The Planted Tank Forum banner

DIY PAR Meter

20K views 59 replies 9 participants last post by  Hoppy 
#1 · (Edited)
Hey, so I know this has been done before so not much new here. After reading the various DIY PAR meter posts though I found that everyone has had to calibrate their sensors. Hoppy's latest iteration uses a sliding tube to adjust distance between the sensor and the lens/filters before securing in place.

This is a DIY PAR meter build from a mechanical perspective and hopefully I can get good readings just after assembly. I have all the components and will try to finish up this build in 2 weeks.

Here's a cross section of my design. The number, order, and distance of the filters to the sensors are all taken from one of Hoppy's designs.
EDIT: I've replaced the original diagram with this. This shows my design intent better. This version of the housing is meant to be a test bed for an integrated 3D printed acrylic housing. The undercut is meant to accept a Tap Plastic 2447 frosted acrylic sheet or a 3D printed acrylic part for testing purposes. Also I've updated it to show the filters I'm using. No big surprise here, it's a copy of what Hoppy used to have.


Here's a picture of the 3D printed housing. It's porous black nylon so it'll have to be sealed before dunking in water. The surface quality turned out much better than I hoped.
 
See less See more
2
#2 ·
I just found this. I really like the idea of using a 3D printed housing, which would make the whole assembly job much easier and repeatable. One suggestion: don't use a lens at the top, use a diffuser - like a piece of frosted acrylic plastic, or make it from white acrylic, or even from clear acrylic that is sanded with about 200 grit sandpaper on both sides. I found that I could get repeatable results by just sanding clear acrylic. Or, you can use the clear lens and add a Rosco diffuser filter, like #116. The best Rosco filter combination I have found yet is #3313 and #373, plus an infrared blocking filter #1995, but the last one is not essential.

How expensive is it to make the 3D printed housing? As long as that is less than $10 this is a very good way to go.

What kind of cement are you going to use to seal the "lens" to the housing, and the electric cable to the housing?
 
#3 · (Edited)
Hey Hoppy, sorry I didn't actually mean a lens literally. Yes, it's supposed to be a piece of frosted acrylic plastic. I was going to keep it as a surprise in case if it didn't work but the idea of the undercut is to make that sheet swappable for testing purposes. See the first post for an updated cutaway.

If I end up building the updated version, I will definitely integrate the new filters you've suggested. I've already bought the Rosco filters so I think I'm stuck with it for now. By the way, this version is optimized solely for the swapable acrylic sheet and ease of adjusting the thickness of the calibration spacer. The idea is to figure out if the acrylic top can be 3D printed as a part of the housing (no sealing on the top!) and if a very precise and repeatable calibration spacer can eliminate the the need for calibration in the future (I bought 2 sensors for this purpose). I have other ideas in mind for future improvements (reduced assembly time, improved sealing, etc.).

I was going to use what you used, Weldon #16, to cement the housing. Supposedly it can bond acrylic to a host of other plastics. As for the electric cable to the housing, I haven't decided yet. My fall back is to mix Weldon #16 with acrylic dust and form a bead around the seam between the cable and housing. In my next version (if it's built), I plan to have a nipple extend from the wiring exit with a small lip and use adhesive lined heat shrink.

By the way, supposedly the 3D printed black nylon is not waterproof so I might even bother sealing it at all unless I can find a suitable potting material. The 3D printed frosted acrylic IS waterproof though but more expensive so it's more appropriate for the next version.
 
#4 ·
How would you assemble the sensor if the top lens/diffuser is an integral part of the 3D printed housing? One problem I find with Weldon #16 cement is that it flows all over the place, and if it contacts the Rosco filters it distorts them and changes the amount of light transmitted. You could probably coat the outside of the finished housing with black nail polish, but that can be a pain to work with too. Weldon #16 adheres very well to the insulation on the wire I use, so it seals the wire pass-thru very well.

Does the 3D printing process always generate a porous solid? Can it print opaque black acrylic?
 
#5 ·
There is a black acrylic available and it is water tight. I'll try the black nail polish this time around. I'll upload a schematic for an integrated housing tonight to show you what I mean. There should be no issues with Weldon getting on the filters with this design.
 
