Help with the path of our sound

[Bruce Cloutier]

RJ45-AB-prelim3D.pngTj, All great points! I appreciate it. I like the 1-8 and 9-16. And, yeah, it was originally going to be OUT instead of C. I went to C in trying to eliminate the directional bias. How about bidirectional arrows between A and C and B and C through the LEDs? I dunno.

Instead of "Switch" we could use "Selector"? Confusion with Ethernet is going to happen just because of the RJ45s.

This is just a logical cable routing gadget for the JNIOR. The automation can then reconfigure as might be needed when there are multiple sources (or destinations). Obviously there is always more involved like power and, well, video. So you could use it at the destination to select from more than one source (stack them - like railroad switches - to route from more than two). Or use it at a source to send signal to more than one destination.

I appreciate the DB25 gender thoughts. Maybe someone who is using the DB25s can confirm. It would probably be 2xDB25F to DB25M. Note that A and B would consist of one DB25 each and there would be one on the output. Or, are those also used in pairs? There is plenty of time for someone to chime in on that.

Makes sense to maybe do a DB9 version. Maybe even a KVM switch for keyboard, mouse and monitor at some point to excite the IT people? How about RCA and 1/4" jacks? XLR?

Anyway, I have the Bill of Materials in the system. We could actually schedule production on this. The relays are too damn expensive. So are the RJ45s. Not in these quantities and from stock. This would have to be $225 to cover the convenient parts and their cost. I need to work that down. It doesn't make sense that it be half the cost of a JNIOR.

Do the RJ45s need to be shielded? Wires basically drop from the cable onto the PCB immediately. The shielding adds a lot of cost.

It never really made sense for me to do this design work in a vacuum. I appreciate the input assuming you guys don't mind me sharing!

[Marcel Birgelen]

What protocols do you want to support with this A/B switch? It's a relays-based switch, so it should be mostly protocol-independent. We're doing gigabit ethernet with unshielded twisted pair cables, so I don't think that the shielding of the RJ45 connectors is going to make much of a difference for most protocols.

[Steve Guttag]

I think there will need to be lawsuits here. Clearly Bruce has gained access to our exclusive A/B switch design:

image.png

Originally designed just for audio to allow a simple A-B choice. The design allowed for basic stereo sources or up to 8-channels (or more). It was designed in 2011...clearly predating this blatant ripoff.

[Marcel Birgelen]

At least Bruce's design has a nice case around it. But I like the minimalist esthetic and the fact that it seemingly can be endlessly extended.

Also, yours seems to be already available (there is a photo, so it MUST exist, right?), so it might actually be a solution to OP's problem?

[Ryan Gallagher]

I agree "RJ45 Selector" makes a little more sense and isn't already a loaded term.

In a perfect world yes, there is shielding on the RJ45... I believe several variations of AES recommend or require shielded cable in the spec? I think our 2016 install came with some shielded patch cables (too short of course). But if it kills the price point maybe people could get by? Our stuff certainly wasn't terminated as shielded in the install, even though the cable used might have supported it, even if it was available, the DB25 adapters that are so common are just the plastic jobbies.

I'm just thinking as a "general purpose" device, shielded capable would be better, but our industry probably doesn't care for only 8 channels each.

[Tj Hopland]

No expert on any if this but from what little I do understand using cat cable and RJ45's isn't really part of the AES3 spec so we are already working in a gray area and somewhat are just getting away with it because for the most part we tend to do pretty short cable runs? Same with the RJ to DB jumper boxes? Some I think the shell is in theory a metallic shielded plastic but many look to be regular old plastic plastic. Based on those RJ to DB's seemingly working fine and likely not being to spec I would think we could get away without shielding if its adding a bunch to the cost.

DB 25 in cinema both analog and digital is just one cable most of the time and 8 balanced lines which is a + - and a shield ground sort of thing often for each pair so 24 pins worth. Analog that's 8 channels. Digital that's 2 channels per pair so all 16 in the same connector. When we use the DB to RJ adapters most of the time its only the 16 pins that are carried through. I think there are a few of the adapters that maybe the RJ shield ties to the shell and DB body? But I don't think the few I have seen have any ties to those extra pins.

As far as other connectors that is where I'm thinking the DB25 should be good enough. If someone is a little DIY they can easily get DB25 connectors and solder what ever leads with what ever connectors they want. You can even find em where it breaks out into screw terminals in both cinema land where it may or may not be labeled to one of our standards or you can get them just labeled by pin number.

If they are in cinema many of the dealer/service companies will do custom cables or Darco/Odyssey has stock and custom options with DB25's. Then outside cinema and more recently in cinema there is the Tascam pinout which is 8 channels and is usually analog but I guess can also be digital. XLR and 1/4 TRS seem to be the most common but maybe there are other flavors of that? And that is where at least the little I looked the DB's seem to all be male. The XLR end looks like it can be either or or mixed. Its the apparent Tascam all being made and the bi directional nature of the selector that made me think maybe all female would be reasonable.

