Typical LOQ peak area, height, noise on Waters PDA (996)

[jjadus]

Hello ALL:

I am a new employee at a certain laboratory and I am evaluating three "used" Waters PDA (996) detectors, with nothing but past experience to campare to.

Using Waters Millennium software, we will consider the criteria for a LOQ (Limit of Quantitation) peak to be a signal-to-noise NLT 10:1.

This calculation is S/N = (peak height (uv))/(baseline noise (mv) *1000)

I understand that precision and accuracy studies at this LOQ level will confirm an allowable S/N across detectors, however...

Generally speaking, from your experience, is there a minimum peak area(uv/s) or minimum peak height (uv) that should be expected to give a S/N of NLT 10:1...

The reason I ask is that of the three Waters PDA (996) detectors I have, the same LOQ solution gives the following results:


Detector--------Baseline Noise------Peak Height---S/N---Peak Area (uv/s)

PDA 1-----------0.019------------------323-----------17------------~10000
PDA 2-----------0.032------------------323-----------10------------~10000
PDA 3-----------0.048------------------323-----------07------------~10000

I understand that in using only the S/N NLT 10:1 criteria, only detectors 1 and 2 would be "suitable" for use with this method. However, I believe this peak height of ~320 and the area of 10000 should give much larger S/N values than 7 - 17.

In other words, I believe these Waters PDA (996) detectors are "noisier" than what I have observed in the past at different laboratories using Waters PDA (996) detectors. Are these Waters PDA (996) detectors performing par for the course or are they noisier than what you've seen?

Thanks in advance for your thought!!

[jjadus]

Sorry,

PDA 3 should read:

PDA 3-----------0.068------------------323-----------05------------~10000

This changes the S/N to 5 and the range for the 3 detectors to 5-17.

Thanks and sorry for the mistake!!

[juddc]

It seems to me that those noise amplitudes are high, but the 996 has been around for a long time and if you have no history on these particular detectors, I'd suggest checking the lamp energies then replacing the lamps and flow cells before evaluating them seriously.

What wavelength are you looking at? What's your MP and does it absorb in the region in which you're taking measurements? What are the respective lamp energies?

My opinion on a used Waters PDA (996 & 2996) is basically this: If the lamp fires, it passes all of the diagnostics, and you can aquire data the machine is basically OK. PM the beast, check the flow cell windows for solarization, replace the cell if needed, and rock on. I've never had to replace anything but those bits in the decade or so that I've been using a 996 (or 2996 or 2998).

CJ

[Dan]

jjadus,

1) It is possible (even likely) that different detectors/systems will yield different S/N values (and thus, different LOQ values) for the same method. For this reason, the USP was considering adding a system suitability criteria for detector sensitivity in General Chapter <621> Chromatography. They did postpone this implementation as older methods may not be able to meet the new requirement; see: http://www.usp.org/USPNF/notices/postpo ... er621.html

2) That said, there may be noise sources in the new lab that cause these detectors to have higher noise values than you have seen previously for this type of detector. Look for possible noise sources such as:
a) nearby unshielded elecrtical motors/equipment
b) power cables wrapped around or near the detector signal cables
c) proper signal cables/wires on the HPLCs between the detector and the CDS (twisted pair wires with a shield are good, but connect the shield to ground only at one device, either at the CDS interface device or the detector, connecting the shield at both ends can actually generate noise)
d) pump noise/fluctuations
e) dirty flow cells
f) aging lamps

One last thing, are you using the same means to calculate the noise from one lab to the other? There are several ways to measure/determine the noise level in the chromatogram and CDS programs can give you options for the noise determination. (You can search for other posts about peak noise, peak-to-peak noise, etc. in this forum.)

Regards,
Dan

[jjadus]

Thanks for the ideas guys.

CJ, the MP is 40:30:30 water:acn:methanol and the wavelength is 260nm; so I'm not expecting MP-induced noise.

You indicated that these noise amplitudes seem high. What baseline noise values (mv) [calculated by Waters Millennium / Empower Software] are you typically observing for Waters 996 PDA's?

Dan, thanks for the input on different sources of noise. Have you any direct experience with Waters 996 PDAs? Are these S/N's low to you as well for peaks of this height and area?

