I've been trying to get a grip on round trip latency (RTL) for a while now testing various DAWs with different interfaces.

I've done loopback tests with a 1 sample long audio click to be able to get sample accurate measurements. I've done the tests with a balanced patch cable on the analog end and a Monster optical cable for ADAT.

When I was doing my initial batch of tests in Spring'12, the UFX USB driver was delivering perfect offset corrections. Exactly 0 samples difference between source and destination. I don't exactly remember the driver version but I think I was on 1.66 at the time. Only with Pro Tools I was reading a 1 sample difference between source and destination.

Now I am on USB driver 1.78. It seems that offset perfection has gone off a bit. In Logic and Cubase I'm reading +3 samples RTL offset difference. With Pro Tools I was initially reading 4 samples. After some experimentation I've come to realize that having Delay Compensation on takes that values down to 3 as well. I guess this was also the reason for the 1 sample offset difference I was reading in Spring'12.

I haven't yet tried downgrading the driver thinking that I wouldn't want to use an outdated driver. I could try though.

Anyhow, as I indicate on the subject line, the FW driver is behaving differently; it's still working perfectly in terms of RTL offset correction.

Finally, with my recently purchased Babyface I'm getting varying readings.

With Logic, both manual measurement and I/O Plugin measurements read 0 samples difference b/w source and dest.

With Cubase I'm reading 1 samples difference.

With PT I'm reading 1 samples with Delay Compensation, and 2 samples with DC off.

Here are some setup details:

Regular MBP Mid 2012 / Ivy 2.3 / 16GB RAM / 10.7.5 up to date 64 bits Kernel
Cubase 7.0.2 / PTHD 10.3.3 / Logic 9.1.8 (all running 32 bits mode)
UFX/Babyface USB Driver 1.78 / UFX Firmware 352-151-341 / Babyface Firmware 203
Test SR: 96KHz w/ 24 bits.

I know these values are negligible. Nevertheless, when I was doing my shopping shootout b/w a UA Apollo and the UFX, the UFX was performing perfectly in this regard as opposed to the Apollo's appalling / very disappointing / huge offset correction error. I'm not even going to talk about the performances of my old stock of M-Audio interfaces. In any case, this reliability and perfection was one of the important reasons for my choosing the UFX. Why not have it go back to it's original glory with a perfect score on both drivers?

So the question: any ideas why I'm experiencing this difference between the drivers?

Thanks!
Derin.

p.s. If you think there might be something wrong with setup and testing methodology, please ask away. However, I honestly doubt this to be the case. And I'm still getting perfect readings on the FW side.
- I've tried a direct USB connection as opposed to my regular connection over a very good quality hub (Digi Hubport 7/c) and that didn't seem to make a difference
- Spring'12 tests were conducted on a Late 2012 MBP / Sandy 2.5 / 16GB / 10.7.3. One obvious difference is USB3 presence on my current MBP.