| Author | Message | ||
     Luca Vicenzutto (Vice) |
Dear all In your opinion, what kind of IT system (Windows Xp, 2000, NT, Linux, etc. etc.) is better to put side by side with NIR Software? | ||
| hlmark |
Luca - you probably won't have too much choice. The instrument manufacturer will provide software that you MUST use to collect data from their instrument. Then he will specify the operating system that the software will run under. If there's a choice at all, it will be very limited, e.g., WINDOWS XP versus WINDOWS 2000. You would have a wider variety of choices for analyzing the data and performing calibration calculations, since that could be done with third-party software, but at that point it would be hard to justify a second computer with a different operating system unless you find you absolutely need a package that runs only under the different operating system. Howard \o/ /_\ | ||
| Michel Coene |
We at Novozymes are actually going an other way. Most systems have severe disadvantages: special software which only works on one kind of analyser, lots of footwork running around the different spectrometers in your plant/lab/... Some people have seen an opportunity and are working towards integration. We have bought a kind of "chemometric server" from the Danish biotechnological institute. The spectrometers have "drivers", not unlike you load a driver for a particular printer. All data is kept in one place and we can use the same routines for fluorescence spectrometers as we do for NIR. I know there is a company in Sweden called proxedra which is working in this area too. On the NIR conference in Kolding there was a paper manufacturor who had simply programmed his own system. Specially for factories who have different kinds of instruments scattered over a wide area, this method definitely helps cutting maintenance. Some instrument manufacturors do this too, but then limited to their own equipment. Our system runs on Linux, and there is no denying that this OS scales very well from the tiny cubes we attach to the spectrometer all the way to the big central server. It is also very easy to reach. Our server has no screen or keyboard, it is completely remote controlled via the internet. Even big software upgrades can be done from the other side of the planet. This cuts cost, as consultants don't tend to sit in a car for several hours free of charge. | ||
| hlmark |
Michel - that's a good approach. I didn't think that anyone would be willing to go to that much effort, though. Or be able to get the information from instrument manufacturers regarding the (usually proprietary) information about how to collect data from their instruments. Your original question was about the "best" operating system. You must realize, of course, that questions like that are liable to start wars! But I think you cannot define a "best" operating system in isolation. The purpose of an operating system is to provide a platform on which to run other programs. Therefore you need to consider what programs you want to run, and which operating systems they will run under. LINUX, for example, has the reputation of being much more stable than, say, WINDOWS, as well as having the other properties you mention. On the other hand, most of the programs written specifically to analyze NIR data, and especially the programs used for calibration of NIR instruments, are written for WINDOWS. Within the WINDOWS-compatible community, most software vendors of necessity follow the lead of Bill Gates, and make sure that their software is compatible with the latest version of WINDOWS, whatever that is. So far, at least, Microsoft has been concerned that users should be able to run legacy software, and so newer versions of WINDOWS largely maintains compatability with older versions (there are exceptions). Howard \o/ /_\ | ||
| hlmark |
Michel - could you send your e-mail address to me. My e-address is [email protected] There are some comments I'd like to make but cannot do it on the discussion group. Howard \o/ /_\ | ||
| David Russell (Russell) |
Echoing Howard's original reply, the vendor that you're working with probably has a recommendation to use a specific version of Windows. Since analyzer vendors are typically small businesses, they vary in their ability to keep up with changes in Windows. So you may find it necessary to run Windows 98 or Windows 2000 rather than XP. | ||
| Tony Davies (Td) |
Hello Michel, It is interesting to hear of programs running under Linux; I have no experience of it. People always seem to say good things about it. If these are small stand alone systems I expect you do not want to run other programs anyway. However, when you do calibrations I suspect that, if you use PLS for example, you will have to transfer data to a program under Windows. How easy/difficult is the data transfer between the systems? As for David's comment I came across one package that would only run under NT! Luckily I did have one NT computer. (It has just been replaced by an XP system so ....!). Best wishes, Tony | ||
