ran more ads in one month than we did over the entire life of my console.

The major improvement I made (back then) was was to switch the current sources off the L/R bus when not in use. There was a small but measurable noise floor contribution from 72 of these puppies always connected (switching them off when not in use completely resolved that).

If i wanted to revisit this today I would use a more modern (lower noise) Bifet op amp than TL07x.  I would also buffer the input feeding the current source, since the impedance of the level pot, degrades the output impedance of the current source that depends on the precise matching of all 5 resistors.

I really really really wish there was some need for quality analog summing today (I have the superior technology gathering dust) but this has been eclipsed by digital summing that actually increases quality (resolution) from combining more stems, rather than degrading like analog does from combining more inputs.

I have looked at the passive summing fad and my conclusion is that people like the warmth from their less than linear make up gain stage. There is no real magic or "there" there IMO.

Again I apologize for this veer...  Current source summing was a thing decades ago... now moot IMO.

JR

I think "bitten" was used in the sense of "having had enough of it" - not that it didn't work.
I remember the Peavey AMR stuff being advertised when I worked in pro audio. Never came into contact with it but I am in the UK.  Anyway - I'm sure I'd have heard if it "didn't work" properly 
And, yes, the big push by Mackie - countered to some degree (in UK at least) by Soundcraft Spirit line.

Digital Summing - yeah - swerving swiftly round the detail of "n bits" / floating point / Sample rate - it's fairly difficult to go against it on a technical level when it's well implemented. Crosstalk is really low unless you fix that with a plug-in  ;D

But there are times when I want analogue summing myself - even in my modest setup I don't have enough Digital inputs for auxes etc plus I want the physical faders. But it's not 'big studio' stuff so not a commercial prospect but DIY is a different matter.

Anyway, thanks for the links and info.

 

[JohnRoberts]

I think "bitten" was used in the sense of "having had enough of it" - not that it didn't work.
I remember the Peavey AMR stuff being advertised when I worked in pro audio.

I wish I remembered it being advertised.... :  I recall talking with the company controller complaining about my lack of advertising and he shared with me how many Hartley had given away to artists he was pursuing for endorsements (but not endorsing consoles  ). In at least a few cases these consoles were then gifted to members of the artist's posse... Since those were the guys we (I) wanted to sell to this wasn't helpful, but I stopped asking for marketing support.

Never came into contact with it but I am in the UK.  Anyway - I'm sure I'd have heard if it "didn't work" properly 
And, yes, the big push by Mackie - countered to some degree (in UK at least) by Soundcraft Spirit line.

We sold units into UK (Coals to newcastle). I was in the trenches back then and Mackie went from zero to several million in sales almost overnight, a lot of those sales came out of my product lines, but all the established players in the mixer business felt the impact from Mackie's market share gains..

Digital Summing - yeah - swerving swiftly round the detail of "n bits" / floating point / Sample rate - it's fairly difficult to go against it on a technical level when it's well implemented. Crosstalk is really low unless you fix that with a plug-in  ;D

But there are times when I want analogue summing myself - even in my modest setup I don't have enough Digital inputs for auxes etc plus I want the physical faders. But it's not 'big studio' stuff so not a commercial prospect but DIY is a different matter.

Anyway, thanks for the links and info.

I have looked at products like the stand alone (Dangerous?) mix boxes and end up scratching my head... I do not have the personality (stomach) to sell products like that.

I prefer to look into real (unsolved) problems.

JR

 

[fripholm]

Just for fun, I also tried the transformer (OEP J30A11C, wired to 1:2.5) driving the same opamp in a non-inverting configuration - and I've found some more interesting things...

To get unity gain at the output, the non-inverting opamp's gain needs to be set to about 2.2 (+6.9 dB).

Compared to the inverting configuration (as discussed above - also set to unity gain at the output), the noise performance stays about the same. Also the crosstalk doesn't change and the whole circuit behaves exactly the same, with regards to terminated vs. open channels (~0.3 dB level drop per one open channel). So far so good...

What I've found most interesting, is the distortion figure non-inverting vs. inverting setup.

Inverting:

This one is the non-inverting opamp at the secondary - using the same opamp and the same transformer, still set to 1:2.5 with the same +4dBu input and output signal:

Also, the frequency response is different, especially at the lowest frequencies. The lower red line is the non-inverting setup:

The transformer seems to be almost 'invisible' when driving the inverting input.  And this is using a mic input tranny, which is not meant to handle large input signals...

 

[abbey road d enfer]

The transformer seems to be almost 'invisible' when driving the inverting input.  And this is using a mic input tranny, which is not meant to handle large input signals...

That's exactly the reason why some designers favor "zero-ohm" inputs for transformers. In this configuration, the transformer sees a very low magnetizing voltage, resulting from the secondary DCR; for this reason, low inductance (and thus low-DCR) transformers are preferred in this application.
Now this technique is suitable for summing amps and line inputs, but for mic inputs, the necessary addition of DCR (for presenting a correct load to the mic) ruins the noise figure.

 

[ruffrecords]

That's exactly the reason why some designers favor "zero-ohm" inputs for transformers. In this configuration, the transformer sees a very low magnetizing voltage, resulting from the secondary DCR; for this reason, low inductance (and thus low-DCR) transformers are preferred in this application.
Now this technique is suitable for summing amps and line inputs, but for mic inputs, the necessary addition of DCR (for presenting a correct load to the mic) ruins the noise figure.

You lost me on the last bit.  If the transformer were perfect (dcr = 0) the impedance seen by the mic would be the secondary load resistor reflected back to the primary. Why is some dcr required??

Cheers

Ian

 

[PRR]

> If the transformer were perfect (dcr = 0) the impedance seen by the mic would be the secondary load resistor reflected back to the primary.

In "zero field", the secondary terminates in a virtual earth.

If all were ideal, this is zero Ohms to the mike. <10r is quite possible, but not good mike (or line!) loading.

 

[ruffrecords]

> If the transformer were perfect (dcr = 0) the impedance seen by the mic would be the secondary load resistor reflected back to the primary.

In "zero field", the secondary terminates in a virtual earth.

If all were ideal, this is zero Ohms to the mike. <10r is quite possible, but not good mike (or line!) loading.

Of course.  Thanks.

Cheers

Ian