#6 ·
Hey Hoppy, here's what I have envisioned the integrated housing to look like. This is by no means my best attempt. One thing I want to point out is that I'd like to get rid of the preload spacer and wave springs altogether. I think that is doable and will measure the thicknesses of my two sensors to see what the tolerances are.

 
#7 ·
You can eliminate the spring if you make the bottom plug without a flange, so it can accommodate the variations in dimensions of the other parts. One thing that might make this configuration hard to work with is that if the calibration spacer isn't quite right it will be a chore to replace it to bring the sensor into calibration. That is a problem I have with my current design.

Another concern: The "aperture" size is also a big factor in the sensitivity of the sensor. I'm using a .25" diameter "aperture" - larger was a pain because it provided too much light to the sensor, making me use heavier filtration, which, since gel filters don't have much affect, if any, on the IR part of the spectrum, left the IR contributing too much of the total light energy hitting the diode.

With 3D printing, can you switch materials during the process - make the top clear acrylic and the bottom an opaque plastic? Could you stack the internal parts and print the housing around them? (Seems unlikely.)
 
#8 · (Edited)
I like your idea of making the bottom plug without a flange. I was also thinking of making it out of brass or some other heavier material since I think the sensor will float in its current design. Yes, I'm definitely worried about the calibration sensor. The spec sheet said I think 1-2% linearity but did not mention about repeatability across sensors. We'll see when I stick the 2nd sensor on there.

I've stuck with the 0.25 diameter aperture. See the picture below. I was really faithful to your design with respect to the working features.


As it is right now, you cannot switch materials. I'm sure someone makes a machine out there that lets you but it's not an option with Shapeway.com. You could stack the internal parts and print around with some limitations but again, not with the company I ordered these parts from.
 
#10 ·
Nothing yet. This week I plan on wrapping everything up related to the current design and get a reading out of it. Then it's onto the redesign. Sorry, for the lack of progress. I'm trying to find time between homework, kids, and work. Really wish I started this before I started classes.
 
#14 · (Edited)


This is a way to make a sensor, using 3D printed parts, so its sensitivity is adjustable, filters are easy to install (but not replace), and sealing it from water is easy.

I bought a great CAD program for my Mac, the "Punch! ViaCAD" program, only $100, (vs about 5X that for any comparable program) and it does everything a CAD program can do, without having to install Windows on the computer first. The learning curve for this is steep, seeming impossible when you first start, but it only took me a few days to be able to make a computer model of the two parts for this. It helped that I first learned the basics of CAD back in the 1980's, using the Dassault CAD program that Boeing started using for commercial aircraft, and this Punch! program is based on that same program. I still have a long way to go to be proficient with the program, but it is certainly fun.
 
#16 ·
I haven't decided, yet. First, this design needs the dome part to be somewhat translucent, so it acts as a diffuser, and I'm not sure what material will do that. Second, I don't know who to have do it, and if I can afford it. I'm really tempted though. I think I still have one of the photodiodes to use in it, but I'm pretty sure I'm out of the filters needed to make it work right. Last, I'm not sure that this is an optimum design yet.
 
#17 ·
I think I have some Roscoe filter from one of your earlier designs. For the dome, I suggest the frosted or translucent acrylic material from Shapeways. You might have to print a dome in each of the materials to test it out.

The design looks good overall though.
 
#18 ·
Shapeways looks interesting, and their "frosted" material looks ideal for a diffuser. I will be studying this more today to see if I want to proceed to a prototype or prototypes. If I do this I also want to do more work selecting the filter(s) to shape the spectral response to be at least as good as the Apogee Quantum meter response.

Since this is hobby activity for me, the more work I have to do, the more fun it is!
 
#19 ·
I just ordered one of each of the two parts, black acrylic for the body and white frosted acrylic for the dome. First, I added the hole for the electric cable in the body, saving the work of drilling the hole, and the risk of breaking it while drilling it. I learned the hard way that you have to be sure to specify that the drawing is in inches, not mm, when you order the parts! (Or do the drawing in mm.)

The cost is about $31 with shipping included. If I had ordered 10 the shipping cost would have been a lot less, per unit. Even so, the sensor is going to cost around $30 for parts only.
 
#20 ·
About 2 years ago I began looking for a combination of Roscolux filters that would change the spectral response of an Excelitas photodiode to equal that of the Apogee Quantum PAR meter. I found many combinations that were usable, but until today I never found a combination that almost exactly duplicated the Apogee response. Here is that Apogee spectral response that is the goal:

The colored area is the ideal PAR meter response, the definition of PAR being all of the radiation between 400 and 700 nm wave length.

