Convoy S2+ is good, but not really 365nm

[ulf]

My second Convoy S2+ arrived the other day.

 

I has been very impressed with the light output compared with all the other crap "365nm" flashlights I have accumulated over the years.

I have thought that the Convoy was based on a real 365nm LED but after I measured the spectrum I don't think it is.

 

 

It is normal that LEDs change the wavelength, normally increasing a few nm when they get hot, but almost 5nm is too much.

The spectrogram was recorded 1-2 minutes after I turned it on.

At first the peak was at slightly below 369nm.

[ulf]

Now I'm more confused than usual.

I let the torch stay on more than ten minutes to get hotter.

Then the peak had moved to 367nm, decreased wavelength for increased temperature!

 

As far as I remember, referring to all data sheets I have read, and LEDs I have measured, this is the first time I see doing so.

Interesting.

[Andy Perrin]

Are there measurement errors that could affect the apparent peak location?

[ulf]

The spectrometer is well calibrated and recently checked against a calibration lamp.

The Hg peaks in that area read 365.29nm and 404.61nm. (D 0.28nm, 0.04nm)

 

The only thing I have noticed is a small wavelength shift when light enter the spectrometer-port directly to the 50um slit a bit from the side.

That is why I always use a fiber.

 

This time the light was coupled into a 100um fiber via my big Avatares UV-Vis collimator.

 

I think the slope caused by the varying sensitivity caused by the grating is too small to shift the peak location.

The torch was hand held but I tried to hold it in a similar way and could not see any big location shift.

 

What else could cause this beside that the LED chip is responding like I saw to increased temperature?

[Andy Perrin]

In cases such as this, a repeat of the experiment may be in order! Paying particular attention to whether you see the "overshoot" and then the decrease back to 367nm again.

 

I don't know much about spectroscopy, so I could not tell you what sort of errors could be present on that side. (Do the spectrometers need to warm up also? Could you be seeing a temperature effect on the spectrometer side as well as the LED side?)

[Cadmium]

John's tests of the Convoy S2+:

http://www.ultraviol...__fromsearch__1

 

LED used in Convoy S2+ (365nm version): Nichia NCSU276A U365

http://www.nichia.co.../NCSU276A-E.pdf

 

In my opinion, the Convoy S2+ is the one to get.

[ulf]

Andy,

 

Thank you for the input.

Yes spectrometers are a bit sensitive to their temperature.

I never turn mine off except when I quickly reboot it.

It has been running for months now and the environment temperature has not changed a lot.

 

I will repeat the experiment in a more strikt setup. It is difficult to know the temperature of the LED chip.

I'll try to plot the peak wavelength during warmup

 

Steve,

 

Thank you for the datasheet.

 

I fully agree that the Convoy S2+ is the one to get!!

I already have two and a third is on it's way.

 

I am not complaining at all over the performance or usability.

It is an excellent light source for UVIF.

 

I'm only looking for the reason for

1.) That I measure an unexpectedly long wavelength when the LED start and is near room temperature.

2.) That the wavelength get slightly shorter after warmup and not as the diagrams show in the datasheet page 16.

 

This is nothing that affect how well it will work for us.

The behaviour bother me as an electronics engineer, nothing else.

 

I might have to change the title if the strange result is due to a measurement error, that I don't understand yet.

[Cadmium]

You might talk to John about it since he is the only other person I can think of that has measured that like you have.

[enricosavazzi]

I tested three Convoy S2+ and found their peaks to be 368, 368 and 369 nm. I did not do any heat stress test.

http://savazzi.net/photography/cheap365nm.html

 

There is always a variability of emission intensity and wavelength among LEDs, and individual LEDs are binned during factory testing into groups with different wavelengths (and intensity). In the case of UV LEDs, those with lower wavelengths and stronger emission command a higher price, so it is natural for most torch manufacturers to skimp on costs and choose from a cheaper "bin".

[ulf]

The data sheet indicate a wavelength span from 360nm to 370nm ±3nm, so the Convoy S2+ is well within spec.

Even if no indication in the data sheet about wavelength binning that is likely to exist and big important customers can cut special deals.

[ulf]

I have redone the measurement in a more organised way and have answers or theories explaining most of the questions.

 

The lesson is to use LiPo-batteries of good quality designed for high current in the Convoy S2+.

Those batteries are better att keeping up the needed voltage during heavy loads.

The one I used is not of that type.

 

The measurements:

This time the torch was mounted well in line with the collimator and I added an iris to limit the light into the spectrometer.

Last time the torch was handheld at distance in a less controlled way.

 

Just as last time the wavelength decreased over time beginning at 367.4nm and ending at 366.6nm.

However at the same time the intensity decreased from 28500 counts to 12500 counts!!

 

Conclusions:

The decrease is much bigger than possible due to decreased efficiency caused by any warmup.

 

An alternative explanation that fit instead is that the result caused by decreasing current through the LED.

With decreasing current the power heating the LED will also decrease.

 

The wavelength decreases, both when temperature and current decrease.

(Page 16 in the datasheet, second link, post #6))

 

The torch is powered by a single 18650 LiPo-cell

The forward voltage needed to drive current through the LED is close to the battery voltage.

 

Sooner or later the battery voltage will be too low. My battery reached that point very soon.

https://electronics.stackexchange.com/questions/32321/lipoly-battery-when-to-stop-draining

 

To select a good brand this link might be interesting:

http://lygte-info.dk/review/batteries2012/Common18650Summary%20UK.html

[Andy Perrin]

Ahhh, yes, I can see how that would do it! So the conclusion is that for doing precise measurements (where you care about a ±5nm error), you should probably operate the LED using a lab power supply that will maintain a constant current during the test. While you COULD try to find a better battery, it seems to me that for testing it should be a power supply.

