TSL245R & TSL237S light to frequency converters

[jwhooper]

I hope to actually add to the Picaxe body of knowledge for a change, and document how to get these sensors working with the Picaxe, and also report back on exactly what they can do.

TSL245R data sheet:
http://www.taosinc.com/images/product/document/TSL245R-e19.pdf

TSL237S data sheet:
http://www.taosinc.com/images/product/document/TSL237-E33.pdf

Why did I choose these particular sensors? Oh, no reason, really. I'm just a curious, like a monkey, and with equal electronics skill.

Once I have them working I will be happy to do whatever experiments the board wants. For instance, I know Dippy will want me to try and record the blood perfusion in an area of my brain, through the skin and skull with the ultra sensitive TSL237. That guy is such a dreamer. Then he will want to see how the infrared TSL245 sensor responds to heat.

Call him crazy, but let's all humor him ;-)

 

[jwhooper]

Oh, well they are utterly simple to hook up. 5v and Grd from the programming breadboard, out to pin 2:

Start:

Main:
Count 2, 100, w1
sertxd(" ,", #w1)
goto main

Up and running in 5 minutes. Read with the terminal.

The 237 has outstanding sensitivity, as advertised, sampling from 0 (dark) to 1,000 (lighted room) in my 100 millisecond increments.

The 245 infrared can't compete, sampling from 0 to 10 with the same program.

The 237 will shoot up to 2,000 if you hold a white envelope near it. It ignores black objects in proximity, which makes sense.

The 245 doesn't change much, but finally goes up a point or two if you get the white envelope very close. It changes (goes down) when you move a black object, or your hand towards it. It is obviously much less sensitive to light and would make a better motion detector, but not a very sensitive one.

The 237 really does seem to be spectacularly sensitive. Stay perfectly still and it barely wavers 1 point, but move or hold up a different color (about 6 inches away) and it varies by 100 points or more. That is in a well lit room, hanging off a wire.

 

[Dippy]

Haha. BBLs.
Oh, go on, call me crazy, humour me

Seriously though, I'm not really surprised about the difference in responsivity;
Data Sheet TSL237 > 2.3 KILOHertz / doodah responsivity.
Data Sheet TSL245 > 500 Hz / doodah. (IR filter).

Anyway, I am glad, that you've got some devices to play with, to get to grips with things.

Don't forget that the TSL237 is sensitive to IR too.

So, what's the next step then? Have you actually discovered from REAL measurements the resolution in these reflected levels of light that you're expecting to measure?

You've now got a lot of work to do.

 

[jwhooper]

I added a red LED and a Visual Basic program to chart the data, then I tried to see if it might make a decent pulse monitor. The red LED shines right through my finger. There is definitely a steady wave when it is reading, but I still have to scale it and figure out if it is my pulse. Yeah, it will take some tweaking.

 

[jwhooper]

It very clearly and easily records the pulse in my finger. This sensor is really very sensitive. On to more challenging experiments with it.

 

[higgeh]

have you tried the light to voltage converters. they are alot easier to use as you can put it into your analog input in your circuit

ive got three running on mine to turn my lights on and off in the shed

 

[Dippy]

Correct me if I'm wrong (more than likely), but it is a lot easier to get high resolution results with light to frequency - just by increasing the effective count period.
With light-volts you are limited by ADC resolution unless you have extra electronics to gain-up a particular range of V.
Something like that. Both are dead easy for general stuff.

 

[leftyretro]

Correct me if I'm wrong (more than likely), but it is a lot easier to get high resolution results with light to frequency - just by increasing the effective count period.
With light-volts you are limited by ADC resolution unless you have extra electronics to gain-up a particular range of V.
Something like that. Both are dead easy for general stuff.

I believe you are most correct. The data sheets show a range of nearly 0-1Mhz, which would be a 20 bit range I think Vs only a 10 bit A/D input resolution for the Picaxe voltage input pins.

Lefty

 

[higgeh]

after initial testing i have put in op amps to increase the range. I only measure about 40 mv total and its up to 4v

 

[Dippy]

Yeah but you're still limited in resolution by the ADC bits.... as described by Lefty.
If you are just looking at a small range of Volt output from sensor you can offset+gain the small range to make that 'window' bigger.
Anyway, horses for courses and if it works for you then absolutely fine. I was just trying to hint at the possible advantages of one type over the other.

 

[jwhooper]

In what I read researching these sensors, the manufacturer says that frequency is the easier and less expensive way of providing high resolution, for all the reasons already mentioned.