>>2946720
>will differences between the transistors mean that it doesn't work IRL?
depends on how well it needs to "work". if youre cool with a 20% difference between the two then this would probably work. it would work even better if you put resistors R_E >> r_E on the emitters so they dont share a common Vbe. 100 ohms should be more than enough. 1 kohm would get you well under 1% matching.
>is there any kind of feedback mechanism here that would cause all the current to come out of one transistor only, for example?
for a given Vbe, current decreases with increasing temperature, so youre good.
>>2946723
>I'm trying to optimize on part count as much as possible
they sell dual op-amps.
>I understand that using a current source like this will be better.
the matching between the OTAs is almost assuredly better than your present circuit; datasheet says 0.3 dB which is either 7% or 3.5% depending on how they did the math. if you really want it tighter than that, i recommend either pic rel left with a dual op amp or pic rel right with a quad. you may have to adjust the gains with a pot to get them to track well.