Op-Amp Help Please

and

sense

the

LM358

to

the

(after a

can

I'd

what

opamp

the

What about anti-parallel zeners across the feedback resistor to kill the gain. As long the output doesn't bury itself into the rails, there is 0V across the inputs of the op-amp. Plenty of op- amps will take Vcm to the rails. In this case, Vcm is really somewhat less than the rails (16.6V) because of the divider on the non-inverting input.

and

sense

the

LM358

the

input

(after a

volts

can

I'd

what

opamp

the

Actually I don't think it's a real problem, despite the abs. max. rating.

If you look at the schematic in Figure 2 (such as it is), there are effectively back-to-back diodes across the inputs (the E-B junctions of Q5 and Q6, with series resistors on each input (of unstated value) internal to the chip).

So it should work reliably without any added stuff, IMHO, provided the input current is limited to a couple of mA.

As a corollary, the LM6132 would make a lousy comparator.

s and

se

the

he

ter a

can

=A0I'd

hat

How about the LT1498? It has clamping diodes to protect against the inputs bing driven too far apart.

pamp

Else I could lose my shirt, literally. You can contractually exclude the consequences of honest mistakes and I always do (or decline the request) but the millisecond something can be construed as grossly negligent for a person skilled in the art you are doomed.

On the lab bench or for hobby there is (usually) nothing wrong with that.

Yes, I did read the entire solution. Your "solution" was to reduce the gain of the diff amp to about 3 and pick up the remaining gain by increasing the value of the sense resistor(s). To get back to a gain of 30 would require increasing the value of the sense resistors by a factor of ten, changing the drop at 1 amp to about 2.5 volts. Since I started off with a rather marginal 2 volts of headroom anyway (17 volts raw supply for a 15 volt maximum output), that would mean that the raw supply would be somewhere around 14.5 for a 15 volt output. Not real easy to do, is it?

I don't hardly think that's "huffy", do you?

Jim

-- "It is the mark of an educated mind to be able to entertain a thought without accepting it." --Aristotle

Jim,

This answer didn't seem huffy, but your first one did. I apologize if you didn't intend it that way.

Anyway, the Linear Tech LT1366CS8 part looks like it's a drop in for your LM358 SO part. It has rail-to-rail input common mode range and 36V max supply. It also drives much closer to the negative supply (GND) than the LM358, so when your supply is putting out no current then your voltmeter will read properly.

Bob

Can you put a voltage divider on the output of the sense resistor with manhattan construction, and increase gain?

I'm also wondering about the LM317 dropout at 15V, as the current rises toward 1 amp. Maybe you could make it into a lesson - watch the regulator fail to maintain 15V as the load current increases.

Ed

How about using a OP77, OP07 ?

I take "magnitude" to mean "absolute value", i.e. a) Vin must be between the supply rails, and b) |Vin| < 15v.

It's a question of the reverse breakdown voltage of the input FETs' gates, and that voltage here is guaranteed >= 30v.

That limit is usually to protect bipolar input stages from reverse breakdown, emitter-to-base. That limits most bipolar input stages to a few volts differential, max. Precision parts often have clamps to protect the input transistors from beta degradation resulting from e-b breakdown.

JFET input op amps aren't usually so constrained, ergo the

Cheers, James Arthur

Those guys have cascode NPN input stages that definitely won't work up to the positive rail, so no joy on those.

Neither those nor the JFET op-amps mentioned have outputs that swing to the negative rail either, though the LM358 does pretty well.

Here's a collection of ASCII, just for fun:

====== FIG. 1 ====== Jim's original circuit requires an op-amp whose output swings to ground, and whose inputs still work at +Vcc:

Rs ___

(created by AACircuit v1.28.4 beta 13/12/04

====== FIG. 2 ======

This approach reduces the CMRR error, but still requires the op-amp handle inputs to +Vcc, and have an output that swings to [Vcc-Vbe(Q1)] (a pull-up to Vcc fixes that). Output to "meter" is hi-z:

Rs ___ .--------.

(created by AACircuit v1.28.4 beta 13/12/04

====== FIG. 3 ====== Here the humble LM358 is fine, just used as a buffer. Accuracy isn't critical, so Q1-2 needn't be especially well-matched.

Rs ___ .--------. +17v >--o---|___|--o----o----| LM317 |------o----> Vout | | | + | | | .-. | --- '--------' .-. R1 | | | --- C1 | | | | | | | '--------->| | '-' | === '-' | | GND | | | === '>| '>| Q2 GND |----------|--. /| Q1 /| | | | | . | o----' |\\ LM358 | | | \\ o---------------------|+ \\ | | | >--o----> 0-10v = 0-1A | | .--|- / | .-. .-. | | / | R2 = | | | | R3 | |/ | 40*R1 | | | | | ' | '-' '-' '---------' | | === === GND GND

(created by AACircuit v1.28.4 beta 13/12/04

====== FIG. 4 ====== If Jim can just hack a few traces and move the sense resistor, this circuit uses the fewest, cheapest parts, and has the best accuracy:

.--------. +17v >------------------------------| LM317 |------o----> Vout+ | | | '--------' | R1 R2 | .-. 2.7k 100k | | | ___ ___ '-------->| | .--|___|--o-------|___|---. '-' | | |\\ | | | '--|-\\ | === | | >---------o---> to meter GND | .--|+/ LM358 | ___ | |/ 0v >--o--|___|--o-------------. .----> GND | | Rs === === GND GND

(created by AACircuit v1.28.4 beta 13/12/04

The LM358 inputs are rated to operate correctly down to GND, making this version especially easy. 2.7k in series with the "+" input would be a good idea, just in case.

Cheers, James Arthur

On Sat, 22 Nov 2008 17:24:48 GMT, James Arthur wrote: [snip]

Sure there is. When specs are so loose that they aren't even measured during manufacture.

Really? Looks to me like they're JFET's, are they not?

...Jim Thompson

-- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona 85048 Skype: Contacts Only | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at

Right, like leakage on CMOS logic inputs being spec'd at +/- 1uA.

Sorry, I could've been clearer. "they won't conduct" = the FET gate junctions; they're reverse biased.

The FET channels would conduct s-to-d, of course, up 'til Vgs exceeds pinch-off.

James Arthur

Yup.

OK. Now I understand you meant the gates.

Yes. That's why the TL08x series of OpAmps are so nice. They were my jelly bean part of choice during my tenure as GenRad's analog guru,

...Jim Thompson

and

sense

the

to

the

I'd

what

opamp

Excuse me, OP max differential input is 0.25 V. And common mode of rail to rail.

Actually, it might be better to force the students to do a hand manufacturing modification. That way they will really learn why that is to be avoided.

and

sense

the

LM358

to

the

(after a

can

I'd

what

opamp

the

Not quite: If there is a short the full input voltage will show up for a very brief moment. I don't know his circuit but if the opamp then fails even just functionally by flipping its polarity or whatever ... *WHADDABAM*

If somebody has an older TL071/TL072 spec sheet, maybe they could check to see if this spec has changed with the 2005 version (Either ST or TI)... Seems I remember it USED to be that the input could exceed the V+ line by a couple hundred millivolts. It has been used it this way fo high side current sensors before. Maybe they have decided this is not such a good idea anymore ??

boB

My "revised 1983" data book says the same thing, but I'd think you can probably get away with as much as 0.3V above supply.

...Jim Thompson

I can tell by the sound effects that it's a low power circuit---no "phssst" first ;)

Cheers,

Phil Hobbs