Hum problem in DAC - weird behavior - need help

kumori

Member
2011-03-04 6:15 pm
Hello

I have a hum problem with my ES9018 based DAC.
After startup, the system is basically hum-free. After about 10 seconds the hum (around 100 Hz by ear) starts. It is only hearable when I turn the pot over half, so it is not that bad.

The funny thing is: when I touch the ground pin of a certain regulator in the DAC, the hum is notably reduced. When I touch the regulator very gentle, only a little bit, bet when I touch the regulator more firmly, the hum gets more quiet.

It also works when the regulator is not even powered...

These are the components used:
->USB feed via waveIO
One regulator for the dirty side, one regulator for the clean side after isolation. The reg for the isolator is also the "funny" reg which - when touched - reduces the hum.
->Buffalo III DAC feed from two independent supplies for AVCC and for the rest of the local regs.
-> SEN I/V stage with four independent supplies and floating ground.
-> transformer coupled to preamp.

So, I asume the fact that the hum starts after a certain warm up time, has to do with something charging. Or?
Why is it that when I touch the regulator PCB (at the backside ground pin) for the WaveIO isolator with my finger, that the hum gets quieter? Can I use that somehow?
The regulator doesn´t even have to be powered.

There should be no global ground loops. The USB side is isolated by the WaveIO, the Preamp side is isolated by transformer. All regulator feeds are galvanically isolated (I use 8 independent secondary windings at the moment)
I tried different combinations of connecting the WaveIO dirty side to the chassis and/or connecting the DAC ground to chassis. The hum stays.

Any help is appreciated.
Thanks
FLorian
 

kumori

Member
2011-03-04 6:15 pm
There may not be a ground loop as you said, but a system usually likes it when a point somewhere is tied to mains earth.

Yes, I had that. The DAC ground was tied to earth by default. I also tried the combinations of earthing the USB dirty side alone and toghether with the DAC ground. No effect. I also earthed the system at the"funny" regulator GND instead of the DAC GND. No effect.

I was thinking: The human body is an RC network, is there any "best practice" approach to use capacitors and resistors to ground to maybe emulate the "human finger touching the regulator GND"-effect?
 
I was thinking: The human body is an RC network, is there any "best practice" approach to use capacitors and resistors to ground to maybe emulate the "human finger touching the regulator GND"-effect?

No, that is not the answer. You either have a design fault or a construction fault, probably invloving ground.

As abraxalito has already requested you to do, please provide a schematic or wiring diagram if you want our help to be useful. If doing that is too much effort for you, then helping you is too much effort for us.
 

kumori

Member
2011-03-04 6:15 pm
No Problem, here is a sketch of the setup. I hope the ground connections become clear. The red voltage-reg at the WaveIO is the one that influences the hum. Only one channel is shown for the SENs and transformer to preamp. The other one is identical. What further informations do you need?

I also attached the SEN schematic that I used. The SEN supplies are floating and not referenced to the DAC ground.
 

Attachments

  • DAC hum schematic.png
    DAC hum schematic.png
    372.1 KB · Views: 125
  • SEN IV ES9018 Schematics .pdf
    163.9 KB · Views: 18

kumori

Member
2011-03-04 6:15 pm
Try 1 supply for all 4 SENs

Or the most straightforward way is to connect whatever you're touching to mains earth.

Each SEN needs its own supply. They may not be referenced to each other. This can not be worked around. The SENs are designed that way.
I tried to earth different things... That has the effect that the hum varies very, very slightly. But I wouldn´t call it progress.
 

wwenze

Member
2008-03-07 12:46 pm
I see. I admit, I haven't seen the SEN circuit until just now.

Say, is the SEN meant to be used in this way to generate +Out and -Out? I see for each individual SEN the Vout = voltage across R_iv, so unless the current itself is humming the output would have no hum. But with your configuration, the output voltage = voltage across both R_iv plus whatever potential difference there is between the grounds of each SEN.

And then you said touching the ground somewhere the problem goes away.

Can you make the grounds within each pair of SEN as close as possible, and then connect them to earth?
 
The SEN circuit features a floating power supply, which is not a problem if the signal has a path to return to ground at some point. Which it does here through that 680ohm Ri/v resistor.

