Agilent 5890 FID Detector problem- small peaks dissapear

[wrightanalytical]

Has anyone ever had a 5890 FID problem where spectra of trace impurities in an otherwise pure volatile compound rise sharply out of an overly flat baseline? In fact a calibration curve with a standard impurity hits zero% even when the known wt% impurity is still 0.05%. When we did this analysis 6 months ago the calibration curve went through the 0,0 origin as it should, and the baseline looked like a real baseline. Same problem with an added inert hydrocarbon impurity. Large peaks are in the correct ratio. It looks like peaks under 0.05% are lost under an artificial baseline! However, peaks are very sharp and well separated. I have balanced all the gas flows, changed all seals, tried changing the column, taken apart and cleaned the detector. I have no leakage current when the flame is off, and when the flame is on my current is about 5-6 pico-amp (everything in the Agilent document http://www.chem.agilent.com/cpdocs/a16024.pdf) . I am wondering if the electrometer or associated electronics is bad. Any way to tell? Any way to correct? Sadly the instrument is no longer supported by Agilent. Is there any way to get replacement electronics? OR have I missed some really stupid thing?

[Schmitty]

"overly flat baseline" ?

Maybe the attenution is set too high, or the range or the zero.

Those are my guesses. Try setting each to zero and try again.

[wrightanalytical]

Thanks, Schmitty, but We have the analogue signal going straight to "peak simple" A/D converter and then to computer. This data acquisition system is working fine on the TCD detector

[GasMan]

It is a long time since I used a 5890, but I remember that there can be problems with the 'Zero' function for the signal. Check the Zero value for the signal assigned to the FID. You may find that the value stored in the GC is higher than your actual baseline value, hence the straight baseline. It may be better to turn the Zero OFF for the FID signal.

Gasman

[wrightanalytical]

Gasman, Thanks. That is a really good idea. Offset is 27. I will set to zero. Will find out if it works Monday. Thanks again! Antony

[wrightanalytical]

Yes. This was the problem! This falls into the category of missing the obvious. I suggest anyone alternating between a TCD detector (where using a signal offset is often convienient) and an FID detector (where setting a signal offset can cause the problem I encountered) on the same analog signal output, put a label on the panel next to "Signal 1" marked SET OFFSET= ZERO FOR FID