Replacement LED light build uses a few tricks

Microscopes have become essential work bench tools for hackers, allowing them to work with tiny SMD parts for PCB assembly and inspection. Couple of years back, mad scientist [smellsofbikes] picked up a stereo microscope from eBay. But its odd-sized, 12 volt Edison-style screw base lamp, connected to a 17 volt AC supply, burned off after a while. He swapped the burnt lamp with the spare, which too blew up after some time. Dumb lamps. Maybe the original spec called for 24 volt lamps, which were unobtanium due to the odd Edison screw base, but those would throw out a pretty yellow-orange glow. Anyhow, for some time, he worked with a jury-rigged goose neck lamp, but frequently moving the microscope and the lamp was becoming a chore. When he got fed up enough about it, he decided to Build a Replacement LED Microscope Light.

Usually, such builds are plain vanilla and not much to write in about, but [smellsofbikes] has a few tricks worth taking note of. He found a couple of high power, SMD LEDs in his parts bin. They were just slightly wider than 1.6 mm across the terminals. So he took a piece of double sided, copper clad FR4, and edge mounted the LED against one side of the PCB piece, twisting it slightly so he could solder both terminals. This works as a great heat sink for the LED while still having a very narrow profile. This was important as the replacement LED board had to fit the cylinder in which the original lamp was fitted.

The LED is driven by a constant current buck regulator, powered by the original 17 volt transformer. A bridge rectifier and several filter capacitors result in a low ripple DC supply, for which he used the KiCad spice functionality to work out the values. The LM3414 driver he used is a bit off the beaten track. It can run LEDs up to 60 watts at 1 amps and does not require an external current sense resistor. This was overkill since he planned to run the LED at just 150 mA, which would result in a very robust, long lasting solution. He designed the driver PCB in KiCad, and milled it on his LPKF circuit board plotter. The nice thing with CNC milled PCBs is that you can add custom copper floods and extend footprint pads. This trick lets you solder either a 0805 or a 1206 part to the same footprint – depending on what you can dig up from your parts bin.

What left us wondering was [smellsofbikes] statement that “The only annoying bit is it (LM3414) also needs to have its enable pin tied high, but the enable pin has a max voltage of 5.5V, so I have to derive 5.5V from the 20-ish volts of the rectified transformer voltage.” Looking at the LM3414 datasheet, there is a VCC output on pin 1, derived from an internal regulator. From the data sheet, “The VCC provides self bias for the internal gate drive and control circuits.” We are not sure why this could not be used to tie the Enable (actually the PWM pin) to 5 volts without having to use an external regulator. Instead, he added a SOT-23 regulator in his design to get the 5 volt to enable the driver. But the regulator he had on hand had an odd, inverted pin out, and had to dead-bug solder the SOT-23 LDO to make it work. Eventually, it all worked out great, so we guess all’s well that ends well.

To fit the LED PCB in side the original cylinder, he designed a 3D printed collar with tapered fingers which hold the PCB without having to use any fasteners. For the main PCB, a simple 3D printed housing was used to fit it inside the transformer box. Even at 150 mA drive current, the LED lamp is mighty bright, and the over-kill driver will make sure he doesn’t have to worry about a burnt out lamp any time soon.