The MiSeq's Excitation LEDs!
The MiSeq is probably Illumina’s most popular instrument and my personal favorite. I’ve written extensively about it elsewhere. But I remain fascinated by it and want to dig further into its construction. So I’ve picked up some old MiSeq parts on eBay and will be slowly documenting them (paid subscribers are appreciated, and help cover these costs).
Today let’s look at the MiSeq’s LEDs! First, here’s a rough (and incomplete) diagram of the MiSeq optical system:
The LEDs sit within their own housing. This contains the LEDs themselves, a photodiode and a temperature sensor:
Let’s take each of these parts in turn. Starting with…
The Photodiode
This appears to be a Thorlabs FDS1001, used for to monitor the optical power output from the LEDs. While the output should be more or less stable I imagine it will vary somewhat with time, temperature and alignment. Using a photodiode for power monitoring will also mean the whole assembly doesn’t need to be calibrated at factory. The photodiode itself will need to be reasonably well calibrated… lucky Thorlabs will sell you a calibrated one for $200!!2
With the part number identified we can take a look at the specifications! This shows it to be pretty meh over the visual spectrum, but likely can be calibrated well enough for power measurements:
The Green LED
The Green LED seems to come from OSRAM. The PCB certainly looks a lot like this OSRAM part:
The die itself looks different to the one shown on the OSRAM LZ1-00G102 PCB above. However the LZ1-00G102 would make sense in terms of specifications. As previously noted Illumina patents suggest the use of Atto532 on 4 color instruments. If we line up the Atto and LED spectra up they’re a pretty reasonable fit. So this is my best guess for the green LED (it costs <$10):
The Red LED
The RED LED appears to come from Luminus. It looks a bit like the PT-39-L51. But the die is smaller (~2mm sq.).
If it’s an old or custom Luminus device similar to the PT-39-L51, the spectra seems about right for a red dye, and a reasonable match for the 660nm laser used in the Genome Analyzer 2:
The PT-39 costs about $100.
Summary
It’s worth noting that each laser in the Genome Analyzer 2 probably cost something like $10000. Here’s they’ve been replaced with a simple and comparatively cheap LED based assembly. The LEDs themselves cost <$100, uncalibrated the photodiode is ~$20 (and Illumina could easily self calibrate at the wavelengths they are interested in).
The most expensive parts are probably the Dichroic, lenses (let’s call it $100) and the custom milled housing.
So you’re easily replacing ~$20000 in lasers (in the Genome Analyzer and HiSeqs) with <$10003 in parts. Another win for the TradSeq!
The part number markings seem to match surplus suppliers of this part and the pictures look similar. In all likelihood Thorlabs source this from another supplier (perhaps Hamamatsu?).
Illumina probably calibrate it themselves, uncalibrated it’s like $20.
Possibly much less, certainly the active components are <$200. With some effort you could probably make this whole assembly very cheaply. But I guess when you’re selling the instrument in relatively low volume for >$100000, only so much cost optimization makes sense.