Soundcraft 1600/800B

[abbey road d enfer]

Just curious, how are you calculating that? I would not think the output resistors on the bus drivers are parallel. They each connect to a separate output stage and the bus.

The summing resistors come from the pan-pot, so the impedance there is variable ftom zero to almost zero, via about 1k; the other side go to a VLZ point (VE summing amp have near-zero impedance), so they are in parallels AC-wise. Since the unrouted channels are actually disconnected from the bus, the actual bus impedance varies between 4.7k (with zero channels routed) to 12-13k divided by the number of routed channels (including FX and group returns).
The mix bus use 7.5k resistors, resulting in a lower bus impedance, particularly considering that there are more channels routed to the Mix master than to the groups.

 

[RogerAF]

VE summing amp have near-zero impedance

Op amps inputs are idealized as having infinite impedance. In real life they are 10s or 100s of Kohms. I'm having trouble wrapping my head around this reasoning. By your reasoning, if only one channel was routed to the summing amp, adding a second channel would halve the input impedance and thus reduce the volume of both channels. This is not how it works.

 

[Brian Roth]

When you use a virtual earth (inverting opamp) summing amp, the impedance at the inverting input is essentially zero ohms.

Bri

 

[Winston OBoogie]

adding a second channel would halve the input impedance and thus reduce the volume of both channels. This is not how it works.

Gain is determined by R fb/source R.
The only reason gain of a single channel isn't reduced when adding a second channel is because source R is halved, therefore gain of the amplifier goes up by 6dB.

 

[Winston OBoogie]

To the O.P.
Abbey's advice seems spot on to me. Do the maintenance stuff first to get the desk working as originally intended so, new caps, switches/pots cleaned etc.

Then, - if the stereo bus really has all channels and returns always assigned (I don't know), then a 990 op-amp, or some sort of hybrid with discrete transistor diff input pair might be worth pursuing for the left/right bus.

Get hold of a copy of Doug Self's "Small Signal Design". It will show you various paths you might wish to pursue in terms of other modifications.

I don't see the point in trying to turn this into an API desk wanna be. As a nice clean desk to run mixes or stems through, it's probably fine as is with a few tweaks. It's a good starting point anyway. Good luck, and have fun

 

[RogerAF]

Gain is determined by R fb/source R.
The only reason gain of a single channel isn't reduced when adding a second channel is because source R is halved, therefore gain of the amplifier goes up by 6dB.

You left off the beginning of my sentence "By your reasoning..." I'm saying you can't use parallel resistor formula and apply it to summing amp input impedance. Context counts.

 

[RogerAF]

When you use a virtual earth (inverting opamp) summing amp, the impedance at the inverting input is essentially zero ohms.

Bri

That is not what literature says. It's called a virtual earth (or ground) because no current flows into the other input; not because it becomes a ground. It is still a high impedance--ideally no current flows into the input, but in the real-world there is small current flow.

 

[Winston OBoogie]

You left off the beginning of my sentence "By your reasoning..." I'm saying you can't use parallel resistor formula and apply it to summing amp input impedance. Context counts.

Sorry Roger, I guess I jumped the gun and only half read/grasped the intent of your reply, my bad.

 

[Bo Deadly]

That is not what literature says. It's called a virtual earth (or ground) because no current flows into the other input; not because it becomes a ground. It is still a high impedance--ideally no current flows into the input, but in the real-world there is small current flow.

He said "the impedance at the inverting input is essentially zero ohms". Yes, the impedance of an op amp input is high. But the AC impedance of the "net" at the inverting input in a virtual earth circuit is near zero because the output of the op amp is going to source or sink current as necessary to force the inverting input to match the voltage at non-inverting input. So it's the feedback that makes the inverting input low Z at AC. Just like feedback makes the output of an amp low Z.

 

[abbey road d enfer]

I'm having trouble wrapping my head around this reasoning.

Obviously.

By your reasoning, if only one channel was routed to the summing amp, adding a second channel would halve the input impedance and thus reduce the volume of both channels.

This is what happens with so-called "passive" mixing.

That is not what literature says. It's called a virtual earth (or ground) because no current flows into the other input; not because it becomes a ground.

What evidence can you show of this assertion?
It's callled Virtual Earth because the node comprized of the inverting input and NFB resistor presents near-zero impedance relative to the non-inverting input, which is supposed to be at "ground" potential AC-wise.

It is still a high impedance--ideally no current flows into the input, but in the real-world there is small current flow.

