2
\$\begingroup\$

To perform a didactic experience, I had to design a double wave rectifier by using op-amps (TL081). I choose to use two stages: a single wave precision rectifier (i.e. ideal diode) and an inverting adder in order to obtain a full rectified Vout starting from Vin.

schematic

simulate this circuit – Schematic created using CircuitLab

The circuit was easily mounted on a breadboard and tested, and it works very well. The problem is its behaviour at "high" frequencies: from 8kHz on, the I/O characteristic starts to modify, thus I consider the circuit being no more reliable. Let me show you the oscilloscope screenshot with a 10Vpp 16kHz sine wave as inpunt voltage Vin.

Test with Vin sine wave 10Vpp @ 16kHz. Down-side: I/O characteristic. Up-side: Vin=yellow line, Vout=light blue line.

As you can see, the I/O characteristic is modified and it is no more V-shaped.

  • I can't understand if is this due to a different behaviour between the rising Vin and falling Vin, or to something else.
  • I can't understand why we experience that sort of "delay" between the Vin (yellow line) and Vout (light blue line), since the switching time of the used diodes is "4ns max" and the slew-rate of the op-amp is 13V/us (and, measuring its output, the max slope reached was 2.4V/us - moreover, the circuit was designed with low resistances to let the op-amp work well).

Simulations performed with LTSpice using the models of the real components gave the same results. Thanks for your replies!

Improve this question
\$\endgroup\$
4
\$\begingroup\$

You are summing two signals together. One of those signals is the input and the other signal is derived from the input and hence, it slightly delayed. It's got nothing to do with the diodes - it is related purely with the relatively slow speed of the TL081 at 16 kHz.

At 16 kHz, the TL081 has an open loop gain of about 100 so it cannot be regarded as ideal and it will impose timing errors on the half-wave rectified output it produces. Why don't you try this out on a linear amplifier (say inverting gain of 1) and watch the effects of this delay as the input frequency gets higher.

Improve this answer
\$\endgroup\$
7
  • \$\begingroup\$ Hi. First of all, thak you for the help. I made two inverting amplifiers in cascade, and the delay is no more negligible only at higher frequencies, like 40kHz. I also tried to perform the same test with the diodes, and they gave a very delayed output with respect to the input signal. Everythink is making me think it is their fault. In this case, how should I interpret the "switching time: 4ns max" given on the 1N4148 datasheet? \$\endgroup\$ – Nicola May 30 '19 at 13:54
  • \$\begingroup\$ Here are the pics for a 50kHz 5Vpp Vin test with 2 inverting TL081 op-amps in cascade (Vout is the light blue line) and a test of a 1N4148 diode with an imposed 50kHz 4Vpp Vin. In the first case, we have a 250ns delay, in the second one it is 2.25us. imgur.com/NHzZmCX imgur.com/lkkskWg \$\endgroup\$ – Nicola May 30 '19 at 14:48
  • \$\begingroup\$ @Nicola All diodes have capacitance that reduces with reverse voltage. Try your finger with ~ 100pF across each part to judge the sensitivity to capacitance. 1N4148 is 4pF , so touch lightly. \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 May 30 '19 at 15:58
  • \$\begingroup\$ @Sunnyskyguy Some time ago I experienced (building a Chebyshev filter) a loss in terms of bandwidth due to the parasitic capacitance introduced by the breadboard contacts, thus are you telling me that the diodes can introduce such a behaviour, too? \$\endgroup\$ – Nicola May 30 '19 at 17:33
  • \$\begingroup\$ Yes the RC = Tau when Vf=0 is large even with 4pF x ?Meg = ? us compared to Vf=0.6V = ? ns Thus you see it converge > 0V \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 May 30 '19 at 17:43
1
\$\begingroup\$

The opamp is heavily loaded, thus internal nodes have to slew further to produce that high output current.

The opamp has to turn OFF one diode and turn ON the other diode.

Improve this answer
\$\endgroup\$
3
  • \$\begingroup\$ Hi, thanks for the answer. I firstly tried with high resistor in order to reduce the current magnitude, but the result is very similar to this one. I thought to lower the resistances to have more current and thus switch the diodes faster. The issue is that the measured slew-rate is not the maximum one of the op-amp datasheet. How can you explain this? \$\endgroup\$ – Nicola May 30 '19 at 15:10
  • \$\begingroup\$ If the error signal is small, the opamp is not quickly driven into slew-limiting. The silicon companies show large-signal slewing behavior with pulses. This rectifier is precision triangles, with slow ramps, not the fast edges of pulses. Takes time to fully charge the internal capacitor of the opamp, thus there is a delay before max-speed slewing occurs. To learn, insert 4 series diodes in place of each of the 2 diodes (8 diodes total), and re-evaluate the performance. \$\endgroup\$ – analogsystemsrf May 30 '19 at 16:08
  • \$\begingroup\$ Thanks for the advice. I'll try that the next time I can access the laboratory. However, I just tried the circuit with a square wave input voltage on LTSpice, and the slew-rate of the op-amp output voltage is now very high! Thank you for the help! \$\endgroup\$ – Nicola May 30 '19 at 17:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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