Has anyone ever thought about designing a DIY Digital Patchbay?

[sahib]

I had designed an 8x8 matrix cross point switch. In the original design I used 4051. We tested it in a friend's mastering studio against his expensive module and could not hear a difference.

A friend (from Norway) wanted to build a relay based version. So, I designed a relay based mux card and he is now using it.

Below image shows the original design with 4051 based mux cards. There are 8 rotary switches that represents the output channels. Using these switches any of the input channel is routed to any of the output channel. Obviously you have to avoid summing. I did not include it due to cost. Illuminated push button switches are for enable/disable. Normally red and channel disabled. When pressed it turns green and channel enabled.
I am now looking into an expandable version. I am also brain storming for a digitally controlled version. Another good friend (from Montenegro) gave me some AD8113 chips to play with.

If there is interest I can put a kit together. However, I will probably design a new motherboard and incorporate in/out connectors to eliminate the hard wiring.

 

[jaddie]

Over 40 years ago I was involved in the design of several matrix routers in various configurations. The idea specifically was to eliminiate the patch bay from a broadcast studio complex of many studios and control rooms. The device of choice was an analog switch series made by Siliconix (now a Vishay company). At the time the alternative was the 4000 series CMOS analog switchs, but they had poor off isolation and limited ability to handle full line levels. Our target was +24dBu differential, so +18dBu single ended, which was no problem for the Siliconx, but no good for the 4000.

An analog switch has an on resistance curve. It's not flat, it's U shaped with downward tails at the ends. Back in that time, IIR, the average on resistance was under 100 ohms, so you had to buffer the switch, but we would do that anyway. Again, from memory, off at 20KHz was over 90dB, 1KHz was over 100dB, which was workable. And it may well have been better, I just don't remember. We used mulitplexer style switches, a group of 4x1 and made up cards for 8x1 stereo which could be stacked. I no longer recall the end size or how many crosspoints of the main routing switcher, but pretty big. We designed the cards without CAD, it was tape and film, double sided cards with plated-through holes. The plugged into the BUD card cages. Control was digital, but pre-computer, so static 4 bit commands.

Today Siliconix makes things like these dual 4x1:
https://www.vishay.com/docs/76516/dg1408e.pdf

They'd be perfectly fine for routing audio. Distortion and isolation are still the issue, but the entire switch is way better than what we had, and what we had worked just fine, so shouldn't be a problem at all.

The relay crosspoint idea was floated at the time and rejected for cost, physical size, and power reasons. The difference in off and distortion was the trade-off, but it was all so low it didn't matter.

I still have a card or two somewhere.

 

[JohnRoberts]

Those probably aren't going to be as transparent as some other options. I believe I recall reading not favorable opinions on the chips used in these that handle the audio switching and, then you have the active output circuits . Not sure but sometimes the utility of something takes precedence over getting the most transparent results.

I used CMOS transfer gates to perform signal routing inside my MAP 8x4 midi controlled patch bay (actually inserts switcher). The signal linearity was good, but one performance limitation was head room. Because of the CMOS TG (and other reasons), I ran my internal audio path from +/- 8V power rails. This was ok with nominal -10dBV audio paths.

A trick that I used to de-click my audio switches was to use HF pre-emphasis before the switches and complementary de-emphasis after the switches. This effectively attenuated the HF clicks.

JR

 

[jaddie]

I used CMOS transfer gates to perform signal routing inside my MAP 8x4 midi controlled patch bay (actually inserts switcher). The signal linearity was good, but one performance limitation was head room. Because of the CMOS TG (and other reasons), I ran my internal audio path from +/- 8V power rails. This was ok with nominal -10dBV audio paths.

A trick that I used to de-click my audio switches was to use HF pre-emphasis before the switches and complementary de-emphasis after the switches. This effectively attenuated the HF clicks.

JR

That's why we picked the Siliconix products, +/-15V rails (actually could go higher).
We solved declick by switching differential. The clicks were always in the same polarity, so a common-mode rejection buffer would take care of them.

 

[JohnRoberts]

That's why we picked the Siliconix products, +/-15V rails (actually could go higher).
We solved declick by switching differential. The clicks were always in the same polarity, so a common-mode rejection buffer would take care of them.

the clicks I was suppressing were from switching the audio waveform away from zero crossings.