[Ryan Gallagher]

Making an alternate that is screw terminals would be more universal than DB25, but coming from film audio world DB25 is pretty common historically... although I think GDC used something other than DB25 on some of their early box server models? As TJ said you can get from DB25 to most anything if needed, doing the same from RJ45 probably requires going through DB25 or some nonsense first if you are trying to use off the shelf breakouts and not able to roll your own like Steve ;-) (but I haven't looked seriously at what RJ45 breakouts are out there).

One use case would be if you were switching at the source, and using two completely different processors. In that case instead of using a cheap Y cable for your HI/VI RCA path, could have a second selector switching that analog output, among other things, to a common amp/destination. Booth Monitor maybe same story?

In that mode it could facilitate a primary and backup processor (or one topology for that anyway).

[Marcel Birgelen]

As a generic hard A/B switch/selector-thingamabob I think I'd prefer a bunch of screw terminals. The only downside is that they require more space than a DBx or RJx style connector. The downside of DBx connectors is that they need to be soldered.

[Steve Guttag]

At least Bruce's design has a nice case around it. But I like the minimalist esthetic and the fact that it seemingly can be endlessly extended.

Also, yours seems to be already available (there is a photo, so it MUST exist, right?), so it might actually be a solution to OP's problem?

Yeah, it exists and has been used several times. It is, obviously, a step-and-repeat job and the generous space between them is so one can bandsaw them apart. When installed, it is, typically, on a rack panel (particularly if it is an 8-channel). For just 2-channels, affix it most anywhere.

Now, as to AES3 and RJ45s. AES3 is based around a 110Ω intrinsic impedance for its transmission line. CAT cable is "close enough." Since the tolerance of AES3 is +/- 20%, CAT 5e and CAT6 is closer to 100Ω. That is why it works, even at reasonable distances. Genuine AES3 cable should work better, particularly at long distances (e.g. Belden 1803F for 4-pair) but for anything in cinema, CAT5e/CAT6 should do the trick, particularly now that most everything is using RJ45 connectors.

Yes, AES3 should be using both shielded cables and connectors. It is to keep the noise in more so than keeping anything out. AES3 is relatively noisy and has decent voltage around it for broadcasting. Do most people (in cinema) use shielded stuff? Not that I've seen but some do and I always try to...particularly if I'm using premade cables (preferred). Now, if you are using an Odyssey board for your installation to a vintage processor...they don't normally, use shielded connectors though the server should be.

[Tj Hopland]

Just did a quick search and DigiKey here in the US has a DB25 to screws listed for about $35 so that would be my base line, should be good quality and repeatable. I see other listing as low as $6 so it seems there are still lots of options if one wanted a DB to screws so it may not be worth a separate screw terminal version? Especially when coupled with the space problem of the current board. To fit on there it would have to be a fairly high density block and or a double stack and those just are not fun to work with. I'm actually a little surprised how many options there are out there for DB to screws. It looks like you can get just about any angle and layout you could want. Must be other industries than us using DB's still.

[Marcel Birgelen]

DB-interfaces have been used for about anything... analog, digital serial, digital parallel... data, audio, telemetry... in the IT industry, telecoms, OT devices, etc. etc.

Still, for signal quality's sake you might want to avoid extra transitions and going from DB to screw terminal is yet another one. Maybe not so important for relatively low bandwidth digital connections with sufficient headroom, but. certainly for low-voltage analog connections, especially unbalanced analog audio connections or high-bitrate digital connections.

[Harold Hallikainen]

I remember seeing an AES paper on AES3 over cat5. It looked good. And unshielded worked over longer distances. St USL, we used Ethernet mag jacks for AES 3. Worked great!

[Marcel Birgelen]

I've seen Tascam AES3 over DB25 working with all kinds of run-of-the-mill DB25 cables, both shielded and unshielded. I doubt their impedance always is 110ohm, but like Steve indicated, most AES3 implementations seem to have quite some headroom.

One downside of AES3 is that, while it does have parity bits, it does not have error correction.

[Steve Guttag]

Early in DCInema's lives, I did see people using ordinary audio snake cable and they did run into issues, for one reason or another. We were paid by a projector manufacture (since they were being blamed) for the audio constantly failing (which wasn't the projector's fault). I ran genuine AES3 snake cable ("Barney cable" owing to its purplish color) and made up fresh ends (DB25 on one end and XLR on the other, as I recall) and the problem was resolved. Was it the cable or the connections, I don't know. But one should start with the correct stuff to rule that out.

[Josh Jones]

The broadcast industry has been using RJ45's for just about everything digital and analog for close to 25 years. All the builds we do now are structured cable, and AES runs of several hundred feet are pretty common. Like everything, this is slowly being supplanted by various AoIP protocols.

The broadcast equivalent is called studio hub, and they make a range of dongles to get you in and out of RJ45's. Naturally there are cheaper alternatives as well, and I find myself making adapters on site if I'm short a dongle.

studio-hub_pinout-768x595.png