Both 740B's are looking good with my hybrid choppers.
I updated the writeup to version 1.1 . Most of the change is a simpler board insulator, LDPE film instead of semirigid PET. The LDPE is easy to cut with a craft knife and you can use the board itself as a template.
Just for fun I measured the generic CdS photocells in my junkbox. Their T5 runs the gamut from five times the HP cells, to 25 times. Chopper cells are super fast. According to a Journal article (I think), their active material is a mix of CdS and CdSe, the latter being fast but unstable. A couple of niche scientific companies make PbS and PbSe cells but they are crazy-expensive.
I also used a tight beam to probe some HP cells looking for fast and slow spots. Nope, each cell is uniform across the whole active area. The slow cells were coated with a bad batch, not an insufficiently-mixed batch.
I put the fastest eight cells in my two 740B's. Here are my notes for the other cells.
PHOTOCELL DATA
Current for 20K, T(20K-100K)
1 3uA 8ms NOT USABLE
2 20uA 3.7ms USABLE BESIDE A FAST CELL
3 50uA 13ms NOT USABLE
4 20uA 3.2ms USABLE BESIDE A FAST CELL
5 45uA 3.2ms MAYBE USABLE BESIDE A FAST CELL (1)
6 15uA 2.8ms USABLE
7 65uA 2.4ms GOOD
8 20uA 2.1ms GOOD
205 350uA 1.8ms GOOD (2)
PC5191 7uA 4.4ms USABLE BESIDE A FAST CELL (3)
(1) Cell resistance rises to a relatively low value asymptotically.
(2) Was A9V1, marked PC5191-205
(3) Was A9V1, marked PC5191
Most cells R vs t curves upward a bit, so T(20K-100K) < 0.5 * T(10K-100K).
T10 10K-100K 20K-200K 30K-300K 40K-400K
2 5.3ms 7.8ms >10ms >10ms
4 4.6ms 5.5ms TBD 9.3ms
5 5.4ms 8.6ms >10ms >10ms
6 3.8ms 5.1ms 6.0ms 7.0ms
7 3.2ms 3.9ms 4.5ms 5.3ms
8 2.6ms 3.5ms 4.2ms 4.9ms
205 2.3ms 3.5ms 4.8ms 6.2ms
PC5191 5.3ms 6.8ms 8.0ms 9.4ms