building a spectrophotometer from off-the-shelf parts

[[email protected]]

In my search for a simple way to calculate ethanol concentrations in
beer (of all things!) I remembered back to my college days - in one
biochemistry lab, we used a spectrophotometer to determine enzyme
reaction rates. Those Shimadzu spectrophotometers cost about $10,000
each, and had an on-board printer to print the resulting graph of
transmission vs. wavelength.

Oh, look, one can be built nowadays for cheap using off-the-shelf
parts:
http://www.rsc.org/Education/EiC/issues/2007Sept/BuildYourOwnSpectrophotometer.asp

I noticed one part is a 3140 op amp. Is this the one?
http://www.mouser.com/search/ProductDetail.aspx?R=CA3140Mvirtualkey55320000virtualkey837-CA3140M

Is there an easy way to do away with the requirement for *two* 9V
batteries and use just, say, +6V (from a wall wart, or 4-pack AA)?

Any input would be appreciated. Just imagine what could be done for
high schools, using/making one of these in labs!

Thanks,

Michael

 

[Carl Ijames]

The problem for you is that that particular spectrometer covers only the
visible spectrum, 450-700 nm, and ethanol doesn't absorb anything over
that range. Do some research and see if there is anything useful in the
near-infrared, where ethanol absorbs and water doesn't, or maybe see if
you can use refractive index instead. [pause] Okay, I found some NIR
info for you:
http://www.axsun.com/media/pdfs/05-02-085a_liquid_phase_quant.pdf and
http://en.wikipedia.org/wiki/Near_infrared_spectroscopy. Of course
water also absorbs in the nir, here are two references (you will have to
handle the wavenumber to wavelength conversion to compare with the
ethanol spectra):
http://cfa-www.harvard.edu/hitran/HITRAN_conf06_presentations/Session4/4.2-Daumont.pdf
and http://www.lsbu.ac.uk/water/vibrat.html Good luck.

 

[[email protected]]

The problem for you is that that particular spectrometer covers only the
visible spectrum, 450-700 nm, and ethanol doesn't absorb anything over
that range. Do some research and see if there is anything useful in the
near-infrared, where ethanol absorbs and water doesn't, or maybe see if
you can use refractive index instead. [pause] Okay, I found some NIR
info for you:http://www.axsun.com/media/pdfs/05-...wikipedia.org/wiki/Near_infrared_spectroscopy. Of course
water also absorbs in the nir, here are two references (you will have to
handle the wavenumber to wavelength conversion to compare with the
ethanol spectra):http://cfa-www.harvard.edu/hitran/HITRAN_conf06_presentations/Session...
andhttp://www.lsbu.ac.uk/water/vibrat.html Good luck.

Yes, absolutely, ethanol solutions are colorless (except for the added
barley and hops). I had thought using the bottom of an AOL CD as a
prism would also produce some near IR, but I'd better double-check
that assumption.

Thanks for the input,

Michael

 

[Allan Adler]

one can be built nowadays for cheap using off-the-shelf
parts:
www.rsc.org/Education/EiC/issues/2007Sept/BuildYourOwnSpectrophotometer.asp

My browser can't open the pictures for some reason. I'll try it from a
friend's computer later this week. The prices are all given in pounds
sterling but look cheap.

I noticed one part is a 3140 op amp. Is this the one?
http://www.mouser.com/search/ProductDetail.aspx?R=CA3140Mvirtualkey55320000virtualkey837-CA3140M

This page says that Mouser is going to stop selling this one. Also, doesn't
mouser have a minimum total price per order?

Anyway, I guess the thing to do is to study the schematics and see exactly
how it works and then consider how to modify the design so that different
parts can be used.

 

[John Popelish]

In my search for a simple way to calculate ethanol concentrations in
beer (of all things!) I remembered back to my college days - in one
biochemistry lab, we used a spectrophotometer to determine enzyme
reaction rates. Those Shimadzu spectrophotometers cost about $10,000
each, and had an on-board printer to print the resulting graph of
transmission vs. wavelength.

