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.
User avatar
mediatechnology
Posts: 3275
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 » Fri Nov 16, 2018 3:22 pm

Paul -

I would hand-match the 10K and 20KΩ "1%" resistors since encode/decode reversal puts them in the M/S pathway.

The absolute values aren't too critical: Just match the 10Ks to each other and the 20Ks to each other.
e.g. if you have a 10K that measures 9982 try and find another one near 9982 Ω.

From a bin of roughly 100 units I usually find a 50 ppm ratio-matched pair pretty quickly.

Gold
Posts: 435
Joined: Sat Dec 01, 2007 6:20 pm

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

Post by Gold » Fri Nov 16, 2018 3:30 pm

mediatechnology wrote:
Fri Nov 16, 2018 3:22 pm
I would hand-match the 10K and 20KΩ "1%" resistors since encode/decode reversal puts them in the M/S pathway.
The 20K are already 0.1% resistors. On the Preview board the 10K are already 0.1%. I'll change these 10K to 0.1% and see what happens. I'm guessing I shouldn't expect to see a 40dB improvement though.

Gold
Posts: 435
Joined: Sat Dec 01, 2007 6:20 pm

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

Post by Gold » Fri Nov 16, 2018 5:52 pm

I changed the 10K resistors to 0.1%. No difference. I'm planning to completely switch the M/S board out of the circuit unless it is being used to alter the M or S gain. So in a way I guess the crosstalk is okay. I'd like to find a better way to do this though.

What about putting attenuator between the + and - legs of the 1646's at the output of the encoder U5 and U6 and adding makeup gain at U9? I'm worried about the noise floor much less than the crosstalk.

User avatar
mediatechnology
Posts: 3275
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 » Fri Nov 16, 2018 6:07 pm

Gold wrote:
Fri Nov 16, 2018 5:52 pm
I changed the 10K resistors to 0.1%. No difference. I'm planning to completely switch the M/S board out of the circuit unless it is being used to alter the M or S gain. So in a way I guess the crosstalk is okay. I'd like to find a better way to do this though.

What about putting attenuator between the + and - legs of the 1646's at the output of the encoder U5 and U6 and adding makeup gain at U9? I'm worried about the noise floor much less than the crosstalk.
I wouldn't add resistance at the outputs of the 1646.
Once you start varying M/S ratio the crosstalk measurement becomes pretty meaningless anyway.
If you want to bypass it when you're not varying M/S ratio that seems to be a good idea.
The noise floor is typically around -100 dBu for full encode/decode so I wouldn't bypass it for that reason.

Gold
Posts: 435
Joined: Sat Dec 01, 2007 6:20 pm

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

Post by Gold » Fri Nov 16, 2018 6:25 pm

mediatechnology wrote:
Fri Nov 16, 2018 6:07 pm
Once you start varying M/S ratio the crosstalk measurement becomes pretty meaningless anyway.
If you want to bypass it when you're not varying M/S ratio that seems to be a good idea.
The noise floor is typically around -100 dBu for full encode/decode so I wouldn't bypass it for that reason.
Even with 0.1% resistors and no gain change between M and S the crosstalk is 40dB worse on the decode side. I didn't expect that. That's caused by a gain error between U9a and U9b, right? Should I put a trimmer in and hand trim the gains to match as tight as possible?

I was hoping with no gain changes the crosstalk would be more like the other section.

If hand trimming the gain won't work it's looking like switching the board out is the only real option. I'm not really worried about the difference between a -100dBu and a -94dBu noise floor.

User avatar
mediatechnology
Posts: 3275
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 » Sat Nov 17, 2018 9:39 am

With 0.1% resistors you should be getting better measurements. Put a trim in one or both sides.

I will point out that if you have more than a few Ohms interconnection resistance in the Mid>Mid Side>Side loop through the crosstalk will also be blown. I'm thinking with 0.1% resistors you should be getting at least 50 dB.

We did show farther back in one of the MS threads a way to make the gain switch have a hard connection to the source at Mid-position. If you don't bypass overall then you'll want that.

I can send you an insert PCB to bypass it.


Gold
Posts: 435
Joined: Sat Dec 01, 2007 6:20 pm

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

Post by Gold » Sat Nov 17, 2018 10:59 am

Thanks Wayne. I’ll see if I can find where the gain error is. I’m getting the same measurements on both Preview and Modulation M/S boards. I already have a relay bypass board(s) in there.

I think I’ll skip doing a gain stage at this point. I’d like to listen to just the encode/decode first. After I hear that I’ll know better how I want to do the bypass. Whether I want dedicated push button In/Out switches or use an extra pole on a rotary switch to engage when gain is applied.

User avatar
mediatechnology
Posts: 3275
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 » Sat Nov 17, 2018 3:05 pm

You might check to see if there is a termination difference between Mid and Side.

The line drivers have a finite 50Ω (balanced) output impedance.
A load difference - one bridging one terminated - could also increase decode error/crosstalk.

The most precise path is using the normal encoder with U9 in the decoded L and R path.

I recall testing an MS one time that had unusally poor x-talk and one of my loop-through connections - a clip lead - had 10Ω resistance.

Gold
Posts: 435
Joined: Sat Dec 01, 2007 6:20 pm

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

Post by Gold » Sun Nov 18, 2018 12:57 pm

I’ll check some connections. The signal path is the balanced output of the JLM Dingo line amps is multed to the input of the MS board and the balanced relay switcher. The relay board chooses the Dingo output or the MS board output.

The build out resistors on the Dingo are 0.1% 47R. I’ll double check that. There are no external connections to the MS board except input and output. I’ll check the jumpers too.

For these tests I have been using the MS board standalone. Input and output directly to and from the AP. The stand alone numbers match what I was getting when the MS board was getting its input after the line amp.

I’ve been exercising all the connections and so far nothing is changing.

Post Reply