In my experience an UFX+ alone with USB3 is better than RayDAT alone, as you need to add the latency of AD/DA conversion of the via ADAT attached preamp or whatever.

If you look at the values in my blog article you will notice that USB3 from pure driver latency perspective is on par with thunderbolt (see picture of excel file below). With my projects I can not see any negative impact of USB vs. RayDAT.

The FPGA for effects surely is a teaser but the disadvantage will be, that I doubt that the CPU power is sufficient enough on the interface to deliver the required performance shall you require the effects on many channels. At least I would expect that there will be limitations in terms of CPU power of the device.

I would lets this processing do the DAW and run VSTs for compression and what not. You can scale performance IMHO easier on the computer compared to recording interface.

As RME has the brialliant steadyclock mechanism I would all in all maybe think about to get an UFX+
and make it to the base of your environment.

Thanks to MADI you have the possibility to connect a lot of preamps and in terms of cabling, MADI with multimode fiber goes up to 2km cable length, so there is a lot of flexibility. Also flexibility in terms of integration of Octamic XTCs as AUX device in the UFX+ TM FX instance ... you can digitally save, gain, panning, PAD, Instr settings, .. for up to 8 Octamic XTCs that you can connect via MADI @44.1/48. And all settings can be saved in 8 Snapshots and these 8 snapshots you can save in up to 30 workspaces, which gives a lot of flexibility to setup and recall routings etc.

Think about it, if you have questions I will happily answer. More information about my setup you can find here:

http://www.tonstudio-forum.de/blog/inde … Cber-MADI/

http://www.tonstudio-forum.de/blog/inde … 8-RME-UFX/

By this you can save the investment into RayDAT and will get better round trip latency (see excel).

The Octamic XTC has MADI built-in, that very practical.

I would at least order both devices plus Octamic XTC to see, what benefits you get with an RME solution over the other.
Also MIDI over MADI is a nice feature, which reduces the amount of cabling.

I simply wanted to present you an alternative, not more not less. I was aware of the price increase,
but not knowing your budget I wanted to give you an alternative.
Also not many people know about the i.e. very nice integration of the XTC as AUX device.
It could have been the case, that maybe such a solution might awake your interest.
Especially also, because with MADI you can scale much better.

If you read my post closely (hint: excel tabular) then you get at least some answer in regards to the RayDAT latencies.
In terms of clocking further answers follow.

Latencies of the RayDAT
=================
Very low. See my Excel. Maybe you didn't find it on the 1st glimpse.

With an ASIO buffersize of 32 samples you have an input latency of 0,771ms and an output latency of 1,519ms.
The values are what Cubase reports (get told by the RME drviver). This is a round-trip time of 2,290 ms !!
The performance of other RME devices you also find there in the tabular.

At the end, all is very close to each other !

Other recording interface as Preamp on RayDAT
=================================

You find in the Excel also values how the Round Trip latency would be, if you connect i.e. an UFX to the RayDAT.
Which I did in the past. Quite nice, the only little burden is the extra work to do the channel mapping.

For my setup I found out there is no real difference between using UFX/UFX+ and the RayDAT.
I am using in my Cubase projects a Mix of 2-3 VST instruments and vocal / guiar tracks.

Further (synthetic) load tests I made with a 400 track Cubase project with a few VSTs in each track.
Also there I didn't find very much difference between UFX+ and RayDAT, only a little less CPU.

Maybe for your projects this can make a difference... you need to find it out on your own.

More to this topic and how to read the excel
===============================

If you connect the Antelope behind the RayDAT then you need to add their latency for AD/DA additionally.

For the RME UFX+ this would be i.e. 0,28ms + 0,16ms = 0,44ms additional.
So the total Round Trip Latency between the UFX+ in this case used as a preamp on a RayDAT would be:
2,290ms (RayDAT) + 0,44ms (UFX+ AD/DA) = 2,73ms

You need to look whether Antelope documents their AD/DA latencies somewhere in the handbook, then you can calculate the latency.

You might notice that the values of the UFX+ connected via USB3 are very close.

If you would have connected an UFX instead of the UFX+ to the RayDAT (which has higher latency for D/A = 0,63ms),
then the Round Trip Time would be higher
2,290ms (RayDAT) + 0,9ms (UFX AD/DA) = 3,19ms

So in this case taking the UFX+ alone would be better than the UFX as preamp in front of the RayDAT.

I tried to find out, whether I can find a significant difference, using the RayDAT or the USB3 based UFX+
on a very large ("synthetic") Cubase project:
http://www.tonstudio-forum.de/blog/inde … cks-de-en/

There was only a merely noticably lower CPU consumption.

So to sum up, yes the RayDAT is a very nice and stable card. If you want you can use it for your purpose.

Clocking
======

If the RayDAT shall become the heart of the solution and you connect the Orion as preamp behind the RayDAT,
then the RayDAT should become clock master. If you change the sample frequency I regard it as
advantage that the directly attached card (in this case RayDAT) sets the clock for all devices behind it.

If its only the RayDAT and the Orion then I would use no WC and would rely on RMEs SteadyClock which
has a very good quality.
https://www.rme-audio.de/english/techin … yclock.htm
http://www.soundonsound.com/people/rme-designs

Yes 32 buffers are the minimum. TBH .. I doubt that less is really required, so don't worry about this.

RayDAT could still be interesting to get extremely low clock jitter by Steadyclock.
Wordclock is not better than Steadyclock via ADAT / SPDIF / AES.

On the webpage RME simply took the Round Trip Time, which you also find in my Excel as ms values.
Its the latency that every ASIO driver has to report to the DAW.
So if you look at my excel you have both: ASIO buffer size and corresponding Round Trip Time.

EDIT:

as RME drivers are of high quality and steadyclock implements a very nice low jitter clock
it could bring to you a real value to consider the RayDAT in your setup.

On top you get a very nice and flexible routing by TotalMix.

Another big advantage of the RayDAT is, that it has an autarc / separate AES channel.

By this you can i.e. create such setups integrating i.e. an ADI-2 Pro into your solution:
http://www.tonstudio-forum.de/blog/inde … ort-DE-EN/

The ADI-2 Pro give you reference quality D/A.

And the best ist, that you can make use of the ARC USB which is a very nice unit.

So .. Orion, RayDAT (Totalmix + ARC USB) could deliver to you a very nice combination.

See my EDIT / addition in last post.

What do you not understand in terms of latency? Please ask specific, maybe I can formulate it then clearer.

> The latencies on RME page are not the full story. The full story is what you have in your excel, right?

No .. this way you cant tell it, RME does not do information hiding.

The values are the same, only that I took the values from Cubase where the latency is reported with
3 digits behind the comma and added the information with which ASIO buffersize you reach the latencies.

The RME webpage tells the same values, but simply the values in ms ... also a valid approach.