current multiplier

Bet you could also do it with an opamp powered off the input. Or, given that it's probably inside a pid loop, why not just...

--------+-----------g fet or bipolar | s | | r r | | gnd gnd

Which raises the idea of making a current mirror that *adds* to the input current, rather than wasting it in a resistor and replicating a copy of it. Yes, there is the compliance issue in this case.

John

Interesting idea--I like the common-base diff pair. I'm a little puzzled, though: Since your TEC controller is always going to be used closed-loop, why worry so much about accuracy in the booster?

For lower-current applications I usually use an unbalanced current mirror--connect the output transistor's collector to the supply, come in the input, and come out via the common emitters (the TL014 was my fave,

Cheers,

Phil Hobbs

It's not very accurate, 2 to 3%, just enough so you can label it x 2.5, instead of "booster," without saying how much. A primary motivation for a semblance of accuracy is so the controller's digital meter still has meaning.

I agree, I used the TL014 in several of my low-power products, buying a few thousand, but still not enough to persuade TI to keep it. I think part of the problem was where to put it in the catalog so engineers would find it. I still have a few left, but they're too precious to use.

Yes, that was my first idea, but I figured a few more parts would give me that "semblance of accuracy," and also let me use a MOSFET, which is better suited for delivering power to a heatsink.

There's probably a good low-V opamp for the job. Its lower offset voltage would mean smaller resistors could be used.

. INPUT x2.5 current booster, 10A max out . CURRENT . o--, IRL2505 ,------o -- TEC ---o . | Q4 | . | |--' | . +--------+------||

Hello Win,

There are quite a few of those "too precious to use" products from unobtanium land. When the last order bell was clanging for the LH0063 high speed buffer I rushed to stock up on those. Plus some crosspoint switches that were vanishing around that time as well, never to be replaced by anything else. That prevented a lot of grief when I had to create a hi-Z test fixture for the HP3577.

In the same way I treasure a little stash of Ge diodes and transistors. It's just too easy to build sensitive passive RF detectors with these. No batteries that can leak.

Regards, Joerg

I don't know which is better, a PNP or an NPN Darlington... (Note the change in opamp sign.)

. INPUT x2.5 current booster, 10A max out . CURRENT . o--, IRL2505 ,------o -- TEC ---o . | Q4 | . | |--' | . +--------+------||

Yep, It'll be unstable.

Since the ratio is not mission critical, why not just do an NPN Darlington mirror?

...Jim Thompson

Hmm, matched Darlington transistors...

. INPUT x2.5 current booster, . CURRENT 10A max out . o--, ,----+---o . | | | . +-----, | | . | | | | . \\| | |/ |/ . |---+---|----| TIP145 . V/| |\\V |\\V . | Q2 Q3 | | Q4 Iout = Iin 2 R1/R2 . | | | . R1 R2 R3 Q2-Q4 - matched Darlington . 0.125 0.1 0.1 R1-R3 - 5W stable WW 1% . | | | . o--+-----------+----+---o

I wonder if matched power NPNs wouldn't be better?

. INPUT x2.5 current booster, . CURRENT 10A max out . o--, ,-----+----+---o . | | | | . Rx | | | . | |/ | | . +---| Q1 | | . | |\\V | | . \\| | |/ |/ . |---+---|----| TIP35 . V/| |\\V |\\V Iout = Iin 2 R1/R2 . | Q2 Q3 | | Q4 . | | | Q1 - 100mA NPN . R1 R2 R3 Q2-Q4 - matched 5A NPN . 0.125 0.1 0.1 R1-R3 - 5W stable WW 1% . | | | . o--+-----------+----+---o

Perhaps resistor Rx would help stabilize the circuit, without bothering the input current source.

There you go, Win, you're learning to be an I/C designer... 3 matched transistors and one "helper" ;-)

Split your stabilizing resistor and put some of it in the emitter of Q1.

...Jim Thompson

They made a pretty nice 5x current multiplier as well as a 4x mirror.

I even have some LH4009s--a very short-lived improved LH0063, sold in

Cheers,

Phil Hobbs

Hello Phil,

Maybe a management shake-up? Anyway, I learned a lesson in the 80's. If a component is almost too good to be true, it usually is. Haven't really been burned by it but seen enough others.

I do miss hybrids though. Designed quite a few.

Regards, Joerg

Matched transistors maybe, but with severely mismatched operating temperatures, due to Vce mismatch. More parts are needed to complete the task and insure matched Vbe, allowing small lower-power Rs. There, this should do it.

. INPUT x3 current booster, . CURRENT 12A max out, 1.8V overhead . o-----, ,-----+----+----+---o . to 4A | | | | | . --> Rx | | | | . | |/ Q10 | | | . +-------| | | | . | |\\V | | | . \\| | |/ |/ |/ . |--+----+---|----|----| TIP35 . V/| | | |\\V |\\V |\\V . Q6 | | |/ | | | Q9 . | '--| Q1 | | | . | |\\V | | | . \\| | |/ |/ |/ . |-------+---|----|----| TIP35 . V/| |\\V |\\V |\\V Iout = Iin 3 R1/R2 . Q2 | Q3 | | Q4 | Q5 . | | | | Q1, Q10 - 400mA NPN . R1 R2 R3 R4 Q2-Q5 - matched 5A NPN . 0.10 0.1 0.1 0.1 Q6-Q9 - more 5A NPN . | | | | R1-R3 - 3W stable WW 1% . o------+---------------+----+----+---o

Hmm, it's not clear to me what would be going on there, why would that help?

I've seen them oscillate due to capacitance on the emitter of the helper.

...Jim Thompson