Current source

No, it's 2.5/R.

It's just a power supply. The current is I = 2.5/Rs, where 2.5 is the bandgap device voltage and Rs is the selected resistor. The 2.5 will be very insensitive to changes in the +15 supply.

John

Le 23/07/2011 19:59, John Larkin a écrit :

Yes indeed.

Le 23/07/2011 19:57, Jim Thompson a écrit :

That is correct. Not a problem in my case. For continuously variation of current Spehro suggestion is much more adapted.

Thanks Jim.

at

Which PMOS FET is that that you can control down to 1nA ? The MOS stuff is generally leaky...

Not the MOS stuff from Calogic:

Everything is wonderful, in Larkin's way :-) ...Jim Thompson

A nanoamp is no sweat at all. You can even do that with a small BJT--BFG25AWs have a decent beta down in the nanoamps, for instance.

Cheers

Phil Hobbs

Small pmos fets should be OK. Gate leakages are not many electrons per second, so there's no "beta" error. All you need is actual drain leakage under 1 nA. The opamp should of course swing close to the + rail. A couple of strategically places diodes could allow the opamp to back-bias the fet gate, if that would help.

Leakages are usually way below the datasheet specs, so you'd have to test a few to make sure they are OK. Something like a BSS84 maybe.

Hey, I can't do all the work, for free anyhow.

John

That would work, but it's expensive, I think. I'd expect that some cheap SOT-23 pfet would work, but you'd have to test some. Something with a high voltage rating, high gate threshold voltage, and high Rds-on, would be a good bet.

John

True, but you'd really have to luck out. There is no market for those anymore, so much has dried up and new P-FETs are all geared towards power rail switching in LiIon driven gear. Everyone wants guaranteed Rdson at 1.8V or 2.5V.

The small remainder of the market is with guys in white lab coats and huge research budgets.

Except for our beloved CD4007, wot has a couple usable p-channel MOSFETs in it for pennies (maybe dimes these days).

Best regards, Spehro Pefhany

I don't think Jim meant a "low-end opamp," I think he was referring to the low end of the range.

Have you considered range switching?

Like Jim says, what's the application? Can you tell us what you're trying to accomplish with this wide range plus precision? How much precision do you need?

What is it you're trying to accomplish?

Thanks, Rich

I've measured their leakage; it's tens of fA.

I'll measure a couple of gumdrop pfets on Monday maybe. Interesting.

John

Sweet femtoamps. ;)

It's the beta and f_T that make the BFG25 (one generation newer than the BFT25) so useful. Next nanoamp thing I do, I'll build a little curve tracer gizmo that works down there. It seems to come up a lot lately.

Cheers

Phil Hobbs

RF relays and Reed switches have always been my choice..

Jamie

Because the reference is with respect to the Vcc (+15V).

Maybe I am missing something how ever, I have to disagree there about not needing a stable 15+ source..

I can see that the whole point of the circuit is to monitor the voltage at the source of the fet as load is changed to maintain a constant current how ever, if the supply voltage is unstable this is also going to change things even when the load does not change.

It look simple to me, I just can't see how you say the 15+ does not need to be stable.

P.S.

Just my 2cents worth..

Jamie

Yes, it will change. Try to calculate the AMOUNT of steady-state change for a supply change of say, 15V to 14V with a fixed load to ground-- it's a useful exercise to get a feel for these things.

You could use page 25 on this reference for the subthrehold Id(Vgs):

You'll need the gain of the op-amp, so assume an LMC6001, for example.

Typically 1.4E6.

And say, an LM385 reference:

Typical Z is 0.4 ohm.

There are two main sources of change with the 15V supply- the reference voltage changes a bit, and the change due to the finite gain of the op-amp when the Vds of the MOSFET changes.

Best regards, Spehro Pefhany

Yes I know that, and what happens when that +15 is not stable?

I see potential issue. Lets say that 15+ drops a bit, it will also drop at the source of the fet, this will lead to less bias on the gate which will place the fet in a higher R point. Also that same reference point goes back to the (-) input of the op-amp and when it drops, it will increase the (+) on the gate of the fet which will also make an increase of R in the fet. SO, end results is, variable +15 source, variable effects on the load.

That's my observation, I am tired and maybe I am missing something that just does not stand out ..

Personally, I would've simply supplied the hole circuit with a precision regulated source.

Jamie