I am using the same formula to calculate the S/N across sites (using Water's Millennium Software).

Before I start trouble-shooting possible sources of noise (i.e. flow cells, solarization, etc) I really wanted to see if others believe these detectors are in fact "noisey"... so what do you guys think?

Thanks again! Your feedback is much appreciated!!

[danko]

Hi jjadus,

The peak area is irrelevant in this case – for the simple reason that the unit for noise is mV or mAU or in some cases μV. So, make sure that the signal and the noise units are compatible.
Another thing one should insure is that the noise and the signal are measured in the same region (roughly at the same “retention timeâ€

[Dan]

Hi jjadus,

Yes, I have worked with the Waters 2996/996 PDAs. Unfortunately, I no longer have access to the relavant data (it was at a previous company).

However, if my memory serves, the noise you gave may be little high. But it is hard to say.

One easy thing to check is the age of the lamps. In typical use, the lamps should be replaced on a 6-month or annual basis. When was the last PM? and did the PM include a lamp change?

One other thing is that while the modern PDA detectors have the same sensitivity as a regular UV detector, they will have more noise than the UV detector (maybe 2-3 times). So don't compare the PDA LOQ to that found on the UV.

Regards,
Dan

[jjadus]

Thanks for your help.

The problem I am having, is convincing the "powers-that-be" that these detectors are not very sensitive - (i.e. they are "noisey") - in comparison with other Waters 996 PDAs out there in use. They seem to believe if they pass the annual PM, they should be able to run any method and all methods developed should be able to be run on any of these systems...

This is why I am looking for comparisons to you all out there who are experienced with or currently using Waters 996 PDAs.

Again, I appreciate all of your thoughts and thank you for your help.

Joel

[juddc]

These are data from 4/21/1999 taken on the day that I recieved my new 996 PDA:

Analytical wavelength: 254 nm
Resolution: 1.2nm
Rate: 2.0 spec/sec
Exposure: Auto
Filter: None

Noise amplitude = 390 uAU (0.00039 AU)

Data from the (to be retired) 991M taken the same day with the same MP, same pump, injector, column, sample, etc:

Analytical wavelength: 254 nm
Resolution: 1.3nm
Rate: 2.0 spec/sec
Exposure: Auto
Filter: None

Noise amplitude = 4120 uAU (0.00412 AU)

If I recall, I integrated the baseline over the course of ca 30 seconds from a valley to another valley in order to estimate the basline amplitude.

Peak areas run for test samples run on both detectors that day are comparable.

Just FYI, the 2998 sitting next to me is running at a noise level of ~100 u AU (0.00010) at 254 right now - I just checked.

So yes, at 3230 uAU (0.00323 AU), I think your instrument is noisy for a 996.

PS - I still have the 996, now upgraded to a 2996...I should check the baseline and see how it's doing 9 years hence.

ope this helps!

CJ

[jjadus]

Hi CJ,

Thanks for that information. I really appreciate it!

The "baseline noise" values I presented earlier are actually in mV (milliVolts) calculated by Millennium (in the results Tab); which equate to mAU (milli AU).

If I blow up the baseline in review and measure the AU (peak-to-peak), this means my PDAs in u AU (micro AU) are:

PDA 1: 0.000019 AU * 1000 = 0.019 mAU * 1000 = 19 uAU
PDA 2: 0.000032 AU * 1000 = 0.032 mAU * 1000 = 32 uAU
PDA 3: 0.000068 AU * 1000 = 0.068 mAU * 1000 = 68 uAU

Our results seem a bit different..by an order of magnitude... are you using Waters software or are you measuring the AU from positve peak-to-negative peak?

You are a great help. Thanks again!!

Joel

[juddc]

I am and was using Waters software & I could very well have miscalculated. I measure the baseline noise from valley to valley and wrote the heights in my notebook as measured by the software.

The amplitudes of the baselines as measured by the software at 254 nm are:

10 for the 2998
39 for the 996
412 for the 991M

A sample run both on the 996 and 991M with a peak height of ~0.014 AU read ~14000 in the software, so I think I am off by an order of magnitude...Sorry!

[jjadus]

Excellent.

With this data, I can see how ours compare. You're the man. Thanks, I appreciate your time. Peace!!