Reducing cross-talk in an 80 series Neve

[nielsk]

They are not backwards, both sides are wired in series. There is a U pad on the input (2x 4K7 and one 200) and the secondary has a 4k7 across pins 7&10

 

[jrowell]

An original vintage 80 series Neve left the factory with very good crosstalk performance on the busses. This was one of Neve's areas of effort and achievement over other consoles of the time and most to this day.

As Rupert Neve said in the Sound City movie, "crosstalk to other busses was paramount".

About the only thing that could degrade that wholesale is changes post-Neve factory.

It sounds like the crosstalk here is quite substantial.

You should be able to route at zero dB to say buss 1, listen to the output of any other non routed buss and measure a signal at least -70 dB down or more, even for the large consoles, with the desk fully loaded.

When returned to original configuration and restored as Neve intended these types of problems tend to go away.

Note I said restored as Neve intended. That means you can make changes to make the desks more useful in a modern production enviroment, and get away from the forced 'A80 in the middle, busses to hit tape machine inputs and tape returns on the monitor panel utilizing the overdub function to switch the Studer's seperate sync and repro outs at the console' style of working, and really do some cool stuff with them.

They are building blocks that can be rearranged to accomplish other goals.

However it must be done in a Neve factory manner.

When the consoles are properly configured, installed and any remaining issues logically resolved (ie making sure that yet another ground wire that was soldered in at 2 am before the next days session is all removed and redone properly) and the console is configured correctly (which wasn't necessarily the case even at the factory - they came with ground loops sometimes - it depended on the individual project engineer in charge and how quickly they needed to get it through final test) they work quite well.

There are no inherent errors in the way Neve did things overall, as long as the original system integrity is maintained and the console installed as intended.

There are more reasons than just ground scheme as to why an SSL is more forgiving of grounding and install issues. One big one is in a single ended +24V system the integrity of the B0, grounding and power supply and distribution is hyper critical as single ended circuitry has inherently less rejection than bipolar powered equipment.

This one sounds like it does it all on its own so forget about interfacing other gear, rack grounding, install, power distro or PSU for now and run the desk on its own into a pair of trusted monitors.

Definitely forget about output capacitors and anything else. You will run around in circles. Work the problem. 

Something has changed to make it operate this way.

Look for unterminated buss resistors. Either in the frame from some change that happened at some point (removed or otherwise changed talkback, slate and line up oscillator panels are a go to here) or what sounds like extensive changes to the original desk routing, mode selection and relay system.

Or in the routing modules from the unselected buss resistors not being terminated back to B0 at the routing module buss selection switches.

I would look for physical things and think about where the issue would be with this in mind rather than even turning it on and testing anything, at least at this point.

Keep in mind, it didn't do this originally, therefore something has changed.

Something that needed to be purposefully changed, not parts selection, ESR or degradation of anything over time.

James Rowell
Owner / Head Technician
Sound Service Electronics
www.ssevintageaudio.com
www.facebook.com/ssevintageaudio

 

[ruffrecords]

They are not backwards, both sides are wired in series. There is a U pad on the input (2x 4K7 and one 200) and the secondary has a 4k7 across pins 7&10

Strange. The 'standard' line input to routing modules began with a 32167 wired 10K:600 which dropped a nominal +4dBu to -8dBu which was a typical internal operating level.

The 10468 is a 1:2 mic transformer and with both windings series connected  is a 1K2:4K8 step up transformer with 6dB gain. The 31267 is designed to take +26dBu at the input. The 10468 is not. The U pad you have fitted to yours makes the input look like 9K6 and drops the input level by 33.6dB and with the 6dB gain should have a nominal output at -27.6dBu. Seems rather low to me.

To be honest, I am puzzled why anyone would decide to do this. As the 10468 has a lower ratio than the 31267 the former might have a better frequency response but knowing the attention Rupert paid to transformer design I would expect it to be very small. There is also the fact that the operating level is now nearly 20dB lower so the gain required to bring it back to 'normal' level will add noise.