| Richard Kramer |
The post by Michael Coene regarding "chemometrics server" software reminds me that there is a new package of this type which was recently announced and should be available around now. It is called Symbion and is supplied by a division of Axiom Analytical in California. Axiom is more commonly know for its sampling accessories. They were showing a pretty impressive demo of the software at EAS. I don't think that many people are yet aware of the new package they have developed. (A google of Axiom Analytical shows them at www.goaxiom.com) | ||
| Michel Coene |
Dear Tony, Our system is actually quite complex. For communications, we use a "Message queue" from IBM. This is used by IBM to communicate with all kinds of computers, including their mainframes. A funny thing is it can work with publication and subscription. A spectrometer could "publish" a result as topic "NIR/spectrometer15/fat/sample26" and somebody who subscribed to "NIR/spectrometer15/*" (all results from spectrometer 15) would automatically get a copy, just like somebody subscribing to "*/fat/*" (all fat - results). This means extra "customers" for the results can be added without needing to tell the spectrometer about it. The data ends up in a set of databases, which we operate via a built in website. It is like buying books on Amazon.com. You drive around with a shopping cart and add spectra, reference values and possibly other things like temperatures. You then tell the system to put everything together in a certain way (e.g. interpolate temperatures so you have an estimate of what the temperature was when the spectrum was taken). Finally you download the composite dataset as an ASCII file. Save it to your PC and open it in the Unscrambler or whatever. Delete after use, as the original is still on the server. Models are uploaded to the server via your webbrowser. For more technical things, like software upgrades and backups, you would use a technology called OPENSSH. It is as if you can open a DOS prompt on your own PC which actually runs on the server. | ||
| Tony Davies (Td) |
Michel, Please can you e-mail me; I seem to have lost your address! Tony | ||
| Dennis Karl (Dennisk) |
We are having speed issues in running WinISI V1.50 (and earlier versions) under Windows XP. The computer is a Dell with 256 RAM, Intel Pentium 4 at 2 GHz. The problem is that, whereas under Windows 95 any toolbar button selected gives an instant response,under Windows XP we wait for a minimum of 20 seconds for each and every function selected and sometimes we see a (Not Responding) message flicker through the status bar. We have set the XP Programs compatibility wizard and are at a loss as to what to do next other than further increase RAM. Any ideas or solutions anyone?? Dennis | ||
| Michel |
1. Make sure you do not have a virus or other spyware. 2. Check with the manufacturor. I once worked with a Siemens program which would only work with Windows NT 4.0 Service pack 5. One extra Windows update and it was dead in the water. 3. If you communicate via a serial port, you could consider buying a plugin-board with an "industrial" RS232. It is really not that expensive. | ||
| hlmark |
Dennis - two possibilities occur to me: first, WIN/XP is notorious for being a memory hog. If you don't have enough RAM, the computer will spend inordinate amount of time in temporarily saving memory contents to disk while it uses the memory for higher priority processes (the term used is "thrashing"). This normally takes only a second or so, however, so I think it's the less likely explanation. The more likely explanation is that the computer is trying to access some hardware device that is not present or not connected. This may be a printer or a floppy disk drive, or maybe something else. When the computer tries to use a device, it send the device a command, then waits for a response indicating that the device has completed it activity. This wait often takes about 20 seconds as you are experiencing. Two things I can think of to do is first, note if there seems to be any sort of activity while you're waiting, especially from the hard drive. Second, use the control panel to see if there are any drivers, or if there is anything in the hardware configuration related to something that does not exist in your setup. Howard \o/ /_\ | ||
| Dave Ryan (Ryan) |
Dennis, We had the same problem and found out that the problem is not XP, it is the speed of the processor. The 16 bit protocols used by Winisi 1.50 have some timing problems with the new fast pc's. XP on an older platform, say 1 Ghz or slower works fine. We decided to go with their 32 bit ISIScan and have not had a problem Dave Ryan |