Chromatography Forum

Hello everybody,

Is there any test can be used to evaluate how accurate the mixing of a gradient pump is?

Thank you in advance

Easiest way is to run a "step" test. You can find a quick description at the end of this thread:
http://www.sepsci.com/chromforum/viewto ... dient+step

More thorough discussions and examples are in several of John Dolan's troubleshooting columns in LC/GC:
Dolan, LC/GC, 6(7), 572 (1988)
Dolan, LC/GC, 17(6), 412 (1993)
Dolan, LC/GC, 13(6), 456 (1995)
Dolan, LC/GC, 13(12), 940 (1995)
Dolan, LC/GC, 14(4), 294 (1996)

Please refer to the previous discussion on this topic "HPLC gradient pump OQ" number 1927. The basic idea it to make sure that your injector, detector and isocratic pump performance are all ok. Then run a stairstep gradient of water versus 0.1% acetone and compare expected versus observed absorbance.

Thank you all I'll try the test.

Dear all,
I applied the Gradient Performance Checks described by Gilroy and Dolan in LC.GC Europe 17(11) 2004 on my quaternary, low pressure mixing gradient pump. The reason for doing these checks was to identify the cause of getting lower resolution and retention times when mixing the eluent in the pump through channel A (water) and B (acetonitrile) compared to what I get normally by using a premixed eluent (40% Acetonitrile/Water).

In the Gradient Proportioning-Valve test (GPV) (Channels A, B= Water, C, D=1%Acetone), the average difference between the maximum and minimum plateau heights was 4.4% which is close to the 5% allowable maximum mentioned in the article.

For the Linear Gradient Test, I got two slopes for the gradient line when running Channel A=Water, B=Acetone. The line started steeper to about 20% B and less after that to the end of the line.

When doing the Step Test for channel A=water, B=Acetone, I got steady baseline at about zero for time lengths of about two steps then a high step but with no clear plateau which represents the 30% B (I think). The chromatogram is shown in this link:

http://www.albumtown.com/showpic.php?ai ... pid=516373


I thought that one of the two channels at least must have a problem. Therefore, I repeated the step test with the other channels as follows:

Water/ Acetone
A/ C
A/ D
B/ D
D/ B

However, all the other previous channel combinations produced clear steps from 10% to 100%. Despite that, all the combinations generated lower practical %B than the theoretical programmed %B. In addition, all the combinations also failed to achieve less than the maximum allowable1% step size difference (the practical step sizes were lower than the programmed from 0%B to about 70%B and higher afterwards).

I am wondering about the following:

What may have caused the previous effect in all the three tests?
Why only A vs B channels had the fused steps at the beginning?
What can I do to solve this problem? (Throw the pump?! )

Thank you all in advance and sorry for this LONG STORY!

You probably did this, but I wasn't sure from your post: did you go back and re-run the A-B step test after you had run the other combinations? (just to check that the problem wasn't an intermittent one that went away by itself).

If the results are the same, then it is definitely time for factor service on your pump.

Thank you Tom,
I had the same feeling after I ran several combinations, so I repeated A vs B again and got the same bad mixing again.

What kind of service do you think it will need?

Thank you again.

Without knowing details of the proportioning system design (one valve, four falves. low pressure metering pump(s), . . . ?) it's hard to say. If you have access to another system of the same model, you could repeat the test on that system to see what it should look like. If you have more time than money, you could then begin exchanging subassemblies between the two systems.

My first guess would be the proportioning valve, but that's only a guess.

I would agree with Tom - get the proportioning valve checked out. It could be you are getting leakage across the A and B ports. Something else could be that because of the way the proportioning works, the valves are sticking open/closed when trying to do fast changeovers (low % B).

Thank you Tom and Tim.

Do you think the proportioning valve may be responsible for getting the lower percentages compared to the programmed ones ( in all the four channel combinations)?

Possible, but you won't know until you fix or replace.