Cheers

Ian

 

[nielsk]

Here is the input section of the 1985 routing module schematic.
I am curious to see if it would be worthwhile to swap the 31267 for a 10468 on a module to see how it sounds in comparison. Eliminating the U pad on the input can only help sonically.
The gain would be affected, typically (from what I have seen) the routing modules that use the 31267 have the output wired in parallel and a 1k resistor and a 2n2 cap in parallel... but they are also feeding a fader that goes to a BA283 amp.
I should supply a bit more information that interacts with any desire that "all things should be kept as original or suffer the consequences": This is a bit of a Frankenstein's monster, as the client had the lot of 1081s that were being use in racks with one of the new little Neve rack mixers & fader packs, and after many years of desiring a real console was able to get this frame, loaded with the center, monitor, and routing sections. Much of what has been done to it was to get it functioning as a unit, and deal with the host of connector and relay problems.
Nothing was altered in the grounding and power scheme (other than adding individual phantom power switching), just all connections cleaned and tightened.
There are some discrepancies between the block diagrams (correct for this frame) and the various modules, as in it is shown with 1948 routing modules but is equipped with 1868. This may have some bearing on how the various grounds are shown in the frame and in the modules, I will have a chance today to get the bottom of this.

Is there any chance someone has a good scan of EB10568 sheet 1? I have a decent sheet 2, but a barely legible cobbled sheet 1...

 

[jrowell]

Look for a change in the area or panel for the oscillator or the slate to buss functions.

Look for a set of resistors that go to every buss and are all tied together.

If the switching for these functions has been moved, removed or altered the tied together ends might not be terminated.

If the key function is not selected the tied together ends would normally go to B0.

Since it sounds like you have quite a bit of crosstalk I would search physically for something like that.

It is simply the most immediate and obvious possible thing that would cause this issue.

B0 and grounding is important but since you are confident of that and the crosstalk is substantial I would look elsewhere first.

The grounding and B0 would have to be pretty bad to cause the levels of crosstalk you are describing.

By that time you would have other issues besides just high crosstalk.

The consoles do not have to be left as original.

It must be assured that any changes are done within the parameters of the Neve system.

Look for unterminated buss resistors.

The change from a 1948 routing to a 1868 routing shouldn't upset things too badly.

The routing modules generally followed a common connector layout, so if there were significant differences here everything would show up in the wrong place.

The grounding and B0 showed up on pretty specific pins almost always. Of course if that wasn't the case the module wouldn't work at all.

The grounding and B0 system was well thought out, although sometimes not as well implemented in the actual particular console, and everything that was done was done for a reason. Those guys knew what they were doing, so something is amiss.

James

 

[jrowell]

My docs show an 1868 to be an early 4 buss module. 1 Foldback and 4 echo.

Perhaps a pic of the routing module.

The 1868 didn't have B208 amps. It is also an 8 3/4" module. The 1948 is a 12" module.

So my assumption that one for the other shouldn't be too much of an issue is quite far off.

You had mentioned 1985 modules, and 1868 modules, a stock 8048 should come with 1948 modules as you had mentioned from the manual you have. Please confirm on all this.

I have an 8048 manual somewhere, I will see if I can dig out anything useful to you.

The frame configuration does look 8048 based. Three 8 trk monitor panels?

James

 

[nielsk]

Good call on the oscillator bussing, I will look into that. This console has 2 oscillator slots (and 2 of the modules).
Rumor has it that the console was originally set up for post work.
I am pretty familiar with the wiring from the mix buss "bars" to and from various places, I will have a very close look through for any un-terminated resistors.
These are 1985 routing modules, not 1868.
There are 1993 modules in the reverb return section (now 8 additional channels with 1066 & 1073 inputs) and in the re-bussing section. I could never really get my head around the re-bussing, what was the thought behind that?
Yes, it has three 8 channel monitor to 4 track panels, for 24 inputs. These are now wired direct to patch, normalled to Pro Tools outs 33-56. This is working very well.

 

[jrowell]

Look specifically for a set of buss resistors that are all tied together at one end. It could be something else but this is a common and easy place to start.

And that tied together end doesn't go anywhere, either because of a change or or a bad keyswitch, or broken off wire, etc.

Usually there was one selectable line up oscillator and a 'slate' oscillator which was fixed 30Hz and had no knobs.

Often the slate system is removed at some point and if it had a 'to all tracks' or all busses key switch on a panel these resistors are left all tied together at one end and that one end would either have been fed from the oscillator out or connected to B0 via the key switch.

The regular line up oscillator also will have a 'to all busses' key switch and that could be an issue too.

Also beware of moved panels as I mentioned before.

I found additional pics of what looks like the same desk online.

It is seems clear from the additional pics of it online that the talkback, line up and slate assign key switch panel has been moved up under the talkback mic, along with another non original looking single width panel beside it.

They are covering up a slot where a 2254 comp would have originally gone. So the key panel has been moved.

Take a close look at that panel and anywhere any of the wires go. Or should have gone originally.

The layout I have also shows the slate oscillator going where your line up oscillator is.

That would just make the key switches labled incorrectly but I would check it anyways.

