TCD Noise - Low Frequency

[mattj63]

I have an Agilent 6890 GC with TCD which has a wandering baseline. The integration picks this up as many peaks.

I have installed a new TCD cell, swapped EPC modules with a spare and plugged the column inlet connection and I still have the same behavior. Therefore it doesn't seem related to the column being dirty. Another GC using the same He source has a beautiful baseline which remains stable for 10s of minutes.

Does anyone have any suggestions on what might be the problem?

[aldehyde]

What is your sample rate for the signal in your method? The signal should be acquired at 5 Hz and if its set for 20 or 100 Hz in the method you will get extra noise.

How much is the baseline wandering? Could we get a picture?

How long ago did the problem start? You mention that another GC is using the same gas supply with no problems, is there a filter on the gas line and if so is it near the supply or is it individual to the GC?

What kind of analysis are you running and have you tried baking the TCD?

[mattj63]

I'm sampling at 20Hz but the noise is low frequency and I've always run at 20Hz. The baseline wanders up to 0.5 uV over the course of a couple of minutes even changing to a slower data rate won't affect this. The high frequency noise is <0.1 uV which is about normal for our GCs - maybe a little worse.

The problem started several months ago. This GC is not used daily so it's hard to put a date on it. Originally I thought it was a dirty detector so I baked the detector. There was no improvement. Also, I have switched out detectors and a new one behaves the same way.

Other attempts: 3 different EPCs, plugged column inlet, swapped Pneumatics control board with a known good one, swapped TCD signal card with know good one, beat on GC with a wrench, mild profanity.

The other GC is located in another room but fed from a common tank. I suppose it could be a leak somewhere in the lines to this GC. Maybe I should bring a tank right to the GC to test.

I'm working getting you a picture. I guess there are instructions somewhere in this forum.

Matt

[aldehyde]

It sounds like you've done a pretty good job eliminating possibilities. I would definitely recommend eliminating the possibility of a leak or something in the lines by plumbing up a known good tank directly (if not horribly inconvenient).

For the picture just hit print screen, open paint (start > run > mspaint) and paste the screenshot in, you can crop out everything but the chromatogram if you'd like and then just upload it to like www.img.cx

It doesn't sound like reducing to 5 Hz sample rate is going to do anything for your problem but the detector physically samples at 5 Hz (the clicking noise) so a sampling rate higher is going to do absolutely nothing but introduce noise. FID and other detectors are have a bit more flexibility for sampling rate.

I've gotta do a bit of driving but I'll check on this thread a bit later once you get the picture posted.

[aldehyde]

Oh one more thing before I go, TCD can be very sensitive to air currents in the room. Sometimes when I am running noise tests in busy labs I have to put signs up so people stop opening and closing the doors (resulting in nice big spikes on the TCD a few seconds later).

It sounds like your wander is more of a meandering baseline than sharp noise peaks but try setting something like a shoebox over the detector and see if it improves at all. --NOTE that you should obviously not leave the cardboard sitting on a hot detector as a permanent solution and I'm not responsible if you put cardboard on your GC and leave it unattended --

If we figure out that it is just a drafty area in the lab Agilent sells a little adapter that you can put on the exit port of the TCD to help reduce any pressure fluctuations.

[mattj63]

It is in a drafty area - actually a hood because we're running a toxic chemial on it. However it's no different from 6 months ago when things were OK. Also, I tried covering up the detector outlet as you suggest and also connecting some tubing to the outlet so it vents in a more calm area. I don't think that is the problem.

I'm hoping it's a leak somewhere in the He. I'm rounding up a spare regulator so I can hook a He tank directly to the GC with no extra tubing.

Thanks for thinking about this.

[aldehyde]

First: what kind of tubing did you connect to the detector vent? The little kit from agilent consists of swagelok reducers and a restrictor, that vent gets pretty hot and if a cardboard box doesn't fix it tubing won't help.

From the image you posted, the initial jump in signal is just from opening chemstation right? The instrument has been on for at least a few hours and is equilibrated? A baseline in the mid 20s is mildly suspicious alone if its not still calming down.

Of the things I can think of that cause this problem there are:

- dirty gas/traps (going to evaluate with clean, close by cylinder)
- column bleed (nope)
- leaks (going to evaluate with the direct cylinder connection but when you connect the cylinder go ahead and block off the TCD with a column nut so we can purely test the detector)
- EPC module/proportional valve (checked by you when you tried swapping EPCs)
- supply gas pressure control (probably not the problem since you have other systems on the supply with no problems)

You have pretty much checked out the detector electronics, what about the actual electronic supply to the instrument? Is this on a shared circuit or is it on the recommended dedicated 20 A circuit? (probably not ) Check the line voltage with a multimeter and make sure you are getting 120 VAC stable. If the instrument is sharing a circuit with other stuff and the line voltage is unstable or insufficient try turning some other stuff off while you're testing.

