Comments on op-amp zero crossing detector

[MrB]

All,

I want to evaluate the following czero-crossing detector circuit:



This circuit takes advantage of the fact that the voltage drop between the base and emitter is always 0.6 volts - for the PNP the emitter is 0.6 volts above the base. So, the circuit uses this fact to bring the zero-volt crossing up to 0.6 volts or so, where an op-amp can operate. In fact, the transistor Q2 with the grounded base can instead be set up with a voltage divider, thus changing the non-inverting trip point. The feedback resistor is for hysteresis.

Question: is the bias current from the op-amp sufficient for the transistors, or should there be pullup resistors to +V applied to both op-amp imputs?

 

[audioguru]

Hi,
Why not get rid of the transistors, feed the AC directly into the opamp and let its hysterisis determine the voltage trip points for the zero crossing?

 

[MrB]

The input signal is bipolar - I want to detect the actual zero crossing point, or even compare slighly negative (like -100 millivolts), hence the voltage translation.

Back to this circuit - is there enough collector current supplied by the op amo bias current for it to operate? Pspice says yes...

 

[audioguru]

The opamp is bipolar too if you use a dual supply for it, or bias-up its inverting input to 1/2 supply then cap-couple the AC into it. Using the opamp alone, you can choose a low hysterisis of 10mV or less depending on the opamp's input offset voltage.

I don't see why you are using transistors. Their current is next to nothing and their voltages cancel each other.

 

[Optikon]

The input signal is bipolar - I want to detect the actual zero crossing point, or even compare slighly negative (like -100 millivolts), hence the voltage translation.

Back to this circuit - is there enough collector current supplied by the op amo bias current for it to operate? Pspice says yes...

Well, I did not see any part numbers........
I can buy an electrometer grade op-amp where the biad currents are <1 pA so your answer could be a resounding NO. Or I could get a high speed bipolar opamp with mA of bias current and that would be a yes.
Some trans & O/A part numbers would be useful. If you used some PSpice model, why not just use that part number for the real circuit?

 

[MrB]

Well, I did not see any part numbers........
I can buy an electrometer grade op-amp where the biad currents are <1 pA so your answer could be a resounding NO. Or I could get a high speed bipolar opamp with mA of bias current and that would be a yes.
Some trans & O/A part numbers would be useful. If you used some PSpice model, why not just use that part number for the real circuit?

Good point! The target part is Onsemi MC34071 bipolar O/A (using this for -40 to +125C temp range, 40+ volt single supply range, and inputs to near negative rail). Datasheet bias current (lowest specified) is 100na. Gut feeling is that this is right on the edge of being too low. I do not have a model of this Opamp for pspice, so I cannot model this directly.

 

[audioguru]

With the input voltages of the opamp operating so close to its negative supply, the hysterisis will be very assymmetrical.

 

[MrB]

The op-amp(s) I am looking at have PNP input stages. Datasheet indicates that the op-amp can operate down to the Vee rail (it just cannot get to the Vdd side) - "Because the input common mode voltage range of this input stage includes the Vee potential, single-supply operation is feasible to as low as 3.0V...." I did not illustrate the required input clipping diodes to keep the input within the supply rails. But, I can see that excessive feedback hysteresis (to the + input) could cause problems, so this would be limited to a very small amount.

 

[audioguru]

I though you wanted a zero-crossing detector. The opamp needs lots of hysteresis to have a fast "snap-action" and a window so that it does not trigger on its own and the input signal's noise.
You need just an opamp that is biased down the middle (for symmetry) and the input signal fed to its hysteresis resistor. If the AC signal is grounded and you are using a single supply, the signal must be cap-coupled. The cap-coupled input signal can't be directly fed to the hysteresis resistor because the opamp's input voltage will go negative at power-up until the cap is charged. So the Schmitt-trigger opamp will need an inverting buffer stage to charge the cap safely. That's OK, a dual opamp costs about the same or even less than a single. Like this:

 

[Roff]

You want symmetrical hysteresis? I'll cho you symmetrical hysteresis.
This assumes your input is relatively low frequency.

 

[audioguru]

Hi Ron,
That's better. You've added R4 and R5 to give the original circuit's input transistors some current (I would like even more current) and you've limited the "opamp's" open collector output voltage with Q3 and Q4 so the hysteresis is very symmetrical, like my opamp circuit.
But its input voltage limit is only about 1.2V p-p and additional resistors or diodes might be needed to reduce an input voltage to this level. My opamp circuit just needs the value of its input or feedback resistor changed to match the expected maximum input voltage swing.

 

[MrB]

O.K. - now I understand what you mean by symmetrical hysteresis and you are right. I failed to state that I am only looking for the positive-to-negative sine transition, the neg-to-pos one is ignored. The goal is to determine the pos2neg transistion as close to zero volts, whereas the other transition can have an "asymmetrical" point, i.e. not looking for the zero point +/- hysteresis, but rather one transistion very close to the real zero crossing and the other containing all of the hysteresis. It does help to state the requirements up-front, but these are both good circuits to keep around. And the original question of bias of the transistors was also answered - thanks

 

[audioguru]

Hi MrB,
So you don't care about half the waveform. Then you are right, you don't want any hysteresis, just high gain and low offset voltage.
Maybe you should also filter-out any noise on the input, but watch-out for a delay from the filter.

 

[Roff]

Here is a zero crossing detector that switches on the zero crossing of the fallig edge. All the hysteresis is assigned to the rising edge. Since it is AC-coupled, it will only work on waveforms (e.g., sine waves) that contain no even harmonics. Any waveform with 50% duty cycle will work. Comparator delay will be significant for high frequency waveforms, or those with fast fall times. This will be even worse if you use an op amp as a comparator, because they are generally pretty slow as a comparator.
In this circuit, U1B operates as a voltage follower, providing a low impedance reference at VCC/2. U1A is the comparator.