LM109 / LM309 Regulator IC - Understanding the schematic

[AG7CK]

On page 2 of http://www.ti.com/lit/ds/symlink/lm309.pdf one finds:




I would like to understand as much details as possible of the schematic. This is what I see or think I see:

D1 reverse voltage protection, D2 reference, D4 clamp, Q18/Q19 series pass element, R14 current sense resistor, Q14 current limiting, Q16 current source for R10, R10 voltage reference for overheating: appr. 0.35 volt, Q15 overheating protection, Q17 bias and more (what more?), R15/D3 current limit level reduction for high Vin, Q13 Q12 bias level reduction / load modification ?, Q10 Q11 protection ?, Q2 sense amplifier ?, ... and that's the end of my guesses.

I would like to understand every single component's function - including the strange multifunction Q17 and the FET Q1.

Thanks.

[T3sl4co1l]

That seems about right.  Q1 is for startup current; the only way they could make JFETs back then was in the substrate, so it's no accident that the gate is grounded, but that's just fine for a (small, very inaccurate) current source.  Afterwards, Q2 pulls up the bias node (Q3, Q4 and such), and R2 and R4 set the bias current.  Which in turn goes into Q17, a dual output current mirror (probably having some ratio of outputs).

I'm not sure where the bandgap is -- a ratio of currents and areas -- but it's between Q4-Q7, with Q7 being the "volt amp stage" (hence the Miller compensation cap).

Q8, 10 and 11 seem to be for limiting pull-down current, Q8-Q10 acting like a Widlar mirror and Q11 somewhat boosting the current (maybe Q9 or 10 is pretty small?).  Q9's pulldown current is limited by I(R8) * hFE, which maybe isn't a great idea?  Still, the pull-up current is given by Q17 period, so I'm not sure.

Q12-13 is probably there as offset to ensure Q2 doesn't run out of bias, even when the VAS is saturated (i.e., leaving ~four Vbe's plus V(R8) between output and ground).

Tim

[AG7CK]

Thanks a lot, T3sl4co1l.

I will study your circuit info and read some generic info about current mirrors on the web. I'll be back when I either understand all you say, or have more questions.

Does Q17 have 3 collectors? and mirror the current through Q1/Q2 at the two rightmost ones?

[T3sl4co1l]

Yes.  It's kind-of shorthand for three transistors with the bases and emitters tied together; except, if you probed the real element in the IC, you'd find there is mutual (inverted*) hFE between the collectors, because it really is made as a multi-collector device.

So, the important characteristic -- which we can emulate with discrete transistors -- is the multiple mirrored currents, but there are more subtle properties which could be used (but I don't think are used here) that we would not be able to replicate with discrete transistors.

*Inverted, meaning, treating each collector in turn as an emitter, and testing the hFE to the other terminals as collectors (including the one labeled as emitter).

But actually it's better than that: this was old fashioned lateral PNP, meaning, every PNP is essentially fabricated from NPN bases.  So they are symmetrical (down to geometry, of course), have high Veb (because of course, Veb = Vce), and, because of the light doping and unfavorable geometry, have quite low hFE (typically 4-10).

That's also why I noted Q11 can boost the pull-down current, but not by much: low hFE.  (Come to think of it, it could also be a parasitic structure around R9, which could be made as, essentially, a long base or emitter but given the values, it's probably intended for some function.

I've probably used a lot of terminology that doesn't mean anything without diagrams and explanations, so, do keep reading on the subject, should be able to find a lot that explains these old designs and fab methods.   If nothing else, keep asking.

Tim

[AG7CK]

Thanks again, T3sl4co1l.

As it was for no important reason that I started out with LM109, it slowly came to me that I should look at stuff like 7805 and 741 and so on. Turns out there is a large amount of tutorial sites on the web. Many includes and discuss partitioned schematics (just a random example - I haven't had the time to study this link yet - it was the first picture that turned up):


https://electronicsforu.com/resources/learn-electronics/7805-ic-voltage-regulator

It even turns out that there is one site that mentions the different versions (equivalent schematics) of the same IC from different manufacturers, and also shows similiarities between the LM109 and 7805: http://www.righto.com/2014/09/reverse-engineering-counterfeit-7805.html It even explains the bandgap circuit:




So I now have a general overview and will study the topic on my own.

Thanks again.