Waters 2998 and Software problems!!!!

[bookoon]

Hi Folks

I am facing a serious problem with the Waters Empower II software recently installed supporting two new Allaince+2998 HPLCs. I have communicated the problems to waters. There unability to solve this issue till date (I am following up for more than a month now) gives a strong hint that they are as flummoxed as I am.


the mail that I sent to the waters people is appended below.

"We would like to highlight some of the problems we are facing with the new PDA 2998/ empower software.

1. Theoritical Plates
This is very significant. The same sample analysed using the same mobile phase/ column on the same day is showing USP plates half or less in the new systems when compared to the older 2996 or 2487 systems. Both the software are empower 2.

2. Peak Purity
The peak which is showing perfect homoginity in the 2996 is being shown as impure in the 2998 detector. One of the causes may be the lower noise in 2998. However we need to understand if there is any issue when a method from 2996 is transferred to 2998.

3. Apply method set vs. Integration giving different results
This is very intriguing. Let’s say we have chosen a bunch of data from the “injectionâ€

[juddc]

Interesting. A few questions:

1. Is the 2998 on one of your pre-existing Alliance units or a new one?

2. For the theoretical plates, has the chromatography changed at all or has only the calculated result changed? Also, is the method isocratic? (I assume it is).

3. Re: purity: Has your data acquisition rate changed? The default rate on the 2998 is 10 spec/sec and the default on the 2996 is 1 spec/sec. I wouldn't expect this to change your purity result, but a higher acquisition rate combined with higher sensitivity might be enough to put you over the edge.

4. Most importantly in my mind, have you done an IQ/OQ on the 2998, an IQ/OQ on the software, and a PQ on the system prior to running your method?

Your integration issue sounds like empower might be applying default parameters in one case and your (custom) integration parameters in the other, but that's just a raw guess. A software OQ might shed light here as data processing is done in the OQ and is supposed to match supplied data.

A system PQ would also illustrate how your system is working relative to a prior PQ with a 2996 (if one was done).

FYI, I had a 2998 installed yesterday on a 2695 and I'm doing some informal work with it, but it will be assigned no "real" work until fully qualified. I'm waiting on my qualification docs and my own requalification certificate from Waters. (I took their course for qualification of LC instruments and need to take a short exam every 2 years to maintain my own qualification).

Good luck!
Chris

[tom jupille]

re the theoretical plates: is there a chance that one system is set to use width at half-height and the other baseline width?

[bookoon]

Hi Tom and Juddc

Tom, the integration parametrs in both cases were "USP calculations". we normally do not change that.

Juddc/ Chris
1. the 2998s (two) are new ones connected to new Alliances. Both are connected to a same PC. My two other Alliances (one with 2996 and the other with 2487) are connected to a separate PC (standalone software)

2. The method is isocratic. Water: ACN 35:65; 1ml flow; col temp 35°C. 254nm. we use this method for testing new columns.

3. we have changed data aq. rate to 1.

4. IQ/OQ/PQ has been done before we started using (nov '07).

we have not qualified our new Allaince with the older 2996 (but that is not a bad idea; let me check if I can manage to do that soon)

Waters engineer has rechecked the IQ/OQ data about a week back; i am yet to get any feedbcak on that.
He had also updated the service pack of the software; but that didn't help.

I hope this gets solved soon.

[AA]

I have been thinking about this one.

Q1. Does the chromatography "look" like you have lost half of the plate counts? I would expect differences in plate count, but not by half, due to differences in the plumbing of the 2 detectors.

Q2. The 2489 will give different peak purity values because it's noise threshold is much lower. And yes, if you develop and validate a method on a 2996 system that has peak purity parameters written in, you could very well have transfer issues. The 2998 is more sensitive and does a better job detecing co-elution, what you did not see on your 2996, you could very well see on the 2998.

Q3. I tried to replicate this one and could not, my Empower II system performed exactly like one would expect.

[unmgvar]

the plate count calculation is a calculated versus 2 variables.
your retention time and your width.

does your peak elutes at the same retention time? such a big difference in your result is indicating that way. if they do not then it means that your peak shape/width is way off, maybe both parameters contribute?

for a retention time shift check if you have a difference in dwell volumes between both systems. lenght of tubing most probably. do you have the same alliance models for both systems (2690/2790)

if the difference is due to peak shape then 3 things can be the cause:
1. a bad connection somewhere in your post/pre column volume.
2. I.D. of your tubing. check that all pre/post tubing have the right I.D.
3. length of tubing. this one can be divided in to 2 possibilities. you need less or you need more.
generally less is better, but when you have a big difference between the temp. of your solvents outside in the room and of the oven you need more in order to bring your solvent to the temp. of the oven before they hit the column otherwise you get poor peak shape even splits.
generally this problem starts to occur when you have more then 15-20 degress difference between your room temp. and the oven temp.

best way to show that the problem at hand for the plates count is in the system and not in the detector is to move your 2998 to the same system where you have the 2996. if the plate count problem still occurs after that there then your detector or connection to it directly are the problem.

also do all detector have the same taper slit type flow cell?

[bookoon]

Hi All

Sorry I could not get back quickly; had some urgent project to complete.

Firstly thanks for all the inputs.
waters guys are scratching their brains out to get me all the answers

Prob 1: Theoritical Plates
Suggestion from Waters (during installation) : In instrument method Use sampling Rate = 1 and Filter Time Constant = Normal
Suggestion from Waters (NOW) : In instrument method Use sampling Rate = 1 and Filter Time Constant = Fast

This has improved the plates. the height of the peaks have increased. The width is still more when comapred to same run in 2996 detector. This still need to be optimized a bit.

Prob 2: Peak Purity
If i have understood this correctly......
In instrument method create the method as per default parameters (sampling Rate = 10 and Filter Time Constant = Normal) if you want to compare peak purity between a 2996 and 2998 detector.

So which parametres do i choose which will satisfy both problem 1 AND 2!!!

Prob 3: Peak Area
No answer as of yet. I am really scared AND angry now. +

I am thinking of reloading the softwre once more and replacing the hardware card. Do you think this will help?

[bookoon]

OK

for prob 3 the solution suggested is the same; sampling Rate = 10 and Filter Time Constant = Normal.

"The data points are not adequate" is what the group of engineers feel.

I had a few chromatograms (three to be exact) which were run with these parameters. Thankfully, the areas for the same are not changing.

We are now going to change all the methods to these default parameters and observe for at least a week/ till we are convinced that this problem is solved.

keeping my fingers crossed!

[bookoon]

Hi all

I think the problems are solved with setting the default parameters in the new PDA. At least I have not faced any unexpected incident since changing all methods to default parameters.

waters is supposed to send me an official comunique on this (since they had earlier instructed us to set a different parameter in the PDA set-up); but waters, being waters, will take their own sweeet time to do that..

Take care folks!