LMC6032 Op Amp Problems

[grapefruit]

Howdy,

I'm waiting on some TS464 rail to rail quad op amps to arrive for a design I'm working on. I used some LMC6032 op amps I had here to test the circuit...

When used as a non inverting amplifier (9V single supply)  the LMC6032 worked fine, almost all the way to both supply rails. But when I use it as a buffer from a 9V supply (with sine wave) once the waveform reaches just over 6V p-p a square wave pops out the top of the sine wave (on the top half of the waveform). It shoots all the way up to V+. The input and output pins are biased at about 4.5V. I tried the methods in the datasheet for driving capacitive loads to no avail.

It works fine with LM833 or LF353 op amps. They clip when they should, but I want more headroom.

Has anyone had problems with this before?
Has anyone used the TS464 or 462 as a buffer with good results, or even the TLV274??

I was going to submit my PCB for manufacture before I got the TS464 op amps but now I'm a bit hesitant. Any advice much appreciated.

Cheers,
Stew.

[grapefruit]

It's not clipping or oscillating by the look of it, but once the signal reaches 6V p-p the top side of the op amp seems to go to open loop gain, causing a square wave to pop out the top of the sine wave.

Stew.

[db]

It may be worth posting your circuit or part of it.
A few of ideas:
Have you checked what happens on the inputs to the op-amp when the problem occurs? i.e. if the output goes full scale then this suggests that there is a difference between the op-amp inputs which causes things to go outside normal linear operation. 
Is your bias network properly bypassed?
Is your supply properly bypassed?

[grapefruit]

I can't post my circuit right now. Maybe tomorrow. I'm testing this circuit segment on its own. It's pretty simple...

I have a half rail splitter consisting of two 10k resistors and a 100n bypass cap followed by an op amp buffer. (the other half of the op amp).
This goes via a 100k resistor to the non inverting input of the op amp (for 1/2VCC bias) , as does the signal, via a coupling capacitor. The inverting and output pins are shorted as per normal non inverting buffer. There's a 100n cap as close as I can get across the PS pins. It *IS* on a breadboard.

This is the only rail to rail op amp I have on hand but LF353, TL072 and LM833 work fine in this circuit, until they reach their threshold.

I'll see if I can grab the scanner out or do a screen shot of circuit after dinner.

Cheer,
Stew.

[grapefruit]

OK, before mayhem happens IN+ IN - and OUT are all 4.49V.

When the signal shoots to the stars (after 6V p-p)

IN +   4.49V
IN-    4.61V
and yes, you guessed it...

OUT  4.61V

But why?
Damn I wish I had some more rail- rail op amps here.

Thanks for the advice.
Stew.

[grapefruit]

Here's a schematic....



Stew.

[R.G.]

I don't know that this is exactly what is happening because I haven't worked with that particular opamp.

But in the past, I have experienced something similar when I exceed the common mode input voltage range on other opamps. What does the datasheet say the common mode input range is for the LMC6032?

[db]

I would try removing/bypassing the buffer from the bias supply just to exclude some unexplained interaction between the two stages (as they are in the same chip).  It's worth noting that the bias current required for this particular op-amp is in the fA range so you could quite happily do without the buffer (assuming that it isn't required to feed other parts of your circuit).

[grapefruit]

T

[grapefruit]

Sorry, not sure what happened last night.

Input common mode voltage range for LMC6032:

TYPICAL
-.4
V+ -1.9

LIMIT
-0.1
V+ -2.3

As I said in an earlier post:
Before mayhem happens IN+ IN - and OUT are all 4.49V.

When the signal shoots to the stars (after 6V p-p)

IN +   4.49V
IN-    4.61V
OUT  4.61V

So, the common mode voltage reaches 0.12 V when the problem happens. Is this cause or effect?
Also, when the problem occurs the waveform has reached 7.6V DC on the peak. Maybe the V+ -1.9 parameter is the problem. I haven't seen this on other datasheets.
But it works fine when used as a non inverting amp with gain of 5....

The reason I'm using the buffered reference is because there are many other op amps in my circuit. I'll try it without this...

Cheers,
Stew.

[R.G.]

