Photometer Test

and with light intensity proportional to the applied voltage.
I don't think this assumption is valid, unless you have checked the light source specification in detail. Change in voltage may well change the colour of the light too.
I would think you are better using a fixed light source and filters of known density for calibration.
Del

I think your cheapest way would be to get one of the photometers calibrated at a certification lab, and use it as a secondary standard.

Next, for your test light source, use a halogen lamp with a stable variable power supply, and note the outputs from your calibrated instrument to build a table of expected outputs vs. lamp voltage over the expected working range.

Now you can use the lamp and power supply to test production instruments - using the calibrated instrument to check the light source. You may have to check before and after every production test to start with, until you can get some idea of drift. May be able to stretch it to before and after a batch, or every morning and evening, once you can trust it.

Make sure the calibrated instrument is re-tested at the recommended intervals.

Using a stable light source is a must, but if you vary the voltage you change not only the intensity of the lamp but its color temperature, too. I.e., it will get 'bluer' when the voltage goes up and 'redder' when the voltage goes down.

It would be best to keep the voltage at a fixed value and allow the lamp to shine onto a opal glass surface or into a 'integrating sphere' device. http://en.wikipedia.org/wiki/Integrating_sphere You can buy a commercial integrating sphere or (if you can't afford one) you can make one from a couple white styrofoam containers like those inexpensive picnic coolers. You can buy opal glass from Edmund Optics: http://www.edmundoptics.com/onlinecatalog/displayproduct.cfm?productid=1671

To vary the intensity of the light, either move the lamp toward/away from the opal glass (the intensity will follow the inverse-square law). Or likewise toward/away from the input port on the integrating sphere. Or you can use a series of apertures at the input port of the sphere to vary the intensity.

By using different distances and the inverse-square law, or by using apertures of known areas, you can precisely control the resulting intensity without changing the lamp output in some unknown fashion.

You then aim your photometer at the center of the illuminated region on the opal glass -- or better, aim it at the output port of the integrating sphere. The integrating sphere has the advantage that the intensity is uniform across the output port, whereas the opal glass will have a peak intensity at the center and then the intensity will drop off away from that point (though there are ways to overcome this).

Just another thought here...

Halogen lamps (like most incandescent lamps) do not have a lot of output at 340 nm. This is well off the blue end of the visual spectrum. You might want to consider using a UV emitting LED, but use it with some eye protection:

http://www.724deal.com/luxeon-lumiled-1w-350ma-led-emitter-lxhl-pw01-7167.html

Thank you very much for your answers!

One additional information about required accuracy of the system: the photometer has a range from 0 to 3Abs (abs: absorbance unit), but for a specific test that I´ve had problems the range is 5mAbs. So, we should have a very accuracy tool for detection of any wrong result failure in this test. Really, varying the applied voltage of the light source not would be the best way to check it. I´ve read about aperture method for checking photometer accuracy. It would be applied to this case?

Thanks

Eduardo

There are a number of LED manufacturers who make 340nm light sources I would use this as my source as they are stable and fairly monochromatic

Try a web search for 340nm Led

Our own experience with using LEDs as reference sources during processing is that they are not as stable as we would like although we are using quite high power LEDs which generate a fair amount of heat.

I think that you are not looking in the right area to resolve your problem. I have worked for spectrophotometer manufacturing companies (CECIL Instruments, Bausch & Lomb, Shimadzu and measuring at 340 nm is a standard wavelength for some biochemistry tests. Normally you would use a Deuterium lamp for the 190 to 340 nm range and then switch to a Tungstem lamp for 340 to 900 or 1000 nm. At the top end you need to start using Infra red detectors. You can use glass cuvettes or cells at 340. Silica cells would be used from 190 nm upwards. Silica cells are higher quality and cuvette to cuvette reproducibility is greatly improved. Measuring 5 mabs is beginning to push the limits of reproducibilty as any small smear or finger mark can greatly affect the value. You also do not give any indication of the stability of the dark current or stray light of the system as this can also give rise to poor stability. Should a client want to measure at 5 mabs I would strongly recommend that an alternative cell or test tube be used - 40 mm and 100 mm being standard values. Remember Beer's Law! Standard interference filters are available and the choice of a suitable value neutral density filter would enable stabilty and reproducibilty values to be determind. A neutral Density filter will not change over time.

