I’m stumped. My Buffalo III DAC is sensitive to EMI. When I first built the DAC, it dropped lock a lot. I changed the internal connections to the S/PDIF board to coax and that improved things dramatically but it still drops lock occasionally in response to electrical events like turning on a light. Other than this problem, the sound is great. Here are the symptoms, my configuration and what I've tried so far. Any ideas?

Symptoms: The lock light goes off and the music stops for about a second. The drops are associated with external electrical events like turning lights on and off. I can consistently reproduce the problem (cause the DAC to lose lock) by turning on or turning off the kitchen light switch that controls a bank of T8 fluorescent fixtures fed by electronic ballasts. Other switches to other electrical loads are less likely to cause a drop. The drops occur with all of my sources but seem less likely to occur when sources are connected by TOSLINK. Drops are more likely to occur when a non-active coax is connected to the DAC but not connected to a source. None of my other electronics in the house has ever had an EMI problem.

Configuration:
BIII with 4-channel S/PDIF input module, stereo
One TOSLINK module connected to S/PDIF input 4
Legatto I/V Stage
Placid HD PS, Vout=5.25v, total current=500ma, shunt current=50ma (measured warm)
Placid HD BiPolar PS, Vout=+/-14.8v, total current=255ma each, shunt current=50ma each (measured warm)
SW1: all switches in the off position
SW2: all switches in the off position except SW2:8 which is in the on position
Power to the stereo cabinet is on a dedicated circuit using 12 gauge wire, home run to the service panel.
The enclosure is grounded to mains ground but not to the buffalo or power supply boards
3 sources, a Sony CD player, Squeezebox Duet and a U-Verse TV box

Key things I tried:
1) I tried removing all internal coax from the S/PDIF board and hardwired one coax from the CD digital coax output direct to one S/PDIF board input (eliminating all DAC jacks and internal coax.) I also tried this direct connection to the Squeezebox. There was no change with either source; with either one, I could consistently cause drops with the kitchen light switch.
2) I added a line filter (Qualtek Q467-ND Power Line EMI Filter). There was no change; I could consistently cause drops with the kitchen light switch.
3) I connected the Buffalo circuit ground to the mains/chassis ground. There was no change; I could consistently cause drops with the kitchen light switch.
4) I replaced the kitchen light switch with a high quality one and replaced all the bulbs. (I wondered why this particular load consistently causes the drops.) There was no change.
5) I tried increasing the shunt current on the 5v supply to 60ma. There was no change; I could consistently cause drops with the kitchen light switch.

Any idea what could cause this behavior? What should I try next?

Thanks Russ, I'll check it out.

Also, does it make any sense to run the mains power through a standard 1:1 isolation transformer or would the DC get through?

Thanks Leon. I stumbled onto this site last night. I'll give it a try.

You wrote that it also happens with the TOSLINK module, but less often.

Because interference can be indirectly transferred also by SPDIF cable from a computer/player.
See my thread on similar issue http://twistedpearaudio.....aspx?g=posts&t=2157

I suspect the 4ch SPDIF input board as the receiver of EMI. Without all R3 resistors even
the TOSLINK without signal interferences with attached SPDIF coax cable.

I suggest you to connect only output from TOSLINK directly to Buffalo DAC (without 4ch input module)
and check if drops will stop.

Edited by user Friday, November 25, 2011 1:48:35 PM(UTC)  | Reason: Not specified

Thanks Russ.

Your thought that it may be a source component prompted to test powering the DAC and all sources from the inverter. Same results - it still loses lock when kitchen lights are switched. I think this reconfirms that whatever is happening, the EMI being picked up is through the air, not the mains.

In his post titled "4-Channel S/PDIF Input - interference between inputs / ground / locking issues", Miero wrote "I've found that unconnected input with R2 is interfering other inputs ... As a workaround I've connected/shorted the "+" and "-" terminals of that input. The SPDIF lock is now stable even on 192kHz." I too have an unconnected input. Could the unconnected input act as an antenna? Should I try shorting it out?

(partially off-topic)

What about the oscillations on Trident regulators I've measured earlier ...
http://twistedpearaudio.....aspx?g=posts&t=2127

If they were real, could they interfere with SPIDF input board?

I've grounded case, all connectors are isolated and the signal GND is connected
to earth GND at one place - from the both GND terminals of LCDPS.

Edited by user Tuesday, November 29, 2011 5:24:57 AM(UTC)  | Reason: Not specified

Way too much speculation. :)

But the answer is no - these thing would not be related. For one the SPDIF receiver is powered directly by VD, not a Trident. For another this is not a problem for everyone. :)

The issue here is EMI from a switched mains circuit finding its way into the DAC.

Edited by user Tuesday, November 29, 2011 10:07:37 AM(UTC)  | Reason: Not specified

One thing I have seen people do is connect the - input and GND together for consumer SPDIF on the receiver side. You might want to give that a try as well.

Russ, I don't follow the part about "on the receiver side". Are you suggesting jumpers between the - input and the ground on S/PDIF board inputs?