Oh, look, one can be built nowadays for cheap using off-the-shelf
parts:
http://www.rsc.org/Education/EiC/issues/2007Sept/BuildYourOwnSpectrophotometer.asp

I noticed one part is a 3140 op amp. Is this the one?
http://www.mouser.com/search/ProductDetail.aspx?R=CA3140Mvirtualkey55320000virtualkey837-CA3140M

Is there an easy way to do away with the requirement for *two* 9V
batteries and use just, say, +6V (from a wall wart, or 4-pack AA)?

Any input would be appreciated. Just imagine what could be done for
high schools, using/making one of these in labs!

There is no good reason why this opamp is better than many
others for this application. I would use half of an LM358
dual. But I would also probably replace the light dependent
resistor with a silicon photo diode. This would improve the
linearity and still give you a spectral range from about 300
nm to 1000 nm. If you use a gallium arsenide photo diode
you can extend the IR end down to something like 1700 nm.

 

[Bill Penrose]

In my search for a simple way to calculate ethanol concentrations in
beer (of all things!) I remembered back to my college days - in one
biochemistry lab, we used a spectrophotometer to determine enzyme
reaction rates. Those Shimadzu spectrophotometers cost about $10,000
each, and had an on-board printer to print the resulting graph of
transmission vs. wavelength.

A less complicated one can be built using filters. For a few bucks you
can buy a little booklet of transparent plastic filters with known
characteristics from Edmund Scientific. Use those to filter the
incident light and avoid the mechanics of prism and slit.
http://scientificsonline.com

The op amp is pretty generic. There are several that will substitute,
like the LM324. If you can't get it from Mouser, try Jameco. Mouser
and other wholesalers get pissy when you want to buy one op amp, one
battery clip, etc, from them. Jameco's a one-at-a-time kinda place.
http://www.jameco.com/

A photodiode will work better than a LDR (cadmium sulfide cell), whose
response time will be slow.

A spectrophotometer can be pretty simple or highly complicated. For
simple one wavelength measurements, it can be wonderfully cheap, but
don't expect Shimadzu quality from the measurements.

Dangerous Bill

 

[Don Klipstein]

A less complicated one can be built using filters. For a few bucks you
can buy a little booklet of transparent plastic filters with known
characteristics from Edmund Scientific. Use those to filter the
incident light and avoid the mechanics of prism and slit.
http://scientificsonline.com

The Edmund Scientific ones tend to be a little less sharp in cutoffs
than other ones (GAM, Rosco, Lee) in my experience. Also, in general,
dye-based filters have gradual cutoffs at the long wavelength end of a
passband, making a narrowband slight source with dyed filters close to
impossible.

Furthermore, all of these filter gels (Edmund, Rosco, GAM, LEE) have a
high tendency to pass infrared. I once took a Rosco booklet, separated
the transparent filters from the diffusing filters and the paper pages,
fired an infrared LED through them at a phototransistor - a fair amount of
the IR made it through all of them!

Most dyes are IR-passing. Furthermore, in stage lighting, it is
desirable to minimize absorption of IR to reduce heating of the filters

(all too often only from "burn up soon" to "last a reasonable amount of
time before getting brittle and/or changing color").

One can easily do better at spectral analysis with a diffractive "clear
CD" that comes in some spindle packs of recordable optical discs, or with
LEDs of known peak wavelength. (Bear in mind that some LEDs are rated by
"dominant wavelength", which is different and usually displaced slightly
towards green-yellow from the peak wavelength.)

The op amp is pretty generic. There are several that will substitute,
like the LM324. If you can't get it from Mouser, try Jameco. Mouser
and other wholesalers get pissy when you want to buy one op amp, one
battery clip, etc, from them. Jameco's a one-at-a-time kinda place.
http://www.jameco.com/

How about DigiKey? They merely have a $5 surcharge for failing to make
minimum order, as well as many items with pricing down to 1 unit.

A photodiode will work better than a LDR (cadmium sulfide cell), whose
response time will be slow.

Keep in mind a couple of things:

1. Cadmium sulfides ain't that slow - milliseconds to a fraction of a
second.

2. LDRs and phototransistors have nonlinearities. Photodiodes tend to be
linear.

3. DigiKey helps again here - it helps to have a spectral response of
your device. It appears to me that nothing easy to get has flat spectral
response.
Closest to flat that I noticed so far is in "blue enhanced"
silicon photodiodes, which have response through the visible spectrum
being roughly proportional to wavelength (a bit less than that
aooproaching UV and IR, falling more rapidly in UV as wavelength
decreases, and gradually flattening as wavelength increases towards their
peak response in IR around 900-950 nm or so).
(Makes me maybe glad to not remember what the spectral response of a
non-blue-enhanced silicon photodiode looks like!)

