I would like to measure the optical transmission of tinted plastic panels "in-process" during a hot-water tinting process.

Ideally, the sensors will indicate that the correct transmission has been reached, and the panels will be automatically withdrawn.

The tinting is carried out as a batch process in a stainless steel tank using a water-based dye. Process temperature is 60 to 85degC. 10 sheets of plastic are spaced out through the tank, with room between them for the dye to circulate freely. Panels are complete when their visible transmission falls to 3-10% of the un-tinted value. This measurement is currently done manually.

I have some ideas, but what do you think? The dimension of the tank is ~500mm in the direction in which any optical measurement could be taken (ie. normal to the panels), which means standard emitter/detector systems may not be suitable (inverse-square law). Or am I wrong?

Many Thanks,

Tom Smith

 

Inverse square law doesn't apply to a collimated beam so that is easily overcome with lenses.

You might get more suggestions with a bit more data.

How thick are the panels?
What is the liquid space between them?
Do the transmission characteristics of the liquid change during the process?
What are the spectral transmission characteristics of the filter?
What are the spectral transmission characteristics of the liquid?
How do you measure when it's done manually?
Can you reduce the variability of the process so that you do not require a measurement? eg Tighten up the temperature control?
Do all the panels tint evenly?

Vince Dooley

 

If the panels must be measured while submerged in the dye, then you need to take into account the optical path through the liquid dye in addition to the path through the panels. Perhaps you could use a baseline measurement made through all the panels and dye at the very beginning as a reference and then wait until the reading fall a specified amount from the baseline. In this way you can factor out the variations in emitter signal strength and detector sensitivity. Certainly use an AC emitter waveform with synchronous detection to factor out ambient light.

Robert Scott
Real-Time Specialties
Embedded Systems Consulting

 

Hello Vince and Robert, and thanks for your replies...

In answer to your questions:

The panels are 1.5mm thick, and are spaced 25mm apart. The dye is pumped around the tank to ensure that the panels tint evenly. The finished panels are a deep green colour.

Transmission is measured manually using an instrument that includes a "white" light source (a bulb) and a detector. The machine is adjusted to read 100% transmission, then the sample is introduced and the reading taken. The spectral transmission of the dye is unknown. Spectral transmission of the completed filter is as follows: 100%T above 720nm, 100%T below 290nm, and approx 0.2%T everywhere else, apart from a smooth 3.8% peak at 510nm (green)(rising from zero at 580nm and dropping back to zero at 460nm).

The time for each cycle is between 40 and 90minutes depending on the age of the dye. After approx 100 cycles the dye is spent, and is replaced with new. Because of this you could say that the transmission of the liquid does change during the process. The change is so slow though that a reading taken at the start of a cycle should be reliable throughout.

The standard operating temperature is 65 degreesC, and is maintained with a PID controller. Other temperatures are used for different setups.

The only significant unconrolled variable is the 'age' of the dye. As far as

I can see there are only two ways of dealing with this:

1) Open Loop Control. Somehow measure the quality of the dye and add small fresh quantities at the start of each cycle. Once this is done the process still depends on everything else remaining constant.

2) Closed Loop Control. Observe the process directly, removing panels when they are complete, regardless of the cycle time. There is also the option to monitor the cycle time and add small shots of new dye to maintain it within acceptable limits. The main difficulty with this approach is observing the transmission directly.

Even so, option 2 has far less potential for scrapped product. Of the two options it's my favourite!

Anyway, your answers gave me some things to think about. Are there any more gems out there? How can I measure the transmission at 510nm?

Many thanks,

Tom Smith

 

Hi,

I am going to take a stab at this one. since you like the closed loop process better i will give a stab at that... though unlike other answers i have made here, this is more off the top of my head.

this one could be automated... but i will give the functional idea here. from the key part you can work out the mechanics.

The idea here is to make sensor calipers. you can make light pipes (j shaped glass rods or plastic that doesn't dye), but i would use small glass jars. if you use the jars you might want to put a silver tube inside... the idea here is to buy ultrabright LEDs in the nm range you want (there are many companies now that formulate them almost to the nm). one side will have wires that will provide the voltage to light the led up... the other side will have a led sensitive to that wavelength (red diodes are sensitive to the color they emit so can be used as sensors)...

now all you need to do is come up with a way of crimping the sensors on and off during the process since you can't leave them on while the dye is operating.

if you have questions or need help feel free to contact me...

Artfldgr

email me at
illuminy
at
h o t
m ai l