Okay, a couple of questions:

1. Does the hum go away with the USB cable disconnected?

2. Are the regulators mounted on the same PCB with the DAC chip, sharing a common ground plane? Or, are they mounted on seperate PCBs? Same question for the SEN I/V circuits.
 
Last edited:

kumori

Member
2011-03-04 6:15 pm
I see. I admit, I haven't seen the SEN circuit until just now.

Say, is the SEN meant to be used in this way to generate +Out and -Out? I see for each individual SEN the Vout = voltage across R_iv, so unless the current itself is humming the output would have no hum. But with your configuration, the output voltage = voltage across both R_iv plus whatever potential difference there is between the grounds of each SEN.

And then you said touching the ground somewhere the problem goes away.

Can you make the grounds within each pair of SEN as close as possible, and then connect them to earth?

Yes, the SEN is the I/V converter after the ES9018. The ground layout of the two SENs is already quite tight. The two phases of each channel are actually on one PCB.
 

kumori

Member
2011-03-04 6:15 pm
The SEN circuit features a floating power supply, which is not a problem if the signal has a path to return to ground at some point. Which it does here through that 680ohm Ri/v resistor.

Okay, a couple of questions:

1. Does the hum go away with the USB cable disconnected?

2. Are the regulators mounted on the same PCB with the DAC chip, sharing a common ground plane? Or, are they mounted on seperate PCBs? Same question for the SEN I/V circuits.

1) No, the USB cable has no influence.

2) The WaveIO regulators are mounted within 2 centimeters of the respective Voltage inputs. The Buffalo has its 5 local regs mounted right on the PCB. They are fed by two separate pre-regs. In my sketch I omitted the local regs. They are just another regulator stage and directly connected to the DAC ground plane.
The regs for the SENs are salas BiB Shunt regs, they are on separate PCBs.

I investigated a little further. The Salas regs take a little while to get to full output voltage because upstream is a CLCRC network after the rectifying diodes. The hum starts exactly at the time where the Salas-shunts reach full output voltage (around 18,94 volts) after around 10 seconds. So I guess I will try to play around with the Shunts a little bit. Maybe add more capacitance directly before the shunt...

But if the shunts might be the cause for the hum, why can it be reduced by touching an unpowered regulator, which is physically 15 centimeters away from them? I am still puzzled...
 
...But if the shunts might be the cause for the hum, why can it be reduced by touching an unpowered regulator, which is physically 15 centimeters away from them? I am still puzzled...

You should eliminate the possibility that one or more of the shunt regs. is at fault. I'll assume that you do not have access to an o-scope. Disconnect all of the shunt regs., then connect a D.C. bench power supply set to the appropriate voltage(s) in their place. If the noise is mostly gone, then something is likely amiss with one or more of the shunt regs.. A finger placed on a component affecting the noise is a haphazard response, and doesn't necessarily point to the source of the fault. You need to methodically troubleshoot the cause.

You likely won't be able to troubleshoot the discrete shunt regs. without a scope, so I would suggest utilizing some simple 3-terminal IC regulator based supply in their stead. Don't worry now about whether you would be losing the best possible supply performance. The only thing that matters at this point is getting the circuit to function without obvious fault.

If the shunt regs. don't apppear to be at fault then you may need access to an o-scope to track down the cause.
 
Last edited:

wwenze

Member
2008-03-07 12:46 pm
Agree with checking the Salas too after reading the hum starts only after 10 seconds.

Since you have 4 of them you can just measure the voltage in-circuit, the one with the stranger DC voltage is... well... strange. Or you can simply measure the AC component.
 
Last edited:

kumori

Member
2011-03-04 6:15 pm
Ahh, yes.

And indeed it was a grounding issue.
I forgot about the Vref ground of the SEN. With a buffalo DAC Vref should be Vavcc/2. The ground reference for Vref is best taken from the Buffalo Vavcc main ground right after the smoothing capacitors (at the PlacidHD or whatever someone is using). NOT from the local ground next to the Vavcc output at the buffalo. Rooting a wire to the main ground creates a larger loop are (or wire length) of the hookup wire, but this does not matter. It took me a while to figure this out, but with this little change I have a completely noise- and hum-free setup.