The peaks in online GC are too small

[Bui Thuy]

Hello,
I am using GC 5890 I with FID like online GC. Before it worked well but this week the peaks are too small. When I injected a standard mixture by hand with the syringe directly into the injertor, the peaks are all good.
I also injected that mixture into the tube of argon during the way argon comes to the multi-port- valve. The carier gas (argon) will bring it through the sample loop and then into the injector and the colume, in this case the result is also good.
My sample is OK because I tried to take it in the gas phase and inject directly into the injector and the result is good
I opened the multi-port-valve, checked and saw it works.
But until now only the first peak (C4) is OK and others either disapear or are very small.
Please help me. Thank you in advance for your guidance.
Bui Thuy

[ketone]

probably some adsorption happened on the injector or valve as you said. If I were you, I would replace a new injector to give a shot to see if this works.

[Bui Thuy]

I have just used "Test signal output" programm. Normally if the GC is working well, we should see 3 peaks, each peak is approximately 1/10 the height of the privious peak, with the first (tallest) heak having a height value of about 125 mv at "range2" =0.
But I only see the first peak with a height about 106 mV, the second is very small, just like a not smooth base line and there is not the third peak.
It is similar with my sample. I saw only the first peak C4 and other peaks seem to become smaller with increasing retention time.
I have looked up again with the standard mixture that I have measured and found that the height of peaks are lower than before even I changed the value of "range2" from 2 to 1.
Could you please show me what I should do when the "Test signal output" is not OK like I have.
Thank you very much
Bui Thuy

[ketone]

all you did is just to test the hardware, am I right?
if you "Test signal output" program test had problem now you can track the original problem arised from the hardware maybe detector.