| Author | Message | |||
     venkatarman (venkynir) Senior Member Username: venkynir Post Number: 69 Registered: 3-2004 |
Dear Sheelagh; I am happy to read alot of suggestion for you.As instrument engineer , I feel if you use pulsed illumination at an angle (suggeted by Mark and others) in sychronous with spectral capturing will more helpful. More over have you thought of illumination correction factor through standard reflectance probe. . | |||
| Sheelagh Halsey (shalsey) Member Username: shalsey Post Number: 11 Registered: 12-2006 |
Lots more useful hints. Thanks again everyone for the input. Sheelagh | |||
| Michael C Mound (mike) Senior Member Username: mike Post Number: 59 Registered: 7-2007 |
Just an additional comment re: Chuck's helpful suggestions... I can tell you with full confidence that my own experiences with high density low frictional coefficient plastics absolutely avoid any fouling or cross-contamination. The flow cell design is carefully engineered to avoid any ridges or sharp corners that attract and retain dust particles. As mentioned, the issue of particle size can be a major or minor issue due to the law of cohesion which basically states that the g-force attractions between particles (related to "Zeta" potential) increases geometrically in inverse ratio to size. This means that smaller particles are likely to agglomerate and can be the source of the pesky issues Chuck is probably referring to. Anyway, the devil's in the detail, and the flow cell design, which is the major obstruction is pivotal. Good luck, again. Mike | |||
| Charles E. Miller (millerce) Junior Member Username: millerce Post Number: 6 Registered: 10-2006 |
Hello Sheelagh: This has been a very useful discussion! I have just a few bits to add to the previous set of responses: - High-wattage illumination might be required for spectral quality, but can cause material quality and safety concerns. I would strongly recommend having the lamp power on a flow interlock to prevent "unpleasant" events from unintended flow stoppages... - Is your process dedicated to a single product grade, or multiple products? In-line flow obstructions are a good sampling idea, but can be annoying during process turnarounds/transitioning between products - Fine-ally, I'd expect lots of "fines"/dust buildup in the sampling area- especially if a flow obstruction is used. This is often exacerbated by static cling. There will need to be some provisions in the design to keep the windows clear of these awful things... Good Luck! Chuck | |||
| Michael C Mound (mike) Senior Member Username: mike Post Number: 58 Registered: 7-2007 |
All, Well, I guess I should contribute something here, inasmuch as I have had some rather extensive experiences with fast moving materials in both conveyor belt (open systems) and closed pneumatic transport of powders (air slides and the like). On the open systems, movement of materials is on the order of 10 m/second and in closed pnuematics, the speed is at least 15+ m/s. All of the suggestions are almost prescient in nature and my hat's off to all of the very clever people who have contributed. Unfortunately, what we have done is in the nature of Intellectual Property and I am not permitted to reveal too much here, nor would it be appropriate to do so for reasons surrounding the ground rules and general communal basis of the Forum's raison d'etre. What I can say is that you are all pretty close to what can be done on a practical basis, so I have confidence that Sheelagh will be successful. I would also like to comment on the flow cell idea. Some time ago, when the idea of presentation and application of CIP (cleaning in place) techniques were combined in a similar operation, we found that using a low-friction high-density material as a flow cell surface placed at a steep angle opposite the flow transport direction solved a number of problems already stated in the discussions. Suffice to say that the angle is as vital to the surface movement and planar orientation of the flow cell impact incidence. Incorrect angles will result in an unsatisfactory saltation (jumping) of the particles, while the correct angle (might have to be as much as 45�, but this depends mostly on the flow dynamics and whether the velocity is constant or variable, as well as particle size distribution). Some of this can be solved empirically, and use of a baffle plate, also known as a scree deflector in these kinds of flow dynamics will, under normal circumstances, if oriented correctly, usually provide the near flat surface approaching a kind of homogeneity that will prove useful in presentation of the moving stream/flow, whether laminar or turbulent in nature. Ilumination of sofficient wattage is also key, as must be evident to permit the optmimum useful reflection and collection. The combination of the components of what is called the optical kernel of this configuration (sample, source of illumination/excitation, and collector/detector) determines key results: the tighter the kernel, the better the chances are for a successful dynamic process repeatabiity and robustness. I will say no more, other than you have received good suggestions from the experts, Sheelagh, and, with some creative closure on your part, you should approach a useful solution to your application. Best of luck, Mike | |||
