Transistor matched pairs

I'm beginning an amplifier project. Some transistor matched pairs are difficult to find. So, I have some questions. How many transistors should I buy in order to end up with one 2 matched pairs? Is the hfe data from a PEAK DCA75 accurate enough for matching? How close should the hfe numbers be, to be considered matched?
 
For differential/LT pair suggest use 2SA798 as a guide, ie, match both hfe and Vbe
1648694316762.png
 
DCA75Pro is a neat tool, but fairly limited. I usually buy a bunch of complementary pairs in suitable hFE ranks and then proceed to curve trace and divide them into narrower hFE groups. I mark them with coloured dots (acrylic marker) and when I need actual matched pairs I search for them within those tighter hFE groups, measuring them in a fixture at the target bias point. Of course if you go SMD you can drastically reduce the amount of matching required because there is still a decent selection of matched arrays available.
 
Matching JFETs can be important but rarely BJTs. That's the biggest reason why JFETs are not popular. I agree with rsavas about little variation in modern batch processes. Few amplifier circuits have a perfectly balanced LTP circuit so even a perfectly matched pair does not give you zero offset. The only place you want to be careful about matching is load sharing the output transistors, but even there a reasonable use of emitter resistors should keep them close enough, assuming the variation is not more than 20% and the design has some headroom. With a few exceptions, experts don't waste time and money being super picky about matching transistors. Unlike resistors, transistors are expected to be all over the block and the circuit is designed accordingly. If the circuit demands tight matching, then you want a better circuit, not better matching. Circuits without local nor global feedback are just nasty crude hacks.

BTW, devices like 2SA798 have their emitters connected internally, making them useless for a modern circuit.
 
Last edited: