Olympus BH2 BHS LED Lighting

[micro_pix]

I'm going to post a few images showing how a couple of LED alternative lighting systems for the Olympus BH2 BHS and the BH2 BHT compare with the factory fitted Halogen bulb.

The BHS was fitted with a 12v 100w system which provides plenty of light for most photography in all techniques, really the only reasons to replace it is if the internal electrics fail, you want to use flash for action stopping photography in light intensive D.I.C. or Darkfield or you no longer want that large hot box on the rear of your microscope. The BHT/BHTU, with its 6v 20w system however can benefit greatly from an LED upgrade, especially if you want to take photographs when using D.I.C polarisation, phase contrast or darkfield.

First for the BH2 BHS. This is a comparison between the standard 100w bulb, a Retrodiode 20w LED and Saul's LED/Flash unit for the BHS. All images were taken on a full frame Canon 5D MkIII using the NFK 2.5 photoeyepiece.

This setup was purely to test the relative brightness on full power so the camera was set to 1/100 sec exposure at ASA 100 in the most light hungry application (D.I.C). All units were at full power. The test was repeated with 20x and 40x objectives. Please don't infer anything from the quality of the image, it is purely a relative max brightness test with the condenser at full open aperture. The full power flash is only shown at 40x, it was obviously even brighter at 20x.
.

.

.

It's clear that the flash is the brightest, the LED on Saul's LED/Flash is the least bright and the Retrodiode 20W is slightly brighter than the 100w bulb.

Now a real-use check with a 20x objective in light-hungry coloured D.I.C. The images were very mildly post processed in colour balance as you would normally do but they haven't been sharpened otherwise altered. The flash and the LEDs were set to full power and the shutter speeds that gave these results are noted.
.

.
While the shutter speed for Saul's LED was slower the results were still excellent.

The next images are to demonstrate the evenness and colour of the light, the halogen bulb, Retrodiode LED, Saul's LED and Saul's Flash all provided good even lighting. The Retrodiode 20W has an integral fan on the heatsink and I have not been able to detect any vibration when using a 100x objective. The shutter speeds are noted but only Saul's unit was on full brightness, the Halogen bulb and the Retrodiode LED could give shutter speeds of over 1/2000sec at 100 ASA. These images were taken through 20x and 40x Olympus SAPO objectives. A blue KB-4 filter was in place for the whole of this test and the camera white balance was set to 5700k.

A correction - the top right image in the first group should say 40x not 20x - and this is how you spell Kohler! I don't have time to redo the images at the moment.
.

.

.

.

My experience is that the Retrodiode 20w unit is an equivalent replacement for the halogen bulb and may give 1/2 a stop more light. I read that going more and more powerful with LED's for microscopes may just result in a lot more heat and not a lot more useful light, the 20w seems right for the BHS. Saul's LED flash has a weaker LED and I wouldn't replace the 100w bulb and use it as the sole light source but with a flash fitted it really does open up a lot more photographic possibilities. In darkfield with a 10x objective I was able to use 1/64 power which is extremely short duration and great for stopping fast moving action.
.
This fast moving polychaete larvae was taken using Saul's LED/Flash unit

I'll add a comparison between the Retrodiode 10w LED and the BHT/BHTU 20w Halogen shortly.

Dave

[Pau]

Excellent and very clarifying comparison! Thanks for performing and posting it.

[viktor j nilsson]

This is such a great post! I actually have a question about flash illumination that I've been meaning to ask.

I have recently put together a combined flash/led illuminator for my Vanox AH, based on the same principle as Charles Krebs' setup - flash tube removed from the flash and placed at the location of the original filament, with a xm-l2 10w led immediately behind the flash tube. The flash tube is sharply visible in the back focal plane. The flash I used is a cheap Yongnuo YN-467 that's a little weaker than yours (the YN467 has a guide number of 33, the EX430ii has GN 43). But since I am not using a beam splitter, I guess that the light output is quite similar.

When I started using it, I was a little surprised that I needed to set the flash at a rather high setting. At 40x 0.95 in DIC (mid-gray background), I needed to run the flash at the highest setting, and I might even have needed to crank up the ISO a bit (need to check).

Yongnuo quotes a flash duration of 1/800s at full power. I think that this refers to the time to 50% intensity (t5), so in reality the flash exposure time will a bit longer. Say 1/250s. I was quite disappointed in this. I really thought that I would get more "stopping power" than this at 40x when using flash.

Is this consistent with other people's experience with flash?

[Pau]

Viktor,
I use a Yongnuo YN565EX (GN 58) with a cube beamsplitter in this setup: viewtopic.php?p=224611#p224611

I don't recall having used more than 1/8 power setting for 40X 0.70 DIC, although our systems are very different

[PeteM]

Terrific post - thank you.

