pmos and pnp pass transistors

A few issues. First, your op amp can't swing its output close enough (0.6V) to the positive rail, so it can't ever turn the transistor off.

Second, even if it could, the circuit would oscillate, because the PNP stage has a lot of voltage gain.

Third, the output impedance of this circuit varies a lot with frequency, because the collector of a BJT naturally presents a fairly high impedance, and feedback is the only thing keeping it following the input.

There are a number of ways to fix this, but which one is best depends a lot on how close to the +20V rail you need the output to go.

What are you trying to build?

Cheers

Phil Hobbs

The LT1001 is only guaranteed to swing up to 3 volts below VCC, so it won't go positive enough to turn off the PNP or a low-threshold mosfet. Add some voltage drop between the opamp output and the pass device. Or use an RRO opamp.

Then expect some AC stability issues.

An NPN or Nmos would be better here maybe.

Another cute trick is to drive a PNP from the opamp Vcc pin, not from its output. Vaguely like the top half of this...

doh! oh yea, makes sense, I can't get to the rail so I can't turn off the transistor. Thanks.

I'll have to think about the oscillations and the output impedance issues...

I'm trying to make a current limiter, I'm using a pnp because the transistor will be on the high side of my load.

I'm playing with using an npn to drive the pnp now, and getting better results

this is what I've got goin on

Thanks

Thanks John, does that work? The base of the transistor is tied to Vcc, how can the op-amp control it?

I can't use npn/nmos cause I'm wanting to use the transistor to drive a low side load, I'm trying this now

When the opamp drives the load more than a few mA, the opamp's Icc goes up and turns the PNP on to help.

Why can't you use an nmos follower? That's simple and more stable than a p-type device.

What are you trying to do?

Some things to try to make the circuit better:

• use a rail-rail op-amp.

• use a bias network off the op-amp: resistor from op-amp to base, resistor from base to +20V.

• Some (many?) older op-amps won't pull the output up to the positive rail, but they'll let the output float to the positive rail if it's pulled there by a resistor to VCC -- which your bias resistors may do. This is usually an off-datasheet application, but it's also reliable on a lot of older op-amps. You can sorta-kinda tell if the data sheet has a schematic of the amp or at least the output stage. I know that you can do this with the el-cheapo ones like the LM324 and LM353, but for the fancy ones it's kind of a case-by-case thing.

• Whatever else you do, you absolutely positively need an external compensation capacitor from the op-amp output to the op-amp negative input. This will limit the bandwidth of your circuit (and increase the output impedance at high frequencies), but it'll also make it stable. The required capacitance depends on the load, so experiment, simulate, model and do the math, or hire me (I'm in a blatant self-promotion mood this week, it seems).

I'm trying to make a circuit that limits current to a load, I was trying to use the opamp to set a voltage across a small resistor that would limit the current

here's my circuit so far

Thank's for the advice, playing with your suggestions now.

The transistor inverts, so this diagram indicates you've given the circuit positive feedback. Yep, it latches up.

If you don't need a lot of accuracy, I've used a low value resistor, and the Vbe drop of a transistor to turn off a FET pass transistor. (the exact current depends on Vbe which changes with temperature...You can get fancier.)

George H.