JR

 

[jaddie]

Oh, that. Switches are too fast. Yeah, we didn't have to switch live audio, it was just routing. You'd have had the same problem with relays. Good solution.

Our clicks were the tiny capacitive discharge within the switch when used on a live console.

 

[Rob Flinn]

I mean patchbays are useful, but the convenience factor and sometimes I notice that I have to turn or re-insert plugs for a solid connection

I always 1/4 turn an back when I insert a patch cord. It helps keep things clean.

 

[JohnRoberts]

Oh, that. Switches are too fast. Yeah, we didn't have to switch live audio, it was just routing. You'd have had the same problem with relays. Good solution.

yup

Our clicks were the tiny capacitive discharge within the switch when used on a live console.

back in the 80s I used TG audio routing for banks of channels at the same time inside a recording console. To improve the linearity I put the TGs inside an op amp's NF loop. Distortion was unmeasurable using my test bench du jour. Clicks were not bad, but it wasn't designed to be banged in the middle of a mix.

JR

 

[mobyd]

Some (35) years ago, I built a digitally controlled patchbay for a restauraunt complex which had multiple rooms and a stack of rotissarie CD players on permanent rotate. The essence was an Analog Devices AD75019 16x16 transmission gate matrix. Any staffer could call up any player by a little button panel on the wall which talked to the engineroom via RS485 (wherein lay the archilles heel of the system - the walls contained numerous mice which peed on the non mouseproof button panels. Mousepiss is extrordinarily corrosive and the controllers gave up the ghost after 20 years or so. As it happened, the restauraunt also gave up around the same time so it was a non issue. Spare mux board in pic.

 

[Brian Roth]

OK....some of us "old timers" have designed and built routers.
Let's go back to the origin of this thread! lol

How would YOU do a crosspoint in year 2024? I mentioned relays..... and showed a TG idea in post #12. Like John Roberts, it has the TG inside a NFB loop.

I DID mention the price of relays earlier in post #11....

With the crosspoint method selected, then just a "simple" design to create the controller. (yeah....right!)

Bri

 

[MidnightArrakis]

I have noticed that there is no digital patchbay DIY projects out there.

Why are they so expensive, like the Flock or Dangerous Music Liaison?

Would an 8 stereo matrix be very expensive or difficult to make that is mastering grade?

I have noticed some AV matrix out there for sale, but they have RCA or unbalanced TS connections for their audio. How bad would it be to put DI's on the ins and Line converters on the outs to make it work with balanced gear? Would this be ok for mastering?
XRL out of outboard > DI > AV Matrix > Line converter > XLR in of outboard

Thank you.

While this chassis shown below isn't a -- patchbay -- per se', it could be used as one.....sort of. This chassis is a -- 12X8 Stereo Audio Router/Switcher" -- that I had both mechanically designed and performed the PCB-layouts of. This unit could take any one of 12 Stereo Audio channels and then route/switch them to one or more of 8 Stereo Audio outputs. However, all of the connectors on the rear-panel are/were RCA connectors and "pluggable terminal-blocks", so this device was not designed for balanced-audio.





Somewhat along the same idea as the above shown chassis, the 12U rack-mount chassis shown below is a "256X256 Video Router/Switcher" that I had also mechanically-designed and performed all of the PCB-layouts of as well. This chassis would allow the user to select any one of 256 video inputs and then route/switch it to any one or more of its 256 video outputs. The LCD screen on the front-panel would allow the user to monitor and/or preview any of the video signals either on its inputs or outputs.



Here is a view of the rear-panel of this chassis just to give you an idea
of what 512 BNC connectors had to look like:


>> A PCB showing 1/2 (256) of all of the BNC connectors on the rear-panel:


>> If anyone has a schematic of an audio=patchbay that they would like to have built, I believe I could help you out there!!!

/

 

[MidnightArrakis]

OK....some of us "old timers" have designed and built routers.
Let's go back to the origin of this thread! lol

How would YOU do a crosspoint in year 2024? I mentioned relays..... and showed a TG idea in post #12. Like John Roberts, it has the TG inside a NFB loop.

I DID mention the price of relays earlier in post #11....

With the crosspoints chosen, then just a "simple" design to create the controller. (yeah....right!)

