6890N noise issue

[Jake]

Hi all,

Seeking your sage advice on a noise issue with our Agilent 6890N. Last time we used it 4 months ago for cis-trans analysis with a SP-2560 column it was working beautifully. We left it on in a low-temp sleep (all thermal zones 30-40 C, low column flow and make up gas). This approach has served us well over many years.

Bringing it up, we're confronted with high signal noise at low temp. with signal decreasing with increasing temperature; that is we see extremely noisy peaks for compounds early, which completely disappear. Background level is typical. Here's what we've tried:

Gases: Changed air from Zero Air generator to tank; N2 and H2 carrier and fuel are from tank and generator, respectively. Those gases are shared with other GCs and they're all performing as usual.

Detector: Swapped jet and entirely collector assembly, which included new teflon collector isolators. Inspected electrometer contact spring, tightened all connections. Nothing obviously wrong.

Inlet: Replaced septum, O-ring, liner. Nothing out of the ordinary. Inlet is very clean.

Leaks: Checked everywhere for leaks - column, all fittings everywhere, etc.

Any ideas? Faulty electrometer? I'm hopefully linking an example of the problem. Green is previous, normal. Ignore the cycling baseline drift. That's from the Zero Air generator cycling. Blue is current problem. This is of the same sample mixture.

Here's the link to the hosted chromatogram pic:

https://imgur.com/a/ykc3svm (couldn't get the image to directly display)

[Jake]

To update: It's a dual-channel system, of which we only currently use the back channel. I don't recall why we abandoned the front channel; presumably due to a performance issue. We swapped the electrometers including the interconnects and FID control boards. The problem persists. We also did the Agilent electronic leak check test, and it passed without issue.

Best we can tell, the spring and interconnector for the 6890 and 7890 are the same (the electrometers obviously are not interchangeable). We have a 7890 FID not in use from which we'll swap the spring and interconnector.

Is there any chance this is due to a faulty detector EPC? We checked all the flows with a Restek flow meter, and they all seem consistent and change appropriately with changing values.

Running out of ideas...

[Peter Apps]

That looks like electical noise - somewhere there is a loose connection. Since it improves as the oven heats up I would suspect the heater coil or maybe something to do with the fan.

Peter

[rb6banjo]

I assume that those injections you shared were real (some liquid was injected into the system).

What do you see if you queue up an injection but just hit start on the GC. No actual injection. Is it still noisy in the beginning and seemingly getting better as the oven program continues?

[AICMM]

Jake,

OOOHHHH this is a good one!

The wave in the baseline makes me suspicious of your gases. I have seen this with a poorly performing generator, in my case it was zero air, but it could be H2 as well. (I did hear you about other GC's.....) So the other thing I would wonder about here is the status of the EPC that feeds the detector and the EPC of the inlet.... Since you have swapped electrometers, consider swapping EPC's...? Also, does this instrument require a higher pressure than others such that line pressure is high enough?

The noise in blue makes me suspicious of the seat of the chimney. Don't overtighten this - it does you no good - and can cause issues.

The peaks in the black trace have width. This is not usually the case with simple electronic noise - it is typically very sharp. Also making me suspicious of the detector EPC.

Some things to think about.

Best regards,

AICMM

[MichaelVW]

For clarification, what's green and what's blue?

The oscillating signal... that's not a test plot, is it? 6890s can be set up to do that.

Do you see the same thing with constant pressure and constant flow mode?

Most my problems with FIDs have been because the spring was touching something it shouldn't. Sometimes I test with a multi-meter to make sure everything is isolated when I put it together.

Baseline/background dropping with increasing temperature could be a leak that gets better as a ferrule or something expands.

[Jake]

Hi all,

Thanks for responses and suggestions. The traces are real standard mixtures, not a test plot. The sample is just a standard mixture from NuChek we've been using for over a decade to evaluate instrument performance. The green trace is from months ago. Please ignore the undulating baseline. This is completely normal for us, due to the Zero Air generator cycling. We can eliminate this entirely as necessary by switching to air from a tank. This isn't related to the problem. The blue trace is the standard mixture this week (on an air tank).

On further investigation, what's really happening is that it's not just the noise decreasing with temperature, it's that the entire signal is being attenuated as the run progresses. Note in the attached file, in the red trace from this week ("Std mix - current"), once the noise decreases to previous levels, all the FAME peaks disappear too. This definitely isn't a problem with the sample. And on repeated injection, the time/temp. at which the peaks disappear varies randomly. As requested, the blue trace is a 'blank' injection (empty vial)

Chromatograms: https://imgur.com/a/xyG2hGw

[Jake]

Thanks all. Mystery solved. The 100-m column had a crack near the detector.

[MichaelVW]

Thanks all. Mystery solved. The 100-m column had a crack near the detector.

Out of curiosity, how did you find it? Did someone actually see it?

[Jake]

Thanks all. Mystery solved. The 100-m column had a crack near the detector.

Out of curiosity, how did you find it? Did someone actually see it?

Right - and on a 100-meter column...

We were pulling the column off to do detector flow/pressure checks when we noticed just above the column install septum on the oven side of the detector, it was a bit at an angle. Upon close inspection, it was clear it had cracked there.

Really bizarre. We'd used for many years without issue. No idea why it cracked.
In hindsight, behavior all makes sense.

Trimmed it up, re-installed, data looks great.

Definitely something to keep in mind.

[Peter Apps]

Thanks for the update on the cause - in my experieince silica columns don't crack, they just break completely.

[Jake]

Thanks for the update on the cause - in my experieince silica columns don't crack, they just break completely.

Right Peter. My experience too. Hence our puzzlement. All Agilent wax fused-silica columns we use snap very cleanly when they break or when you bend them. We had one break not long ago, and it was obvious what the problem was.
This is a Restek RT-2560 column, also fused-silica. But physically, it's very odd. You can severely bend short runs of it without it seeming to break. When the person re-installing showed me this on cuts from the column, she and I couldn't believe it. I even had to convince myself this is actually a fused-silica column. Maybe Restek is using an extra-thick polyimide coating or something?