jfet as current limiter

Il 15/10/2012 16:27, John S ha scritto:

Well well well, the simulation is very nice. Tomorrow I'll change the pcb to try this version.

Thank you Marco

I tested the LM317L constant current source circuit using a 51ohm resistor and mine limited at 24mA, changing by less than 0.1mA from 3V to 24V with no visible temp sensitivity. Only downside is that as Phil Hobbs pointed out it drops about 2.9V just before going into current limit. At 10mA the voltage drop is 2 volts.

I think whit3rd's version is much better. His will not drift with temperature noticeably. I would use the LM431. See the data sheet for applications and you will also see whit3rd's version there.

Il 15/10/2012 21:25, Rocky ha scritto:

The problem is not the voltage drop, but it should work down to few mAmps without affecting in any way the linearity. That is if the current source sources 6.84 mA or 7.53 mA I must get 6.84 and 7.53 mA at the output. This over the whole 4-20 mA range.

Marco

Il 15/10/2012 22:03, John S ha scritto:

But his solution needs a third connection. Unfortunately, I should rely only on the two terminals (series connection).

I just need to limit the current below 30 mA and above 20 mA. If the real world knows the result of the spice simulations of your suggestion it should work fine ;)

Marco

I don't understand, Marco. Where is the third connection?

John

Oh! I just realized you were looking at a different post by whit3rd. His

(warning: bad ASCII art, use monospace font)

(+) | +-------+ | / | / R1 | +----| | |> | \ | | | | | | /---/ | / \ | / \-----+ --+-- R2 | | +-------+ | (-)

I_limit = 2.5V /R2 + (minor correction for zener bias)

which can be redone with an LM4041 for smaller Vref, using a PNP transistor... the range of 23... 30 mA should be easy to hit with 2V or so of voltage burden.

It's strange that you experienced that much of a change in Beta in the short time, I've never experienced that much of a change, but then again, you did add that extra R that I didn't put there.

All is not lost however. Below is a circuit that was posted by Phil H which is something of an old relic but it seems to work nicely with out the use of a reference. It uses the Vf of the base and also with the way it's configured it'll counter balance the beta while self heating takes place.

As Q1 self heats, the Vf (base) will drop and thus Beta will increase. When this happens however, current limit will increase, but it'll also force Q2 to pull R1 down, causing a - feed back effect.

R1 should be selected for lowest Loop supply, in this case I picked 5 volts but you can do what you wish. R2 is the base line of the limited value. You can replace Q1 with a low Vgs(th) NMOS and still maintain a desired value with out the worries of the NMOS process variations. I have sims for both..

Well, its 2 components, easy to build and cheap - maybe hard to simulate if you don't have the LM317L model, but being 2 terminal it won't affect the current from a current source. For 'guaranteed' stability I put a 100nF cap across the circuit.

Il 15/10/2012 23:13, John S ha scritto:

Yep, I saw the wrong post :) Sorry!

Thank you! Marco

supply

voltage

give

that

get

batches.

Tech, not engineer; but then again you admit that don't you.

?-)

Ahhh. Then what you want is a 4-20 mA transmitter IC. Readily available.

?-)

Il 16/10/2012 07:09, josephkk ha scritto:

Nope. I want just what I asked :) A bipole which limits the current over 20 mA and which is completely transparent for currents below that.

Marco

Il 15/10/2012 18:10, Marco Trapanese ha scritto:

It works like a charm up to 75 °C! The current changes only a little: from 24.1 mA @ 18 °C to 26.8 mA @ 75 °C.

This time the correspondence between the simulation and the real circuit is quite impressive!

Power of the emitter resistor.

Marco

Il 16/10/2012 02:22, Jamie ha scritto:

If you're referring to the R in the bottom is the simulation of the load.

If you add ".step temp 0 80 202 to the spice's directives you will see how much change the Ilimit w/ temperature (about 5 mA). Using the circuit proposed by JohnS the current changes about 1 mA only.

It makes the things easier to match the constraints dealing with tolerances and differences among batches.

Bye Marco

That last circuit is not a simulated run, that current limited is a classic and i've used it many times over the years. I don't know why you're having such wandering current problems like that, it does not happen in real life when I use them and they get used in many places that we employ control units that have no current limiting.

I beginning to think you are using different parts and different scenarios..

Use what ever works for you,.

Jamie

available.

It has volts of overhead but it will do just that. Just setup the inputs for full scale (20 mA).

?-)

Both of us, definitely. But the circuit is older than I am.

Cheers

Phil Hobbs

I put together the 2 btj version here at home base using 2222's, 10k and 31 ohm I r.

I did this early this evening, I applied enough to generate 100mA and it started at 20.6mA using a 15 volt supply. It has been 1.5 hours and I just looked at the amp meter, it's reading 20.7ma and flickering back down to 20.6 mA. that maybe no more than 100uA at most..

When I first tested this, I put my hot air wand on it for a short time to get the btj's hot to the point of near destruction, I saw no more than 2mA increase. The - feed back (not voltage but effect) seems to work nicely as expected.

I just can't see how a few here are seeing outrageous readings as they are? Makes me wonder at times.

I think maybe some ot to go back to the actual proto testing and stop believing what the sim states. The sim is not always correct.

Its like saying it has to be true if it's on the internet..

I only did this for my own curiosity because we have used that type of current limiter for many things over the years and still have them operating in circuits made years ago. I don't remember ever doing a temperature test, ambient or junction, on this limiter.

I am satisfied with it, the rest can do as they please :)

Btw, this was the one that Phil H originally posted if I remember.. old relic. The circuit, not him :)

Jamie