DAD Reference Wavelength: do we need a separate channel?

[Multidimensional]

My boss (who provided much of the info I cross posted) has been involved in the design of the hardware and software used in these instruments for decades. He has detailed knowledge of these systems. No need to assume.

Btw: You appear to be very new to these instruments and appear to have no practical, hands-on chromatography knowledge. You appear to be a programmer trying to figure this out. This explains many of your false statements and assumptions because they are not based on how these instruments have been designed and evolved over the many decades. Many of these instruments were designed before personal computers were used.

Another area that continues to drive your thought in the wrong direction is that DAD/PDA only acquire a full spectrum of data. This is false. Most of the time, the detector is used to collect only discreet wavelengths (rows) of diodes. Scanning only a range of wavelengths is an optional feature. All of these features utilize the array of diodes to collect data. No need to collect a full spectrum most of the time. If a range of data is stored, for a scan, this data is usually kept in a separate register. Discreet channels (usually 1,2,...7) are also in separate registers too. They are not extracted. This is because you can run a full scan, collect individual wavelength channels, all separately or both. They have to be kept separate. BTW: Retention times are not part of this (no ide why you thought they would be). Abs data only is stored. Rt are separate for each peak stored in another area (because RT's are made-up, that is determined by the software settings used, not real, but calculated based on integration values).

Anyway this is FAR TOO much info to try and teach you on a web forum. If you really want to know how these systems work then you should go work for one of instrument manufacturers (you will not find the info in a book). They hire lots of software engineers to write their code. Also, this is so far off topic from your original question where you did not understand what "Reference Wavelength" feature was all about and why we collect data in Channels (SIGNALS) too. This has been explained, but would make a lot more sense to you if you have decade or so of chromatography experience using DAD/PDA's because these are related to setting up the detector to run analysis by HPLC. Your question has been answered.

[sbashkyrtsev]

Btw: You appear to be very new to these instruments and appear to have no practical, hands-on chromatography knowledge. You appear to be a programmer trying to figure this out.

Correct. That's why I asked the question - because I hoped to learn from people who know better. I'm just having hard time following your thoughts - I can't find any confirmation in the docs and the data that I have contradicts what you're saying. And at the same time you don't explain why what you're saying should be true or why what I'm saying should be false. You just keep repeating that it is. And I can't simply take your word for it.

Retention times are not part of this (no ide why you thought they would be).

Sorry, I didn't mean Retention Time of peaks. I meant the time axis of the 2D data. It's not stored within dad1A.ch file that I have.

[vmu]

You assume that collecting a WL in DAD happens separately from acquiring full spectra. May I ask what's the source of this information? I went through Agilent and Waters operators guides on DAD - none of them mention how they do it.

They do it by allowing the user to check or uncheck the corresponding checkbox "store spectral data" in the software window. This is sufficient knowledge for a chromatographer.

[lmh]

Shimadzu don't even offer the option of collecting individual wavelengths as signals. During data-processing, you extract whatever wavelength chromatogram you want from the full spectral data (and subtract the value at another wavelength if you want to use a reference). And of course there's some processing going on between the diodes of the PDA and the signal it sends to the chromatography system: there's a combining of signals from adjacent diodes according to the desired spectral resolution, and there's averaging of signals over time, according to the response-time you've specified.

I'd endorse the comments about not collecting signals with reference wavelengths already subtracted. If you can do this later in software, do it there. If you do it at the time of data-collection, you cannot undo it; if something appears that absorbs at the reference wavelength, you'll be contending with negative peaks.

[James_Ball]

Shimadzu don't even offer the option of collecting individual wavelengths as signals. During data-processing, you extract whatever wavelength chromatogram you want from the full spectral data (and subtract the value at another wavelength if you want to use a reference). And of course there's some processing going on between the diodes of the PDA and the signal it sends to the chromatography system: there's a combining of signals from adjacent diodes according to the desired spectral resolution, and there's averaging of signals over time, according to the response-time you've specified.

I'd endorse the comments about not collecting signals with reference wavelengths already subtracted. If you can do this later in software, do it there. If you do it at the time of data-collection, you cannot undo it; if something appears that absorbs at the reference wavelength, you'll be contending with negative peaks.

As others said I prefer to use no reference or one with a very narrow band width. I had an operator once who set the band width of the reference to something like 200 and could not figure out why he has so little sensitivity. He was subtracting out more signal than he was gathering at the wavelength of interest.

On Agilent if you uncheck the collect spectra box, it does only collect the wavelength range of interest, because usually those data files are smaller, which makes long term storage of data take up less space and is why we don't collect spectra unless we are doing method development. At least on older A and B versions of Chemstation, not sure what it might do on the newer data systems.

[lmh]

... going wildly off-topic, of course there's a bit of history here. Back in the days that file-size mattered so much, and Agilent were offering all those options of collecting spectra only during a peak, or every second time-point etc., computer-processing-power was also much lower.

I tried, a while back, doing a simple PCA-based compression on a full set of spectra from an LC-PDA run, and was able to throw away at least 90% of the data while still getting extracted wavelength chromatograms that overlapped the un-compressed chromatograms so perfectly that even zoomed in, it was difficult to discern any difference, and spectra where you could see 4nm changes in wavelength maxima on related compounds. It made me realise that if only the computing-power had been available for compression, back then, we could have had the full spectral data with little file-size; and now, there's no incentive to compress because no one is scared of big files any more. And of course mine was a clumsy compression by someone who is not an expert at compression techniques, so it is likely much better could be achieved. PDA data-handling is an area characterised by lack of ambition or creativity. Sad!