3
\$\begingroup\$

Well, I'm working on an oscillator topology to excite a piezoelectric device that works at high voltages. For this I have a high frequency signal generation stage followed by a push pull amplifier to provide a current gain before injecting this signal into a ferrite core step up transformer.

The signal is 35kHz, 7Vpp.

The problem is with the push pull output..., I'm using a TIP31 and TIP32 pair, at first I'm not interested in removing the crossover distortion. As the base current of the TIP is relatively high and to generate the signal I am using a TL084 operational amplifier that does not provide large output currents, I use another push pull stage formed by a pair of BC337 and BC327 to then inject this signal into the base of the pair of TIP's, then a strange behavior occurs when I place a load that drains more than 20mA, the signal is completely cut off leaving me with zero output on the oscilloscope, I know that I need resistors on the emitters of each pair of transistors to compensate for the ratio of hfe by temperature when the transistor is conducting, but theoretically shouldn't this arrangement work?

enter image description here

\$\endgroup\$
7
  • \$\begingroup\$ When the output is "completely cut off", what do the voltages at other nodes in the circuit look like? What frequency is your oscillator, and what is the reactance of the various coupling capacitors at that frequency? Especially: how much current can you push into the bases of that TIP31/TIP32 pair? \$\endgroup\$
    – Hearth
    Nov 15, 2023 at 15:49
  • \$\begingroup\$ I'm not sure where you expect the recombination current to come from. Just via the capacitors? That's it? \$\endgroup\$ Nov 15, 2023 at 16:07
  • \$\begingroup\$ How big is the output of U3B? You have about 2.5V of dead space on your input signal when it changes direction when you're pulling nonnegligible current. You might be able to compensate for this by running the U3B feedback rheostat from the push-pull stage output rather than from the op amp output. \$\endgroup\$ Nov 15, 2023 at 16:51
  • \$\begingroup\$ @CristobolPolychronopolis ..., I already did this, I used the push pull output as feedback for U3:B, but I believe that this is not the reason for the problem, as I am using a function generator to do the tests and the problem persists, therefore, the part of the circuit that is not highlighted is irrelevant. \$\endgroup\$
    – ever_pi
    Nov 15, 2023 at 17:06
  • 2
    \$\begingroup\$ "Cutoff" has a specific meaning with respect to devices. Please show the waveforms you have measured to arrive at this determination so we can confirm it or offer alternative explanations. \$\endgroup\$ Nov 15, 2023 at 20:57

3 Answers 3

5
\$\begingroup\$

Try the complementary Darlington configuration, aka "Sziklai pair" for the output stage, rather than the Darlingtion as shown in the OP. Refer schematic below, just connect the bases of Q3 & Q4 together, then connect that node to the output of the opamp. No need for DC blocking capacitor.

The schematic is from a low-power audio amplifier, for your application simply change the transistors to match your requirements, and adjust the resistors as necessary. This configuration has less voltage drop (better) from opamp output to load compared to Darlington.

enter image description here

https://en.wikipedia.org/wiki/Sziklai_pair

\$\endgroup\$
2
  • 1
    \$\begingroup\$ thanks, it worked for me \$\endgroup\$
    – ever_pi
    Nov 16, 2023 at 14:24
  • \$\begingroup\$ Would be good to show the waveforms before and after applying this remedy. Cheers. \$\endgroup\$ May 12 at 22:23
3
\$\begingroup\$

Your driver and output transistors have no DC bias. Since your input signal is a squarewave from the oscillator then you do not need the coupling capacitors. push pull

\$\endgroup\$
3
\$\begingroup\$

At that point you may as well put the whole op-amp together :)

The circuit below has +/-1V (2Vpp) input and +/-7V (14Vpp) output. Increase C1 should there be any instability with your particular transformer load. Output current is mostly limited by the dissipation of Q10 and Q11.

schematic

simulate this circuit – Schematic created using CircuitLab

The DC offset on the output can be trimmed out, but generally should be fairly small - especially if the input pair is matched. No need for capacitive coupling, usually, even though I've shown it.

Adjust R10 for say 20mA idle current through Q10-Q11 without load and with input at 0V.

\$\endgroup\$
2
  • \$\begingroup\$ Confused: is the base of Q1 tied to the collector of Q12 and both tied to GND? Seems that way in the schematic. Perhaps the location of the "dot" associated with the node "IN+" may be misplaced. \$\endgroup\$ Nov 17, 2023 at 23:41
  • \$\begingroup\$ @FabioBarone IN+ is at ground potential, and so is the collector of Q12 and base of Q1. Bases of Q1 and Q2 are op-amp inputs. Q12,D1,D2 are a bias current reference (lousy, but should be good enough). \$\endgroup\$ Apr 21 at 17:33

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.