# Op amp clipping well under voltage rails asymmetrically

I've been trying to figure out what looks like a clipping issue in the linked circuit.

Our input is a 8 kHz 120 mV sine voltage signal coming from an AD9850. We are trying to amplify it to a 5V signal. The signal is clipping at 1.3 Vpp well below our rails of +/-5V.

We've upped the supply to 20V (+/-10V) but the "clipping" remains (the clipped signal peak to peak slightly increases to approximately 1.8V). We have a slightly uneven splitting for the supply voltage on the op amp at 5.5V and -4.5V, but from what I read that should not be causing that much of an issue. We were having the exact same problem with a OPA2134PA. The problem persists up to +/-30V and adding parallel resistors to the virtual ground capacitors did not help, either.

The negative peak is clipped less than the positive peak. We are pretty sure our capacitors are oriented correctly (we have $$\C_1\$$+ facing the positive side of the supply, $$\C_2\$$+ facing the ground and $$\C_3\$$+ facing the AC input).

simulate this circuit – Schematic created using CircuitLab

• That is not a virtual ground. You need to add two resistors in parallel with C1 and C2. Dec 27, 2017 at 15:28
• We need a "Go stand in the corner.." button to click every time someone uses a 741.... Dec 27, 2017 at 15:34
• In that corner there should also be a huge stack with copies of Opamps for everyone: web.mit.edu/6.101/www/reference/op_amps_everyone.pdf for everyone to read and take home after reading. Dec 27, 2017 at 16:19
• And read Andy aka's answer to this question: electronics.stackexchange.com/questions/304521/… for a list of reasons not to use a 741 while standing in the corner. Dec 27, 2017 at 16:25
• 741 problems can usually be solved by carefully removing the 741 from the breadboard and placing it in a suitable rubbish bin, and replacing it with a decent op amp of more recent design than the late 1960s. Dec 27, 2017 at 17:18

The big problem with the schematic as drawn is that your zero-volt reference is free to float anywhere between either supply rail. Try adding low-ish value resistors << 1k, if possible, across C1 and C2. This will hold the ground reference voltage a bit better.

If that improves things then consider either a split rail supply or replacing the lower resistor with a 5 V Zener diode to hold the zero-volt line at +5 V relative to the negative supply.

simulate this circuit – Schematic created using CircuitLab

Figure 1. Resistive and Zener ground generation.

• Or, and I hate to say it.. add another 741 to buffer the virtual ground. Dec 27, 2017 at 15:32
• I'm afraid we think too much alike. We could take alternating shifts on the site or arrange holiday cover for each other. I'll be in touch! Dec 27, 2017 at 15:36
• LOL.. no doubt ;) Dec 27, 2017 at 15:37
• Hey guys we added the resistors in parallel to the capacitors and our output is still clipping starting at around 1.1V when we put a 10k resistor on the feedback. We are grounding everything but the AC input on the virtual ground. Should we be separating the OPAMP +IN ground from the virtual ground? (No zener diode on hand to see if that configuration would fix it)
– Paul
Dec 27, 2017 at 16:21
• @Paul maybe you have used 10x scope probes and didn't realize. Feb 14, 2018 at 14:07

The AD9850 circuit has (supplied with positive voltage) a DC offset voltage at the output, and the 500 uF capacitor used for coupling the signal to the amplifier probably also has a low leakage resistance, hence the DC offset and the different behavior on the positive and negative side.

On page 2 of the AD9850 datasheet it states that its output impedance is typically 120kΩ.

I find that hard to believe, but that's what it says. It just doesn't sound right, for a 125MHz device. Even 1pF at the output would cause heavy attentuation at 1MHz. I think that must be a typo.

Anyway, even if they are an order of magnitude out, it doesn't stand a chance driving your 1kΩ input impedance amplifier.

I'd buffer that DAC output with a voltage follower, prior to any gain stage. Or perhaps you could employ the opamp in a non-inverting configuration, with high input impedance:

simulate this circuit – Schematic created using CircuitLab