Filter design for longer cable runs

[MCUdude]

Thank you for helping out making the StageConnect open source!

Analog Devices suggests cable-length of up to 15 meters, while Behringer allows longer cables. The filter-network of the AD242x-chips is quite tricky and needs caution on adjusting the hardware-parameters. Use the schematics of the AD2428MINI as a reference.

I've read about the A2B solution before, and read about the hard limit of 15M between the nodes and a hard limit of 16 nodes and a total of 40 meters. However, Behringer "allows longer cables". Why can they do this, or why would this work when Analog Devices are has a stated a hard limit in their datasheets?

It would also be very interesting to see how the input and output filters are designed on StageConnect compatible products and compare them to the AD2428MINI reference design. Personally, I would be interested to know which cable length would be possible over an SFTP CAT6 cable.

[xn--nding-jua]

Hi,

I do not have detailed information, but in one of the Behringer Forums they stated 40 meters as a maximum cable-length: https://ideas.behringer.com/p/stage-connect-cable-spec-clarification

There they mentioned:

Using 110 ohm DMX cables keeps things under reproducible circumstances and allows you to get up to around 40m runs.

I guess they optimized the passive filter of the A2B chip to extend the maximum length from 15 meters to up to 40 meters.

[MCUdude]

Thanks! Feel free to share a picture of the filter input/output if you ever tear down one of your StageConnect devices 🙂

I've been thinking about a similar system for multi channel digital audio over a single pair. My approach is to use a 10:1 LVDS serializer/deserializer (app note here, datasheet for the chips here). These chips have no built-in encoder to make sure there's no DC component, so I'll have to "waste" half of the lines used for data to send an inverted signal so the sum adds up to zero. But if I go for TDM8 streams I should still be able to send a total of four streams (plus LRCLK and BCLK), which makes this a 32 ch solution. And the neat thing about a uni directional system like this is that you can use "traditional" LVDS transmitters, buffers and receivers. The DS15BA101/DS15EA101) claims that 100+ meters of CAT5e is easy to achieve as long as the data rate is less than 500 Mbps.

[YetAnotherElectronicsChannel]

Hi @MCUdude - I can tell you at least a bit about the backgrounds of A²B as far as I understood:
Originally there was never the intention being used in a pro-audio manner. The original purpose was automotive use-case only.
In automotive, everything is super cost-sensitive and the expectation of automotive OEMs was, that this A²B system should run with cheap unshielded twisted-pair wires throughout the car. At the same time, automotive also has very tight EMI requirements. Now with A²B - running basically an ~50 MHz LVDS signal through an unshielded wire (even though it's twisted pair) - you can assume that's not fun. From what I heard, ADI was struggling a lot getting A²B EMI compliant (and an OEM approved solution) in the beginnings of A²B. Later they came up with a very dedicated filter-circuitry in their reference-schematics which everyone just adopts 1:1 - and so, also Behringer did just from looking on the Wing's PCB....
As typical in automotive, parameters are always specified very conservatively because everything is designed for long lifetime, huge temperature-operating-ranges etc... So you can most likely see the 15M cable-length-limitation a bit more relaxed in this constellation here...

[MCUdude]

@YetAnotherElectronicsChannel Interesting. Thanks for the additional details! I've also looked at solutions like the DS90UA101, and a pattern I see in these two-way communication systems is reduced cable lengths compared to unidirectional. It seems like unidirectional LVDS is simpler to get across longer distances, especially with active EQ receivers such as the DS15EA101.

Do the A2B nodes have reclockers, or just (automatic) pre-emphasis/EQ?