[viktor j nilsson]

Viktor,
I use a Yongnuo YN560EX (GN 58) with a cube beamsplitter in this setup: viewtopic.php?p=224611#p224611

I don't recall having used more than 1/8 power setting for 40X 0.70 DIC, although our systems are very different

Well, that suggests that there is room for improvement. I can see at least two areas worth exploring: 1) the image of your flash tube appear somewhat more magnified and fills the BFP a bit more than mine; and 2) you have the rear mirror in place.

I guess that in combination with my weaker flash, this might explain why my system is less efficient. I wonder if I could incorporate the mirror somehow. Since i only use the LED as focusing light, I might get away with having the LED shine through a small hole in the mirror.

I'll post a new topic on this eventually. Don't want to clutter micro_pix's excellent thread.

[micro_pix]

I thought it would be useful to add this information. The dimmers/drivers used with Retrodiode LED's use pulse width modulation for dimming i.e. at full brightness the LED is fed a constant current and dimming is achieved by switching the current on and off rapidly - as you turn it down, the off periods get longer and the on periods shorter - so in effect it is flashing. On a full electronic (rolling) shutter the CMOS chip is read from top to bottom (or bottom to top) in sequence, to build the image. The read process takes time and during this time, if the LED is in anything other than full brightness, it is flashing on and off. It follows then that, as the CMOS is read, some rows may receive less or no light and this shows up as banding in the image.

Using high frequency PWM dimmers goes some way to minimising this issue - on the slower shutter speeds the PWM is switching the light on and off so fast that each row will get its share but on higher shutter speeds it's possible that some rows will still get less or no light. This is most apparent on Full Electronic Shutters and still somewhat apparent on Electronic First Curtain Shutters - which is what I am using here.

The main issue is - if you wanted a really bright LED to be able to stop the action at high shutter speeds, then, unless you are at full brightness (and you probably will never be when using a 20x objective and a 20w LED in brightfield or Darkfield but will always be if you are using D.I.C), you may well get banding in the image.

Here are two images in brightfield with a partially dimmed Retrodiode 20W LED at 1/320 and 1/1000 sec exposures. I reduced the exposure in post processing to make the banding more visible - there was no banding visible at 1/200 sec.

Dave
.

[micro_pix]

At full brightness the Retrodiode 20W LED is too bright for brightfield work at the lower magnifications and as I showed above, you can get banding with a dimmed LED at higher shutter speed so it struck me that it may be best to operate the LED at full brightness and use ND filters get the light level to match the shutter speed required. I don't have any ND filters but I do have a couple of high quality linear polarisers that are the same diameter as the illuminator opening in the base. With the 20x objective I set the light to full brightness, put both polarisers on the base and partially crossed them until the camera showed correct exposure at 1/1000 sec. With the 40x I used just one polariser which gave me a shutter speed of 1/400 sec on full brightness. The very pleasing side effect was that the resolution of the image was much better than when using a dimmed LED with no filters, and of course there was no banding at high shutter speeds. Using plane polarised illumination is obviously not suitable in all situations but it worked well here.

Dave
.

[Scarodactyl]

Neat! You can always add a quarter wave plate after the second linear polarizer if you don't want plane polarization.

[micro_pix]

Good point!

I’m thinking that the fact that the LED is still flashing on at full brightness when dimmed by PWM is responsible for the lack of contrast and resolution apparent in the earlier images. Saul’s LED performs better than the Retrodiode there because, being less powerful, it’s un-dimmed by PWM and giving constant light at the correct intensity for the camera settings - as is the Retrodiode when turned to maximum and dimmed by the linear polarisers. When operating on full brightness, and intensity is controlled by ND or Polarising filters rather than PWM dimmers, the LED light quality looks much closer to the factory halogen bulb results.

Here is crop of the yeast cell in brightfield using a 40x objective with the Retrodiode 20w LED on full brightness with an Olympus 50% ND filter (I found one) and a linear polariser over the illuminator. It's a stack of three images with a shutter speed of 1/250 sec. I think the result looks as good as you can get with the original lightbox and 100w halogen bulb.
.

Dave

[Duke]

On my LED-modified microscopes I've used Adjustable ND Neutral Density ND2 to ND400 filter.
These come in various sizes, so can be easily fitted to almost any filter holder.

[micro_pix]

What a great solution! So you could power an LED directly with its rated current from a bench power supply and use one of those variable ND filters, which would be cheaper and give better results than using a driver/PWM dimmer.

Dave

[viktor j nilsson]

Those variable ND filters are just two polarizers, one of which can rotate. So it does the exact same thing as the two polarizers you used.

[Duke]

Those variable ND filters are just two polarizers, one of which can rotate. So it does the exact same thing as the two polarizers you used.

They are circular polarizers back-to-back to be exact, not linear. I've got 2 of 30mm for epi, and 2 37mm for condenser.

[micro_pix]

Those variable ND filters are just two polarizers, one of which can rotate. So it does the exact same thing as the two polarizers you used.

They are circular polarizers back-to-back to be exact, not linear. I've got 2 of 30mm for epi, and 2 37mm for condenser.

I did wonder if they were using polarisers.

Dave