Hi amirm,
Thanks for your insight in this clock option. I am surprised you couldn't measure nor hear the difference.
Looking at you spectrum analysis showed a very clean performance even without the clock option.
For those who are skeptic, the idea of an external clock makes a lot of sense in a few applications:
The first is when you have a disk player and a DAC. If these devices are connected via spdif, toshlink or AES/EBU it is almost impossible to get a really low clock jitter. These connections are flawed in getting the data and clock across. Even the best PLL will always show some residue of data induced jitter.
20 years ago I was involved in a connection called i2s.e, developed by Ultra Analog for Sonic Frontiers. It was system that transported clock and data over separate wires inside a multi coax cable. This connection always sounded better than any other.
So if you clock the DAC and clock the transport from the same clock you can remove this problem. This could be done by using a nice clock in the DAC and slave the transport or slave both to an external clock (= more expensive, and not necessarily better unless the internal clock is poor)
The second application (which I think Auralic tried to solve here) is when you have a streaming DAC that can also lock to a transport. This kind of DAC uses 2 VCXO crystal oscillators. One for the 44.1kHz family and up, and one for the 48kHz family and up.
They are VCXO's (voltage controlled x-tal oscillator) because the incoming clock from a transport can be relatively off. So a circuit measures the difference between the incoming clock and the VCXO clock and adjusts the latter with a PLL to lock to the incoming clock and it wil try by using a low frequency filter to clock with as little jitter as possible.
This is a good method, and better than most VCO's which are a lot cheaper but less accurate. (the X is missing from X-tal) The original digital audio input chip from Crystal Semiconductor, the CS8412 and CS8414 were such solutions. They were used in virtually all DACs 20 years ago and didn't do a good job with respect to jitter.
But now this same DAC is used as a streaming DAC. This is an asynchronous method. The DAC is master and the data from the streaming engine (USB or Ethernet) is asynchronously put into a small buffer and filled in the rate the DAC is eating the data.
Now the VCXO of the DAC is not in a locking mode, but in a free running mode. Often they just supply half the supply voltage to the VCXO control pin to put it in this stable middle control position.
But the fact that the VCXO can be controlled makes it less precise. Noise on the control pin will result in jitter of the clock. Now the use of VCXO is not the best method, but a fixed X-tal would be, but will cost extra.
So feeding it with a clean external clock could be beneficial. Using a rubidium (atomic) clock is BS, as it doesn't say anything about the jitter. It only makes sure that the track is finished at exact the run-time and not plays 20ms longer or shorter.
The trick of the external clock is that it should have an extreme low jitter and the absolute frequency if it is within a few ppm doesn't really influence the sound.
Your test proves to me that the Auralic DAC (in standalone not locked mode) has a very clean clock. The external Leo wil most likely only make the absolute time more precise but didn't improve on jitter.
The fact that the unit locked up when you removed the clock proves to me that their VCXO was in a bad controlled state until you rebooted.
Just my 5 cents in this discussion.