An Excelitas VTB8441BH photodiode has built in filters to give this spectral response:


Today I found a 3 filter combination that works very well, using the spectrum plots shown on the Rosco website, http://www.rosco.com/filters/roscolux.cfm Here is how I calculated the response:


These are the three filters, along with their transmission at each of 20 wavelengths. I used a spreadsheet to figure out the sensitivity of the diode at each wave length by multiplying the transmission percentages at each wave length together and multiplying that by the relative sensitivity of the diode at those wave lengths. Then to be able to plot that on a graph I converted each of those sensitivities so the the highest one is 1.0 and the others are fractions of 1.0.


When this is plotted on a graph, along with the Apogee Quantum PAR meter spectral sensitivity it looks like this:


You can see that a PAR meter using that photodiode and those filters would cover the whole PAR range better than the Apogee meter, but at the cost of including a little of the radiation outside the PAR range. I think that is a good trade off.

If the 3D printed parts here work out right, and if it is possible to adjust the assembly to calibrate it to read PAR accurately with just geometry changes, this should make a very good PAR meter.
 
#54 ·
Have you verified the spectral response with Excelitas? This is the reply that I received from them:

From: Desfonds, Eric [mailto:Eric.Desfonds@excelitas.com]
Sent: Thursday, July 10, 2014 12:12 PM
To: James Howard (JHN)
Cc: Larsen, Ella; Ruiz, Luz; Chepulis, Dennis; Charlebois, Alain
Subject: RE: Response Curve

I agree it is a bit off. I took the liberty to digitize the curve, see attached.

The first datapoint is around 330nm. The 720nm corresponds to about 10% relative to the peak. We more routinely would have use the 10% points in the low- and high-range, but given the steepness of the low wavelength side, we used 330nm. And we used the 10% point on the long-wavelength side.

The peak in the specific curve is closer to 530nm than 580nm, which is around 85% of the peak response. Apologies of the confusion.

That being said, the datasheet with the graphs is a more general sheet, and some products like the VTB8441BH have varying Silicon thickness, so I would trust the datasheet values MIN/MAX/NOMINAL (330, 720, 580) as valid for the VTB8441BH part. Other VTB parts also have varying peak wavelengths, the graph was meant as a reference really.


Éric Desfonds


There is a big difference between a curve of 330,720, 580 and the general curve Absolute Spectral Response “B” Series Filtered curve published in their attachment which you quote.

Thanks,
Jim

About 2 years ago I began looking for a combination of Roscolux filters that would change the spectral response of an Excelitas photodiode to equal that of the Apogee Quantum PAR meter. I found many combinations that were usable, but until today I never found a combination that almost exactly duplicated the Apogee response. Here is that Apogee spectral response that is the goal:

The colored area is the ideal PAR meter response, the definition of PAR being all of the radiation between 400 and 700 nm wave length.

An Excelitas VTB8441BH photodiode has built in filters to give this spectral response:


Today I found a 3 filter combination that works very well, using the spectrum plots shown on the Rosco website, http://www.rosco.com/filters/roscolux.cfm Here is how I calculated the response:


These are the three filters, along with their transmission at each of 20 wavelengths. I used a spreadsheet to figure out the sensitivity of the diode at each wave length by multiplying the transmission percentages at each wave length together and multiplying that by the relative sensitivity of the diode at those wave lengths. Then to be able to plot that on a graph I converted each of those sensitivities so the the highest one is 1.0 and the others are fractions of 1.0.


When this is plotted on a graph, along with the Apogee Quantum PAR meter spectral sensitivity it looks like this:


You can see that a PAR meter using that photodiode and those filters would cover the whole PAR range better than the Apogee meter, but at the cost of including a little of the radiation outside the PAR range. I think that is a good trade off.

If the 3D printed parts here work out right, and if it is possible to adjust the assembly to calibrate it to read PAR accurately with just geometry changes, this should make a very good PAR meter.
 
#24 ·
Gentlemen,
Have you looked into what is available in smart phone. I have a Galaxy S4 and an app, Galaxy S4 Sensors, that has a Lux meter in it. I think it uses the front camera as the sensor.

