8
\$\begingroup\$

I'm building a simple waveform generator based on the Raspberry Pi Pico, a DAC0800 IC and a TL072 op amp. The circuit works, but the slew-rate is - I believe - unreasonably low for the components I'm using, making higher frequency functions basically impossible to generate. The problem is not the RPi Pico, as I can make it output values so fast that the DAC + op amp can't even follow anymore.

This is the circuit (and test setup): enter image description here

And this is the capture of the oscilloscope, capturing a square wave signal at 8 kHz, where you can see how slow the voltage jumps between the two extremas, taking around 30 microseconds: enter image description here

This 30 microseconds slew-rate seems to be constant, regardless of the speed I feed data into the digital pins, this same delay is present in lower frequencies too.

I've tried reading the datasheets, but admittedly I might not understand important parameters. For the DAC0800 for example, it says "Fast Settling Output Current: 100 ns" (since it's a current output DAC), which in my head would translate to "1 000 000 000 ns in a second, so 1 000 000 000 = 10 000 000 (= 10 MHz) output current sets".

For the TL072, it says "High slew rate: 20 V/μs", which in my head would translate of "being able to follow a 20 volts of change in a microsecond", so following at least up to 1 MHz of voltage changes. Since the amplitude is around 15 volts, that seems to be way in range as well.

So theoretically, both components should do way-way better, unless I'm missing something important. The datasheets of both components:

What causes this slow slew-rate and how could I improve it with the current circuit setup, if possible?

Edit: Some additional signals (ignore the magenta channel, I ended up not being able to fit in the probes at once), yellow is the output. Negative input (pin 2) of the op amp in blue: enter image description here

Positive input (pin 3) of the op amp in blue: enter image description here

Edit 2: I was asked for the origin and a picture of my TL072 chip. Unfortunately, I have no idea where they are from, my father gave me a handful of TL072s and TL074s from his stash when I started out with electronics. Close-up picture of the IC: enter image description here

Edit 3: It was pointed out that my TL072 was a counterfeit one, so I've tried a TL074 from my stash that seemed to have a more legitimate marking and the slew-rate problem disappeared completely: enter image description here

Interestingly enough, there is now a slight overshoot problem (mostly downwards) that is more apparent, when going higher frequencies: enter image description here

The overshoot seems to take a little longer to compensate than what the datasheet suggests (around 10 microseconds instead of 0.4): enter image description here

Edit 4: Turns out, the slow overshoot was because of the imbalanced feedback loop. By shorting the feedback resistor, the issue is completely gone.

\$\endgroup\$
6
  • 1
    \$\begingroup\$ A low slew rate corresponds to a high rise (or fall) time. I think the latter is the word you're looking for. \$\endgroup\$
    – Hearth
    Commented May 19 at 23:08
  • 1
    \$\begingroup\$ Please show the scope waveforms for the two inputs of the op-amp. \$\endgroup\$ Commented May 19 at 23:16
  • 1
    \$\begingroup\$ @FabioBarone Added them, I hope they are adequate. \$\endgroup\$ Commented May 19 at 23:47
  • \$\begingroup\$ Where did you purchase the TL072? Could you show a picture of it? \$\endgroup\$ Commented May 19 at 23:57
  • \$\begingroup\$ @JonathanS. Added a picture. Unfortunately, I don't know the source, it was from my fathers stash who doesn't remember. \$\endgroup\$ Commented May 20 at 0:03

2 Answers 2

7
\$\begingroup\$

That TL072 is counterfeit. You can tell by the completely misshapen TI logo that misses the "i" and the upper left part of the Texas shape. My guess is that it's a relabeled LM358, given that you're observing a slew rate of approximately 0.3V/µs, which matches the 358.

Here's a (hopefully genuine) TL074 for comparison:

TL074 close-up

You might be able to rub off the fake markings with acetone as these fakes are often made by applying a black paint to the top of the chip before new markings are lasered or printed.

\$\endgroup\$
3
  • \$\begingroup\$ That is unfortunate, thank you. I'll try to verify this by removing the covering paint from one of them. I found that 2 of my TL074s seem to have a legit marking on them (if I have to guess, not TI but Microchip Technology? Not sure), I'll try to somehow slot one in into the DIP8 socket tomorrow, to try if it makes a difference. \$\endgroup\$ Commented May 20 at 0:36
  • \$\begingroup\$ Hey, I've edited the post with more details. I'll accept your answer as this answered my direct question, but if you have ideas about the overshooting correction time problem, I'd gladly accept additional input about that. If advised, I can also open a new question for that. Big thanks for identifying the counterfeit. \$\endgroup\$ Commented May 20 at 7:39
  • 1
    \$\begingroup\$ @PeterLenkefi Nice to see that a new OpAmp solved it! Regarding the overshoot, try adding a small capacitor (i.e. 22pF ... 100pF) between the output of the OpAmp and the inverting input. If that doesn't help, you might need to open a new question as this one was specifically about the slow slew rate. (Actually nevermind, I just saw the edit #4, glad it works!) \$\endgroup\$ Commented May 20 at 13:07
0
\$\begingroup\$

Looks to me like you have connected a decoupling capacitor to pin 2 and/or pin 4. The circuit works fine, but it’s not the circuit you show us on the schematic. It just has a lot of capacitance on the nodes where there should be as little of it as possible. Assembly error I bet.

\$\endgroup\$
1
  • \$\begingroup\$ Pin 2 and 4 of the DAC chip are in direct contact with the op-amp with jumpers, there is no capacitive component between them. I'll try to verify this with measurements as well, but reinspecting that portion of the circuit, I don't see a capacitor accidentally wired in there. The schematics is as close to the built thing as I could get it. \$\endgroup\$ Commented May 19 at 23:51

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.