Mbox studio latency

[Dale-c]

As a USB interface, how is the latency compared to firewire or thunderbolt interfaces? Do you have to use a zero latency monitor option like on old interfaces?

[albee1952]

Just personal experience having used the following:
002/003, Mbox 2 Pro, HD/Native(PCIe card) with 192, and Mbox Studio
I record at a 64 buffer setting with no high-latency plugins(but LOTS of EQ's, compressors, channel strips and reverbs). With the old firewire boxes(002/003), the latency was barely noticeable(over 4 years with many clients, I had maybe 1 or 2 mentions). I record myself with the Mbox Studio and do not notice any latency. Thunderbolt will likely be faster, but will you hear the difference? I don't know that answer, but I know its not an issue for me

[dominicperry]

Not good. See here: https://duc.avid.com/showthread.php?t=426381

The internal mixer has Avid FX—EQ, Delay, Reverb, Tuner as per the Avid product page ....https://www.avid.com/products/mbox-s...specifications

Dominic

[Darryl Ramm]

It would be nice for vendors to humor some of us and just publish latency data. (Including, and that's getting more into the weeds, what the latency of the hardware monitoring is).

If anybody has a Mbox Studio and has a few minutes some RTL measurements with would be nice. Easy to make with RTL Utility https://oblique-audio.com/rtl-utility.php Anybody wants to do that and has questions/needs help I'll help them out.

And yes, all this is not too useful by itself, the real question is how small a H/W Buffer can you run at for a real session where you need to track. And yes that may change with the quality of the interface/driver itself but some of us nerds do like to see the raw latency information.

[junkgear]

[Darryl Ramm]

Hey Thanks Will

It's concerning that the actual and reported RTL measurements do not agree. And this is a good example of why it's great to measure/report these values. I don't know enough about the Mbox Studio or it's driver to know if there is anything that might explain this, and don't have one to test.

A long shot, it's maybe possible that Pro Tools knows more about the MBox and will correct these latencies or change how the driver is behaving... so one interesting test might be measuring RTL latencies from inside Pro Tools (e.g. with a click though a HW insert and hand measuring the latency with ADC disabled and then enabled and seeing that the reported ADC latency is etc.).

The latency values are not "great", but maybe they don't need to be with an interface so clearly intended to use hardware monitoring (and what the H/W monitoring latency is is also interesting and I wish would be reported). And again none of these measurements matter unless users can run sessions they need to track in at these HW buffer sizes.

By comparison here are some values for a RME Fireface UFX+ connected via USB 3.1 (Gen 1) to a Mac.

Sample Rate    Buffer Size    Reported RTL    Measured RTL
48000	                32             157             157
48000	                64             221             221
48000                  128             349             349
48000                  256             605             605
48000                  512            1117            1117
48000                 1024            2141            2141
48000                 2048            4189            4189

These were using an older RME kext driver but AudioDriverKit Drivers should have similar performance, I just don't have those numbers with me at the moment.

---
Measurements made with RTL Utility and my own tools.

Intel MacBook Pro 15", macOS Big Sur 11.6
RME UFX+ Driver Firmware v54 Driver v3.27
RME Fireface USB Settings utility Version 1
RME Fireface UFX+ connected via USB 3.1. Fireface UFX+ USB in "Auto" mode.

The safety offset in the driver is the default (using the short safety offset option in the drivers can cause disagreement between reported and measure RTL).

All RME interfaces used had RME TotalMix running/enabled. In some cases TotalMix can add ~1 sample of latency, and that increase is reported by the driver.

[basslik]

That is really cool to know.

With my DIGI 003 @ 48K/24bit, 18th tracks at a time with a few plugins at a 32buffer and no one ever mentioning latency.

I guess that perhaps it's still around fro me. But would like better convertors.

I also like to just pull up a session template I have already for bands.

On the Studio Mbox do you have set all the parameters every time, or can you save templates that one would regularly use as myself.

I'm really curious if anyone tried recording 18 tracks @ 48/24 @ 32buffer ?

[junkgear]

Here is another run, only this time I disabled the Safe Mode option in the MBox Control settings. Not sure what else can be disabled, but the numbers are slightly better.

[basslik]

Originally Posted by junkgear

Now would that be at all 18 tracks ?

[Darryl Ramm]

Originally Posted by basslik

Now would that be at all 18 tracks ?

Changing the number of inputs actually being (edit:) used changes nothing to do with latency in the interface or driver.

But here you have to use ADAT to get that many inputs, and the exact latencies for those ADAT inputs are going to be different than that of an analog input. And that difference is not reported by the driver as I've explained many times before on DUC.

But again the high-level question is what can your system achieve, how small a H/W buffer can your setup run at with the sessions you want to track into.