I've added a size/drilling template, please let me know if there can be improvements.JohnSG wrote: ↑Wed Jun 21, 2023 6:05 pmYes, at least for me. My guess is that most of the LED heat is transferring to the baseplate, so I think using a 3D printed baseplate will result in substantially higher LED temperature. There will be very little convection inside the base cavity, so the fins on the heat sink won't be doing a lot. The clamp arrangement doesn't help with this. But, if the heat sink has good thermal contact with the baseplate, the latter has a lot of surface area to transfer heat to the surroundings, so this is probably a better path. This is a guess, but I have a number of years of power electronics experience and a big part of that is experience with thermal management, so it is an educated guess. If I'm correct, a little thermal grease between the heat sink and the base plate will help. If you try it someday, use a very thin layer of thermal grease.
I can be pretty sure I won't be building this soon, because I can barely carve out time to get some microscopes going. I have just about gotten my bright field AO 10 into usable shape, and it's taken months. Just planning ahead.
John
American Optical Series 10 LED replacement for 1036A
Re: American Optical Series 10 LED replacement for 1036A
Re: American Optical Series 10 LED replacement for 1036A
Much background work, but not much progress.
In an attempt to gather some usable metrics I acquired a TSL2591 light sensor to measure lux..a few hours of learning how to use it with a microcontroller and now I've got a working light meter; however, it seems the lamp's intensity is higher than the sensor's capability at the bottom of the condenser. Time to see if I can a higher range, get a sensor capable of higher levels, until then I'll probably use an ND filter, probably the built-in 10%(?) ND filter, just so I can get some comparative results.
In an attempt to gather some usable metrics I acquired a TSL2591 light sensor to measure lux..a few hours of learning how to use it with a microcontroller and now I've got a working light meter; however, it seems the lamp's intensity is higher than the sensor's capability at the bottom of the condenser. Time to see if I can a higher range, get a sensor capable of higher levels, until then I'll probably use an ND filter, probably the built-in 10%(?) ND filter, just so I can get some comparative results.
Last edited by dtsh on Mon Jul 17, 2023 9:10 pm, edited 1 time in total.
Re: American Optical Series 10 LED replacement for 1036A
Some quick estimates, I setup the sensor 25cm away from the collector lens on the illuminator.
That correlates well with what I see when I use it.
Code: Select all
6550 lux for 1036A with 17w bulb
17450 lux for LED retrofit in 1036A
Re: American Optical Series 10 LED replacement for 1036A
This is not surprising. Since the rated lumen output of a lamp is measured in an integrating sphere, nearly all the light is captured no matter what direction it is emitted, but in practice not all the light is usable. One often forgotten advantage of LEDs is that the total lumen output is emitted into half a sphere maximum (2*pi steradians), and usually substantially less. With a filament lamp, the total lumen output distributed over a sphere (4*pi steradians). Hence, much more of the generated light is used by the optical system of the microscope when using an LED. With an incandescent filament, much of the optical output of the filament is lost. While a reflector can help, in practice they are often surprisingly poor, especially small reflectors.
This is all in addition to the basic efficiency (lumens/watt) of a typical white LED being an order of magnitude higher than incandescent lamps.
John
This is all in addition to the basic efficiency (lumens/watt) of a typical white LED being an order of magnitude higher than incandescent lamps.
John