Godox Witstro AD200 for UV photography

I discussed on another page the Godox AD200 electronic flash and its use in general photography. The modular construction of this unit, and in particular the quickly replaced flash head, make this model a prime candidate for UV photography, since a custom-modified head and UV reflector are all that is necessary for converting an ordinary AD200 to a UV flash unit.

For this modification, I decided to use a fused quartz xenon tube. Quartz (a crystalline form if silicon dioxide) is highly transparent to UV down to about 220 nm. Fused quartz is mechanically stronger than Pyrex and other types of UV-transparent glass, and this allows fused quartz xenon tubes to have a lower diameter than ordinary xenon tubes at the same power ratings.

At low current densities, the xenon emission consists mostly of distinct spectral lines, which is undesirable for photography. At high currents, the xenon emission approximates a black-body emission spectrum with a color temperature around 9,800°K, but at this point the arc inside the tube becomes opaque and absorbs most of the emitted radiation. It is therefore pointless to drive a xenon flash tube at very high current densities. Xenon tubes used in flash units for photography are operated at intermediate current densities (roughly 25-30 A/mm2) sufficient to stimulate a more-or-less continuous emission spectrum, but insufficient to make the discharge opaque to radiation.

Considering that UV imaging with a Bayer sensor is only possible down to about 280 nm (the usable limit is actually closer to 320 nm), and that radiation below about 300 nm is particularly dangerous for eyes and skin, I decided to use additional filtration in the form of a UV-grade acrylic sheet. This material is essentially non-stained acrylic (do not confuse it with UV-stabilized acrylic, which contains staining chemicals that absorb UV to limit the amount of UV-induced yellowing).

UV-grade acrylic is broadly used in suntanning beds, and cuts wavelengths below approximately 290 nm. In addition to UVA (315-400 nm), also some UVB (280-315 nm) is necessary for suntanning. Without the erythema induced by UVB, there is no darkening of the skin even after exposure to UVA. By manually sanding this filter, it also works as a diffuser.

The AD200 comes with modular heads. Both the speedlight-style head and a bare-bulb head contain a xenon flash tube (and a resistor to tell the unit which type of head is attached), and therefore either head, in principle, is suitable for modification.

Modifying the speedlight head involves a few unknowns. The plastic window that covers the tube and its reflector, in most speedlights for generic photography, cuts wavelengths below 400-420 nm. This is often, but not always, accompanied by a visibly yellow color of the plastic. This is not the case in the speedlight head of the AD200. It is also an open question whether the reflector of the speedlight head (usually a piece of plastic coated with a thin film of vacuum-deposited metal) is suitable for UV. Finally, the interiors of speedlight heads are usually cramped, their reflectors specifically shaped to accept a tube of a given length and diameter. It may be difficult to modify one of these reflectors to accept a physically different flash tube.

The bare-bulb head, instead, appears easier to modify. It accepts a 4-pin flash tube. Three of the pins appear to be standard 4 mm banana plugs, the fourth is 3mm and not connected (it is only used as a key to prevent inserting in the wrong way). Replacement heads and compatible tubes are available as a source of parts.

The original xenon tube is wound into a short helical spiral, and I have been unable to locate a quartz tube of the same shape and construction (a quartz helical tube is available, but it has a double anode and requires different electronics). However, my plan is to always use a parabolic reflector around the UV tube. Therefore, a straight xenon tube can be mounted coaxial with the reflector. In this orientation, only a little radiation from the tube directly strikes the subject. Most of the radiation emitted by the tube strikes the reflector and is then reflected toward the subject, potentially avoiding the formation of a central hotspot caused by direct illumination coming from the tube.

The internal surface of the metal reflector must be tested for UV reflectance and coated with aluminium foil if necessary. The reflector comes with a plastic diffuser that may need replacing with a sanded sheet of UV-grade perspex. However, test the existing diffuser for UV transparency, just in case.

Emission spectra of xenon flash tubes

I recorded the spectra displayed on this page with a Lasertack 200-1200 spectrometer with external triggering option spectrometer and the ASEQ Spectra software, then processed the data and generated the graphs in Excel. This spectrometer records data in an interval slightly exceeding 200-1,200 nm, with a data resolution of approximately 0.3 nm (but 1 nm is the precision rating specified by the maker).

In order to compare the spectra of different flash models and different intensity settings, I decided to normalize the data as:

Ln = (L - Lmin)) / ΣVIS(L - Lmin)

Ln normalized level value
L measured level value
Lmin minimum among L values (subtracting this from L eliminates the occasional negative values caused by noise, or systematic bias in either direction, so the minimum becomes zero)
ΣVIS(L - Lmin) sum of all adjusted L values between 400 and 700 nm (i.e., area under the curve in the visible range, in arbitrary units). This is enough to compare among spectra, if all spectra are taken at the same wavelength interval between adjacent measurements, in this case about 0.3 nm. Some extra processing would be necessary if this interval varies among spectra (e.g. one spectrum is recorded at 0.3 nm intervals and another at 1 nm intervals).

This normalization scales the area under the curve so that it is the same between 400 and 700 nm (i.e., the visible range) for all graphs. This provides a way to visually compare the amount of emitted UV, relative to VIS. If a graph shows consistently higher values than another in UV (i.e., the first line is largely above the second in the graph across the UV), then the first emission contains proportionally more UV than the second, relative to their respective VIS emission. In the following graph, for example, the red line is consistently above the green line, indicating a higher relative proportion of UV.

Spectra of electronic flash units.
Blue line: Metz 52 AF 1, full power.
Green line: Bowens Gemini 500R with non-coated tube, full power.
Red line: Bowens Gemini 500R with non-coated tube, lowest power.

The output spectrum of electronic speedlights for photography is typically limited to wavelengths above 400 nm by a UV-cut coating on the tube (or sometimes a stained glass used for the tube), often together with a yellowish color of the plastic window of the unit, which cuts UV and lowers the proportion of blue wavelengths. The above spectrum (blue line) is typical of modern speedlights. This is obviously useless for UV imaging, since there is no measurable amount of emitted UV.

"Non-coated" tubes are sometimes available as replacement parts for studio strobes. This is the case, for example, of non-coated replacement tubes marketed by Bowens for their studio strobes. Bowens units are sold with coated tubes that supposedly make the xenon emission more similar to daylight. Coated tubes display a characteristic yellowish or brownish tinge, suggesting a vacuum-evaporated coating with gold. Some photographers, however, prefer to use the slightly "colder", or more "neutral", emission of non-coated tubes.

Although these non-coated tubes are visually transparent and colorless, they are designed for use in studio environments, where both photographers and human subjects are routinely exposed to the emission of these units. Therefore, as shown in the above graph, emission does contain some UVA, but very little UVB and no UVC. Although UVA is, strictly speaking, not safe by any means, daily exposure to UVA from these strobes as part of one's work is only a minor fraction of the solar UV one would be exposed to by working outdoors.

The data shows that either the glass of these non-coated tubes absorbs the wavelengths shorter than about 300 nm, or the tubes are given an invisible coating (despite their name) to cut this part of the spectrum. It also shows that the relative proportion of UV is higher when the unit is fired at lower power. This agrees with the statements in the literature about the initial part of the discharge being UV-richer than the rest of the discharge. Since firing at low power shortens the discharge time, its UV-rich portion lasts longer, relative to the whole discharge.

L-6070Q quartz tube by Xenon Flash Tubes, in AD200 at full power.

The above graph shows the emission spectrum of a quartz tube fired by the AD200. Here we can finally see some UVB and UVC.