Latency from Active Thunderbolt and or USB4 cables?

[Jeff Sloan]

Hi Craig,
(And any others in the know!)

 

I just read your Thunderbolt 3 vs USBC article on Sweetwater. I realize it was a 2018 article...

 

But I can't seem to find any information about active TB cables and any latency they may introduce. 

 

If there are active transceivers in the cable, even though the throughput may be able to be higher, these things must have to take time to do their magic.

 

So when I am trying to get the absolute lowest round trip latency for a singer through a DAW and interface, such as the Presonus Quantum 2626, would a non-active cable at 20 (I need the longer length) have less latency than an active cable capable of 40?

 

From what I can tell, I wouldn't be pushing bandwidth issues with 48k sessions anyway.

 

I am sure lots of people would love to hear about this, and I can't find anything on it, such as real numbers, comparisons, tests, etc.

 

Thanks, 

 

Jeff

[Anderton]

I have not worked with, or analyzed, active Thunderbolt cables but they use low-power GN2033 transceiver chips. I don't believe they are performing conversions, like an A/D or D/A converter, but are basically amplification and relaying elements in a network. As such, they are subject to propagation delays from signals going through cables. There, you're looking at something 5 nanoseconds per foot. You're also going to have propagation delays through the transceiver chips themselves. Again, I don't know about the GN2033, but similar devices are in the 250-500 nanosecond range. So worse case, the system will probably have 0.0000005 seconds (0.0005 ms) of delay. In other words, about the same delay as moving your ear 7/1000ths of an inch from your speakers. Bottom line: don't worry, be happy!

 

Disclaimer: again, this is a best guess based on what I know...but trust me, I don't know everything!

[analogika]

8 hours ago, Anderton said:

I have not worked with, or analyzed, active Thunderbolt cables but they use low-power GN2033 transceiver chips. I don't believe they are performing conversions, like an A/D or D/A converter, but are basically amplification and relaying elements in a network. As such, they are subject to propagation delays from signals going through cables. There, you're looking at something 5 nanoseconds per foot. You're also going to have propagation delays through the transceiver chips themselves. Again, I don't know about the GN2033, but similar devices are in the 250-500 nanosecond range. So worse case, the system will probably have 0.0000005 seconds (0.0005 ms) of delay. In other words, about the same delay as moving your ear 7/1000ths of an inch from your speakers.

 

Doing some napkin math, that equates to about 1/45 of a single sample at 44.1 kHz or about 1/10 of a sample at 192 kHz. 

So in terms of monitoring latency, it's well below what even the highest-resolution systems could reproduce. 

(A digital systems engineer is probably going to wince and pipe in with "yeah, kind of…but it really doesn't help to think of it like that, because it doesn't work that way", presently.)

[Anderton]

9 hours ago, analogika said:

(A digital systems engineer is probably going to wince and pipe in with "yeah, kind of…but it really doesn't help to think of it like that, because it doesn't work that way", presently.)

 

Since I was just guessing...all digital system engineers who want to make fun of me are welcome to join in :) 

[RABid]

I'm not a digital engineer, but can I make fun of you? I hear it is a blast.

 

Seriously, I just ran the possibilities through my head and came up with the same conclusion as the two of you. At most, if you ran the right channel of a stereo signal through one of those cables and the left channel through anon-processing cable it might get thrown out of phase just a bit.