[updated 2019-02-13] [I will update this post again after testing the sensor]
Rather recently Vishay announced a miniature sensor under the name VEML6075 with two channels nominally centred at 365 and 330 nm. The peak width at half maximum is 20 nm. So, in practice it is a sensor measuring two regions within the UV-A band with the tail of one of the two channels extending into the UV-B. It is not a sensor capable of separately measuring the UV-B and UV-A bands. However, under sunlight it collects enough information to obtain a reasonable estimate for the UV Index (see the application note from Vishay for deatils).
It is not just a sensor but instead a sensor module with a digital interface. It has all the electronics for temperature compensation and for converting the analogue signals from the sensor into digital data with a rough calibration applied. The package of this sensor is 1 mm thick and 2 mm times 1.25 mm in area. The sensor itself is much smaller and it follows reasonably well the cosine law without any diffuser. Price? Less than 2€ as a component… and between 4 and 7 € for a breakout board.
A breakout board is a small printed circuit board usually containing a single, or very few components. The components included are only those needed for a single complex integrated circuit or sensor module to function, and given the small size of the components, the board allows easier soldering by hand of wires. I have bought two different breakout boards with the same VEML6075 sensor. They differ in size, the smaller boards has components on both sides, while the larger one only on one side. (Drag the slider to see the bottom of the boards.)
The images above cover an area of 23 mm × 17 mm. I took a pair of photographs at higher magnification, and as it fits a UV sensor, I photographed it both in visible light and in UV-A radiation. The whole image is 3 mm tall by 4 mm wide.
Technical information about the photographs
All photographs were taken with an Olympus E-M1 digital mirrorless camera, tethered to a laptop computer and controlled using Olympus Capture 2 software. A camera converted to full spectrum was used.
For the images at lower magnification I used a modern M.Zuiko 60 mm f:2.8 Macro objective, a Sunwayfoto FL-96 LED light source. I took focus-bracketed stacks of between 15 and 35 images, depending on the depth of the electronic components on each side of the boards. I merged the stacks of raw images using Helicon Focus 7 and edited and converted the images to compressed JPEG format with Capture One 12.
For the higher magnification photographs I used a Zuiko 38mm f:3.5 macro objective (Olympus OM-System ca. 1972-1975+, single coated early version). The visible source was the same, and the UV-A light source was a Convoy 2+ 365nm UV-A flashlight filtered with a visible blocking filter. For the UV-A photographs a used on the objective a zwb1 2mm thick filter.
All the images are slightly cropped from the full frame, most to better align them for the slider. The photograph below shows the nearly 50-years-old objective. It is very small and its mount is the same as used for microscope objectives.