I have on of Hoppy's DIY Lux meter and will try it as soon as I can find it.

Of course, I worry about putting a $600 smart phone under water.

Joe
 
#27 ·


The parts arrived today. The translucent dome seems to pass more light than I expected, and the fit isn't what I expected, so I will have to do a little sanding to get the dome to fit over the body part. Other than that they look very good, and should work fine.

Now I just need to assemble this, and experiment to see how I can alter the design to make it always work without laborious testing, adjusting, re-testing, etc. Ultimately, I hope it will be possible to make this work with a single 3D printed part, so the cost is much less.
 
#28 ·


It works, but is far too sensitive to be usable as it is now. It reads about 3500 when the PAR is about 41, so on the 50,000 scale it reads a little low, and on the 20000 and 2000 scales it reads way too high. Adjusting the height of the translucent dome above the diode works for about a +/-5% adjustment, which is good, but the overall sensitivity is just too high. It sure looks good though!

The three fixes for this would be making the dome from white acrylic instead of translucent acrylic, or making it much thicker above the diode, from ,050 inch to perhaps .50 inches, which is impractical, or making the dome much higher so it is much farther from the diode. I think the best approach will be using white acrylic, which is still a bit translucent. This material just transmits too much of the light.

I think I will try a different design for the next iteration, and use white acrylic for it. The design I'm thinking of is much closer to Jalopy's original design, but a one piece design.
 
#30 ·
The diffuser dome is made of a translucent acrylic that Shapeways uses. It will never reduce the intensity of the light striking the diode enough - just too transparent. The one I just ordered will be made of white acrylic, which may end up being too opaque, but I like this design better, and it "only costs" me $18 for one, with over half of that the shipping cost. This will be a one piece body, with the walls, other than right over the diode, thick enough to block most of the light. But, only trial and error will work to find a final design.

The acrylic I was using before, for diffusers, is the "frosted acrylic" rod from Tap Plastics. It works very well, but has to be cut with more precision than I can apply, in order to make the sensors reproducible without lots of adjusting and testing - very labor intensive.
 
#31 ·
Oh, btw. Thanks to your hard work with finding the 3 roscolux filters, I'm going to add them to my sensor. This will make it more accurate. I just bought the roscolux swatch, hopefully, they'll have all 3 filters in there. It's cheaper than buying a big sheet plus shipping.
 
#32 · (Edited)
Those little sample books are supposed to be free! I just picked up another one yesterday, and they had a whole carton of them to give away. My original book of samples is pretty well shredded from all of the pieces I have cut out to try over the past 2 years.

Jalopy mentioned casting housings for PAR meters, as a cheaper alternative to 3D printed ones. I've been looking at that too - TAP plastics has all of the resins, mold making materials, dyes, microballs, etc. needed, and they have a store near me. It is an interesting DIY method, but with a part having detail on both inside and outside it isn't the simplest thing to make a mold for. And, getting the right mix of resin, dyes and/or microballs to make a diffuser work right, plus being able to duplicate the mix easily will be a challenge.

EDIT: With a little more googling, it looks like making the mold would be pretty easy, so this could drop the price of the housings to a few cents each! (But, only after the trial and error process to get the proper mix for a translucent diffuser.)
 
#33 · (Edited)


My latest attempt to make a housing for a PAR meter through Shapeways has arrived. It is the all white one shown above. The basic design looks promising, but finding the right combination of dimensions is going to be very difficult. As is, this one is too sensitive except in the 50,000 range, where it is a little too insensitive. So, it takes more than minor tweaks of the geometry to make it work as it is now. And, I'm not sure that having the whole housing somewhat translucent is going to work. It looks like much of the light getting to the diode is from areas that are thick, where I expected to get only trivial amounts of light.

I will be doing more experimenting to see what to try next.

EDIT: After using very fine sandpaper to "polish" the diffuser part of the housing, I made it read almost exactly 2X the correct readings, with the readout meter set at the 50,000 range. I can work with that! My next step is to drill the housing for the electric cable, do a little more "polishing", and try to cast one with clear epoxy http://www.tapplastics.com/product/...ing_products/easycast_clear_casting_epoxy/386 colored with white pigment http://www.tapplastics.com/uploads/pdf/TAP Premium Pigment.pdf I should be able to change the sensitivity of the diffuser/diode combination by adjusting how much pigment I use.
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top