| Ralf Marbach (ralf) New member Username: ralf Post Number: 4 Registered: 9-2007 |
Hi Sheelagh, Ralf Marbach here at VTT Optical Instrument Center in Finland. We routinely develop new types of optical probes here, maybe half dozen per year, for various industries and always customized for some kind of on-line PAT application. From the discussion so far I understand that you don't need to do spectral imaging on INDIVIDUAL flakes, which is quite a relief. (At a speed of 15 m/s you would be at the borderline of technical feasibility for this. In other words, the tradeoff between shortness of illumination flash -to freeze the motion- and number of photons-collected-back would be quite a narrow ridge to work on, if at all. It would require some detail thinking to estimate feasibility in that case.) If, however, all you need is a diffuse reflection spectrum off of a sample which happens to be moving, then to me it sounds like just another case. From an engineering point-of-view, the goal in optical probe design is to (a) illuminate a represenative-enough portion of the sample while (b) getting enough photons back to the photodetector and (c) meeting a multitude of application-specific practical requirements. You have not given enough info to comment in detail, but anyway, some general remarks follow. The two big things you have to trade-off are (1) and (2) below: (1) Hardware noise level From (i) the amplitude of the absorbance response feature(s) of your analyte of interest and (ii) your time resolution requirement and (iii) your analyte-precision requirement follows the maximum hardware noise level you can tolerate. The hardware noise floor goes down when the number of photons collected back to the detector goes up. It is usually a waste of time to make experiments for (1) -- with hardware knowledge, everything can be estimated using a spreadsheet-like "photon budget" calculation, which is also the #1 tool to tell you about the best design tradeoffs. (2) Sample Presentation "Noise" There are a number of things to consider here and pretty much all turn out to be tradeoffs vs. (1) above. Once you have decided where to measure in the process, important practical things to consider inlcude: First, the area of the illuminated spot, i.e. what spatial freqeuncies of the sample non-uniformity you like to average out. A bigger spot is better, obviously, but too large a spot will backfire on (1), so check the photon budget. The thing that counts in diffuse reflection measurements is the re-emitted radiance [Watt per area per steradian]. So, if you starved for light, then illuminate your sample with as much light as you can (i.e. don't use fibers for illumination unless you have to and can afford enough of them) and don't reduce the radiance of your collected beam (i.e. don't use integrating spheres or other diffusive components unless you got plenty of signal-to-noise ratio to start with). Also, it can make sense to make the spot wider in the cross-direction than in the sample-flow direction because, as Howard just said, the averaging happens "automatically" in the latter direction. Second, passline insensitivity -- optical probes are usually non-imaging and thus can be designed to allow for a suprisingly large "depth of focus." Fiber probes, incidentally, tend to be worst in this respect because the collected signal is highly sensitive to distance variations whenever the sample is close. Third, given the speed and "granularity" of your sample, you need to measure all wavelengths simultaneously in this application. A diode-array would thus be a good choice of instrument whereas an FTIR would not be. And then, there is the plethora of practical things. I briefly mention some more common ones, but the list is not complete: (3) representative sampling -- Big picture issues here concern the orientation of the flow direction, the use of deflector plates, etc. BTW, for small enough tubes this is often possible also when measuring through windows, by shaping the window accordingly on the inside; (4) probe cleanliness, i.e. resistance to surface fouling -- .. too long to talk here ... (5) "effective" optical pathlength through the sample -- basically, the geometry of the illumination and pickup spots ... (6) time-stationarity