There is also another key switch and a red panel lamp below the oscillator area. That should be below and associated with the line up oscillator, it looks like. Again from the other pics online the line up oscillator appeared to be in the wrong spot. Check that too.

The slate and or talkback keys are a good place to start and perhaps there was a second panel (usually) that repeated all these functions for the producer.

If the panels have been removed and the wires cut the feed resistors tied ends don't go anywhere.

There may be up to 4 sets or more so look for them all, and make sure the tied ends go somewhere sensible. It all depends on what custom options the desk had been ordered with and what changes have been made since 1973 or 1974, give or take.

Let me know what you find. PM me if you need to. I don't have notification for regular thread updates set and am at a service call all day tomorrow.

James

 

[nielsk]

I have found something very promising, there appears to have once been a producers desk that had a duplicate set of oscillators in it that was removed and the wiring hacked off. The bridges across all the busses is still there, un-terminated. Of course, very difficult to get into with the iron, but I will either remove it or ground it.
I have gone through the frame & routing module wiring completely, there appears to be no ground loops. One lead goes from the B- buss bar to each module, and hits all audio grounds and internal module shielding. The module case and all external cable shielding goes to the frame, no connection between them.

 

[jrowell]

I have found something very promising, there appears to have once been a producers desk that had a duplicate set of oscillators in it that was removed and the wiring hacked off. The bridges across all the busses is still there, un-terminated. Of course, very difficult to get into with the iron, but I will either remove it or ground it.
I have gone through the frame & routing module wiring completely, there appears to be no ground loops. One lead goes from the B- buss bar to each module, and hits all audio grounds and internal module shielding. The module case and all external cable shielding goes to the frame, no connection between them.

That will be it. I am glad the issue has been located, and it was as expected. Permanently ground the tied together end with a thick blue wire directly back to the bussing B0. This will preserve the original bussing structure in terms of the number of drop on points. Although once you get up to 32 channels plus it wouldn't make too much difference to remove a few feeds, however you might as well keep things as they were by grounding them, as they would have been via the keyswitches on the missing panel.

Check for any other tied together buss resistors in this area. There might be up to 3 more for slate and possibly they repeated the engineers key for talkback and slate at that panel location as well.

Get them all or the desk will still exhibit excess crosstalk.

Please post back with the results after you have made the changes. It is excellent that the suspected cause has been located in the actual console and that it was not something more obscure.

It pretty well had to be that cause, based on the amount of crosstalk you indicated the desk had.

I wonder how many people spent how many hours before you received the console, trying to trace down this issue, all caused by an improper removal of a producers switching panel.

Check the engineers keying panel as well, as that has been moved.

James Rowell
Owner / Head Technician
Sound Service Electronics
www.ssevintageaudio.com
www.facebook.com/ssevintageaudio

 

[jrowell]

Any news on the results on the 8048?

Thanks,


James

 

[nielsk]

It is just about all back together, should be firing it up & testing today / tomorrow.
While I can't see how un-terminated resistors on the mix buss bars can be a problem (other than acting as a tiny antenna, how can a resistor not connected to anything cause a problem? there are many un-used rows of resistors for various options, as well as un-used rows of ties to the busses) I do believe that the groups of resistors that were tied together and not terminated are the problem. Feeling very optimistic....
My current line of thinking on the 10468 as a line input is that the 208 amps can't take much level, it is stomped down to -29. Maybe it was determined that it works better to do this by the input pad then by attenuating the transformer output? Or just sorting through the various options to get the right level?
In my experience, as long as the transformer can take the level,  it sounds better to attenuate level after a transformer than by padding the input, are there cased where this is not true?

 

[jrowell]

It is just about all back together, should be firing it up & testing today / tomorrow.
While I can't see how un-terminated resistors on the mix buss bars can be a problem (other than acting as a tiny antenna, how can a resistor not connected to anything cause a problem? there are many un-used rows of resistors for various options, as well as un-used rows of ties to the busses) I do believe that the groups of resistors that were tied together and not terminated are the problem. Feeling very optimistic....
My current line of thinking on the 10468 as a line input is that the 208 amps can't take much level, it is stomped down to -29. Maybe it was determined that it works better to do this by the input pad then by attenuating the transformer output? Or just sorting through the various options to get the right level?
In my experience, as long as the transformer can take the level,  it sounds better to attenuate level after a transformer than by padding the input, are there cased where this is not true?

No, unused drop on points that are isolated will not cause an issue.

As long as the feed ends are not all tied together to create a feed to all busses.

Make sure to get all of those or you still will have crosstalk.

Unless of course they are tied back to a keyswitch for an oscillator or talkback etc. function.