- Don Klipstein ([email protected])

 

[Paul Mathews]

In my search for a simple way to calculate ethanol concentrations in
beer (of all things!) I remembered back to my college days - in one
biochemistry lab, we used a spectrophotometer to determine enzyme
reaction rates. Those Shimadzu spectrophotometers cost about $10,000
each, and had an on-board printer to print the resulting graph of
transmission vs. wavelength.

Oh, look, one can be built nowadays for cheap using off-the-shelf
parts:http://www.rsc.org/Education/EiC/issues/2007Sept/BuildYourOwnSpectrop...

I noticed one part is a 3140 op amp. Is this the one?http://www.mouser.com/search/ProductDetail.aspx?R=CA3140Mvirtualkey55...

Is there an easy way to do away with the requirement for *two* 9V
batteries and use just, say, +6V (from a wall wart, or 4-pack AA)?

Any input would be appreciated. Just imagine what could be done for
high schools, using/making one of these in labs!

Thanks,

Michael

Old spectrophotometers go really cheap on ebay. You'll pay more for
the shipping than for the unit, unless you can find one nearby.
Paul Mathews

 

[John Larkin]

In my search for a simple way to calculate ethanol concentrations in
beer (of all things!) I remembered back to my college days - in one
biochemistry lab, we used a spectrophotometer to determine enzyme
reaction rates. Those Shimadzu spectrophotometers cost about $10,000
each, and had an on-board printer to print the resulting graph of
transmission vs. wavelength.

At the Tulane chem department, they had an IR absorption spectrometer
that they used to check for benzene in the lab ethanol supply. If they
got a safe batch, they'd have a party.

John

 

[[email protected]]

At the Tulane chem department, they had an IR absorption spectrometer
that they used to check for benzene in the lab ethanol supply. If they
got a safe batch, they'd have a party.

John

Hahaha! Straightforward enough to add yeast to sugar water, then use
the lab glassware to distill.

Michael

 

[[email protected]]

There is no good reason why this opamp is better than many
others for this application. I would use half of an LM358
dual. But I would also probably replace the light dependent
resistor with a silicon photo diode. This would improve the
linearity and still give you a spectral range from about 300
nm to 1000 nm. If you use a gallium arsenide photo diode
you can extend the IR end down to something like 1700 nm.

That's what I was hoping to hear. Soldering SOIC chips is a bit
beyond my ability.

Thanks y'all,

Michael

 

[Clifford Heath]

That's what I was hoping to hear. Soldering SOIC chips is a bit
beyond my ability.

Hey, I've probably got a few CA3140's here in cans if you want.
ISTR I used them as I needed the very high input impedance that
was difficult otherwise to obtain in 1975.

SOIC's aren't that hard to solder - you mainly need good light and
magnification.

Clifford Heath.

 

[G]

At the Tulane chem department, they had an IR absorption spectrometer
that they used to check for benzene in the lab ethanol supply. If they
got a safe batch, they'd have a party.

Thats why they use the 95% stuff.

It would real nice if we had an updated, beer, alcohol, and calories, test.
The last documented test was many many years ago, which still circulates
the internet. I would settle for just the alcoholic content.

greg

 

[Phil Hobbs]

Hahaha! Straightforward enough to add yeast to sugar water, then use
the lab glassware to distill.

Michael

Keeping the fusel oil (propyl and butyl alcohols mostly) out of the
distillate is the hard part. Bread yeast makes some *nasty* hooch.

Cheers,

Phil Hobbs

 

[Marvin]

In my search for a simple way to calculate ethanol concentrations in
beer (of all things!) I remembered back to my college days - in one
biochemistry lab, we used a spectrophotometer to determine enzyme
reaction rates. Those Shimadzu spectrophotometers cost about $10,000
each, and had an on-board printer to print the resulting graph of
transmission vs. wavelength.