If you read correctly my point: "VE summing amp have near-zero impedance ", you would have noted that I'm not talking about an isolated opamp, but of a whole circuit.
An opamp input by itself has a high input impedance.
When submitted to NFB, the circuit, with the help of the NFB network, presents near-zero impedance to the rest of the world at its inverting input, as long as the circuit operates in its linear zone.
You can't neglect the rest of the circuit.


Let's be clear. I should have said writ: "the bus source impedance is about 250 ohms". It is the impedance the summing amp input sees, whatever the technique (VE or "passive"). The importance of this parameter is in order to optimize the noise factor and OLG/slew-rate of the summing amplifier.
The actual bus impedance with the VE summing amp connected is near-zero.

 

[Matt Syson]

There are many 'swings and roundabouts' involved in summing and it up to the designer, constrained by the finance department who tries to get the 'optimum balance' between performance and profitability. Having to contend with a large frame, maybe a couple of metres long with attendant capacitance between the bus and ground and whether sources are really switched on and off the bus thus affecting noise and frequency response (if taken to extremes) can make devolved summing quite attractive in both VE and voltage summing realms. Like 'star grounds' a concept that falls down when physical reality kicks in. once you have decided which atom in the planet is 'THE' star ground having to connect it to other atoms that are some distance away loses some of the 'benefits'.

 

[abbey road d enfer]

There are many 'swings and roundabouts' involved in summing and it up to the designer, constrained by the finance department who tries to get the 'optimum balance' between performance and profitability.

I agree; using a DOA and a xfmr is not my idea of cost-effective design.

Having to contend with a large frame, maybe a couple of metres long with attendant capacitance between the bus and ground and whether sources are really switched on and off the bus thus affecting noise and frequency response (if taken to extremes) can make devolved summing quite attractive in both VE and voltage summing realms.

Yes; in addition to providing theoretical 3dB improvement in bus noise for each doubling of branches, it also results in even more significant reduction of longitudinal noise (because most of it is correlated).

Like 'star grounds' a concept that falls down when physical reality kicks in. once you have decided which atom in the planet is 'THE' star ground having to connect it to other atoms that are some distance away loses some of the 'benefits'.

Thank you for helping de-bunking the star-ground myth.

 

[JohnRoberts]

You guys are so cute (sorry).... It is obvious that several here have killed a lot of brain cells thinking about this (like I have) over the decades, and some have not. I wrote about this in my 1980 article about console performance, and the physics hadn't changed since then.

Using a quieter sum bus amp is impacting only one of the several noise terms. Hint: even a perfect sum bus (like digital) would still have a noise floor from all the individual channel's noise floors combined. In practice the noise coming from a sum bus will likely be dominated by even near perfect mic preamps, not to mention room noise.

In my research decades ago, the larger benefit from improved summing approaches that reduced noise gain or improved loop gain margin was the subsequent reduction in phase shift and distortion.

@ Paul.... the 990 op amp very cleverly (RIP Deane Jensen) uses inductors in series with the input LTP (long tail pair). This creates a unique open loop transfer function with specific stability behavior. It is not practical to integrate these inductors onto an IC, but modern high performance op amps are quite respectable.

JR

 

[RogerAF]

Sorry Roger, I guess I jumped the gun and only half read/grasped the intent of your reply, my bad.

Accepted, no problem. Appreciate your input on this subject.

 

[RogerAF]

What evidence can you show of this assertion?

