Is usb reclocking necessary?

That depends. Re-clocking reduces jitter. If you have a sufficiently good DAC ($$$ is not always good or sophisticated), the jitter is already handled - maybe better than the re-clocker. I did see some test data that indicates that many reclockers are not all that good.

However, jitter makes a big difference. Since I am beginning the process of DAC design, i started by adding USB interfaces to two old, but excellent in the day, DACs that were S/P DIF only (one step ahead of smoke signals).

Holy smoke signals batman! What an improvement. OK, OK, low bar, yea, but it goes to show the fundamental issue we’re dealing with.


Now if you reclocker is <<< than your mega-bucks DAC, it ought to, uh, worsen jitter.
Its neither magic nor rocket science. Its jitter, which is the X-axis in the Cartesian system that translates a digital code to an analog waveform.

I will point out that the clock anomalies ARE analog :-)
.. and have direct analog implications.
I will also add something i ought to have said earlier but neglected to:  proper clocking inside the DAC (either between the USB interface and the internal oversampling or as the actual system clock itself) is vastly better than an external box.  An external box can only reduce the incoming jitter.  Keep in mind that, in an idea world,that jitter ought not matter since the DAC syncs on that, and generally either terminates or enacts a PLL to reduce jitter and then  typically clocks things in an out of whatever register arrangement exists.  But clearly, it does matter.
Since the only thing that matters is the timing coming into/out of (they better match!) your DAC chip or resistors, the latest possible point to execute a proper clock is the best.  his makes any box you can plug in less than ideal. Unclear what the outcome is if it is better than source but worse than internal.
That would be an interesting experiential some day.

@mijostyn says:

No. Only unclocked USB signals need to be clocked. With modern clocks reclocking is unnecessary.

Well, that appears true on the surface. And most people don't specify enough configuration data to rally know. But let me be very clear: with systems running with the source as slave and the DAC as master (clocking), and with stuff that was considered well-clocked when it was new, adding a USB interface with isolation and re clocking has made a huge difference several times.

Note that my data point is NOT a box in the middle, which brings its own can of worms, but a replacement of the coaxial SPDIF provided originally with a custom PCB with independently powered (line powered) USB; full ground, signal and power isolation; clean power to the processing stage, and tight clocking beginning with a read buffer --> then spit out to the DAC.  Yes I'm being a little vague.

So while your point is well taken it comes down to "how well is the internal clocking/timing implemented?", "how dependent is it on source timing?" and "how good is the magic box being added".

I tend to agree that i would NOT just add a magic box.