Yes, it does sound better to pad a transformer on the secondary as Neve usually did, but in this case
you have to build out the input impedance of the 10468 to create a 10K ohm input.

I have not confirmed on my docs about the use of a 31267 with a b208 amp.

I hope everything goes well with testing.

Talk soon.

James Rowell
Sound Service Electronics
www.ssevintageaudio.com
www.facebook.com/ssevintageaudio

 

[jrowell]

What is meant by unused ties?

Any unused ties between resistors going to even two busses will cause crosstalk on those busses if the tied ends are unterminated.

Just want to make sure they are all accounted for so you don't have remaining issues.

Bus to bus should be at least -70dB or more. If not keep looking.

They had very low crosstalk. The busbar assembly with each strip individually screened and everything shielded is one key part.

This was lost when consoles went over to ribbons for the bussing. Sure, ground every other conductor but it still isn't the same.

James

 

[nielsk]

To attempt to clarify, the buss bars have 3 types of connections, the first (coded yellow) are the resistor legs poking out that are tied to the buss traces on the other side. This is where is "input" signals from modules and so on connect. Second are "outputs" (coded red), direct connections to the buss traces where the lines to the buss amps connect, and where other buss area join. Third are the ends of the traces, where sections of buss "bar" join.
There are extra yellow and red connections available on each buss bar section, they are more or less generic so that the same part can be used in different areas. This means that not all connections are used, but as noted previously, I can not see that this is a problem (as long as nothing is connected to the unused terminals).
I was able to fire it up last night, other than finding a few broken wires everything is working great, "hum" seems much lower after fixing several internal (and factory) ground loops, and routing all B- and chassis grounds to "star" terminate at the tech ground point.
I will be testing further today.

Thanks again for all the help!

 

[jrowell]

Yes, the standard Neve setup on the busbars always had extra drop on (yellow in your case) and take off points (red)

These will not cause issues.

It sounds like you have got all the tied ends ones taken care of.

The crosstalk should now be very low. Test it, assign a 1k tone to one bus, use an external generator.
Put the tone all the way to the group 1 output. Measure it to be 0dB. Open the path for group 2. Check it with tone so you get 0dB on the group out. Unroute the tone to group 2 while leaving everything up and the tone coming out of group 1. Measure the group 2 output. It should be -70dB or better. Post the results if you could.

The individual resistors behind the unused yellow drop on points will not cause issues if not connected to anything.

Good that you found the individual ground loops (always seems to be some factory ones as well).

Happy holidays,

James Rowell

 

[jrowell]

How is the patient? Back to making crosstalk free mixes?

James Rowell
Owner / Head Technician
Sound Service Electronics
www.ssevintageaudio.com
www.facebook.com/ssevintageaudio

 

[nielsk]

The control room is fully functional now and I am into testing. So far so good.... Noise floor is lowered, part of which is due to a much improved infrastructure, but some due to improvements in console ground paths and elimination of faults... overall buss crosstalk is much improved but there are a few problems still to sort out. I will get into that in the next few days

 

[nielsk]

As the cross talk problems have been solved I have been delving into lowering the noise floor. Other than isolating modules that have higher "hiss" than others and rebuilding them, with helpful guidance from Jrowell I have removed all audio B- leads from the power switch block (why would they put that on a switch in the first place have no idea) and re routed them to the b- buss bar, and removed the multiple leads from the buss bar to tech ground, replacing the with one heavy high quality lead. Triple checked for shorts between chassis and B- (last one found was a meter bulb touching the panel!)...
The single ended mix busses are prone to picking up noise, considering options here... it seems like a transformer input mix amp with the return line going to a shield is a very strange concept, it would have been so easy for them to make a balanced mix buss, everything is there to do it but the buss traces. There is even an unused pole on each buss assign switch...

 

[Orpheus]

Frame is back together in it's new home (still under construction!)
I will measure at the routing module connector between B- and audio ground, but it ties inside the routing module (clearly shown on the schematic, on the 1993 module pin E1 is tied to pin H1, and on the 1985 unit pin V1 is tied to pin A1)
All the B- runs from the edge connectors to a (maybe 1/4 " square) copper buss bar that runs the length of the console, and the buss bar is tied to the chassis.
All the shield lines tie to the chassis.
I need to fully explore the path the audio (referred to as "audio B-) grounds run.

I think he said check between B- and CHASSIS which is not of course audio ground.

I don't think the B- bus bar should be tied to chassis at any point in the console except really close to the technical earth terminal. Check out this Neve missive:

Cheers

Ian

I'm not sure I understand this paper. Is the B- the shield of an unbalanced audio signal?

Why would they call it B+/B-?