opamps coming off the rail

LM324.... just kidding :-)

Can't you add a clamping circuit to make sure it never saturates, like a Baker clamp, just wound around the opamp and with very small clamp voltage? (maybe adding a small bias)

Regards

Klaus

"Connecting the load resistor to ground and operating from a single supply, the outputs swing completely to ground without saturating. The outputs can also drive to within 1.2V of the top rail. The EL2150 and EL2157 will outp ut ±100mA and will operate with single supply voltages as low as 2.7V, ma king them ideal for portable, low power applications."

clamp, just wound around the opamp and with very small clamp voltage? (maybe adding a small bias)

CMOS doesn't "saturate", i.e. there's no storage. Although compensation caps can "wind up", even in CMOS OpAmps.

Solutions would depend on how fast John needs the current to change, and/or whether the object is to switch on and off a fixed current. ...Jim Thompson

Ummm... "However, special circuitry built into the EL2150 and EL2157 allows the output to follow the input signal to ground without recovery delays."

"Special circuitry" performance should be verified by the user.

They do spec the recovery from upper limit clamp at 7ns.

clamp, just wound around the opamp and with very small clamp voltage? (maybe adding a small bias)

With a negative supply, I'd just hang a diode across (or instead of) the cap in my circuit. The opamp output would hang at -0.6 or whatever, still closed-loop. It takes more parts to do that without a negative supply, to hang at +0.3 or something just below the mosfet threshold. It's a parts count tradeoff against just making a negative supply voltage with a charge pump or something.

outputs swing completely to ground without saturating. The outputs can also drive to within 1.2V of the top rail. The EL2150 and EL2157 will output ±100mA and will operate with single supply voltages as low as 2.7V, making them ideal for portable, low power applications."

Nice amp, but it seems to be "retired."

clamp, just wound around the opamp and with very small clamp voltage? (maybe adding a small bias)

Doesn't the feed-forward compensation (for stability) guarantee that there will be a recovery time from off to on? ...Jim Thompson

John - take a look at the CMOS 4000 series "buffers". They looked ok on a test jig for driving 250 volt mosfets rated around 20 amps. If I remember correctly, the results weren't blitzkreig speedy but ok. A single mosfet could be driven by all the "buffers" in the package, which might match the

Hul

John Lark> I need to drive a mosfet in current-sink mode. It's similar to this

Baker clamp, just wound around the opamp and with very small clamp voltage? (maybe adding a small bias)

Then add the gate charge plateau. Interesting problem with so many interleaved issues. ...Jim Thompson

ly, the outputs swing completely to ground without saturating. The outputs can also drive to within 1.2V of the top rail. The EL2150 and EL2157 will o utput ±100mA and will operate with single supply voltages as low as 2.7V, making them ideal for portable, low power applications."

The good ones always retire young.

"As shown in Figure 41, the AD8051/AD8052/ AD8054 recover within 60 ns from negative overdrive and within 45 ns from positive overdrive."

the outputs swing completely to ground without saturating. The outputs can also drive to within 1.2V of the top rail. The EL2150 and EL2157 will output ±100mA and will operate with single supply voltages as low as 2.7V, making them ideal for portable, low power applications."

That's a nice amp, not very expensive either. Its compensation caps are connected to the output, like an LM8261, instead of being on some node buried inside.

We stock the LM8261, so I tried one:

Not bad, recovers in well under a microsecond for sine or square wave signals.

Chopamps are notoriously bad, because as you say, the autozero mechanism winds up when they're railed. I'm a fan of the AD8605 for that sort of job, on account of its very low open loop output resistance. Unfortunately it's a 5V part, so it probably isn't what you want.

Cheers

Phil Hobbs

or2.JPG

What kind of big power FET in linear mode, which should mean some kind of e levated temperature operation, gets you 0.000 current? If say 10V across R2 is 150A, then that means 10uV gets you 150uA. I don't think the circuit as shown is capable of 120dB range.

elevated temperature operation, gets you 0.000 current? If say 10V across R2 is

The Magic FET from Wizard of Oz >:-} My guess is a 150A MOSFET will leak more than that.

And, with a FET that size, the gate charge plateau is going to rape any fast response. ...Jim Thompson

As I noted, I can add some tens of millivolts to the input (say, add a resistor from some V+ to the opamp inverting input) to overcome the opamp offset voltage and force its output to rail low.

elevated temperature operation, gets you 0.000 current? If say 10V across R2 is

R2 will be 10 milliohms, 1.5 volts at 150A.

One candidate fet is an IXYS part, huge SOA, that has Idss of 5 uA max ar room temp, 50 uA max at 125C. Sounds like power mosfets aren't as leaky as they used to be.

First problem needs to be solved is based on the fact that any opamp configured that way has an undefined output voltage with zero in. So close the loop; ditch Mister Big Cap and use a resistor in its place; start with 10K but no larger than 100K and work from there.

irror2.JPG

f elevated temperature operation, gets you 0.000 current? If say 10V across R2 is 150A, then that means 10uV gets you 150uA. I don't think the circuit as shown is capable of 120dB range.

Okay, so this is not really an analog circuit, it's switching between two e xtremes, 150A and 0A.

elevated temperature operation, gets you 0.000 current? If say 10V across R2 is

extremes, 150A and 0A.

No, it will be pulsed but it is linear. It can do any current from 0 to 150A, with moderate accuracy. The customer programs the current. Of course, it can't do 150 amps for long, but the duty cycle requirement is low.