| Author | Message | ||
     Eric |
Hi everyone, A few questions: 1)What is the relationship between resolution and data interval? 2)Can we compare the data/results at different resolutions but same data interval? 3)How important is the resolutioin in the NIR measurement? 4)How to choose resolution before the experiment, for example, measuring the protein solution. 5)After deciding the resolution, if we change to a new data interval from the fixed one, how much it will effect the spectra? 6)How to choose data interval, if you do not like the fixed one, which is recommended by the instrument. Thank you very much!!! Yiwu | ||
| David W. Hopkins (Dhopkins) |
Hi Eric, Good to hear from you! You asked good questions, that could be covered by a longer course. I'll try to answer briefly. 1) There is no relation between resolution and data interval, it depends upon the particular instrument and if you have no choice, it depends upon the choice made by the instrument manufacturer. 2) Not easily. You can try to deresolve the higher resolution spectrum, but there may be problems knowing what exactly is the resolution of the lower resolution spectrum, unless you have been using a FT instrument with selectable resolutions. It is best to compare spectra recorded at the same resolution. 3) It depends on what kind of NIR measurement you are making. If you want to look at the spectroscopy of the materials, you need to select the resolution carefully. For good spectroscopy, you should select a resolution as fine or finer than the natural bandwidth of the narrowest band in your sample. However, if you are doing chemometric analysis of the spectra, Karl Norris, Peter Griffiths and others have shown that you can employ significantly lower resolution data and still get perfectly good results. 4) To measure the natural bandwidths of materials, you need to scan the materials at succesively narrower instrument resolutions, and note when the apparent bandwidth of the materials stops decreasing and is no longer limited by the bandpass of the instrument. In general, the band widths of proteins, starch, fat and water are 10 nm or more in plant materials at 8% moisture or more, and certainly in animal tissues and solutions, so that 8-10 nm bandpath instruments give good spectra for spectroscopic investigation and chemometric analysis. In dried proteins and crystalline materials frequently found in pharmaceuticals, the band widths of the absorbers are often much less than 5 nm, so lower resolution may give better spectra. However, lower resolution may not be required for good chemometric qualitative and quantitative measurements. 5) The amount of change depends upon the natural bandwidths of the absorbers. The effect could be minimal in the case of broad absorbers, or marked, in the case of sharp bands. 6) If you have an instrument that allows you to change or select the data sampling interval, follow the manufacturer's instructions. Many instruments do not allow you to change the data interval, it has already been set to be equal or less than the bandwidth of the instrument. If the the sampling interval is significantly lower than the bandwidth of the instrument, this can be an opportunity to achieve noise reduction in smoothing or derivatives using convolution functions. Hope this helps. Many of the points could be expanded upon, but this is a start. Best regards, Dave | ||
| Eric |
Hi Dave, Thank you so much! Would me ask another stupid question? If we measure low concentrion proteins, What resolution is better? low or high? or does resolution affect the sensitivity of the spectrometer? Best regards, Eric | ||
| David W. Hopkins (Dhopkins) |
Hi Eric, When you are measuring low levels of an analyte, you are usually going to optimize all sources of error. That means the reference method too! On the spectrophotometer side, minimizing the error may require operating at the maximum possible bandpath, or minimum resolution. Yes, the sensitivity (noise level) of the spectrometer is improved by lowering the instrument resolution. However, it is not possible to give a general answer to your question, you have to see how your best overall chemometric model is produced. Best regards, Dave | ||
| Tony Davies (Td) |