of response, i.e., usually the stability of the effective pathlength against scatter changes of the sample -- ... too long for here... (7) chemical resistance, pressure rating, space limitations, sanitary requirements, ... (8) need for built-in reference and/or dark (9) cost, availability of parts ... (10) Servicability, How the users would handle the probe-- often important in batch processes ... In summary, if all you tried so far is an off-the-shelf fiber probe then you have plenty of options to improve. If a factor of 2 or so in SNR is all you need, then consider alternative off-the-shelf probes. For full potential, however, customization is required. My gut-feeling is that a fiber probe may not be good enough in your case, because fibers don't bring much light to the sample in the first place and then they only collect a tiny fraction of the solid angle into which a DISTANT sample remitts. So even if you do something to these probes to get (2) right, you probably will be lost on (1). Have you considered buying a commercial flow cell and then using that to construct your sampling interface around it? Hope these remarks are helpful. Regards, Ralf --------------------------------------------------------------------- Dr.-Ing. Ralf Marbach VTT +358 20 722 2249 (office) Optical Instruments Center +358 20 722 2320 (fax) Kaitov�yl� 1 +358 40 770 8375 (mobile) FIN-90570 Oulu mailto:[email protected] Finland http://www.vtt.fi/oic ----- VTT - Technical Research Center of Finland -------------------- | |||
| Sheelagh Halsey (shalsey) Junior Member Username: shalsey Post Number: 10 Registered: 12-2006 |
Hi all Thanks again for the extra ideas, there are certainly a few more possibilities here. I will have another look at the process and see which of these might work the best. Sheelagh | |||
| Art Springsteen (artspring) Junior Member Username: artspring Post Number: 9 Registered: 2-2003 |
Hello again, Sheelagh, You didn't mention how big the tube was (diameter) or whether it was transparent. If it were small enough, you could run it through the middle of an integrating sphere and use an external source to diffusely illuminate the stream. The fiber collector could then be used in a passive (non-illuminating) mode to measure the stream. We've done this before but you would obviously need a reference channel to monitor the light source. Not totally practical- Howard's last suggestions are far better- but a possibility none the less. Regards, Art S. | |||
| Howard Mark (hlmark) Senior Member Username: hlmark Post Number: 201 Registered: 9-2001 |
Sheelagh - there are at least two other approaches, that may or may not involve major hardware changes: 1) Measure over longer times. This may mean measuring multiple times while the sample is moving, or it may mean stopping the sample and measuring the spectra, then repeating that. In either case, average together the multiple spectra. 2) Set up an automatic sampling system, that automatically removes samples of the necessary size from the stream, then measure that under whatever conditions are optimum (I suppose this qualifies as "at line" rather than "on line", but it's another way to address the problem). \o/ /_\ | |||
| Bruce H. Campbell (campclan) Moderator Username: campclan Post Number: 112 Registered: 4-2001 |
Why not angle the light beam to maybe a 45 degree, or larger, such that the beam is close to parallel with the pipe? Bruce | |||
| Sheelagh Halsey (shalsey) Junior Member Username: shalsey Post Number: 9 Registered: 12-2006 |
Mark That is an interesting suggestion. I will need to speak to an engineer about the solution as it may change the flow dynamics, but it may be a way to get a more repeatable sample presentation. Thanks for the help Sheelagh | |||
| Mark Murphy (qamarkm1) New member Username: qamarkm1 Post Number: 5 Registered: 6-2008 |
Sheelagh, Have you considered modifying the pipe to make your sample more presentable to your probe? It may be something as simple as putting a flat plate in the centre of the pipe at a 45 degree angle so that your sample will hit the plate and 'run off' it. This would give you a 'set platfom' from which you could work. Alternatively, a more complex approach may be to mount the plate perpendicular within the pipe and also mount the probe within the pipe. This would need deflection plates to ensure that the sample is correctly projected onto the plate infront of the probe. (See options 1 and 2 in the attachment)
Please note that I have no personal experience of such a solution. I have heard of this type of sample presentation being used for standard FTIR in-line analysis and your best bet would be to talk to an Engineer before starting down this route. The metal plate in the pipe may be more suitable as a highly reflective surface like a mirror in order to better suit the optics and analysing instrument. | |||
