Rectifying 100khz signal

[Fiorenzo]

I am trying to half wave rectify a signal of 100khz but without success because the sinewave get distorted.

I have tested many different diodes I have but every one doesn't give good results.

As you see in the attached photos the halfwave is someway ok up to 10khz  but with 100khz it is not.

It seem to me a combination of leakage current and recovery time of the diode.

In the setup you see on the photos I have used an ultra fast diode that it's datasheet says it has no recovery time.

Should I use the small 1n4148 signal diode or is the way I am going that is wrong?

Thank you!!

[CatalinaWOW]

If the only load in your circuit is the oscilloscope input impedance you may just be fighting the rc time constant from the diode and your leads.  If you can set the scope to 50 ohms that would be one solution, but the real solution is the set up your diode as it will be used in whatever your application is.  Make the oscilloscope a negligible observation tool, not a major circuit component.

[Fiorenzo]

Actually I have put a 1kohm resistor from the output of the diode to ground, without it the output is equal to the input (diode not biased?).

Still the problem stays here... Also with a 100ohm resistor...

[Kleinstein]

In addition to the reverse recovery time the diode capacitance can also load down the source. The fast power diodes are usually in the range of 50-200 ns reverse recovery time. The slower ones may alreday start to show some recovery current at 100 kHz. The 1N4148 is at some 5 ns and relatively low capacitance, so definitely a suitable choice.

It the source signal is very high in impedance, it may be better to first buffer or amplify the signal and only than use rectification.

For a sensor signal with a known / available excitation signal one could also consider a phase sensitive rectification: so no using simple diodes, but more like CMOS switches or similar as a kind of multiplier / mixer. This would also allow to include the phase information and get good suppression of noise from other frequencies.

[Fiorenzo]

Thank you Kleinstein.

What I don't understand is why on the datasheet of the diodes I have tested is written on many of them that they are suitable for frequency over 1mhz.

I know that the capacitance of the diode is a problem with high frequency, but it doesn't seem to me so extreme a 100khz signal.

Do you think that the distortion seen on the oscilloscope are due to the too much "high frequency"? or because I should implement around the diode a more "complex" circuit?

As you know, the input signal is the one from the torque sensor and it is of high impedance... If it has to deliver more than few mA the signal collapses.

[Kleinstein]

With a high source impedance the signal will be distorted, but this may be actually not that bad. The rectifier may still work. It is just that the signal one is looking at than is not longer the input AC, but something in between.
How much the capacitance matters also depends on the signal impedance. With higher source impedance it gets more important.

For the torque sensor it many still be better to use phase sensitive detection, as this also captures the phase and thus gets a sign to the amplitude.
The simple rectifier with diode gets tricky if the amplitude gets much below 100 mV as it gets no longer linear. An active rectifier with OP is comparable effort to phase sensitive rectification, possibly even more tricky with the relatively high frequency.

[Fiorenzo]

Thank again.

Is the phase sensitive rectifier the same as it is used in modern smps power supply?

Because I don't now where to start with this kind of circuit.

I may go as you suggested but I was also keen to experiment with diodes and op amps for learning reasons.

So.... yesterday I desoldered a small smd signal diode that was on a faulty electronic board and used it for some tests.

The shape of the rectified signal is way better than with the other diodes I have tested.
What I don't understand is why diodes rated for switching applications give a so big distorted output. Also the diode stated with an insignificant recovery time did similarly...

I have many power supply and DC-DC converters that rectify with diodes the high frequency square wave that is done for the voltage conversion without apparent problems.

Is maybe because in my setup there is a too small current flowing out of these diodes? Or are they using some strategy to mitigate the capacitive, leakage current and recovery time?

[CatalinaWOW]

A phase sensitive rectifier is the first half of a synchronous filter.  Googling synchronous filters should give lots of information.