Dear David and Eric, Your discussion is related to the problem I have been trying to sort out today. A JNIRS author had written that spectra were collected over the range 1000 � 2500 nm at a resolution of 2 cm-1. A referee had noticed this and said �I assume you mean 2nm resolution� The author said �No we are using a Perkin Elmer Spectrum ONE NTS, FT-NIR and this was the set-up we used�. The spectra are plotted in nm and the discussion is nm so I am confident that while the instrument measured spectra in wavenumbers the authors never saw this data. The instrument outputted a file in wavelengths. The resolution of this data is quite difficult to determine because the values at wavelength intervals are probably determined by interpolation. If Peter Griffiths knew the algorithm I expect (from the Birth Award paper JNIRS, 11, 229-240) he would be able to work it out. It is possible that Perkin Elmer have been more clever than I am giving them credit for (I have talked to them) and have actually done the complex conversion to constant wavelength resolution discussed by Peter. Unless someone can suggest something better I will suggest to the author that the scan details should be given in wavenumbers: 10,000.00 � 4,000 cm�1 at a resolution of 2cm-1, followed by �the data were converted to wavelength data at x nm intervals by Perkin Elmer software�. Comments? Are any Spectrum One users listening? Best wishes, Tony | ||
| David W. Hopkins (Dhopkins) |
Hi Tony, Yes, I agree that your phrasing would be most accurate. Depending on what their application is, 2cm-1 resolution may be much more than they need, and they may therefore have much more noise to remove in their modelling than need be the case. An interesting question comes up. If they apply a Savitzky-Golay derivative calculation to their data presented in nm scale, is it legitimate? I think that the data is equally spaced in cm-1 space, but the units don't enter into the calculation, so the derivative is valid. The units for the purist would still be "absorbance units per cm-1" for the first derivative, even if the wavelength scale is presented in nm. What a mess. But the chemometric models would still be valid and useful. Regards, Dave | ||
| Tony Davies (Td) |
Dave, I believe the data has been interpolated so that it is at 2nm intervals. Tony | ||
| venkynir |
sir, we seek the help from NIR lovers , we have procured polarisation insensitive AOTF with fiber pigtailed ,Model:TEAF4-1.2-2.0-H-2FPmm we are unable to detect output from aotf test setup Test setup used Instruments: BRIMROSE:- VCO(50-90MHZ) APLAB POWER SUPPLY , FUNCTION GENERATOR, CRO[100MHZ],100W TUNGSTEN LAMP SOURCE, INAS detector,Judson preamplifier Testing Arrangements: A 100W tungsten lamp source was used to illuminate the plastic material,reflected light from the plastic is collimated and focused to the input fiber of the AOTF VCO driver has two inputs,one for frequency generation depending upon the voltage provided (1v to 10v )other input is analog modulation with 50 ohm impedance (0 to 1 v) Output fiber of AOTF was focused to INAS detector which has sensitivity with in our requirement (1200 to 2000nm).output of the detector is coupled to the preamplifier to measure the output. Test Conducted 1.A +0.8v positive pulse at the rate of <1KHz (we also tried +0.8v(max 1v)DC)was given to analog modulation input of the VCO driver with 50ohm output impedance system. 2.we applied positive voltage between (1v to 10V)to the frequency input of the VCO driver. 3.using 100watts tungsten lamp reflected light from the samples is focused to input fiber of AOTF. 4.Output fiber of AOTF was coupled to detector followed by premplifier to measure the output OBSERVATION: 1.If the light focused to the input fiber of AOTF is chopped at a regular interval ,we are able to measure changes in the voltage level with CRO 2.we have varied voltage (1 to 10v),but the output voltage level remains(actually it should not to be same for the entire wavelenghth range(1200 to 2000nm)as in general. QUERIES: 1.Eventhough the voltage given to the frequency input of the VCO driver is ZERO and analog modulation input is also ZERO,we are still able to measure change in the output voltage level from the detector if input light to the AOTF is chopped at a regular interval. 2.Ouput from the AOTF fiber has got only diffracted ray( +1 and -1)or it also consist of undiffracted ray along with diffracted ray. 3.How much minimum light intensity is required for focussing AOTF input.currently we are using 100watts tungsten lamp source. 4.What made the analog modulation distinct from TTL moudlation,because even using anlaog modulation we can shutter the input light at the required frequency as in TTL MODULATION. | ||
| Renato |
Hi venkynir, I did not understand everythuing but I have alredy built one of these (Guchardi R, Pasquini C Evaluation of a dual-beam near-infrared spectrometer based on acousto-optic tunable filters APPL SPECTROSC 55 (4): 454-457 APR 2001 ) 1.Eventhough the voltage given to the frequency input of the VCO driver is ZERO and analog modulation input is also ZERO,we are still able to measure change in the output voltage level from the detector if input light to the AOTF is chopped at a regular interval. AOTF does not "let go" only the selected light, it difract and module the selected one but it does not absorve the other. It happens what your experiment have showed. 2.Ouput from the AOTF fiber has got only diffracted ray( +1 and -1)or it also consist of undiffracted ray along with diffracted ray. The AOTF that I worked, had no fiber (I had setup my fibers... may be old time) sorry I do not know 3.How much minimum light intensity is required for focussing AOTF input.currently we are using 100watts tungsten lamp source. It depends on your measurements for example light direct (absortion) I have worked with 40 w, reflection 50 W, 100 W (it depends on the sample...). 4.What made the analog modulation distinct from TTL moudlation,because even using anlaog modulation we can shutter the input light at the required frequency as in TTL MODULATION. Analog modulation is a TTL modulation. If you module the light, every wavelength is modulated... but if you use this analogic modulation the AOTF will module only the selected wavelength... Good Luck! Renato | ||
