help with opa1656 unity gain inverting amplifier

[Audiofanman]

Have a circuit with an OPA1656 set up as a unity gain inverting amp.




Feeding the opamp with a 2Vpp 1k Hz signal provides millivolt ripple on the output.
V+ 24V,
the output of a tle2426 (The virtual ground) of 12v going to +In
OPA1656 pins,
Pin 1, 2, 3, 5, 6, 7 = 11.8V
Pin 4 = 0v
Pin 8 = 23.76V

I almost feel like the OpAmp is creating the inverse signal which is nulling out input...
But, LTspice gives me an inverted output sooooooo Im lost...

[jomas]

Did you put decoupling capacitors on pin 8 to gnd (as close as possible to the ic)?
If not, I suggest you do, because the circuit might be oscillating. 100nF ceramic and 10uF parallel will do.

[Audiofanman]

Thank you, and Yes I did, C5 100nF.
So, this is the 2nd OPA1656... I must have toasted the 1st one, now I think I just messed up the tle2426 
I have been troubleshooting this one on and off for days...
Was hoping to conclude before our weekend getaway but now believe just dug my ditch a little deeper.
The only positive I had in the last 3 days was finding that if Im carefull enough I can use my full size heat gun to solder the opamp to the 8 pin SOIC board!!! 
Strange to get 150mv 1khz out at pins 1 and 7....
And yes, I checked the resistors in the divider...

[dmendesf]

For unity gain without inversion signal must enter by the + input of the amp op and output goes directly (and only) to - input, unless you want to mess with the offset.
All other instbilities are probably caused by lack of decoupling / noise from power supply / noise from neablr sources (like beck lights... some are terrible and emit a lot of RF noise).

[Audiofanman]

So this is in fact an inverting amp at unity gain. That being said, as shown in the included snippet, there are coupling caps on both the input and output.. Since I have checked to make sure both the signal grounds and opamps share the same net (they do) Replaced the opamp with another, no change. Checked for shorts, bad caps, resistor values etc... Going nuts!

[sourcecharge]

The problem is when you offset the inverting opamp.

It turns into a IA differenetial opamp problem.

Edit:

This needs all the same resistors 4.99k or 10k, either one.

For example, replace the 4.99k with a 10k.

Add a 10k ohm resistor to ground hanging off of the + input, and drop the input decoupling capacitor.

This makes your 12V to 10V 1khz output AND keeps the output signal proportional to the input.

[Terry Bites]

It does exactly what you'd expect for an inverting amp. The phase is shifted by 180'.
If you don't use a coupling capacitor, the voltage at the non-inverting input (12V) will be multiplied by 1+(R1/R2).
The output will swing to the positive rail without it.
Reversing power pins will usually (definitely) fry the op-amp.

My advice. Breadboard with super cheap op-amps like LM324s first.

BTW 741's are not worthy of your precious current, they only exist because people (still) buy them by mistake.
Or because they've been looking at a text book written by a superannuated hermit.
All jelly beans eventually dry out and go stale. Belly button fluff starts to appear on them.
Join my cause to have them banned and build a (s)hit list.

[Audiofanman]

@Terry Bites,
Too bad the circuit in real life wont chooch... And I will be picking up a bunch of different types of opamps, I just have to verify pinout configs, hoping to keep one layout for a few different types...

[sourcecharge]

@Terry Bites,
Too bad the circuit in real life wont chooch...

I assumed that the input was supposed to be 0V to 2V, because you have only a single source.

Is it possible that because you are trying to operate it passed it's common mode voltage input range that it's failing?

[partsplanai]

Hi Audiofanman,

The U1B stage looks right as drawn: R20 = R21 = 10k with C8 across R21 is a unity-gain inverter, and pin 5 is biased from the 12V splitter through R19. Your DC readings (about 11.8V on pins 1/2/3/5/6/7, half of 23.76V) are consistent with correct mid-supply bias, although they don't rule out oscillation or a wiring fault.

On the common-mode question: per the TI datasheet (SBOS901C, p.6) the input common-mode range is V- to (V+) - 2.25V, so roughly 0 to 21.5V on 24V. 12V is well inside. The output swings to within 0.25V of either rail (p.7, RL = 2k), so 2Vpp around 12V is fine too.

One thing I would check first: if the 150mV reading was peak-to-peak and R-3 really has 2Vpp, the numbers point at C8. With R20 = R21 = 10k, that is what you get at 1kHz if C8 is around 220nF instead of 22pF: 10k in parallel with 220nF is only about 720 ohms at 1kHz, so the gain drops to about 0.07, i.e. about 144mVpp. Unmarked 22pF and 220nF chip caps look identical, and LTspice would not show it because the schematic says 22pF.

Two quick tests:
1. Measure at pin 7 with the external output load disconnected, using a 10x probe with matching scope attenuation and a 1 Mohm input. Keep R-3 at 2Vpp and sweep down in frequency. With 220nF there you would see about 1.2Vpp at 100Hz and close to 2Vpp at 20Hz, and at 1kHz the output would be about 90 degrees out of phase with the input instead of the 180 degrees an inverter gives.
2. With power off, lift one end of C8 and measure its capacitance, or replace it with a verified 22pF part before retesting.

If C8 checks out, measure with the generator connected, scope ground on circuit 0V (pin 4), same probe setting everywhere:
- R-3 side of C10: should be the full 2Vpp. If not, something else is loading or driving R-3, for example another op-amp output on that net.
- C10/R20 junction: full 2Vpp. Pin 6: about 11.8V DC, with very little 1kHz AC (virtual ground).
- Pin 5 and the TLE2426 output: about 11.8V DC, with very little 1kHz AC. At 1kHz the output is about 2 x (any 1kHz on the 12V reference) minus the input, so ripple on the 12V net matters.
- Keep the generator return and scope grounds on 0V, not on the 12V net. If circuit 0V is already earthed through another connection, clipping an earthed ground lead onto the 12V net shorts the TLE2426 output to 0V.

About pin 1: channel separation is -135dB at 1kHz (p.6), so normal crosstalk cannot put 150mV on OUT A. If both channels use the same circuit and each receives 2Vpp at 1kHz, the same incorrect feedback capacitor value could explain similar output amplitudes. Otherwise, knowing how channel A is wired would help find the path.

On keeping one layout for several op-amps: the single OPA1655 in the same datasheet (p.3) is not pin-compatible. Its V+ is pin 7, where the OPA1656 has OUT B. Check each pin table before swapping.

TI datasheet: https://www.ti.com/lit/ds/symlink/opa1656.pdf

[Audiofanman]

SOLVED!!!!
😁
C6 and C8 were 900nf and NOT 22pf!

[ledtester]

SOLVED!!!!
😁
C6 and C8 were 900nf and NOT 22pf!

I see where C8 is in your first schematic but where is C6?