0
\$\begingroup\$

I'm using a TI F28379D to generate a sinusoidal waveform from a DAC channel. The output of the DAC is shown in Figure 1 with a Vpp of 688mV. This signal is then connected to a resistive network of 1k ohms with a 1V DC source as shown on Figure 2. The DC source is a power supply.

Unfortunately, the DAC output signal gets clipped once the DC voltage source is connected. If I use a 1.5V alkaline battery same issue occurs. It is a simple circuit, yet I do not fully understand why the clipping is happening.

Figure 1. DAC Output DAC Ouput signal

Figure 2. Superimposed circuit

Superimposed circuit

Figure 3. Clipped DAC output enter image description here

The datasheet for this particular MCU, states the maximum resistive load is 5k ohms. Since I am using 1k ohm would this be the root cause of the issue? I do not believe so since I remove the 1V DC source, the clipping goes away.

I am confused why the DC source is causing the clipping of the DAC output. If I were to increase the DC voltage from 1V to 2V, the clipping increases.

Lastly, if I increase R1 and R2 to 3.9k or 5k ohm, the clipping disappears. This means the current is decreased throughout the circuit, and somehow the DAC output is affected. If I must use 1k ohm, what can I do?

The following are key observations:

  • Tried isolating the DAC output by adding a voltage follower and the clipping persists
  • Different frequency output does not affect clipping
  • Simulations do not show any kind of clipping
  • If I change R1 and R2 to 5k ohms, clipping disappears
  • If I change R2 to 5k ohms, and keep R1 to 1k ohms, clipping persists
  • Removing DC power supply removes clipping
  • Simulation shows approx. a current of .667mA flow through R1 when it is 1k ohm

Solution: The voltage follower with LM324A I had was powered by a single power supply. I decided to try the dual-supply operation of the op-amp and it solved the issue.

The root cause was trying to drive an AC signal close to 0V using an op-amp in single supply operation. By using dual-supply operation, the output range of the op-amp can handle 0V.

enter image description here

\$\endgroup\$
4
  • \$\begingroup\$ Buffer it with an opamp that can drive 1k. \$\endgroup\$
    – user16324
    Feb 18, 2020 at 19:37
  • \$\begingroup\$ I wonder if the DAC is only a single quadrant. The DAC does not have circuitry to drive the signal low from a high state or at least can't drive it well. When it is only the DAC the 1k discharges the signal. When the battery is in play the DAC only pulls it as low as it can. Try adding a DC blocking capacitor between the DAC and circuit. \$\endgroup\$
    – vini_i
    Feb 18, 2020 at 20:34
  • \$\begingroup\$ @BrianDrummond I used a voltage follower, LM358P and also LM324A. It seems neither can drive the 1k load. I will investigate more buffers and see which one I can use. \$\endgroup\$
    – JC203
    Feb 20, 2020 at 2:38
  • \$\begingroup\$ @vini_i If I add the DC blocking capacitor, I remove the DC offset from the DAC output. It does in fact resolve the clipping but I need the offset from the DAC. I can try adding the offset back with a circuit. This can be an alternative I had in mind. \$\endgroup\$
    – JC203
    Feb 20, 2020 at 2:40

1 Answer 1

1
\$\begingroup\$

OP oscillograph shows clipping of the bottom peak of the sinusoidal wave. This suggests that the internal buffer can only source current, and cannot sink current. A load resistor is required to pull the buffer output down. An internal pull-down resistor is included.

From the data sheet spec, the 5k minimum pull-down resistor combined with \$V_{REFHI}\$ of +3.3V, the buffer could source at least 0.66 mA. No data sheet spec of how much current it can sink.

Data sheet does specify a settling time of 2 microseconds when slewing from 0.3V -to- +3V with 100pf load. This spec might be used to estimate charging current...
\$ I=C \frac{dV}{dt}\$
\$ I = 100\times 10^{-12} \frac {2.7}{2 \times 10^{-6}}\$
\$ I = .135 mA \$...this is within the 0.66mA calculation above.
It is interesting that a discharge spec is not included. It is also interesting that output voltage linearity is not specified between ground and 0.3V. Lack of these specs is consistent with this suggestion that active pull-down current is lacking.

What can you do if a load of 1k is required?
Spec sheet says "don't load with less than 5k". So you need another external buffer. And you should use one that lightly loads this DAC, and doesn't try to push current back into the pin.

\$\endgroup\$
3
  • \$\begingroup\$ Thank you very much for the details. I've tried using a voltage follower with a LM358P and a LM324A but had no luck with it. I may need to look into the specs of these op-amps and understand if they are adequate for this DAC. \$\endgroup\$
    – JC203
    Feb 20, 2020 at 2:36
  • \$\begingroup\$ Both those opamps push bias current back into the DAC. If my hypothesis is correct, you'd need to add a pull-down resistor to the DAC output pin with these voltage followers (gain of +1). \$\endgroup\$
    – glen_geek
    Feb 20, 2020 at 5:25
  • \$\begingroup\$ The issue was I was running the LM324A in a single supply operation. A +5/-5 supply to the op-amp resolved it. \$\endgroup\$
    – JC203
    Feb 27, 2020 at 2:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.