Light meter on the head of a match.

I wasn't sure where to plant this thread. So if a moderator considers it out of place here, feel free to move it.

While searching for photo-sensors, I came across this amazing little chip from Osram.

Link: Datasheet for SFH-5711.

It's an eye-response photo-sensor, amplifier and logarithmic converter all in one. In short, it can easily be used to make a lightmeter or densitometer with very few additional components.

"Wow! That's got to cost a bit." - you might think; but no. They cost just over UK £2 each in small quantities.

One small drawback. A very small drawback. The thing is absolutely tiny. Like, one mistimed sneeze and you'd be on your hands and knees for several hours searching for it!

It's too small to handle on its own, so I soldered one to a small offcut of 'Veroboard'. Thus:

Not a giant matchstick BTW, just a normal sized one.

Here's a closer view of the chip with its pinouts. Osram kindly omit to tell you whether it's a top or bottom view they're showing you.

I shall probably use one to rebuild my densitometer, since I discovered that its spectral response leaves a lot to be desired.

That aside. Don't you think it's an incredible feat to get such a tiny light sensing area to give a useable signal? Let alone almost a complete light/luxmeter in such a small package and at such a low price.

If you read the specification, it becomes even more amazing. And these things are sold just to control the brightness of a mobile phone LCD screen. Unbelievable!

Makes you wonder what else is out there for peanuts that would replace something costing a small fortune just a few years ago.

The range is 3 to 80K lux it's kinda from EV0 to 15 or so. Too bad it's too small.

Too small?

Too small for what?

It outputs 10 microamps per decade, which is plenty for almost any purpose. In fact it could drive a 50 uA meter movement directly and give a readout in LVs.

The sample I tested was extremely close to the specified 10 uA/decade, and tracked my luxmeters almost exactly. It appears to have better linearity than my analogue moving-coil Sekonic luxmeter, but closely agrees with a no-brand digital Luxmeter.

Looking at the spec. sheet, the maximum lux figure it can handle is load-dependent, and 80K lux is Osram's most conservative top figure. The maximum reading I've seen from midsummer open sunlight is 120K lux, which isn't much of a stretch from 80K. Just another microamp or so. There's no reason why you couldn't use an aperture or ND filter to extend the range either.

I had to wrap black vinyl tape over the sensor to get minimum output, which was 0.3 uA. Taking log0 = 1 lux, that's only 1.07 lux. Pretty good by any standard. And look how closely the spectral response follows lambda V.

What else can you buy for two quid that comes anywhere close? A basic BPW21 photodiode costs over £7 these days. And only outputs a fraction of a uA per lux linearly.

Too small?

Too small for what?

It outputs 10 microamps per decade, which is plenty for almost any purpose. In fact it could drive a 50 uA meter movement directly and give a readout in LVs.

 

The sample I tested was extremely close to the specified 10 uA/decade, and tracked my luxmeters almost exactly. It appears to have better linearity than my analogue moving-coil Sekonic luxmeter, but closely agrees with a no-brand digital Luxmeter.

 

Looking at the spec. sheet, the maximum lux figure it can handle is load-dependent, and 80K lux is Osram's most conservative top figure. The maximum reading I've seen from midsummer open sunlight is 120K lux, which isn't much of a stretch from 80K. Just another microamp or so. There's no reason why you couldn't use an aperture or ND filter to extend the range either.

 

I had to wrap black vinyl tape over the sensor to get minimum output, which was 0.3 uA. Taking log0 = 1 lux, that's only 1.07 lux. Pretty good by any standard. And look how closely the spectral response follows lambda V.

 

What else can you buy for two quid that comes anywhere close? A basic BPW21 photodiode costs over £7 these days. And only outputs a fraction of a uA per lux linearly.

Too small to handle. If it's bigger it would be easier to use.

Thanks.

I'll probably never do anything with it, but I confess to being delighted with such things. More fun than an office supply store:)

After 'playing' with two samples of the chip for most of the day; I'd say it exceeds the published specification by some margin.

It agrees with two commercial luxmeters quite closely. In fact it sits in the middle of readings between an analogue Sekonic (reads high) and a digital meter (reads low). So I tend to believe the chip, rather than the meters! However, the meter sensors have integrating diffusers, while the chip obviously doesn't.

The upper detection limit of the chip appears to be in excess of 250 K lux. I managed to get an output of slightly more than 54 uA out of it, into the low resistance of a digital multimeter set to measure current. The sun didn't oblige today, so I had to stuff a high-brightness COB LED light right up against the chip.

Output current seems independent of supply voltage between 2.4 and 5 volts - practically the entire specified supply voltage range.

I take my hat off to the designer(s) of this device. So far it can't be faulted.

Too small to handle. If it's bigger it would be easier to use.

Well, yes. That's why I soldered it to a bit of old Veroboard. However, it didn't need any specialist equipment. Only a fine tipped and low-power soldering iron.... plus a bit of experience wielding one. The assembled module is now a similar size to the shallow TO-5 transistor can, like the BPW21 photodiode package.

Still doesn't have the panache of a Weston Master V.

Actually, you've got me thinking of how one might use it to remake the Weston Ranger 9. I always loved the thing, with it's 9 degree viewfinder, but it used mercury batteries and had accuracy problems that couldn't be calibrated out. Replacing the cell with the Osram, the batteries with regular alkalines and making a few other mods might turn it into a useful thing. Or not. The meter was probably marked to match the nonlinearities of the rest of it. I'll keep my Gossen Luna Lux SBC. Ordinary batteries and no mechanical movement.

Probably make a great enlarging meter too, if anybody still uses those.

For enlarging perhaps the 3 lx low end limit isn't sufficient.

Very interesting!

I don't have any ideas for projects popping to mind, but I think I'll head over to Digikey and buy a few to have for a rainy day.

Thanks as always for this, Joe.

It will absolutely work, I have already diy a reflected light meter using VEML7700 because it has side-view sensing, and a incident light meter using VEML6030. Light Meter / Exposure Meter see this link about reflected version. SFH5711 is great too, I'm considering make it into a spot-meter

It will absolutely work, I have already diy a reflected light meter using VEML7700 because it has side-view sensing, and a incident light meter using VEML6030. Light Meter / Exposure Meter see this link about reflected version. SFH5711 is great too, I'm considering make it into a spot-meter

[ATTACH]1351231[/ATTACH]

[ATTACH]1351232[/ATTACH]

Good link. Thanks!

Yes, the tiny sensor area would make a spotmeter 'easy' to fabricate. However, the main failing of spotmeters is that excluding adjacent light and flare from hitting the sensor isn't trivial. Usually an entire sequence of baffles and other anti-reflection measures need to be used. Although I'm sure a better job of it could be done than that of many commercial spotmeters.

The main point of a spotmeter would seem to be measuring contrast ratios. Well, a digital histogram does that better and quicker, and takes account of the actual camera+lens flare factor.

Good link. Thanks!

 

Yes, the tiny sensor area would make a spotmeter 'easy' to fabricate. However, the main failing of spotmeters is that excluding adjacent light and flare from hitting the sensor isn't trivial. Usually an entire sequence of baffles and other anti-reflection measures need to be used. Although I'm sure a better job of it could be done than that of many commercial spotmeters.

 

The main point of a spotmeter would seem to be measuring contrast ratios. Well, a digital histogram does that better and quicker, and takes account of the actual camera+lens flare factor.

You're right, flare issue is complecated. Matting sponge and sawtooth structure would help.

For a spotmter, a bigger active area of sensor would make it easier for the lens to focus on it, but using a large photodiode and amplifier greatly increase the complexity.