If I remember right the total current setting for the Trident is set by the resistor across a voltage of 1.2v Hence at 20 ohms the current would be 1.2/20 = 60mA. I'm doing the same set up as you and the Crystek maxes out at I believe 25mA. I used a 24 ohm resistor thereby setting the current draw as 50mA, ie shunting 25mA at the clocks max power. I am sure Russ or Brian will chip in if this is wrong.

Edited by user Monday, November 14, 2011 9:45:13 PM(UTC)  | Reason: Not specified

Thanks Martin and Brian. Just got it up running with a dedicated Trident for each of the two CCHD-957s on the SOtM Async USB board. I have a little 5VDC pre-reg running into it, with a random transfomrer I had lying around. I do not have a lot more room in my case, so just wanted to figure out the current consumption before I purchase a small dedicated transformer for this supply.
I put 100 uF Nichicon polymer at each Trident's input (to help compensate the longish wires from the supply) and .1 uF ceramic right at the input pins on the oscillators. I am going to all this trouble because I really like the sound providing masterclock from the USB board running the ESS synchronously, and figured I should get the clock phase noise as low as possible. This seems to result in really smooth sound, but without losing any details or dynamics, and instrumental timbres seem more real.
I'll measure at R7 to get the actual consumption, set resistors are 20 ohms for now, and the max current use for the CCHD-957s is indeed 25 mA.

The SOtM board has DIP 14 sockets for the oscillators. I removed the sockets, and added legs to the pads on the Crysteks. Then I soldered the Crysteks into the (now socketless) board. I did not solder the power pin to the board, instead extending it out, and directly soldering the output of a Trident to it (also added .1 uF ceramic right at the power pin, grounded to the SOtM ground plane at that point). Felt good getting rid of the sockets. One thing I noted on the SOtM board is that the clock outputs go both to the XMOS chip, and to another IC right at the board output. Looks like SOtM reclocks the data right before the output-I know Gordon Rankin's XMOS based solutions do the same thing.
Then I applied a separate +5V supply, with a Dexa linear reg to feed the two Tridents. All is working, and sounding, fantastic. I just needed to figure out he consumption to purchase a small transformer to fit in the case.
~1.16 V across 20 ohms, so figure ~60 mA per Trident. So 120 mA load on the pre reg, requires, say 240 mA in max. Looks like 5 VA or 6 VA transformer will be plenty.

Edited by user Wednesday, November 16, 2011 6:22:23 PM(UTC)  | Reason: Not specified