You can put low leakage diode between BD136 and 5.1k
good point! without considering the reverse leakage current they have, i had wanted to use a schottky blocking diode (on the left) to minimize losses due to voltage drop - there is very little power available from the mains transformer in the fluke 45 as it has been designed for operation on anything from 90v to 264v
without switching of windings. across 8 cells using schottly instead of regular silicon diodes would have saved nearly half a watt.
i'm now warming to the idea of just using a silicon diode on the right,
after the 5k1 resistor, and putting up with the (small) variable voltage drop as the charging current goes from 200mA down to 100mA. this then allows me to throw out the 2N7000.
Alternative solution is use two back-to-back mosfets as 'low leakage' bi-directional switch
problem is, anything other than a 2N7000 that has reasonable R
SD(on) is relatively expensive. i'd prefer the associated components to be well below the cost of the Li-Ion cells they are charging.
just did a quick test of V
f on a random 1N4002 diode:
50mA - 0.74v
100mA - 0.78v
150mA - 0.79v
200mA - 0.80v
250mA - 0.81vi think i can live with this. revised schematic:

You can think of small relay too, working only when charger is enabled. There is no discharge protection if I see
i did consider small relays, but with 8 of them the coil current starts to mount up. also there is the expense.
each cell will have a DW01A protection circuit (the assembled boards for this are around 50 cents each) that provides for under-voltage lockout.