Tony

Thank you very much for your answer!

Considering that 5mAbs is a low absorbance value and any small noise could affect the measurement in the detecton board, it would be very dificulty to evaluate the photometer in 340nm using this absorbance delta and respective voltage values, could you suggest another way for checking the photometer reproducibility?

If I have some absorbances standards (Ex: NIST standard) in the range of the biochemical instrument values (0-3Abs), it would be enough to read these standards too many times for checking the reproducibility of the photometer, of course in a dedicated environment?

Thanks

Eduardo

I will need to do some research - my experiences were a long time ago. The use of neutral density filters or Potassium dichromte standards could be useful. You do not indicate whether you are using test tubes, cuvettes, flow cells or the means for measuring. There are many variables before the elcctronic, cleanliness of cells, repeatability of cells or test tubes, (if you use test tubes, use only 1 and mark it so that it always goes in in the same orientation). Make standards at say 1.0, 0,5, 0.25. 0.1, 0.05 and do checks at those values. Put a sample in at say 0.5abs and measure at say 5 minute intervals to check the stabilty. Make sure the light path is not contaminated by smoke. I was doing a demoonstration and the readings began to wander. This was due to the technician smoking and the cooling fan was bringing the smoke into the instrument and the smoke was going into the light beam which was then measuring different value. Check stability of bench to ensure no vibrations as this can also change readings on more basic instruments.

Look at http://www.starna.co.uk/ukhome/d_ref/neutden.html

Tony

Answering your question, I have used plastic cuvettes in this biochemical instrument. I will use a potassium dichromate solution as you suggest. My question now is: if I check the voltage on the detection board during five minutes, how will I evaluate if the photometer has a good reprodutibility or not based in the measured values. We think that I will have to measure a lot of photometers for determinated a acceptable standard desviation for it.

Thanks

Eduardo

The specifications for ALL instruments is based on the readings it produces - NOT on electrical stability or measurements. Plastic cuvettes are not the best on which to establish inaccuracy and inprecision measurements. If you are going to use plastic -use the same one being careful not to create any scratch marks - always put it in the same way round.

Tony

All said is correct.

When you start with the setup, you HAVE TO HAVE familiarization runs on your equipment. Plenty of familiarization runs, to gather the base confidence data, on which you can support your later experimental work. Please, consult a statistician, as you need an absolute certainty as to the reliability of the base. My rough estimate is in the range of 10 - 100 base runs, with standards and the same "stuff" you will run, and repeat, repeat, .... Same temperature, same voltage, even the same time of the day(?!?). You may have to become extermely fastidious until you know better. Glass only? maybe. For standards only? Maybe. Cleaning the instrument before every run. No fingerprints. And every other you and colleagues can think of. Remember Murphy's law: the one you ignore will come back to bite you.

Since the instrument is assumed not stable, every effort is worth stabilizing it. You can assure the environment being the same every time. I would suggest to use a "Power Conditioner". That is usually a ferroresonant transformer. It reduces the normal variation of the power in the building by a factor of 10 to 50 feeding your instrument.. It saves a lot of headache later.

Keep us posted.

I have measured in electrical points because the automatic biochemical analyzer has 60 cuvettes in a tray and makes a measurement of each cuvette in each 15 seconds (four times) when the tray rotates. So, we positioned only one cuvette in a optical path. In a detection board it´s possible to measured during all time. Of course, we must to have a very good instrumentation for this measurement.

Eduardo

Our own photometer uses a deuterium lamp which gives a fairly broad output (200-900nm). We use an opal diffuser which attenuates some of the lower wavelengths but gives a more even distribution & before each run the photometer is referenced to a known, calibrated detector as has been suggested earlier.