Well gee, a bit snarky aren't we? I didn't think I'd have to, and I'm not prepared to give lessons in electronic engineering here. I'd suggest that you give Walter Jung's "The Op Amp Cookbook" a (nother(?)) read. And perhaps "The Art Of Electronics" by Paul Horowitz and Winfield Hill as well. Unfortunately both of my copies of those are in storage and not accessible at present. So I was forced to consult Google--that source of all knowledge great and true (snark, snark) and I referenced the following two articles when researching to post my original comments.
I'd like to clear something up first. When I claimed that "It's called a virtual earth (or ground) because no current flows into the other input; not because it becomes a ground." I was basing that on the articles and hadn't thought it out much. I now believe the term "virtual earth" refers to the idea that when connecting various points of any circuit to earth (ground) we don't expect any one of them to reflect anything back to any other one. We consider earth to be a giant hole that swallows up all signals/voltages/currents.
This is how a summing amp functions.
Take an inverting configuration with +In earthed and -In connected to RF and RIn. If RF = RIn, then whatever Vsource is applied to the other end of RIn will appear as -Vout at the output. OK so far?
If RF's value is changed, then -Vout will also change proportionally. This is why an inverting gain stage can have a negative voltage gain whereas a non-inverting gain stage can only reduce to a voltage follower when Rf = 0 ohms.
In the case of a summing amp, given that RF = RIn, and each subsequent RIn = RF, then -Vout will be the sum of all Vsource applied to source ends of all RIns.
The reason (as I now understand it) for term "virtual ground" is that no Vsource will affect any other Vsource. The same as if each was driving into ground.
So that was my idea of how perhaps I should have explained it. Sorry.
In any case, the input resistance as seen from the Op Amp has no relationship to how many RIns are connected. The output gain is still a condition of RF's value. For RF = any single RIn, (and all RIn being equal) then Av = 1.
This holds true for a 4 input mixer or a 64 input mixer.
I hope this helps you understand my posts. I have a feeling we may be trying to say the same things, but human interaction via text is not very efficient. I'm sorry for any misunderstandings on my part.
Finally, here are the two articles I found on which to base my asstounding assertions:
https://www.electronics-lab.com/article/the-summing-opamp-amplifier/https://www.electronics-tutorials.ws/opamp/opamp_1.htmlCheers

 

[JohnRoberts]

Thank you for not trying to give Abbey a lesson in circuit design.

I also wrote about virtual earth topology in my old 1980 console article.

===

For todays (perhaps amusing) TMI,,, back in the 80s when I was looking into patenting my current source summing approach, I got a response from the patent examiner citing a textbook explanation of how an inverting op amp works (sums currents at minus input) as prior art predicting my invention. Back then I didn't have enough money to be able to afford teaching basic electronics to a wet behind the ears patent examiner, who didn't understand the difference between a current source and a current. I am fan of education but not with me paying my lawyer's hourly rate.

JR

 

[Bo Deadly]

I now believe the term "virtual earth" refers to the idea that when connecting various points of any circuit to earth (ground) we don't expect any one of them to reflect anything back to any other one. We consider earth to be a giant hole that swallows up all signals/voltages/currents.

No. Virtual earth / ground refers to how the amp drives the inverting input to match the non-inverting input which is at ground. So if you were to look at the inverting input with a scope it would look just like the non-inverting input which as at ground. There would appear to be no signal as it it was at ground potential. Currents are not being swallowed up. In fact the currents are the signal. It is currents and not voltages that are being summed.

The reason (as I now understand it) for term "virtual ground" is that no Vsource will affect any other Vsource. The same as if each was driving into ground.

That's not a reason, that's just a simple formulaic observation. The reason that a signal of one source does not affect another is because the summing amp is supplying an equal and opposite amount of current to exactly cancel the currents of all sources thus holding the non-inverting input at "ground".

In any case, the input resistance as seen from the Op Amp has no relationship to how many RIns are connected. The output gain is still a condition of RF's value. For RF = any single RIn, (and all RIn being equal) then Av = 1.
This holds true for a 4 input mixer or a 64 input mixer.

Actually the source impedance seen by the amp does depend on how many RIns are connected. They are effectively in parallel. The more RIns connected, the harder the summing amp has to work to supply said equal and opposite currents. But because they're separate sources, it does not change the overall inverting gain equation. At least not for voltage gain. It does affect the noise gain.

I hope this helps you understand my posts.

It does.

 

[ruffrecords]

3

Just curious, how are you calculating that? I would not think the output resistors on the bus drivers are parallel. They each connect to a separate output stage and the bus.

In practice all the bus feed resistors are effectively connected in parallel. Obviously one end of them all are connected together on the bus. The other end of each one is usually fed from an amplifier with a very low output impedance. So at this end each of them is as near as makes no difference connected to 0V as far as ac is concerned. Hence, in ac terms they are all connected together at this end as well. Since they are connected together at both ends they are effectively in parallel. Bus impedance is simply the parallel combination of all the bus feed resistors.

Cheers

Ian

 

[JohnRoberts]

It is common practice to consider all the bus feed resistors in parallel for noise analysis... Some folks who have been working with the stuff for decades don't explain every step along the way.

Perhaps ask a question.

JR

 

[Winston OBoogie]

The original starter of this thread hasn't logged into the forum since November 17th last year.
I'm wondering who exactly we're trying to educate in the matter of Soundcraft desk up cycling these last coupla days?
Haha