Earlier you say you "switched out the detector," do you mean you replaced the entire detector cell including the filament etc and this is still happening?

One more thing about the sampling rate:




peaks only .01 minutes wide can SLIGHTLY benefit from running at 10 Hz but you should never run at 20.

[mattj63]

Dropping the data rate to 5 Hz certainly helps the high frequency noise but the signal still drifts around. I always figured oversampling can't hurt (plenty of disk space) but from what you've shown me, I'm going to drop the sample rate to 5 Hz on all our TCDs. The scan above was with a tank connected directly to the inlet of the GC manifold (There are some Ts to split He to inlets and other detector but basically just copper tubing). I double checked all the connections to make sure they were tight.

Yes, I am waiting for the temperatures to stablize (usually). I've compared signal right after starting TCD with waiting 1 hour and it's basically the same.

I'm bothered by the high (>20) signal also. Our other TCDs run 10-15 uV. The He line is shared with another system (unbeknownst to me) but the direct connection to a new tank should have diagnosed that. Also, if the helium was gettng contaminated, I would think it doesn't really matter since the reference and sample flow is from the same source. BY THE WAY - the scan above, I changed the reference flow to 10 mL/min which dropped the signal to 10 - it goes back to 20+ if I use the normal reference flow.

When I plug the column inlet (just run reference and makeup flow) the signal still looks the same.

I'm beginning to think electrical noise also. I looked at the FID signal and it has some step changes also, but not to the extent that the TCD does. My next step I think is going to be to take the GC out of its current location and put it somewhere else (don't know where - no bench space in my normal lab).

Oh - I connected some teflon tubing to the TCD outlet. Just temporarily to see what effect it had. None really. We've done this elsewhere and it doesn't cause any problems. Teflon can take 250C.

Thanks again. I'm pretty close to calling in service but I'm not sure what they would do. I don't want to spend $$$ having someone try all the stuff I've already done. Unless I've had some sort of brain fart and done something really stupid. Going over things here helps since I'm double-checking things and maybe I did overlook something.

Matt

[aldehyde]

Do try using a multimeter to check your line voltage, if you see spiking/cycling/insufficient voltage it would explain things.

If you haven't changed out the detector cell it could be contamination. When you baked out the TCD what temperature did you use? What temperature do you keep it at for analysis? Acids and particularly sticky samples can affect/damage the filament and change its resistance. Your baseline is slightly elevated, if you run sample through do you see reduced response/sensitivity?

You can check the resistance of the filament with a voltmeter if you are confident in your abilities: http://www.chem.agilent.com/Library/Sup ... a16093.pdf (remember, resistance is measured with the instrument off/unplugged).

[mattj63]

Well, I moved the GC to a different room. Less drafty, independent circuit (nothing else on the circuit) and I'm still getting the same behavior. Could there be a power problem in the GC itself?? Almost time to call for service

[aldehyde]

There are voltage test points on the mainboard that you could check to make sure you're getting the right voltage but at this point you should have a service person come out so they could try isolating the problem further by swapping a few parts. Its really difficult to say whats causing this since you have tried so many things without any difference. Since you have tried swapping the signal board and the pneumatics control I would suspect either the main board or the EPC itself (maybe water absorbed into the frit and is messing with the flow through the detector). Make sure the service engineer brings a flow meter and checks the flow out the TCD vent.

[aldehyde]

There are voltage test points on the mainboard that you could check to make sure you're getting the right voltage but at this point you should have a service person come out so they could try isolating the problem further by swapping a few parts. Its really difficult to say whats causing this because you have tried so many things without any difference. Because you have tried swapping the signal board and the pneumatics control I would suspect either the main board or the EPC itself (maybe water absorbed into the frit and is messing with the flow through the detector). Make sure the service engineer brings a flow meter and checks the flow out the TCD vent.

[mattj63]

I finally called in Agilent service. Stumped them also. They tried the same things already suggested and tried here.

After Agilent left, I swapped the ECD control board for one from another GC. The problem went away. Baseline is low (around 12 uV) and there is no more wander. It seems something on that board is amiss and causing some flow instability. However, that instability doesn't register in any of the pressure or flow readouts that one can monitor in the Chemstation software.

This is the second time I've had a mysterious problem that was solved by replacing the ECD control board. Next time it'll be the first thing I try - too bad they cost $1500+.

[aldehyde]

Yeah once you troubleshoot to a point its good to just try swapping the signal board, can be overkill for a lot of problems but when all else fails its worth a try.

glad to hear you figured it out.