A couple of thoughts (coming from what's been posted so far):

1. As your photometer uses a halogen lamp itself, we have to assume that a halogen lamp will produce enough output in the 340nm region to do the test. Maybe use a higher power lamp - say a 50W car headlamp bulb - as your reference. A 12V vehicle lamp will actually be rated to run at about 14V, so if you limit it to 12V max. it will have a very long life.
2. If you're using a variable voltage supply, something with numeric setting (thumbwheels or a keypad) will make setting the test levels easier.
   
3. As far as I can work out, you're using optics - a diffraction grating - to separate the wavelengths before detection. If this is the case, the fact that the colour temperature changes as the lamp voltage varies is irrelevant - you're only interested in the part which gets through the grating at the correct angle for detection. If there is a problem with "swamping" (from the relatively much higher output at longer wavelengths), a filter centered on 340nm should solve it.
4. A UV LED would be a good idea (getting around all the problems mentioned in (3)), if you could get enough light output from it.
   
5. As others have suggested, some means of producing uniform illumination is a good idea, to reduce errors from position changes. You should also use a rig so that the position of your instrument relative to the light source is fixed and repeatable.

Please don't stomp on me for plagiarism - I'm just trying to bring some ideas together (and chuck in a couple of my own). Credits to whoever first posted each suggestion I may have mentioned.

John, I think you have summed it up well and would like to see a response.

All previous notes I agree with in essence. I am also willing to stipulate, that the lamp puts out proportionally (at whatever ratio) steadily on all interested wavelengths.

You still have to stabilize that output by measuring a selected wavelength. You need a bandpass filter, followed by a photocell. That has to be kept at an exact temperature (the specifics do not matter much) to get reliable measurement. From there via negative feedback to the lamp (most likely via power MosFet) the light output at the particular wavelength is stable, precisely. At other wavelength you can hope, that the output is proportionally stable.

All that amounts to a full fledged instrument development. No shortcuts, I am afraid.

While I generally agree with the tenor of all participants, I would like to continue focusing on the last two: mine and Tony's.

First of all, I am plenty familiar from a visitor's point of view with the instrumentation in a bio lab. My daughter now finished biochemistry PhD. I was visitor to the lab, many times. From an electronic engineers point of view, the instrumentation's stability and calibration is underwhelming. Repeatability? What is that? And that for a top Columbia Uni lab in New York. We simply speak different languages.

So, I completely switch tracks. It is difficult enough to keep one instrument calibrated. But, Tony's note indicates, that two different light source with two different instrument, in all likelihood is required. Light output consistency, and measuring consistency cannot be maintained across.

But, there is more than one way to skin a cat. And I bet, the manufacturer's engineers will be eager participant in the problem. Once they "finally" encounter somebody, understanding 'their" problems. It entails one or two overlapping standards. These are standard cuvettes filled with something stable and very well known and measured by the manufacturer. You can do it too, but will find it very painstaking. Including this standard in every run in the instruments, you can correct afterward every measurement by the measured, known factor.

When published, it will give you a firm reputation. Keep me posted!

It seems to me, that Tony's and my assessments converge.

So here it is:

1,. Now, you either make an instrument more exact, or, produce a standard (or two) to carry the "normal" value, or at least a correction factor with you.

2,. Any factory have to have standards,against which the equipment measured in production in a go / nogo fashion. That is a good start for you.

3,. When you have "normal" values included in every run, you have the way to correct for errors.

4,. The maker's technicians have plenty of experience. Use them.

5,. Other than that, my previous notes, and Tony's apply.

Yes and YES! ALL the notes before are useful. And as inconvenient it is, you have to learn plenty about statistic. You will hate it first, then it will come quite good for the rest of your life.This is something you are capable of. Developing a stable instrument, you are not in a position. That takes an electronic developer with biochemistry help.