828mk3 hybrid ADAT bank B ticking noise

[HCMarkus]

ADAT does not do sync on it's own, definitely the problem. Has to have WC also. You can set a device receiving from one ADAT to resolve sync from the incoming signal, but multiple incoming ADAT will be trying to resolve to different things.

Can't one send clock via ADAT from a single master device with multiple ADAT Outs, then bring synced audio back in thru multiple ADAT Ins?

[EMRR]

ADAT does not do sync on it's own, definitely the problem. Has to have WC also. You can set a device receiving from one ADAT to resolve sync from the incoming signal, but multiple incoming ADAT will be trying to resolve to different things.

Can't one send clock via ADAT from a single master device with multiple ADAT Outs, then bring synced audio back in thru multiple ADAT Ins?

Sure, in that case it's one device, not multiple devices as are giving the OP trouble.

[HCMarkus]

ADAT does not do sync on it's own, definitely the problem. Has to have WC also. You can set a device receiving from one ADAT to resolve sync from the incoming signal, but multiple incoming ADAT will be trying to resolve to different things.

Can't one send clock via ADAT from a single master device with multiple ADAT Outs, then bring synced audio back in thru multiple ADAT Ins?

Sure, in that case it's one device, not multiple devices as are giving the OP trouble.

One Clock to Rule Them All. Where's Frodo?

[EMRR]

Haha!

[Mythh]

ADAT does not do sync on it's own, definitely the problem. Has to have WC also. You can set a device receiving from one ADAT to resolve sync from the incoming signal, but multiple incoming ADAT will be trying to resolve to different things.

Can't one send clock via ADAT from a single master device with multiple ADAT Outs, then bring synced audio back in thru multiple ADAT Ins?

Sure, in that case it's one device, not multiple devices as are giving the OP trouble.

Well my (admittedly limited) understanding was that for my setup the 828 should act as the master clock and both ADAT expander devices should resolve to it. Obviously not the case in practice though! As someone else mentioned, seems odd that the 8pre wouldn't include WC if it was documented that you couldn't slave two 8pres to an 828 via ADAT alone...

Just curious, anyone know then how something like the RME Digiface USB works? It has four banks of ADAT I/O, but no WC. Seems similar to the MOTU LP32, though that does have WC. Just better clocking tech in the more expensive RME?

[EMRR]

Maybe I’m still missing something in the description.

828 clock source set to SPDIF (so Saffire14 acting as master), and 8pre and Saffire 40 both slaved to 828 via ADAT

The Saffire14 can't be both master and slave.

Whatever is upstream has to be on the same clock for an ADAT receiver to resolve all of them. If they aren’t, there are competing clocks. I have never seen anything to suggest a return path ADAT lightpipe can act as a master clock; I don’t think it’s so. That would require a device to somehow be both slave and master. It would work if you were sending the other direction and the upstream devices were set to slave.

The only bulletproof thing I do in that realm is ADAT tape machines feeding a 2408mkIII. The ADATs feed each other clock, and feed the 2408 in parallel. An ADAT sync protocol cable goes from last ADAT to the PCIe card in the computer (card manages the DAW/2408 relationship) so I can do multiple sample accurate passes from the ADAT tape machines and the computer will chase the ADATs precisely. EVERYTHING is locked to the same master ADAT clock, which is the first machine in the group, the one I hit play on. If i want to go the other direction and get from DAW to tape machine I have to change DAW clock to internal and tape machines receiving ADAT to external sync, move the ADAT sync cable from the last ADAT machine output to the first ADAT machine input, with the first one being the master slave, and the others being slave slaves, clocking to the first ADAT. I'm not even sure it would work well to disconnect the ADAT's from each other and set them all to external sync; it might, but they'd all run from different locations...but that's tangential to the situation at hand.

I've got word clock connection between THAT computer and ANOTHER computer, from 2408 to 16A, so I can run the ADAT capture computer and monitor through ADAT lightpipe through the 16A on the newer computer.


A tangent - if you're ever able to hook up a device that has AES/EBU, SPIDF, and ADAT light pipe outputs and listen to them through another device with the same sets of digital inputs, switching between them apples to apples, you'll possibly be severely weirded out by what the clock recovery errors of each do to the sound. There is no pure transfer (like moving a WAV file from one place to another), and ADAT is the worst of them. Sometimes an analog transfer going DA to AD can sound better! ADAT light pipe may be a necessary evil in some cases, but It's one I'd avoid if I could. At least spend up on higher quality optical cables, keep them short, don't let them get kinked, etc.

