Hi guys, I've been reading this thread for long time, and heavily modded my DAC according to the suggestions here. Now I'm using UTC A-20 transformers. And one thing that is bothering me, is 50Hz hum. When DAC is off it is really audible, when it is on, hum is barely audible, at the same volume, but is still there. If I touch both DAC and amp cases it goes away. It is ground loop isn't? Trafos are not connected to ground, + and - terminals go to the RCA which are isolated from case. Maybe I need to isolate whole DAC board and power transfomer from the case?
Bill, if I remember correctly you wrote several posts about this, that we don't need to connect negative terminal to the DAC ground, but what will happen if we do?
Thanks,
Fedor
Bill, if I remember correctly you wrote several posts about this, that we don't need to connect negative terminal to the DAC ground, but what will happen if we do?
Thanks,
Fedor
That's a real can of worms. Thinking logically it doesn't make sense, but I know strange things happen. Noise can be capacitively coupled through the trafos, but unless you have a connection to ground on the secondaries it cannot be a ground loop. Have you tried moving the dac away from all other equipment, the trafos could pick up stray magnetic fields from other sources. I would suspect your interconnect cables are at fault, too long maybe, or a problem with your preamp.
If you connect one side of the output to ground you will be creating a path for a possible ground loop, but it will work OK. Do not ground the output to the dac board, and don't ground the dac board to the chassis.
Let us know how you make out.
If you connect one side of the output to ground you will be creating a path for a possible ground loop, but it will work OK. Do not ground the output to the dac board, and don't ground the dac board to the chassis.
Let us know how you make out.
Thank you for your answer. Actually, I've tried few things, like taking power trafo out of the case, or holding output trafos in the air, so they won't touch the chassis, nothing helped. The interconnects are changed too, now they are 20cm long, so it can't be length. Interesting thing though, this problem appears with both of my DIY amps, but not with Yamaha RX-350 receiver. Maybe it's indeed problem with amplifier...
Anyway, thanks for the suggestions, I'll try them in a couple of days and post back.
Anyway, thanks for the suggestions, I'll try them in a couple of days and post back.
I don't know anything about the UTC transformers, but in a general sense, nickle core transformers, and most signal level transformers are nickle core, are VERY susceptible to power transformer fields and really, any AC wires within about 2 feet of them. What you describe sounds like stray field coupling to the core of the buffer transformer. Start turning things at different angles, with respect to power and buffer. If it is an air borne field coupling between them, or between the buffer and any other source, a vector change will show up in the scope. You have to think in 3 dimensions for the transformer rotation with respect to outside fields. If a change does not show up on the scope, then you probably have a ground loop in front of the transformer primary.
Good point about a possible ground loop if the board is grounded to the case.
The A-20s are in mumetal cans and are pretty well shielded, I got to mess around with a pair.
The A-20s are in mumetal cans and are pretty well shielded, I got to mess around with a pair.
Of the half dozen or so trafos(shielded and unshielded) that I tried, under the same circumstances, the A-20s were the ones which produced the least amount of hum.
Cans are your friends! Too bad that the only supplier left in the US requires a $1k purchase. You can use regular cold rolled steel, but you will want to have it heat treated to the Marstenite / Osteonite temperature to make it ductile again.
22 gauge buys you about 30 db of suppression and 16 gauge will get you 100 db.
22 gauge buys you about 30 db of suppression and 16 gauge will get you 100 db.
I'm using those Tams I bought in my 2nd dac, I rotated them 90* and that got rid of the tiny hum I had originally.
In my old drag racing days we used to cook ring and pinion gears in self cleaning ovens to soften them up a little, stunk like hell, but it worked pretty good. A little OT.
In my old drag racing days we used to cook ring and pinion gears in self cleaning ovens to soften them up a little, stunk like hell, but it worked pretty good. A little OT.
Thanks for the answers guys. I will try rotating the trafos. And will check for any ground loops. As well as taking away the power trafo.
Last edited:
Comparison before after modifications
I've put up an comparison between an unmodified Gigaworks DAC (with a DIR9001) and a modified one where the second opamp was eliminated, input caps removed and filter changed to same values as Cirrus Logics example. The opamp is also replaced with a GD-Audio Moon discrete opamp. You can check it out here: Gigaworks DAC before after compare | Jacken's Blog
I've put up an comparison between an unmodified Gigaworks DAC (with a DIR9001) and a modified one where the second opamp was eliminated, input caps removed and filter changed to same values as Cirrus Logics example. The opamp is also replaced with a GD-Audio Moon discrete opamp. You can check it out here: Gigaworks DAC before after compare | Jacken's Blog
That looks like a decent alternative to store bought opamps, very reasonable pricing. I think I would go for a better PS for it too.
Have you tested the CS8416, most are using it.
Have you tested the CS8416, most are using it.
Nope, I don't have one unfortunately. But I'm in the process of buying a newer version (with digital oversampling) of the Gigaworks DAC, so I will test that one as well.
On my version the opamps get their juice from 78/7915, so changing to better ones is on my list. I'm very curious if this will show up on the measurements.
If you follow the 78/7915 with a capacitance multiplier stage and you decouple also the OpAmps power pins to the ground with good capacitors you will get the best performance.
