Spectrometry: Neon/Xenon calibration lamp

Neon/Xenon calibration lamp

It has been known since some years that fluorescent light glow starters contain gas discharge lamps with various mixtures, depending on the vendor, that are more suitable for spectrometer calibration than regular neon bulbs. Most famous is the Relco 470, but I could not get that in Germany, so I bought well available Philips (now Signify) S10 ecoclick starters, which sounded promising for containing a mixture of Neon and Xenon and indeed they cover a large wavelength range.

This calibration lamp is powered by USB-C and uses an internal DCDC converter to power the lamp, which light is coupled into a fiber.

Mechanics

Like for the Neon calibration lamp, I use a case of two 3D printed shells. Inside is a tiny integrating sphere with a SMA905 port, a DCDC converter and a USB-C connector:

Everything is designed with OpenSCAD, if you want to change something. I printed the shells with a 0.4 mm nozzle, a layer height of 0.2 mm and an extrusion width of 0.4 mm. The quality is OK, but more fine settings or a more fine nozzle would have been better.

Optical SMA 905 connectors are very popular, but also very expensive for hobbyists. The subtypes SMA 905 A/B use 4 mm for the ferrule, like the metal part of electrical SMA connectors. Most common is SMA 905 C using a ferrule of 3.175 mm, also called F-SMA I. Devices have female connectors and cables have male ones.

You can save money by buying a female electrical SMA connector and pushing out its inside. The result is a connector with the right thread and flange, but it is typically 9.4 mm long, which lets the 12.0 mm long ferrule look out by 2.6 mm. It is also too wide at the end. This design uses the length to let the ferrule end up at the surface of the integrating sphere. The sphere port is narrow enough to center the ferrule. The SMA connector has 2.6 mm diameter holes and is made to be mounted with 2-56 UNC-2B screws. That is not recommended for M2.5 and M2.2 is rather uncommon, but M2 works as well. I designed a M2 core hole and simply cut a thread into the print. Adjust the connector to be centered to the sphere port.

The integrating sphere is painted white after mounting the SMA connector. The tip of the tube is fixed with a small drip of epoxy glue so that the electrodes are centered to the sphere port.

If you never used OpenSCAD before: Install it, load the part-* files, click on render and export them as STL.

Electronics

The DCDC converter has a shutdown line. Contrary to many seller specifications, this line must not stay open (see MAX1771 datasheet). I connected with it with some wire to the GND pin next to in on the PCB.

The high voltage ground must be connected to the lower electrode of the Neon lamp, which sits near the SMA port, because only that one glows.

In its intended use case of starting a fluorescent lamp, the starter is powered with about 30 mA for half a second, which at 20,000 starts yields a life time of 10000 s or 2.7 hours. If you read about those starter lamps having a life time of just some hours, that is why, and the answer is: Don't run them at such excessive loads. That current is high enough to heat up the bimetal electrodes until they close. I want the bulb to live longer than that and I do not want the bimetal electrodes to close, because that would short circuit the DCDC converter.

I used a 27 kOhm resistor in series with the lamp and adjusted the voltage until I saw a resistor voltage drop of 27 V = 1.0 mA, which is 30 times less than usual. The lamp shows a voltage drop of 194 V, which is significantly more than small Neon lamps. If I reduce the voltage some, the back side of the electrode does not glow, and that is the most suitable spot to couple out light. Interestingly, its voltage drop appears not to be as constant as for Neon bulbs, perhaps due to its asymmetric electrodes? The life time is supposedly much more than linear to the current reduction, which should yield a sufficient life time as a calibration lamp which is only ever used for a few seconds to take its spectrogram. At 27 mW, a 1/4 W resistor is plenty. If you reduce voltage during adjustment, then check if the lamp still starts. Its glowing voltage is lower than its starting voltage!

The DCDC power supply is soldered to the USB-C breakout board. The solder spots on the underside must be fairly flat, otherwise the board no longer fits in the case.

The wires of the tube are crimped to the socket. Cut them as close to the crimping spots as possible. They appear to be coated with something that refuses flux and solder, but that coating can be scratched off with a sharp blade and then soldering is no issue. Isolate them with heat shrink tube. You don't want 220 V getting in contact with anything.

Optics

The light at 1 mA is rather weak and the large tube makes it very hard to couple it into the fiber. I placed the lamp near the fiber entry, yet only a tiny fraction of the light will couple in directly and the rest goes in all directions. The electrodes glow in a rather small area near the end of the tube, which requires to move the SMA905 port off center to keep the integrating sphere small. Overall, it is not a great solution that requires 10 times more exposure time compared to the Neon glow lamp.

I bought two starters and they differ significantly in how much blackening their tubes show. I guess that is from burn in at the factory. I used the one with the most clear tube at the back of an electrode and connected that electrode with GND to glow.

Visually, I saw only the major line from Neon, nothing else, but with 2 s exposure time the image changes a lot. Neon shows a characteristic spectrum that is easy to identify: A very bright orange peak (shown yellow by the camera, but visually looks orange) starts a series of peaks towards longer wave lengths that begins with two pairs of lines, then a single one, followed by two more narrow pairs. Xenon has a similarly nice pattern: Three blue lines followed by a larger one in between other groups. Together, that's all you could ask for in the visible spectrum.

The dominant lines of Neon are (data from NIST Strong Lines of Neon) in nm:

540.05618
576.44188
585.24879
588.18952
594.48342
597.46273
597.55340
602.99969
607.43377
609.61631
614.30626
616.35939
621.72812
626.64950
630.47889
632.81646
638.29917
640.2248
650.65281
653.28822
659.89529
667.82762
671.70430
692.94673
702.40504
703.24131
717.39381
724.51666

The dominant lines of Xenon (and possibly something else) are (data from NIST Strong Lines of Xenon and Newport) in nm:

450.1
452.47
458.26
460
462.43
467.12
469.70
473.4152
480.702
482.971
484.433

It is interesting that NIST does not list 452.47–469.70, but Newport does. I see those lines in a few other spectrums on the net. Is Xenon possibly commonly contaminated with something else or is there something missing in the NIST data?

This spectrum was acquired as a single shot of 64 ms exposure time with an Ocean Optics USB 2000+, which calibration has long expired (data set). Note the asymmetry in peaks towards the right end, which supposedly is coma from that type of folded light path. That asks to check the spectral PSF.

This spectrum is a single shot of 2 s exposure time with a 100 µm slit, 1000 lp/mm transmission grating, magnification 0.5 and a Canon EOS 350D. The higher resolution is obtained by a much smaller range and the need for more light, but in this case, the first spectrogram was not well exposed.

BOM

Lower and upper case shell: 3D printed from OpenSCAD files
SMA connector, female: Reichelt but much cheaper from Ebay. Press the inside out with a pin.
2x M3x8 mm screws with countersunk head or 2x M3x6 mm screws with cylinder head
2x M3 threaded insert
4x M2x5 mm screws with cylinder head
Philips (Signify) S10 ecoclick starter. Buy multiple, because the tubes differ in blackening.
27 kOhm resistor 1/4 W
Heat shrink tube isolation
White paint
Clear glue, e.g. epoxy to fix the tip of the Neon lamp
USB-C breakout board (see below)
MAX1771 DCDC converter 150–220 V (see below)