Why do these precision current sinks use both a JFET and BJT instead of one FET?

Question

I was looking through the AN-31 Op Amp Circuit Collection from National Semiconductor (now TI) and saw this precision current sink:

I understand that the 2N3456 JFET is there to remove the beta error caused by connecting the BJT base directly to the op amp output (as in this circuit). It also seems to be the case that the BJT is necessary to provide enough output current, as the 2N3456 datasheet indicates that the JFET can sink at most \$I_{\text{DSS(max)}} = 1\text{ mA}\$. Why not just use a JFET (or a MOSFET) that can sink more current?

It's not just National/TI that has this circuit -- Linear Tech's LT1001 Precision Current Sink uses the JFET + BJT combination as well:

Again, the datasheet for the 2N3685 JFET indicates that it has a low \$I_{\text{DSS}}\$ and the BJT is needed to increase the output current capability. Is that the only reason why the JFET + BJT is used, or is there some other benefit to using a JFET + BJT instead of a single FET that can sink a higher current?

score 7 · Answer 1 · answered Feb 10 '17 at 21:10

7

The important point here is that all the current goes thru R1. That's what makes it a "precision" current source. There is no error due to base current.

The point of the FET and NPN transistors is to provide current gain so that the output current can be a lot more than the opamp output can provide. In fact, in this case the opamp output current is ideally 0.

If the opamp is powered so that its output can swing to 12 V or so, then just a single modern MOSFET would suffice, even for very large currents. Note that the LM107 is very old.

answered Feb 10 '17 at 21:10

Olin Lathrop

310,974
36
428
915

You stole my mind LOL+1 – Andy aka Feb 10 '17 at 21:17
So MOSFETs from back in the day when these circuits were designed weren't good enough? Why -- was it an electrical limitation or cost or something else? – Null Feb 10 '17 at 21:20
The size of the silicon slab used was very limited. – Jack Creasey Feb 10 '17 at 21:28

score 2 · Accepted Answer · answered Feb 10 '17 at 21:16

2

Both of the example circuits are old; the 2N2219 transistor is a metal-case unit no longer easily available. Power MOSFET devices would be a good choice today, but were not economic thirty years ago.

answered Feb 10 '17 at 21:16

Whit3rd

7,177
22
27

I suspected it might have something to do with the limitations of old MOSFETs, though I'm surprised that these are still the example circuits given on the TI and LT websites. I can sort of understand why TI wouldn't bother updating an old app note, but I find it strange that LT wouldn't update one of its "solutions" on its website. – Null Feb 10 '17 at 21:24
Modern MOSFETs would be good from a component limit point of view. However, you'd need to be careful about the gate capacitance causing stability problems. – WhatRoughBeast Feb 10 '17 at 21:45
MOSFETs not used as switches don't have effective capacitance-to-ground loading; it's a capacitance-to-channel, so depends on sense resistance as much as MOSFET characteristic. – Whit3rd Feb 10 '17 at 22:48
@Null: one of LT's app notes (AN-5) mentions LT1001 and is dated 1984. Probably updating solutions is not a business priority. The old solutions still work! – Whit3rd Feb 10 '17 at 22:58

Why do these precision current sinks use both a JFET and BJT instead of one FET?

2 Answers2