| Renato |
Hi venkynir, I did not understand everything but I have alredy built one of these (Guchardi R, Pasquini C Evaluation of a dual-beam near-infrared spectrometer based on acousto-optic tunable filters APPL SPECTROSC 55 (4): 454-457 APR 2001 ) 1.Eventhough the voltage given to the frequency input of the VCO driver is ZERO and analog modulation input is also ZERO,we are still able to measure change in the output voltage level from the detector if input light to the AOTF is chopped at a regular interval. AOTF does not "let go" only the selected light, it difract and module the selected one but it does not absorve the other. It happens what your experiment have showed. 2.Ouput from the AOTF fiber has got only diffracted ray( +1 and -1)or it also consist of undiffracted ray along with diffracted ray. The AOTF that I worked, had no fiber (I had setup my fibers... may be old time) sorry I do not know 3.How much minimum light intensity is required for focussing AOTF input.currently we are using 100watts tungsten lamp source. It depends on your measurements for example light direct (absortion) I have worked with 40 w, reflection 50 W, 100 W (it depends on the sample...). 4.What made the analog modulation distinct from TTL moudlation,because even using anlaog modulation we can shutter the input light at the required frequency as in TTL MODULATION. Analog modulation is a TTL modulation. If you module the light, every wavelength is modulated... but if you use this analogic modulation the AOTF will module only the selected wavelength... Good Luck Renato Guchardi | ||
| krishna (Krishna) |
HI RENATO, From ur reply to venkynir, i has certain queries In general AOTF has three outputs +1,-1,and undiffracted ray.Diffracted ray from AOTF is seperated from undiffracted ray by certain degree(eg:5 degree)and this degree depends up on the wavelength it diffracts.Then, 1.What is meant by polarisation insensitive AOTF and it's advantages over other AOTF? 2.please suggest a general procedure for testing AOTF 3.only array detector has to be used for testing with AOTF ? 4.From venkynir mail,what i understood is output from AOTF also consist of undiffracted ray along with diffracted that's why he got ouput eventhough there is no input voltage and modulation input to the AOTF.Whether iam correct? looking for ur reply | ||
| Renato |
Hi Krishna, "1.What is meant by polarisation insensitive AOTF and it's advantages over other AOTF?" In an "normal" AOTF the output (+1,-1) are orthogonally polarized and this can be explored as advantage in some methods with polarized light like CD (Circular Dichroism Spectroscopy),� or be a disadvantage if you need to analysis something that has different iteration with polarized light, the AOTF itself is a example. If the input light is polarized it will be only one diffracted output (+1 or -1). An insensitive one does not have this. The output light has random polarization like the input. �2.please suggest a general procedure for testing AOTF� Build a �spectrophotometer�� Source (light) -> AOTF -> detector -> lock in Do not forget voltage input, modulation input to AOTF card and to lock in (to reconstruct the signal) Be patience to find the correct position to the detector (you can see the NIR just guess a position on the space�, I have tried some cards with inorganic salts that show fluorescence with NIR light. It would be easy to find the NIR (AOTF with input voltage and without modulation) but I did not see anything�) �3.only array detector has to be used for testing with AOTF ?� No, I have always worked with single detectors. You cannot work with Diode array detectors because they cannot be modulated, they give you a signal after a accumulation time but you need a �real time� signal to make a deconvolution (with a lock in). �4.From venkynir mail,what i understood is output from AOTF� also consist of undiffracted ray along with diffracted that's why he got ouput eventhough there is no input voltage and modulation input to the AOTF. Correct. The AOFT will always let the undiffracted light go like a quartz glass, even without input voltage and modulation. These are necessary only to have the diffracted beams. Regards, Renato | ||
| Jane |
Hello How to do experiment on stability of naxolone using the analytical methods? |