OP amps clipping way below rails

[poiureza]

I made a very basic amplifier using a TL072CN ("CN" probably stands for "China" lol).

I used a 7660 chip to double the voltage range : I get +8 and -8 volts out of it (measured)

The Spice model predicts the same voltage values that I read on the scope.
It also predicts that clipping starts at around 6.4V amplitude at pin 1.
That's in line with the TL072 spec sheet which states +-8V swing for a +-10V supply and 2K load.
That could probably translate into an available +-6V swing for a +-8V supply and 3.3k load.

However the scope shows clipping starting at +-4.1 volts already (that's about 65K in the op amp feedback resistor).
I tried other chips and the clipping threshold looks all the same. I even tried other resistor combinations but that 4.1V holds true whatever I do.

I'm very surprised. I expected to be able to get a clean +-6V swing from my +-8V supply and the available +-4V falls pretty short for my project.
Am I missing something or is that limitation real ?

[poiureza]

Some update

I added positive voltage doubling from the 7660 chip : now I have +16V and -8V.
I added a resistor network at the non-inverting input (2x 1M with ends connected to the +16V and -8V pins ) to bias the input at midrange i.e. 4V.
Spice tells me that I have now a maximum clean output swing of +-10V before clipping, thats a swing from +14V to -6V at pin #1
(clean output +14V/-6V for a +16V/-8V supply makes more sense. I can imagine that it needs to be 2 volts off the rails. I still don't understand why it needed to be 4 volts away from the supply rails in my initial +8V/-8V setup ...)

The scope tells me something slightly different : the output clips as soon as I hit 10V on the upper swing now.
As a matter of fact this might actually make sense since the TL072 spec said +-10V max for a 2k load (I suppose this includes the DC offset)

+10V on the upswing and a DC offset of +4V, that's a swing from +10V to -2V : a 6V amplitude swing.
I can get away with that but tommorow I'll try a lower input bias (+2V maybe). Maybe I can get an extra volt before it hits either rail or the output limitation.
Stay tuned.

[phatt]

Pin 3 (the pos input) of any of these opamp circuits needs a ground reference (a DC path to ground) try a 220k up to 1Meg then you might get better results.
In the case of single supply that reference is usually half way.
Phil.

[poiureza]

Mmmmh ... that was the very first thing I did when the initial setup gave strange figures, but it made no difference at all.
My understanding is that this DC path clamps the voltage at pin 3 to ground when no input is connected.
It doesn't make any difference once a ground referenced signal is applied at the input.
Though I might be wrong ...


As for biasing lower ... +4V was the optimum, no cigar here.

[Dirk_Hendrik]

Could you link to the spec sheet?

And more specific, could you link to the datasheet of the exact type you use. I want to see the performance figures you refer to. I also want to see the meaning of the "CN" suffix. Commonly suffixes apply to the package. I have never seen a suffix refer to the country of manufacture.

[Dirk_Hendrik]

ANd how is the performance if you add a 100pF in the feedback loop?

[poiureza]

Sorry, "CN" was an attempt at a bad joke. I bought that chip from a chinese seller on eBay. And it actually says "TL072CP"
No specific data sheet, I found the figures in the TL072 spec from TI http://www.ti.com/lit/gpn/tl072a

I picked the 8V max peak output for Vcc=10V and KL=2kohm from figure 3, pg 21
Scaling down on that graph I guessed 6V max peak output for VCC= +-8V and K=2kohm
The (what I suppose to be the overall maximum) output limitation of 10V peak is VOM on page 13 (given for VCC=15V as stated on the very top of that page)

Actually, both information are called VOM and somewhat contradicting :
- on pg 13 it says 10V max output for VCC=15V and K=2kohm
- on pg 21 it shows 12V max for VCC=15V and K=2kohm


I'll try adding a cap in series with the resistor in the FB loop, though the cap to ground from that FB loop already prevents DC at the non-inverting input from getting amplified methinks. If you meant in parallel with FB resistor, well I can try that too but then I wouldn't understand if that made a difference.

[phatt]

I'll try adding a cap in series with the resistor in the FB loop, though the cap to ground from that FB loop already prevents DC at the non-inverting input from getting amplified methinks. If you meant in parallel with FB resistor, well I can try that too but then I wouldn't understand if that made a difference.

That is not how it works, you may need to research opamps to gain a better perspective. As for the RC leg off the neg input, It's not the FBack. The 3k3 sets gain along with the FB resistor while the 10uF cap sets the bass roll off point. (smaller values will roll of even more bass)
Meanwhile a small value cap (as Dirk mentioned above) in parallel with the 100k FB resistor will roll off the treble, larger values will roll off more treble.
If you want to stop the circuit hitting the rails just put a voltage divider in front. As you have no series resistance in front then the circuit is prone to problems and may even squeal if input left with no DC path to ground when unplugged.
Hope it helps a bit, Phil

[poiureza]

Thank you for your input Phil.
I must admit I'm a bit confused about what the cap to ground in the leg outside the FB loop does.
Beside the eq'ing (though with 10uF it has almost no effect) it must do something else. For instance, if I remove it and connect the 3k3 direct to ground instead, the opamp doesn't work AT ALL, zero output.

[phatt]

Well as you have no blocking capacitor at the front end and are relying on the previous unit to set the Q point then the signal could be offset and could give misleading results.
I'd put a 10~22nF cap and a 1k Resistor in series with the input then a 220k~1meg to ground as per just about every other circuit ever made,, (HINT!!!)
Then you might get better results.

Always have some series R in front otherwise you may find you pickup the taxi rank radio out the front of your gig which I've seen happen because the cheap mixer a chap purchased had NO RF filter on the front end. The guy that built that brand went broke,, I was not surprised?

BTW, the series 100k on the output will reduce the signal dramatically it only needs to be ~1k but signal out does depend on what it drives I guess.
If you are having trouble with the power supply just use two 9 Volt batteries to make sure the psu you used is not messing things up.
Phil.

[poiureza]

Thanks again

I'd really need to update the picture of my circuit.
It has changed a lot since my second post were I added negative PS (so I have +16 and -8 volts) and DC offset the input signal to +4V.
And I have an input cap and series resistor now ...

Both 100K resistors are coming from a dual 100k pot btw
I plan to use this here as a base with unity gain for an actual tone shaping circuit snippet that's still to come.

I'll scrape this breadboard and make a fresh new start, I must be overlooking something ...