Also, when the problem occurs the waveform has reached 7.6V DC on the peak. Maybe the V+ -1.9 parameter is the problem. I haven't seen this on other datasheets.
But it works fine when used as a non inverting amp with gain of 5....

This is a classical description of the problem. The bias point is in the middle of the supply. You're feeding it several volts peak of signal, and when it gets over the least positive Vcm, all bets are off. Notice that the instantaneous voltage on the + and - inputs follows the signal up above (and down below) the bias voltage. Both inputs remain together because feedback forces them to. So when you get above the Vcm spec, the output goes to the positive rail. It's backed up when it works fine with a gain of 5. With a gain of 5, the input and output move only 1/5 as far before the output clips, so the Vcm is not exceeded.

I think you're exceeding the input voltage range. You need either a higher voltage supply or a smaller input signal. Or to run that input buffer at a gain of maybe two, then divide it back down. Or to get an opamp that has rail-to-rail input voltage specs.

[grapefruit]

Thanks RG,

So are you saying that the V+ - 1.9V is the problem?
That's what I thought because the inputs are seeing the same voltage DC and AC, until this problem occurs.

The op amp I'm planning on using is the TS464. I don't have any on hand.
The input Vcm is min -1.35V  max +1.35V
It doesn't say anything about V+ -X Volts, but it also desn't specify rail to rail input, only rail to rail output. I guess I'll have to wait until I can get some and try it...

Stew.

[R.G.]

Yeah, I think so.

I went and looked up the datasheet.

+V - 1.9V is the "typical" value. What that word means in a datasheet is "We kinda think most of them will work like this, but no guarantees." The words min(imum) and max(imum) mean "We'll give your money back if they don't do this and we can't figure out how you could have been applying them wrong to make that be that way."

The actual limits they will guarantee are +V -2.6V - that word "max" appears next to that one.

So if you're using a 9.0V supply, biased at 4.50V, you may only count on the thing working right if you use a signal less than what moves the inputs up to 9.0-2.6 = 6.4V. From your bias point, that's 1.9V peak. Get over that and... well, they don't say what will happen.

That's another thing to read in a datasheet. What do they NOT say. The datasheet doesn't talk about what happens when you exceed the positive input common mode range. It used to be that opamps pushed past Vicm went to the nearest power supply rail and latched there, pulling the power supply through the chip. Some of them went into SCR latchup which meant that by the time you got the power supply turned off, you had a smoking crater there. Newer opamps are better behaved, but since the datasheet doesn't say it, they don't guarantee what happens when you go there.

The moral of the story is - don't go there.

[Paul Perry (Frostwave)]

+1 on that advice to notice what data sheets DON'T say. Plenty of times I've been burned when I just assumed that if a chip did X and Y, then surely it would do Z as well...... and this lesson (emphasised in Bob Pease's troubleshooting book as well) applies outside electronics as well

As for exceeding common mode, worst trouble I ever had with that was an envelope follower, it only went out of range occasionally on heavy hits, but when it did, the chip locked up & you had to turn off the unit & wait for bias caps to discharge....really sucky & it took me a while to work out.... at least one doesn't forget a lesson like that.

[db]

Almost sounds like an argument against using non-inverting buffers with certain op-amps (the LMC6032 included)?
I had a quick look at some op-amp specs yesterday (these apply for +/-15V rails except for the LMC6032 which is spec'd for a single +15V rail with gnd reference at +7.5V):

Op-amp	Output swing min:	typ:	Common mode range min:	typ:
NE5532	+/-12V	+/-13V	+/-12V	+/-13V
TL072	+/-12V	+/-13.5V	+/-11V	-12V to +15V
LMC6032	+1.75V to +12V	0.79V to 13.9V	0V to +12.4V	-0.4V to +13.1V

So, with a non-inverting buffer, the LMC6032 certainly could swing outside the positive common mode voltage range while trying to follow the input but it looks like other op-amps could do the same thing - it's not that clear from the spec.

It's even worse in theory as we work with a single 9V(ish) battery supply more often than not, so there's more chance of hitting the rails.  It just seems that some op-amps are good in this respect and others are not.  Perhaps the LMC6032 is just one offender and to be avoided in unity gain followers.