Voltage Follower

of

Sorry, I messed up the schematic the first time - here it is again:-

IC=CA3130 +5VDC | 20VDC 1.25V |\\| o----18K-+--+-----|+\\ | | | >-+---o Vout (Measures 3.97VDC) | | +-|-/ | 0.1uF--- | | |/| | --- | | 0V | | 1K2 +------+ | | | | | | 0V o-------+--+--------------o

You appear to have forgotten to add a compensation capacitor. At unity gain, you will definitely need one. The output is at positive saturation (it will go to within a volt or so of the positive rail).

I suggest you read the datasheet:

See figures 8 and 9, read the text.

If that fails, you probably have something miswired.

Cheers

PeteS

There are several possibilities:

1. you blew the input with the high voltage, when the 1k2 was intermittent
2. your gnd connection is not continuous between the 2 supplies
3. The op-amp is oscillating, put a ceramic cap 0.1u across the supply pins.
4. The O/P is capacitivly loaded, put a 220R inseries with the Out and take the feedback from there via 1k, put a 10n from out to +in.
5. This opamp is very weak in driving higher current (1mA only).

Change the opamp to LT1637, very rugged, DIP8, works down to 3V and has +22V common mode range. It drives/sinks 15mA with a 5V supply and you can shut it off to HiZ.

hmm. did you for get to put the 0V to your supply common ? the (-) side?

pins.

take

it

Pete: I checked out the datasheet and added a 56pF compensation cap and the recommended 100K offset null adjustment. The cap helped a bit and brought the output down from 3.97 to 3.6V and the offset null adj helped, but if I adjust it for a given input/output voltage, the output varies with other inputs and doesn't stay linear..

Ban: 1. I haven't blown the input of the 3130. I tried a replacement and the problem persists. 2. I doubled up on the ground connection. (I'm building on a breadboard) I also re-built the circuit on another breadboard, with similar results. 3. As well as adding a 56pF compensation cap, I have a 0.1uF cap across the supply at pins 7 and 4 and an 0.1uF cap from pin 3 to ground. - No change 4. Tried the 220R resistor etc. No change. There is no load on the op-amp except my multimeter. Certainly no extra capacitance. 5. As above, I have negligible load on the output.

I haven't tried an LT1637 as I don't have one on hand. I didn't think I'd strike these problems with a simple voltage-follower. There's almost nothing to go wrong. I tried a '741 in place of the 3130 and it performs beautifully as long as the input voltage on Pin 3 is above 1.9V. (In this application, I need accuracy from 0V to 20VDC input (0-1.25VDC at Pin 3). Maybe both of my

Jamie: The 0V rails are connected correctly. Both the 5V and 0-25V supplies come from the same multiple-output power-supply and have a common ground, so this isn't an issue.

In the morning I'll put the 3130s aside and try my planned circuit using an LM324. Hopefully it will perform the way I expect.

... Winfield

Did you look at the output with a scope? If not I'd start there.

Don

No, I haven't. It's midnight here, so I'll do that first thing in the morning

... Winfield

intermittent

shut

the

brought

and

similar

cap

ground. -

I'd

nothing

beautifully

application, I

my

followers

from

are

will

for

supplies

ground, so

an

Why did you depart from figure 9 on the datasheet? Have you tried it as shown - output tied back to pin 2 through 2K, bypassed by .1 uF?

Ed

of

I didn't have the datasheet at first. I now have the 2K resistor and cap in place, but it makes no difference. I've been busy this morning, but I'll take a look at the output with a scope shortly.

... Winfield

Bingo - I just scoped the output - high frequency oscillations. I increased the compensation capacitor's value from 56pF to 150pF and now everything works fine. I also tried 100pF but that wasn't high enough. Thanks, Don

... Winfield

be careful with LM324s. The rarely work the way one expects, unless you were expecting the Spanish Inquiistion. (Apologies to Monty Python.) They have a flaky output that switches between class A and B

whenever it feels like it, which tends to give huge nonlinearities at odd times. I'd say, try them, and test thoroughly under all the real-world loads they are going to get. If they work, then fine; they are certainly cheap enough. Just be careful.

Best regards,

Bob Masta dqatechATdaqartaDOTcom D A Q A R T A Data AcQuisition And Real-Time Analysis