| Sheelagh Halsey (shalsey) Junior Member Username: shalsey Post Number: 8 Registered: 12-2006 |
Thank you all for your comments so far. However, my problem is sample presentation. I am measuring in a pipe not a conveyor, so I can not control the sample surface or distance. I agree with Howard's comments that I need more sample area, but I am not sure how I can do this in a fast flowing diffuse powder. I think extra optics or an integrating sphere is this application would not be feasible - unless anyone can suggest how! Any further comments would be appreciated. Many thanks Sheelagh | |||
| David Donald (ddonald) New member Username: ddonald Post Number: 5 Registered: 11-2006 |
Dear Sheelagh, In response to your quire to obtaining good spectra from a moving powdered sample, my experience in measuring crystallized sugar on a conveyor belt at speeds over 1m/s, I hope, are useful to you. Firstly and fore mostly is sample presentation; a relatively flat sample at a fixed distance will aid spectra collection enormously. This can be achieved by mounting a glass/quartz plate between the probe and sample or alternatively you can precondition the sample with a roller. Secondly, reduce light. Remove all other light sources from the sample area since the wattage from a remote probe is typically very small thus any environmental light sources will play havoc with any spectral collection. If you have the ability to illuminate the sample independent from the probe, a nice big lamp (50-80 watts) will increase your sample area and decrease the effect of environmental light. Your concern of the nitrogen concentration of 1/50 is not much of a problem (with sugar cane/crystals/other) as I routinely calibrate constituents with lower concentrations � variation is king here. Online spectral collection follows the same principals as laboratory methods: prepare your sample well and 99% of your problems disappear � the other 1% is usually political I�m afraid. Best regards, David Donald. | |||
| Lez Dix (lez_dix) Junior Member Username: lez_dix Post Number: 9 Registered: 10-2006 |
Sorry guys yes. I knew Sheelagh when we both worked for FOSS (NIRSystems)and had lost touch, I simply took advantage of seeing a post from her to get back in touch. Appologies for the confusion. Lez | |||
| Howard Mark (hlmark) Senior Member Username: hlmark Post Number: 200 Registered: 9-2001 |
Venkatraman - I suspect that Lez's reason for asking for a private response is that the friend is probably in the business, so that discussions with him would be considered "commercial" and therefore not appropriate for the discussion group. Otherwise, I agree with your comments; it would be good for all of us to hear the advice given. Hopefully, Sheelagh or Lez will come back and let us know what happenned. Sheelagh - my own advice about your problem is that you want to maximize the area of the sample that you measure. If you can't increase the area, then at least you might want to increase the amount of light captured. In either case, auxiliary optics would be called for. To increase the area observed, you would need to move the probe back, to increase the field of view. To avoid losing too much light, you might need to incorporate a telescope-type optic in front of the detector. If you can't do that, then you might try capturing light leaving the sample at high angles; the device that would help you do that is an integrating sphere. I hope one of these suggestions works out for you. \o/ /_\ | |||
| venkatarman (venkynir) Senior Member Username: venkynir Post Number: 68 Registered: 3-2004 |
Dear Lez Dix; I may be permitted to point out the following points with good intention and affection with NIRS .I honestly believe that the objectibve of this NIR work groups is to share the experience and expertise knowledge one and other and spread the warness of NIRS.Please avoid this as personal developemnt page.I hope our moderator will agree with me . | |||
| Lez Dix (lez_dix) Junior Member Username: lez_dix Post Number: 8 Registered: 10-2006 |
Hi Sheelagh I don't know the answer to this but may know a man who might. It would be good to hear from you and catch up, drop me a line on my private e-mail [email protected] Lez | |||
| Sheelagh Halsey (shalsey) Junior Member Username: shalsey Post Number: 7 Registered: 12-2006 |
I have an application looking at the active content of a fast moving, 15m/sec, powder of the density 10kg/500kg in nitrogen. The sample looks like snowflakes. I am having trouble getting good spectra with a standard reflectance probe located in a pipe. Does anyone have any ideas on how to get good spectra from this type of sample? Any comments appreciated, Thanks Sheelagh |
Diagrams