Auditronics Grandson Summing Topology

[bjoneson]

I'm in the process of bringing a mid 70s Auditronics Grandson  (24 channel, 8 bus) console online.

I noticed it has an intriguing bus Topology.

The busses are fed via 9.1k resistors. Then hit a unity gain buffer (5534), which feeds the master fader, which in turn hits a line amp with 20-30dB of trimmable gain. The line amp is a 5534 in non-inverting configuration, with a pair of output transistors feeding a jensen 600:600 transformer. The monitor section is also fed from the output of this stage.

I'm used to seeing the virtual earth Topology, so this kind of caught me off guard. I'm curious as to the tradeoffs, and if there's any obvious opportunity for improvement. When the console was designed these would have been 301 opamps as opposed to the 5534s. They were incorporated in a later revision along with associated compensation network changes.

I'm comfortable with almost any level of modification all the way to etching and drilling PCBs if needs be. I have a number of 2520 and 1731 DOA clones laying around.

The board honestly performs astonishingly well as is. But I can't help but wonder if there's opportunity here.

 

[ruffrecords]

The unity gain buffer seems the odd thing to me. If it is passive mixing then you would expect this amp to have some gain. If it is VE mixing then unity gain is what you would expect. Can you post a schematic of the relevant section?

Cheers

Ian

 

[abbey road d enfer]

I'm in the process of bringing a mid 70s Auditronics Grandson  (24 channel, 8 bus) console online.

I noticed it has an intriguing bus Topology.

The busses are fed via 9.1k resistors. Then hit a unity gain buffer (5534), which feeds the master fader, which in turn hits a line amp with 20-30dB of trimmable gain. The line amp is a 5534 in non-inverting configuration, with a pair of output transistors feeding a jensen 600:600 transformer.

It looks like "passive" mixing followed by a make-up gain amplifier. I find it somewhat strange since I would put some gain in the pre-fader buffer in order to minimize the noise contribution of the later stages. With "passive" mixing, attenuation depends on the number of channels, so gain adjustment is neede along the chain.
Now, with the buffer at unity gain, noise is increased by about 3dB, which may not have been deemed an issue.

I'm used to seeing the virtual earth Topology, so this kind of caught me off guard. I'm curious as to the tradeoffs, and if there's any obvious opportunity for improvement.

There is certainly room for improvement, in terms of noise and x-talk, using VE. You have to watch the bus capacitance; it may or may not be an issue. A decoupling inductor may be necessary. And you'll have to cross the wires somewhere in order to restore polarity.

 

[JohnRoberts]

The ugly secret about signal combining is that there is no free lunch***.

The passive combining will suffer an insertion loss based on number of stems combined, due to a voltage divider with other bus resistors. This make up gain is almost identical to the noise gain of a virtual earth bus topology so no S/N advantage.

There is little difference (on paper) between running the bus buffer at unity gain or higher gain (in practice it would be better to put some gain there), while there could be headroom issues from too much fixed gain there if stems are only connected to the bus when assigned. If the bus resistors are back grounded you could expect a constant insertion loss (level doesn't drop when more stems are assigned) and higher fixed gain could be tolerated at the bus.

[veer] Back in the 90's I used a vaguely similar gain topology (?) in a small sound reinforcement family of mixers. While virtual earth (inverting) topology, I operated the channel post fader buffers and master sum bus all at unity gain, with some +20 dB of gain available post master fader. My design rationale for this mixer was to maximize channel post fader and master bus headroom. You could literally run the input strip hot at +20 dBu, channel fader full up (labelled +10dB but actually unity), master fader at  -10dB ( again actually unity gain) and output that +20dBu signal cleanly without clipping. No other comparable mixer could deliver that kind of headroom. For modest numbers of input channels bus noise was inconsequential.  [/veer]

JR

*** while there is no lossless combining I came very close back in the '70-'80s by replacing the bus resistors with electronic current sources. I made one console at Peavey (AMR) with 112 stems feeding the L/R master bus. This would result in a noise gain of more than 40 dB with conventional virtual earth topology.  I was able to realize a noise gain 20 dB lower than that with current sources in place of 72 of those resistors. Nowadays with digital combining this is all moot. 

