Advice Needed For Test Bench Setup

[jazbo8]

While working on my DIY curve tracer project, I discovered that my shop bench has a serious problem with EMI, for an example of what I was seeing, please see: DIY Curve Tracer

The EMI is really bad during the "business hours", 8AM-8PM, which is just unacceptable, so I need some advises on how I can isolate the source of the problem and improve my bench setup, particularly AC mains filtering, power supply and perhaps RFI shielding. Here is what I have tried so far (none seemed to help):

- AC line filter (Corcom) and/or isolation transformer for the adapter, the scope (use as the XY display for the adapter);
- DC supply for the filaments;
- Turning off all the electrical appliances in the vicinity of the shop;
- Ground connections in various ways.

And a follow on question - what is the generally recommended practice for AC wiring for the test bench? Isolation transformer/filer for the whole bench? Individually isolated/filtered outlets for the test equipment? etc.

TIA,

[bear]

I saw your scope pix in the other thread... looks like the jitter in the display is being passed through the grid drive circuit. Suggest a passive LP filter on the grid drive first. It's a pretty slow signal, yes? See if that gets rid of the crap or not.

The other thing is that the crap has to be getting into the grid drive circuit somewhere. I'd set the grid circuit to a *fixed* level, as far back in the circuit as you can go, and look at that now DC voltage on a scope during the time when it is worse. Try to see what it looks like frequency wise, and if it is periodic or random.

IF it's in the audio band, perhaps you can capture and run an FFT of it... even *listen* to it!!

If then, it is *not* there when you set it to DC, meaning that is after the ramp generator (assuming that is what is being used, I did not look back at your circuit), then the place to look is in that circuitry...

Assuming you have the circuit in an enclosed and grounded metal box??

Just my random instantaneous thoughts on the matter...

[cihtog]

What's on the other side, or within the walls, closest to your bench?

Have you tried using a pickup to locate the source?

[jazbo8]

Originally Posted by bear

I saw your scope pix in the other thread... looks like the jitter in the display is being passed through the grid drive circuit. Suggest a passive LP filter on the grid drive first. It's a pretty slow signal, yes? See if that gets rid of the crap or not.

I am dealing with two sets of problem - one is as you pointed out, a design flaw of the the tracer adapter, which I have been working on and off for awhile and your suggestions will come in handy. The other issue is the EMI/RFI - I would like to find out where all that crap is coming from, then how to get rid of it - I mean, how can the exact same circuit have such drastic different responses? The only difference is the time of the day as far as I can tell and the experience is repeatable and not just a random event, something has to be dumping all kinds of crap onto the line (or RFI as well I suppose), no?! It's really driving me crazy...

Jaz

Originally Posted by cihtog

What's on the other side, or within the walls, closest to your bench?

Have you tried using a pickup to locate the source?

I was thinking about that as well, there is a security office next door with surveillance monitors and small communication (walkie-talkie) system installed, but they are on 24/7, so I don't think the EMI or RFI is coming from there, but worth looking into I suppose.

What pick up do you suggest? Will a portable AM radio do? I may have one somewhere...

[Speedskater]

An AM radio tuned off station is very good for a wide band of RFI. Some legacy guitars and microphones are also good at lower frequency noise.

See page 31 of Jim Brown's paper:
http://www.audiosystemsgroup.com/SurgeXPowerGround.pdf

[cihtog]

Originally Posted by jazbo8

...What pick up do you suggest? Will a portable AM radio do? I may have one somewhere...

Any humbucking guitar pickup will do - they cancel out typical hum/noise, but aren't immune to EMI. The closer they get, the louder they HUM/BUZZ.

Old CRTs are notorious for EMI emissions, and if they have a wall of them, you may have to resort to changing the layout of your shop, or maybe some kind of EMI screen on that wall.

EDIT: You may even have luck making one, with a bar magnet and some wire - just make sure they are wound correctly.

[jazbo8]

Originally Posted by Speedskater

An AM radio tuned off station is very good for a wide band of RFI. Some legacy guitars and microphones are also good at lower frequency noise.

See page 31 of Jim Brown's paper:
http://www.audiosystemsgroup.com/SurgeXPowerGround.pdf

Found it on p.30:

"MAGNETIC FIELD PROBE
Effective probes for audio-frequency magnetic fields can be made from a CRT degaussing coil (the kind sold for use in a TV service shop), or from a good dynamic microphone that has no hum-bucking coil and relatively little magnetic shielding (an EV 635A, for example), or even a telephone pickup coil from Radio Shack. Simply connect the probe to the input of a battery powered microphone preamp that can drive headphones, and walk around listening for the fields. A user receiver for an inductive loop hearing impaired system is also a convenient and inexpensive probe."

Thanks,

[bear]

I'd suggest a more empirical and analytical approach.

As I said, find a place in the circuit working backwards from the display that you can *kill* the "jitter" (interfering signal)... or the other way, a place using a scope where there is no jitter apparent...

also try to figure out what that signal looks like and what the approximate frequency range it is in... your 'scope is a good sniffer too...

Ok, IF there is an external signal entering at *one* point and at a low level you have the option of creating a nulling circuit. What you want is an antenna of some sort to pick up the signal, then inject the *inverted* signal via an opamp of the right type, adjust the output level, and voila! Nulled... when there is no RFI signal then the circuit injects nothing.

Of course it would be better to identify exactly what the RFI looks like, what freq it is, and then see if it can be eliminated by preventing it from entering, or perhaps redesigning that part of the circuit so it is RFI immune, and/or balanced operation, so that common mode signals are nulled.