[James Steele]

Reading this thread is why I haven't sold my Lucid GENx192 and have kept in just in case I will need it at some point as a Master Clock. Right now I am using an 828es and a 24Ai and they're connected to each other via AVB. I assume at some point if I add more analog i/o, the GENx192 is likely to come in handy.

[Mythh]

Maybe I’m still missing something in the description.

828 clock source set to SPDIF (so Saffire14 acting as master), and 8pre and Saffire 40 both slaved to 828 via ADAT

The Saffire14 can't be both master and slave.

Ah yeah, sorry I probably didn't explain this clearly. I may still be misunderstanding things myself, but I don't *think* I had any one device trying to be both master and slave. Possibly some of the confusion comes from talking about both the Saffire14 and the Saffire40...? To be clear, those are different devices. And I tried a few different configurations to simplify the rig as much as possible and pinpoint the problem.

In my original setup I did have the Saffire14 as clock master (set to "internal"), going into the 828 via SPDIF (RCA), and the 828 set to clock from SPDIF signal. I set it up that way based on how the 828 manual says to handle a SPDIF input. This worked fine--even with the 8pre also connected to the 828 via both ADAT in and ADAT out and set to clock via ADAT.

So, I assumed that the master clock signal (from the Saffire14) was being successfully distributed "through" the 828 to the 8pre via ADAT. Again, no audible issues with that setup. It was only when I attempted to add a second ADAT expander (whether the Saffire40 or the 2408) that the ticking/sync problems came up.

My first troubleshooting guess was indeed that having the master clock signal coming into the 828 via SPDIF and out to both ADAT banks might be too complicated. To test this, I disconnected the Saffire14 altogether and reset the 828 clock source to internal. So, setup 2 was: 828 as clock master, distributing to both 8pre (ADAT Bank A slave) and Saffire40 or 2408(ADAT Bank B slave). As far as I can tell from MOTU's manuals and general info about ADAT connections, this should theoretically work... Unfortunately, though, it did not resolve the ticking/sync problem.

Ultimately, I don't doubt that ADAT syncing is just less than ideal in general! Which is too bad, because at first glance some of these older interfaces with dual ADAT banks seem like great options to get a lot of i/o on a budget. Luckily there are some other workarounds as well.

Thank you all for the interesting discussion!

[EMRR]

The wikipedia entry has lots of bad info about the way ADATs work and doesn't address sync outside of this:

For straightforward digital audio transfer, the receiving device can synchronize to the lightpipe's embedded clock signal, achieving a 1:1 digital copy.

This matches everything I've ever heard and experienced with ADAT sync over the last 32 years:

https://emastered.com/blog/what-is-adat

What If I Want To Hook Up More Than 2 ADAT Accessories?
Things get a little more complicated if you're daisy-chaining multiple devices together.

For example if your interface has two ADAT inputs, you can in theory make use of two external preamps to boost your channel count by 16.

But since there's no primary clock source in this scenario you'll need to use a dedicated Word Clock signal to sync everything up.

Some equipment has Word Clock outputs and inputs. As long as all the gear in the above scenario can receive a dedicated Word Clock signal, and your primary device can transmit it, you'll be golden.

I don't have any proof of the concept I've always had in my head. I've taken it that the clock was recovered from the audio stream, and a loopback cable to another upstream device isn't sending audio, so no clock to recover. Is it sending a blank data stream? I've never heard that. If you route audio from the 14 through the 828 to the 40 or the 2408 outputs, does that pass correctly? When you get it to do so, do the 40 or 2408 inputs then pass correctly to the 828?

I then think of the various other hardware sync devices with choices to "pass through" or not, and I would think it would be down to the coding in the device as to whether something automatically passed through, if it even could.

The Saffire40 manual seems vague to me, the flow chart implies a lot without really stating it. Then there's possible differences in the way it works in interface mode versus standalone mode; don't have one, don't know. I saw mention of an indicator for received sync, maybe there's a clue there.

The 828 manual chart on master sync shows only what I'm used to seeing, I've not seen a flow chart like your path that didn't involve WC.

I can confirm I've experienced clock idiosyncrasies in ADAT devices depending on whether they were send or receive at any given moment, with more behind the scenes settings to change for each.

Aggregate device works, but (dangit) there's a warning there too. Anything set for drift correction is being resampled in real time, which makes for drifting phase artifacts in the audio, small but there. Big enough to not get away with a stereo input split across a master device and a drift corrected device.

[HCMarkus]

Word Clock

Word.