[ulf]

Well I was mainly interested in if the Convoy S2+ was a real 365nm lamp or not.

 

When I saw the unexpected wavelength change I wanted to find the reason.

You are correct that a better way of driving the LED would be needed if you need to characterise the LED more precisely.

That was not my goal this time, only to solve the mystery I thought I saw.

 

The important lesson learned, that might be useful for others, is that the Convoy's intensity performance depend strongly on the battery quality.

[Oliver Krohn]

As far as I know, the difference between the Convoy S2+ and the Convoy S2 is that the S2 has a deeper reflector leading to less spill. Unfortunately I couldn't find any S2 with Nichia NCSU276A 365nm. Anyway, both flashlights can be equipped with filters to reduce visible light, but maybe that's not necessary for most applications. I have seen ZWB3 filters with diameter ~20mm, which should fit into the Convoy S2/S2+.

[Andrea B.]

The ZWB filters are not highly thought of here due to excessive visible and IR light leak in those tested. Also quite some few have been seen to have pitting and other flaws.

 

http://www.ultravioletphotography.com/content/index.php/topic/1313-filter-transmission-charts/page__view__findpost__p__13766

[Cadmium]

http://www.ultravioletphotography.com/content/index.php/topic/2553-another-red-seaweed/page__view__findpost__p__19538

[Andy Perrin]

Also be careful of (random) eBay sellers. The glass that is offered on there is often suspicious. Make sure you know who you are dealing with!

[ulf]

IR leakage and optical surface quality is not that important for filtering an UV-LED.

 

A cheap filter of ZWB1 type might work for fluorescence photography if the attenuation is strong enough in the visible area combined with the much lower light levels from the LED in that area.

For weak fluorescence it is more critical with correct filtering.

 

For scientific work it is important to use well documented quality equipment and a quality filter made by Schott or Hoya from a trusted supplier is recommended.

However just as many of us are using lenses that are not specially designed for UV it is no harm in experimenting with cheap filters if you are aware of their possible limitations.

 

Those of us that has access to a spectrometer :) can measure and verify the performance.

 

I got a few very cheaply from a big Chinese site Al.....s, just to measure how bad a SWB1 is .

They happen to fit in my Convoy S2+ too.

 

I will return with measurement results of the filters and filtered light from my Convoy.

 

I would expect that there can be a big spread in performance between different batches and makes of these filters.

My measurements might not be valid for other filters named ZWB1 ZWB2!!!

 

Edit: the filter I tested is more likely a ZWB2 not a ZWB1!!

[ulf]

Here comes my measurement results for my ZWB1 ZWB2 and Convoy 2+

First the transmission of the filter:

 

It has a rather big leakage in the IR range above 700nm

 

However the Convoy do not emit any light in that area:

Linear plot

Log plot

 

Then we can focus on how the filter and LED performs at the wavelengths around 400nm.

Filter (blue), torch (green) and filtered torch (red):

Zoomed in at the lower levels

 

At the peak of the LED (367nm) the filter has a transmission of ca 66%

At 405nm the filter has a transmission of ca 0.6%. => 1% relative transmission there.

The intensity of the LED at the same wavelength is less than 1% of the peak intensity.

 

The resulting emission of the filtered torch at 405nm will be < 10e-4 and this is decreasing rapidly with longer wavelengths.

At the slope I can measure the decrease is 1:10 för 8nm of wavelength increase.

 

If a long-pass filter with a cutoff wavelength > 410-420nm is used on the camera lens, I think this light source filtering will work reasonably well, at least if the intensity of the fluorescence isn't too weak.

However I do not know the relations between fluorescence and excitation intensities.

My guessing about this might be completely wrong.

 

I might have missed something.

Please let me know then.

 

I have not yet started with practical exercises in this area.

Reality always win over flawed theories.

 

Now I guess I have been boring enough and forced some readers to fall asleep. :)

 

(I have seen my typo in the name of the green curve. @#%...)

 

Edit: the filter I tested is more likely a ZWB2 not a ZWB1!!

[Andy Perrin]

Interesting. The Convoy emits a lot less visible than I would have expected, but I suppose that's relative to the UV it emits. The conclusion seems to be that since the Convoy isn't putting out all that much visible, almost any old filter will be enough here?

[Cadmium]

I have seen ZWB3 filters with diameter ~20mm, which should fit into the Convoy S2/S2+.

 

You would definitely not want to use ZWB3 for filtering 365nm torch for fluorescence photography. ZWB3 is like UG5 and U-330, and passes blue and green.

http://www.ultravioletphotography.com/content/index.php/topic/1313-filter-transmission-charts/page__view__findpost__p__13766

[ulf]

almost any old filter will be enough here?

 

No, I think that is not a quite correct conclusion.

 

The Convoy is very strong at the peak wavelength and the spill into visible violet is still rather strong.

You need a filter that has enough attenuation around 400nm.

 

My ZWB1 happened to be good enough.

It looks like a slightly flawed Schott UG1 2mm.

 

That might just have been a lucky coincidence that I got a good ZWB1.

Other cheap ZWB1 filters might have less usable transmission graphs.

 

If you could trust the graphs, the other ZWB-filters, ZWB2 and ZWB3 are less usable, having cutoff wavelengths that are further into the visible area.

Their peak transmissions are also not as good for 365nm LEDs.

They are blocking more of the light used for exciting the fluorescence.

 

Naturally it is always about how much light contamination you tolerate in your pictures.

It is all about tradeoffs.

 

With weak fluorescence and if you want to record more of blue and violet fluorescence, then the demands on the filters gets more stringent.

 

I fully support Steve's claim that Hoya U-240 U-340 and Schott UG11, 2mm are the optimal alternatives as filters for the Convoy S2+.