Mid Side M-S Matrix Uses No Precision Resistors

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
Gold
Posts: 677
Joined: Sat Dec 01, 2007 5:20 pm

Re: Mid Side M-S Matrix Uses No Precision Resistors

Post by Gold »

I think I understand what you are saying Wayne. The loading issues you bring up are the kind of esoterica that generate the wildly different impressions of gear that abound. That possible 1.2dB error is significant enough that it shouldn't be possible to have this happen.

I vote for the soft bypass. It would be easy enough to jumper out the soft bypass and add a hard bypass but not the other way around.
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Mid Side M-S Matrix Uses No Precision Resistors

Post by mediatechnology »

Thank Paul - I'll draw us a quick picture in a little bit so we're all reading from the same page. Fortunately the scheme I'm proposing is the simplest too.
emrr
Posts: 570
Joined: Sun Mar 02, 2008 10:27 pm
Location: NC, USA
Contact:

Re: Mid Side M-S Matrix Uses No Precision Resistors

Post by emrr »

in a quick morning scan with no coffee yet, I say go for the option which minimizes possible gain errors.
Best,

Doug Williams
Electromagnetic Radiation Recorders
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Mid Side M-S Matrix Uses No Precision Resistors

Post by mediatechnology »

Thanks Doug. Speaking of quick morning scans here's this:

Image
https://www.ka-electronics.com/Images/j ... iagram.jpg

This is the basic flow diagram of the combined encoder/decoder. Notice the interstage points labeled A-H.

Bypass discussion

Two THAT1246 or (1206) receive L and R and convert them to single-ended to drive the encoder section. (Shown at the beginning of this thread.) The encoder (identical to the decoder) are two THAT1240s. The output of the encoder is converted to a balanced drive by two THAT1646s. Those stages drive the external mid/side processor(s). This provides the insert send.

The return from the external processors is received by two THAT1246 and fed to the decoder. The ouput of the decoder is then re-balanced by two more THAT1646s to produce reconstituted L and R.

Points A, B, G and H are in the "left/right" domain. Points C, D, E and F are in the "mid/side" domain.

Bypassing external processing from A to H, or B to G has limitations. One is gain error in the L/R domain, the second is that monitoring through matrix end-to-end, without external processing, is not possible. It doesn't really matter if that bypass is half or full.

Bypassing the insert send and return from D to E also has limitations. Minor gain error occurs from C to D due to the 1646. Major gain error, from loading, occurs at the interface from D to the external processors. (Up to 0.7 dB just from 600 ohm loading.) The gain error by the external processor is significant, but it is the only intentional one. WIth points C through F being in the mid-side domain, very small gain errors reduce separation.

The best place to perform bypass is from C to F. This is the most gain-accurate approach because it eliminates the gain errors at D through E. (It's how the crosstalk tests were performed.) This allows monitoring through the mid-side encoder/decoder without processing. When a bypass is performed at C to F, point H is as close as it's going to get to being the same as the signal at point A. Only this way can the bypassed separation crosstalk figures posted earlier in the thread be attained. Bypass from C to F can be done single-ended using a DPDT relay in a half bypass configuration.

Monitoring Considerations

There really needs to be a second output pair with non-destructive solo capability allowing monitoring of points C-mid, C-side, F-mid and F-side. This allows one to hear the unprocessed mid and side components as well as the post-processed results. How else you gonna really fine-tune the external processors?

Separation Anxiety

The premise of the whole precision encoder-decoder was to be able to monitor through the matrix's entire encode-decode process and have it be absolutely transparent without processing applied. Because very small gain errors affect performance in bypass significantly I've had to give a lot of thought to this. I may be wrong.

Does it make sense to you guys?
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Mid Side M-S Matrix Uses No Precision Resistors

Post by mediatechnology »

I don't think so since the whole thing will probably be patched in when needed and removed when not. I may be wrong about that though and defer to Paul and Doug.

FWIW the solo output could probably be mono. Three switches would be required for that: Mid/Side, Pre/Post and solo On/Off. If destructive solo is OK, then the main outputs could be used. But it seems like there is a need to hear either mid or side before and after to dial in processing. I think hearing both at the same time would be distracting.
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Mid Side M-S Matrix Uses No Precision Resistors

Post by mediatechnology »

Here's the solo switching logic:

Image

https://www.ka-electronics.com/Images/j ... iagram.jpg

There can be destructive solo, non destructive or both...

The "to 1646" lines lead to the final outputs.
Gold
Posts: 677
Joined: Sat Dec 01, 2007 5:20 pm

Re: Mid Side M-S Matrix Uses No Precision Resistors

Post by Gold »

My first thought about the bypass was to have the choice between full hard bypass and the soft bypass. This is fine with me although it adds complication. I doubt most would want both. I imagine one set would be jumpered out on most builds. Maybe the full bypass would be more useful as a separate utility board.

That solo switching logic seems like a can-o-worms to me and unnecessarily complicated. What about just putting in pads for a mute switch on the both the M and S channel?
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Mid Side M-S Matrix Uses No Precision Resistors

Post by mediatechnology »

What about just putting in pads for a mute switch on the both the M and S channel?
Thanks Paul. That does seem to make more sense. It would only allow post-processing monitoring though. Wouldn't you want to hear mid (or side) both pre and post? Or, would you do that switching in the processor?
Gold
Posts: 677
Joined: Sat Dec 01, 2007 5:20 pm

Re: Mid Side M-S Matrix Uses No Precision Resistors

Post by Gold »

Wouldn't you want to hear mid (or side) both pre and post? Or, would you do that switching in the processor?
I don't think it's too much trouble to mute what you want and hit bypass on the processor. I bought the CMOS cookbook and it scared the daylights out of me. I have a feeling fancy logic would add a lot of design time for little benefit. Not to mention all the extra relays and parts.
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Mid Side M-S Matrix Uses No Precision Resistors

Post by mediatechnology »

Actually there's no logic involved - just the switches as shown with solo audio flowing through them. I didn't mean to use logic in the "digital" sense.

If the pads for them were put in there, they simply become auxillary tie points on the board. The only one where I might suggest a relay to be used was for the L&R output buffers. If that relay wasn't used, it could be jumpered out.

The tie points that the switches connect to bridge the M and S outputs of the encoder and the M-S insert returns from the output of the line receivers. Other than the switches (and possible DPDT relay for the main output) would be a buffer for the dedicated solo output. That one could just be a 5534.

Roger without "hard" bypass you only need two DPDT relays: One for "soft" E-E bypass and the second optional one for the 1646 outputs. I'd just keep the solo switches as off-board switches with audio flowing through them. They're not in the primary signal path anyway.

EDIT: Also Roger I think a "half" hardwire bypass (around the whole encoder/decoder enchilada) would be only two relays (4 poles). I'm not as keen on that one because it adds relays and real-estate.
Post Reply