Fluorescence-signal 35 times to low Waters 2475

[bisonium]

Hi, i've got an odd problem. After the fluorescence-detector waters 2475 by misstake was reset during the yearly waters-service, the signal is about 35 times lower than befor. The main compound peak (SEC) usually (for several years) reaches about 800 000 EU (arbitrary units). But now we only get at a maximum ~27 000 EU (arbitrary units). We use Empower software. The instrument metgod has not changed, therefor the EUFS set to 10000 and Gain=100 is locked. As the method is validated for GMP, it is not allowed to change those values. The Waters service man seems confused about the problem, and he says that fluorescence detectors are prety rare out on the field and that he kant seem to find anyone experiencing the same problem. Is it posible that the detector cell is ruined or full of protein?? Or why this low signal?

We have tried the "generate peak" option in the W2475 and the conclution is that: whatefter the detector signal is, Empower shows the exact output it the chromatograms. So it could possibly be some kind of output-factor that shoud be enabled in the detector to get the high signal, but as I described above the EUFS and gain is locked and we are not allowed to change those values...

Please help if anyone has some kind of idee for a solution of the problem.

[tom jupille]

Arbitrary units are just that: arbitrary. If possible, take a look at the baseline noise. If that is down by the same factor, then it is a question of scaling and therefore irrelevant -- unless the method explicitly specifies a particular value (which would indicate a *very* bad method!).

Remember that you will generate your calibration plot on the same instrument, so scaling cancels out.

If, on the other hand, the baseline noise is *not* reduced by the same factor, then you have a signal/noise problem. A dirty flow cell, misaligned monochromator(s) or lamp, or a weak lamp could all be possible causes.

[HW Mueller]

Also check whether the detector has the possibility to check itself via a water Raman line (S/N. etc.).

[bisonium]

Thanks for fast replies!

Even if the outoput isarbitrary, still, the signal has reduced compared to earlier runs. The 800 000 is not mentioned to be absolute in the method, but since there is an example chromatogram with that EU-signal enclosed in the method, an FDA investigator could propably be interested in "where the rest of the protein gone" I guess.

RAMAN test is perfomed. The noise is somewhat bigger compared to earlier runs, but not 35 times bigger though.

The Lamp is excanged during the service just performed, so it should be good (I expected a rather higher signal compared to runs performed using the old lamp actually).

[sam.pedraglio]

I'm agree with Tom Jupille, arbitrary is arbitrary.
Did the Waters service engineer make a firmware upgrade of the detector?
If so, this could be the reason for your rescale problem.

[bisonium]

He didnt make a firmware upgrade. But he just now, as part of the investigation, reinstalled the same firmware to totally reset the FLD. But no success what so ever.

/Mattias

[danko]

You better perform excitation- and emission- wavelength accuracy before you give up solving the mystery.
Also, check your detector instrument method, to make sure the parameters (excitation and emission settings, including slit) haven’t been changed by some joker. What are these wavelengths set to - by the way?
If I was an FDA inspector, you’d never got away with 35 times weaker signal (arbitrary or not).

Best Regards

[bisonium]

Hi danko, thanks for all your tips! My emission/excit.-settings are 340/280. Those values are set in the instrument method, and have not been changed for, i'd say 8 years (method birth). The slit width on the other hand is to be 18 nm (for ex. and em.). I cannot find anywhere to set the slit width. Du you know where I find it?

Regards Mattias

[elsico]

I think problem is in built-in calibration data. It is changed or missed. Try to do procedure of manual wavelength calibration described in manual.
Same problem we have before with Waters uv detector.

Or another possible reason not properly aligned new lamp. Looks like they do now ready-to-use solution - not best idea.
So if you have old one try to reinstall it back...

[bisonium]

Elsico, thanks. I willlook into the built-in calib. data. I think that the waters support person has allready checked it, but absolutely worth a try. The old lamp was wasted since he succesfully had performed all RAMAN- and calibration-tests. I will check the procedure of manual wavelength calibration described in manual. I am "glad" someone else actually seen the problem, even if its was an UV and not the FLD.

Regards

[danko]

Maybe the slit is controlled by the so called bandwidth parameter. I’m not sure, but I think the default value is 4 nm. If so, you can figure out that the energy you’d get with 18 nm would be much, much higher.
Also, Internal/built-in wavelength calibration is all very fine and that’s what you should utilize initially, but if it doesn’t result in meaningful evidence, then you might like to check the matter independently. I think most people utilize antracine for the purpose.
The chosen Exc. and Em. wavelengths seem OK (in protein context), so no need to worry about that – unless the detector wavelength accuracy is lousy.

Best Regards

[HW Mueller]

Since this appeared after a lamp change it seems obvious that something was reset or messed up. A higher noise in the Raman test (which is also a wavelength test) is also indicating something was changed in a way that was not in your intention. The mentioned slit is a good candidate, can´t imagine a detector that doesn´t allow its adjustment. Still since the signal is arbitrary, it should be able to be calibrated and give the right concentrations of analyte. If this is not the case someone did a really good bodge job.

[bisonium]

Thank you all for all input and tips. I will take contact with the Waters the following days and see if we together with your valuable tips can find what the problem is. I'll be back with more info about the outcome and hopefully a possible solution.

Thank you all for now!

/Mattias

[bisonium]

Hi again!

Nothing seemed to work in the solution hunt. Everything looked at: motherboard, battery, lamp and front panel board exchanged. Many days of investigation and numerous of injections seemed like a waste of time and material.

But 10 minutes before the waters guy was about to leave last friday, the signal jumped up. And now I tried to run a known sample, the signal is very high (even about 30% higher than before). So I have three questions left (the Waters guy almost didn't believe me having such high signal before he saw the digits on the FLD front panel, guess if I was lucky?!):

1. Have anyone experienced this?
2. Is the problem posible to has been caused by something stuck in the detection cell, and suddenly just washes out followed by the much higher signal output? The Raman tests (and other tests) are outside the detections cell , and therefor they would not have shown this type of problem, right?
3. Is the 30% higher signal to be awaited after a lamp exchange?

/Mattias

[danko]

Mattias,

30% signal increase after replacement of the lamp is not an unexpected observation. So, no reason for panic!
Regarding the tests:
The Raman test is performed by sending a beam through the cell. That’s why you need to fill it with the purest water, prior to the measurement.
The wavelength accuracy test is performed outside of the cell. So if you didn’t perform your own test (f.ex. with antracene) you can’t be completely certain of what went in the cell and what came out of it.
Your idea of potentially dirty cell is quite viable, so maybe that’s what happened.
Another explanation could be loose electric connection, either to the lamp or the PMT. If that is the case, you’ll probably experience the problem again – especially after shut down of the detector.
The last idea that comes to my mind is a defective lamp. If that’s the problem it’ll probably “dieâ€