<snip>
Years ago, I was the co-inventor of a simple colorimeter.
The reference is
Simple Ultraviolet Photometer.
R. E. Thiers, M. Margoshes, and B. L. Vallee
Anal. Chem. 31, 1258-61 (1959)

If you can't get the publication, I can probably send you a
copy. It was designed specifically to measure the coenzyme,
NADH. Ralph Thiers developed a list of chemistries for the
instrument, which was on the market for some time as the
Coenzometer, along with packaged reaction mixes called
Determatubes.

There are many enzyme reactions that can be used for various
analytes. For ethanol, an assay with yeast alcohol
dehydrogenase will work. It is an inexpensive enzyme that
is easy to work with. Your students will be painlessly
introduced to biochemical analyses.

Marvin

 

[[email protected]]

<snip>
Years ago, I was the co-inventor of a simple colorimeter.
The reference is
Simple Ultraviolet Photometer.
R. E. Thiers, M. Margoshes, and B. L. Vallee
Anal. Chem. 31, 1258-61 (1959)

If you can't get the publication, I can probably send you a
copy. It was designed specifically to measure the coenzyme,
NADH. Ralph Thiers developed a list of chemistries for the
instrument, which was on the market for some time as the
Coenzometer, along with packaged reaction mixes called
Determatubes.

There are many enzyme reactions that can be used for various
analytes. For ethanol, an assay with yeast alcohol
dehydrogenase will work. It is an inexpensive enzyme that
is easy to work with. Your students will be painlessly
introduced to biochemical analyses.

Marvin

Was it patented in the US? US patents are now freely available
in .pdf format at www.google.com/patents. Do you have the patent
number? I'll look it up.

Thanks,

Michael

 

[Marvin]

Was it patented in the US? US patents are now freely available
in .pdf format at www.google.com/patents. Do you have the patent
number? I'll look it up.

Thanks,

Michael

At the time, Harvard University (where I worked in the
Medical School) had a policy against patenting inventions in
the field of human health. There is no patent, but there is
the reference I gave.

 

[Bill Penrose]

the reference I gave.

Send the man a reprint, Marv! Or at least a scan.

What was your UV source?

DB

 

[Mark Thorson]

Send the man a reprint, Marv! Or at least a scan.

What was your UV source?

Betcha a nickel it's the biggest one in the solar system.

 

[default]

In my search for a simple way to calculate ethanol concentrations in
beer (of all things!) I remembered back to my college days - in one
biochemistry lab, we used a spectrophotometer to determine enzyme
reaction rates. Those Shimadzu spectrophotometers cost about $10,000
each, and had an on-board printer to print the resulting graph of
transmission vs. wavelength.

Oh, look, one can be built nowadays for cheap using off-the-shelf
parts:
http://www.rsc.org/Education/EiC/issues/2007Sept/BuildYourOwnSpectrophotometer.asp

I noticed one part is a 3140 op amp. Is this the one?
http://www.mouser.com/search/ProductDetail.aspx?R=CA3140Mvirtualkey55320000virtualkey837-CA3140M

Is there an easy way to do away with the requirement for *two* 9V
batteries and use just, say, +6V (from a wall wart, or 4-pack AA)?

Any input would be appreciated. Just imagine what could be done for
high schools, using/making one of these in labs!

Thanks,

Michael

Why not just use specific gravity or polarimetry to measure sugar
content and calculate alcohol from converted sugar?

That link you show is somewhat idiotic. An LED is not a white light
source but a mixture of primary colors (which you can see with a
compact disk used as a diffraction grating).

And he shows in the diagram the light passing through a diffraction
grating - that isn't how it is done.

A SCANNING spectrophotometer or just an adjustable wavelength spectro
is still a fairly complex instrument and I don't think the average
DIY device will do it. They use dual beam balanced instruments to
compensate for spectral response of the source and detector and
subtract them from the total absorption.

More goes on in the UV range than the visible range when it comes to
analyzing chemical compounds.

Build a liquid chromatograph - spectro with a flow cell and separation
column added - much more accurate when you can single out the alcohol
from the rest of the mixture.

It may be a valid device for explaining the operation of black boxes
to chemistry students - demystify the instrument - but a few drawings
on a black board and popping the cover on a real instrument will
achieve the same thing with better understanding and accuracy.