50 Hz interference with DSP

[drval]

I have a very sophisticated DSP package developed in LV that analyzes biomedical signals at 256 sps. In the US the code works perfectly but in Europe, with some laptops, the code does not function correctly and 50 Hz "noise" is present. There are a few interesting wrinkles. On some machines 50 Hz "static line" is punctuated by rhythmic bursts up and down the entire spectrum every 4 seconds. On other systems this bursting occurs every 10 seconds. It requires approx 20-40 seconds of recording to begin to see the bursting. On my laptop I cannot capture this phenomenon. I'm wondering if the issue might not be a disk-write problem, rather than a more pure DSP issue. The other idea is that there is a capacatance discharge occuring due to so
mething hardware related. Any further ideas would be helpful.

The DSP includes real-time JTFA with 128 bins and multiple arrays of time-based filters, generally high order Parks-McClellan FIRs.

[PdB]

I would first look for the obvious. In the US powerAC is 60 Hz, here (Europe) it's 50. Try to look for notch-filters (hardware/software) that you have to change from 60 to 50 Hz.
Good luck, Patrick de Boevere

[drval]

I should have been a little more explicit. There are no notch filters being used -- either for the US or Europe -- as the 3rd party hardware has a hard rolloff for anti-aliasing at 60 Hz. Now under other proprietary software (read I can't see directly what they're doing) the same hardware does not show the 50 Hz activity in Europe. Also mine doesn't show the activity on my laptop, but does when running on other laptops, and the 50 Hz shows up with bursts that "flood" the spectral analysis. What is especially interesting is that the time-period of these "spectral floods" is unique to each laptop. On one laptop the "flooding" happens every 4 seconds, on another every 10 seconds. But, in both cases it takes time for the flooding to occur (generally 20 or mo
re seconds of data recording).

So it's a very interesting situation which I think may be related to the timing cycles of the deployed laptops but I can't figure out why.

[PdB]

Another idea,
Some time ago when I tested a prototype serial-powered experimental sensor design. I also got strange noise signals when connected to my portable, compared to my desktop model.
The signal changed between battery power or mains-power and when the battery was charging or charched. The ventilator-activation could also be seen in the signal. Maybe this is your problem.

Patrick de Boevere

[MattBradley]

My first thought also has to do with the 50 Hz power signal, but in a different way. I have seen this when the PC, instruments, and Device being tested do not share the same exact ground. (In one case it went away when I put them on the same outlet). The effect varied the further I moved the two apart.

Not sure if this helps, but I figured it's worth the try.

[drval]

The serial instrument is not powered by line current but by internal batteries: viz 4 AA batteries. The device being tested would be another human being, as the serial-based instrument is an interface for EEG recordings.

So I don't think that this idea would really apply but thanks for the thought.