Bri

[How would YOU do a crosspoint in year 2024?] -- The two chassis shown within my post actually used "Matrix Router/Switcher" IC chips and that was 15+ years ago!!! But, don't ask me what they were or who made them cuz "I dunno"!!! I just took care of the mechanical and PCB-designs of this equipment. The electronics engineers were responsible for all of the circuitry design of things.

/

 

[cdc3oo]

Below image shows the original design with 4051 based mux cards. There are 8 rotary switches that represents the output channels. Using these switches any of the input channel is routed to any of the output channel. Obviously you have to avoid summing. I did not include it due to cost. Illuminated push button switches are for enable/disable. Normally red and channel disabled. When pressed it turns green and channel enabled.

Nice work !

Another good friend (from Montenegro) gave me some AD8113 chips to play with.

I didn't knew this IC. I have an old project lying around that is based on the MT8809 8x8. When I compare the specs, it's clear that the AD chip is much better. Thank you for the information !

 

[Brian Roth]

Interesting chip, also new to me:

https://www.mouser.com/ProductDetail/Analog-Devices/AD8113JSTZ?qs=/tpEQrCGXCwC8S77E59luQ==

256 crosspoints for $60...a bit less than 24 cents/point. That many points using the "proposed" Panasonic TQ2 relays would be $245.

/EDIT Typo!!!.....$645......although each of the 256 relays have 2 poles apiece.

Just sayin.....

Bri

 

[sahib]

Nice work !


I didn't knew this IC. I have an old project lying around that is based on the MT8809 8x8. When I compare the specs, it's clear that the AD chip is much better. Thank you for the information !

Thank you too.

 

[Errit]

Digital controlled patchbays with analog in- and outputs for studio’s are already available for a few years now.
Take a look at this website:
https://anatal.io/index.html

 

[hildy]

I had designed an 8x8 matrix cross point switch. In the original design I used 4051. We tested it in a friend's mastering studio against his expensive module and could not hear a difference.

A friend (from Norway) wanted to build a relay based version. So, I designed a relay based mux card and he is now using it.

Below image shows the original design with 4051 based mux cards. There are 8 rotary switches that represents the output channels. Using these switches any of the input channel is routed to any of the output channel. Obviously you have to avoid summing. I did not include it due to cost. Illuminated push button switches are for enable/disable. Normally red and channel disabled. When pressed it turns green and channel enabled.
I am now looking into an expandable version. I am also brain storming for a digitally controlled version. Another good friend (from Montenegro) gave me some AD8113 chips to play with.

If there is interest I can put a kit together. However, I will probably design a new motherboard and incorporate in/out connectors to eliminate the hard wiring.

View attachment 135143

Jeez man. That looks nice.
What's your take on modern ICs in terms of ability to handle real-world signal levels and THD? Would you rely on solid state switching logic for critical applications. (mastering in particular)

 

[sahib]

Jeez man. That looks nice.
What's your take on modern ICs in terms of ability to handle real-world signal levels and THD? Would you rely on solid state switching logic for critical applications. (mastering in particular)

Thank you.

I am not a mixing/mastering engineer. So, I can not really say that I have trained ears. But I believe I have good ears. (for a 63 year old) However, we did A/B and could not hear a subjective difference between his expensive matrix unit and this one (at that time I was 60 and my friend who is a mastering engineer is/was 20 years younger than me)

The ICs were powered at + - 8V We passed music material at +4dBu, +12dBu and +20dBu and again observed no subjective difference. The THD at 20kHz was 0.003% (measured on APx520).

In terms of relying on solid state switching or not, well, I have never designed such a thing but throughout the music production history thousands of consoles utilised it. So, we can safely say that they must be ok by now. As long as the device performs well subjectively and objectively then I see no problem.

 

[Newmarket]

Solid state switching has been much used to switch / bypass in analogue desks with excellent specs. Often by means of jfets. You can then minimise audio clicks by ramping the gate voltage. You do need a relatively large negative voltage rail for good offness.
For a more convenient implementation - recall the now obsolete SSM range of analogue audio switch ics.

 

[Matador]

Using these switches any of the input channel is routed to any of the output channel. Obviously you have to avoid summing.

If any of the input channels can be assigned to any of the outputs, then how was summing avoided? Did you rely on the user only switching any given output to be from a single choice of input?