 

[bjoneson]

Thank you guys for all the thoughtful replies.  Here is the schematic for the bus assignment / pan:

 

[bjoneson]

Unity Gain buffer (now a 5534 instead of 301)

 

[bjoneson]

Lastly, Line amp (hard to read, but trimmable 20-30dB gain, also with 5534 instead of 301)

 

[JohnRoberts]

Unity Gain buffer (now a 5534 instead of 301)

That is actually +6dB (2x).

JR

 

[abbey road d enfer]

Unity Gain buffer (now a 5534 instead of 301)

So it's not unity gain, it has 6 dB...

 

[abbey road d enfer]

My design rationale for this mixer was to maximize channel post fader and master bus headroom.

Looks like the same motivation here. Depending on applications, the compromise between noise and headroom can vary substantially. Pathé-Marconi had commissionned a custom-built live mixer specifically for one French singer who could go from whisper to thunder in a couple of seconds; operating level was -16dBu, it was so noisy it proved unusable for recording.

 

[bjoneson]

So it's not unity gain, it has 6 dB...

Gah, you're right guys. I was just looking at the resistor ratio. Total brain fart.

Still much less gain than I'd normally expect to see at that stage, but makes sense in terms of internal gain staging / headroom.

 

[JohnRoberts]

Looks like the same motivation here. Depending on applications, the compromise between noise and headroom can vary substantially. Pathé-Marconi had commissionned a custom-built live mixer specifically for one French singer who could go from whisper to thunder in a couple of seconds; operating level was -16dBu, it was so noisy it proved unusable for recording.

-16dBu seems low, but may still be below mic preamp noise. Increasing bus headroom will not solve an input dynamic range problem (perhaps add some compression or limiting). 

I used to run consoles at -2 dBu inside. As mentioned the mixer I ran with lower bus gain, only had a modest number of channels and inexpensive ICs were already pretty quiet by the '90s.

My design brief was to provide a mixer where they never ran out of channel fader with hot input signals. Not a concern for recording, but for live use (sometimes mixing yourself) you don't have time to reset your channel gain structure if creaming the bus. 

JR

 

[abbey road d enfer]

-16dBu seems low, but may still be below mic preamp noise. Increasing bus headroom will not solve an input dynamic range problem (perhaps add some compression or limiting). 

The operator's hand on the fader was supposed to be the compressor... 8)

 

[ruffrecords]

1970s Neve consoles all run at -8dBu internally (relative to a 0dBu input)

Cheers

Ian

 

[JohnRoberts]

1970s Neve consoles all run at -8dBu internally (relative to a 0dBu input)

Cheers

Ian

I also wrote one of my old Audio Mythology columns, comparing and contrasting +4 dBu vs. -10 dBV audio paths. The -10 dBV paths were surprisingly good in almost every metric except max output level.  The reduced nominal voltage swing scaled down the slew rate required allowing cheaper op amps to deliver adequate power bandwidth. Lower voltage rails meant lower power and less heat.

I was surprised by how well the -10 dBV path fared (at least on paper), while being held in such low regard among professionals.  ;D

JR

 

[ruffrecords]

I also wrote one of my old Audio Mythology columns, comparing and contrasting +4 dBu vs. -10 dBV audio paths. The -10 dBV paths were surprisingly good in almost every metric except max output level.  The reduced nominal voltage swing scaled down the slew rate required allowing cheaper op amps to deliver adequate power bandwidth. Lower voltage rails meant lower power and less heat.

I was surprised by how well the -10 dBV path fared (at least on paper), while being held in such low regard among professionals.  ;D

JR

At Neve in the 70s it was driven simply by the fact that the class A preamps on a 24V supply could only output +18dbU max. Running at -8dBu gave the required 26dB headroom.

Cheers

ian