Direct output from Neve channel switching unit

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

atavacron

Well-known member
Joined
Jan 28, 2009
Messages
344
Let's say you have a 33752 doing all its normal things from its 5534-based -6dB unbalanced output – sending to four aux busses, assigning to eight groups through an optional panner, and providing signal to solo, all through appropriate summing feed resistors. 33752 output stage schematic attached.

If you wanted to derive a balanced direct out from that same output, the most Neveish way to go about it would be with an LO2567 transformer, with a 1k5 resistor and .01µF cap in series across its output to control ring. Transformer spec in reply below.

If wired series/series, you get 4dB of free gain, and your direct output lives at -2dB. You've achieved a great signal, but it's a bit impractical if you're sending both these direct outputs and +4dB post fade outputs (groups, mainly) to tape, and later trying to use them at the same level in the mix. You end up needing to add 6dB of noise gain at some point, and there's really no way to align tape to accommodate both levels. It's less of an issue in a DAW, but still creates a bummer of a monitoring situation during tracking. It would be a shame to have to attenuate post fade outputs down to the direct output level.

It looks like the easiest solution is to mimic the output of a 1081 (whose 340 output amp is pretty similar in function to the 33752's output stage) and run the trafo in parallel/series for a healthy 10dB of free gain, putting your balanced output at +4dB. 1081 output stage schematic in reply below.

Here are my questions:

Is it an issue to ask this shared output to drive a 50 ohm load? Is the 200 ohm load similarly problematic, given the multiple feeds?

What options do you have at this stage to take advantage of the free gain other than a unity buffer, and would the best option for that be another 5534?

Would it be preferable to place a THAT 1646 here, before the transformer, and leave it wired series/series? This would also bring the output to +4dB. If that's the way, how would one best derive an appropriate supply for the 1646 from a +24V supply with a 0V reference?


A Geoff Tanner anecdote on the topic:

On most vintage Neve consoles there is no separate Studio Loud Speaker amplifier… it’s wired off the unbalanced output of the monitor amplifier using an external transformer to balance the signal. On the 8058/68 the external transformer is a LO1173 with a primary impedance of around 70 ohms and 8db step up. On a 1271/2 you would use an LO2567 with a primary impedance of 200 ohms and 4db step up (connected with a 100uF coupling capacitor).

What Mister Neve fixer forgot to take into consideration was that, by retaining the original LO1173, the Monitor 1271 was driving three times the original intended load and dramatically changing the conditions around the 2N3055 driver stage.

The effect was that, at 0VU, the console sounded reasonably “Neve like” but the minute the signal was driven into the red on the VU’s (above +7dBu) the Monitor 1271′s went into premature clipping and sounded like crap. Obviously, the rest of the console could be working famously well but you listen to the mix on the monitor system and this will be what the artist/producer hears.

A famous LA producer was the first to use the console and, needless to say, this first session did not go well! I was called in and we swapped the LO1173′s for LO2567′s… to produce a much sweeter sounding console with no headroom issues.





 

Attachments

  • Neve 33752 Output Stage.png
    Neve 33752 Output Stage.png
    937.2 KB · Views: 98
Because all the outputs need to be at the same level, I suppose, and because the direct outs are already there.

There are post fade inserts (from the back half of a 3416 off of the fader, before going back to the front half of the 3416 and on into the back half of the 33752 for routing) and ALSO direct outs. The direct out is off of the unbalanced output from the 33752, straight into an LO2567 series primary. Thus -2dB off the series secondary.

The direct outs are the easiest to use for tracking due to the way the patchbay works. But for mixing one needs to be able to put the converters or tape deck back into the same point in the chain for recall, thus will probably be patching from and to the post fade inserts. Lucky to have those. Is that what you mean by “the channel amp”?
 
atavacron said:
Because all the outputs need to be at the same level, I suppose, and because the direct outs are already there.
Sorry, I still don't get this. All classic Neve channel amps are transformer balanced in and out., The balanced out goes to an insert on the patch bay. You simply use the sniff of the insert  for a direct out

Cheers

Ian
 
Fair. The additional direct outputs, which were not part of the original build, go to a patchbay row that is  half normalled down to an Elco. That Elco can be easily patched to either tape or A2D.

I’d been thinking that the add’l direct outs might be unnecessary...there’s also the fact that they are an extra transformer, opamp, and transformer down the line from the post fade inserts.

Okay so the question is, with the caveat that the add’l LO2567 direct outs may be superfluous, what would be the best/simplest/most transparent way to bring them up to +4dB?
 
atavacron said:
Fair. The additional direct outputs, which were not part of the original build, go to a patchbay row that is  half normalled down to an Elco. That Elco can be easily patched to either tape or A2D.
Workflow has certainly changed, Back in the mid 70s at Neve, most of the big mixers we made had a group per track. Direct outs were rare. But later I think there were quite a number that had channel outs wired to an Elco as well as the patch bay. Then we started adding big relays to switch 24 tracks of line inputs to tape outs and soon after that the inline consoles appeared so everything was in a state of flux. Not surprising they don't exactly meet the needs of today

I’d been thinking that the add’l direct outs might be unnecessary...there’s also the fact that they are an extra transformer, opamp, and transformer down the line from the post fade inserts.

Okay so the question is, with the caveat that the add’l LO2567 direct outs may be superfluous, what would be the best/simplest/most transparent way to bring them up to +4dB?
If you want to retain the single 24V supply then something from THAT would seem to be a good approach.

Cheers

Ian
 
If you want to retain the single 24V supply then something from THAT would seem to be a good approach.

Cool, so I’m not barking up the entirety wrong tree. The THAT 1646 takes a bipolar supply, I think +/-20 max...but again I believe the point is to buffer, not amplify, and I am unsure what will drive the transformer appropriately without discrete transistors. I wonder if the BA340/440 can be run inverted at unity gain. I don’t think the BA406 would do the job.

I guess I’d need DC-DC conversion and boosting to get in the range of bipolar 18V. I’m a bit of a PSU noob, so I’d take pointers.
 
OK, I think I need to explain what I had in mind. First, all op amps, including THAT ones, can be operated from a single supply. You just need to provide a bias equal to half the rail voltage and capacitively couple the inputs and outputs. This means you do not need an additional power supply

The THAT 1646 has two driven outputs. The +ve op can be driven to near the rail voltage while the other is at 0V (in a single rail arrangement) and vice versa. So the total available output swing is 48V peak with a 24V rail so there is no need for a bipolar supply. From the data sheet, it can swing to within 3V of the rail so with a 24V single supply the peak to peak output swing is (24-6) x 2 which is 36V pp which is 12.7V rms which is  +24dBu.

Hope that makes sense.

Cheers

ian
 
Great, thank you! Maybe I will draw something that accomplishes this and post it for correction.
 
On a related note, what is the modern equivalent of the single conductor shielded wire that was used alongside FST? I'm not seeing an unbalanced version on the Canford site.
 
OK, first you need to bias the Gnd pin to half the rail volts. So connect two 10K resistors in series across the 24V and connect the junction to Gnd pin. Put a 100uF 25V cap between Gnd and B-. The input is now biased to half the supply so you need to add a blocking cap in series with the In+ input with its + side to the In+ input.

What is FST?

Cheers

Ian
 
abbey road d enfer said:
Why a xfmr after a balanced line driver?  :eek:

Agreed, would rather keep this unbalanced. The LO2567s are wired and not coming out – I'm basically finishing the job. So I need to get them up to +4 on the secondary. That could mean unity gain into 50 ohms, or 6dB gain into 200 ohms. The former is preferable, unless it requires significant current draw. I suppose if there were no way to do this without a large transistor pair, it wouldn't hurt anything to go with the latter option.

Suggest an IC that is stable at unity, cool with single 24V, will take a non-inverting input (ideally), and can drive 50 ohms with enough current to let the LO2567 do it's 10dB-of-free-gain thing?

Also, Ian, thank you! I will redraw just to make sure I'm getting it. Canford FST is the balanced wire in this desk.
 
ruffrecords said:
OK, first you need to bias the Gnd pin to half the rail volts. So connect two 10K resistors in series across the 24V and connect the junction to Gnd pin. Put a 100uF 25V cap between Gnd and B-. The input is now biased to half the supply so you need to add a blocking cap in series with the In+ input with its + side to the In+ input.

Second try
 

Attachments

  • 2nd draft of THAT 1646 into LO2567.png
    2nd draft of THAT 1646 into LO2567.png
    182.3 KB · Views: 62
atavacron said:
Agreed, would rather keep this unbalanced. The LO2567s are wired and not coming out – I'm basically finishing the job. So I need to get them up to +4 on the secondary. That could mean unity gain into 50 ohms, or 6dB gain into 200 ohms. The former is preferable, unless it requires significant current draw. I suppose if there were no way to do this without a large transistor pair, it wouldn't hurt anything to go with the latter option.

Suggest an IC that is stable at unity, cool with single 24V, will take a non-inverting input (ideally), and can drive 50 ohms with enough current to let the LO2567 do it's 10dB-of-free-gain thing?
But why not operate the 1646 with some additional gain?
 
You second try looks basically OK to me. However, as Abbey points out, the THAT chip gives a balanced output so the transformer is superfluous. Also, I think the THAT chip will only frive a load as low a 600 ohms so it probably won't drive a standard Neve step up output transformer anyway.

If the transfprmer are already in there and you just need a means to drive them then the original Neve sinle ended class A amp is probably the best way to go. It may not be unity gain stable but that should not matter - just pad the input.

Cheers

Ian
 
Right..the 1646 starts getting loaded down around 600 ohms on the output; spec sheet shows a little drop as compared to 10K, and no minimum is stated. Unsure what would happen at 200 ohms, probably not the right things. Thanks for that exercise though, Ian, really appreciate it!

Abbey, I wouldn't know how to set up the 1646 for more than 6dB gain. It seems fixed, at least according to the documentation. Is there some trick? It would be a viable solution here at 10dB gain, as the sound would be just a 31267 and a 5534 away from what's coming out the Post Fade Inserts.

The Tech Info Sheets are pretty clear on not running an opamp at less than 10dB gain...hence the padding, I get that. But they also detail voltage mixing into the gain boost pin of the rear of any module (aka the inverting input), and mention that you can ignore the 10dB minimum in that case. I suppose if one had a bevy of BA489 cards, one could run a double bevy of output amps, inverting, at 6dB gain (unless there's a caveat to using the inverting input?).  I'm guessing that you could also run the amp non-inverting at 12dB gain, and wire the LO2567 series/parallel with a 2dB drop, to come out at +4. Now that would be a low impedance output.

The BA489 is not hard to ape in a custom board for however many channels. One could make the Neve equivalent of CAPI's Inv-ACA, four 640s on a card pumping out 1.5W, with the ability to drive four 15 ohm loads, wouldn't that be swell. There appear to be no modern reproductions of the 340, 440, 640, or 512 (save for Hairball's neat little 512 for bipolar 16V). They are all massively overpriced on the used market. I wish someone would do a run of faithful 640 or 512 PCBs and heatsinks – count me in for a double bevy.
 
Back
Top