How to rectify a 10 - 100 kHz pulse signal?

Hi people,

I was wondering how to rectify a bipolar RZ signal like fig. 1 in order to get an output like fig. 2.

I tried to use a full wave rectifier circuit using 4 diodes 1N4148, but I got it all distorted for frequencies above about 100 Hz but I need faster reaction diodes because my input signal has 12.5 or 100 kHz frequency.

Is there any diode which I can purchase to do the job like, maybe, an ultra fast one? Or perhaps an IC that would make it work for me?

I appreciate very much for your attention. Excuse my English, I’m not used to write in international forums.

:smiley:

brz1.jpg

brz2.jpg

Can use a circuit like this.
May a have to AC couple your pulse stream into the op-amp inputs & use resistor divider to get into the same voltage range.

Use fast op amp and high speed logic family.

You might try a "precision rectifier" which puts the diode in the feedback loop of an op amp.

http://www.play-hookey.com/analog/feedback_circuits/full-wave_rectifier.html

1N4148 diodes switch in 4ns, the reason you didn’t get above 100Hz is
perplexing - perhaps you didn’t have a load resistor?

I think, you need a limit comparator, and than invert outputs.

LM393.pdf (73.7 KB)

MarkT:
1N4148 diodes switch in 4ns, the reason you didn't get above 100Hz is
perplexing - perhaps you didn't have a load resistor?

Or your signal source impedance is mis-matched to your load ie high source and low load.

Also if you use a simple bridge rectifier using signal diodes the output voltage lines ( signal and ground) must not be commoned or connected to the source lines signal and ground. ie they must have separate grounds

I tried to use a full wave rectifier circuit using 4 diodes 1N4148, but I got it all distorted for frequencies above about 100 Hz but I need faster reaction diodes because my input signal has 12.5 or 100 kHz frequency.

Hello. It's a digital signal. You can't rectify it because then it becomes a dc signal (no longer digital). It's apples and oranges. Crossroads circuit is what you need. And FYI, next time , if you don't really know, don't do anything until you find out. That kind of approach would get you in a lot of trouble if were some other kind of circuit. I know you're thinking "but I can't hurt it so why not ?" The answer is that it is not about whether you can damage the circuit or not, it's about doing things without knowing what you are doing. For hobbyists, the justification could be made that you have nothing better to do with your time. I'm an electronics technician. If I did that at work and the management found out I was doing stuff without knowing what I was doing I wouldn't last very long. I'm trying to break you of that habit and encourage you to stop and think and ask yourself "Where can I find the answer to this problem?" instead of just blind experimentation.

a bipolar RZ signal

RZ ?

COMPARATOR

Look at the circuit. V+ is Level-2, V- is the signal. To get a LOW for a negative polarity signal, V+ needs to be the signal and V- needs to be Level-2.

If you combine the two polarities all you will wind up with is the clock signal. You have to decode the RZ
by inverting the B signals (swapping the op amp inputs) so that a negative B signal yields a TTL LOW.
"When input-2 goes below Level-2 , the output goes HIGH" would change all the 0s to 1s. You need
"When input-2 goes below Level-2 , the output goes LOW" . That would give you HIGHs wherever input-1
is above Level-1, and LOWs wherever input-2 is BELOW level-2. 10101010 would be converted to the
TTL equivilent which would be HIGH/LOW/HIGH/LOW/HIGH/LOW/HIGH/LOW. Otherwise , no matter what
RZ signal you put in, the output would just look like an RZ clock signal because each bit can only be a 0 or a 1 and the output would ALWAYS be 11111111. If you are TRYING to extract a clock signal from an RZ signal then Crossroads circuit would certainly do that.

IvanRocha:
I tried to use a full wave rectifier circuit using 4 diodes 1N4148, but I got it all distorted for frequencies above about 100 Hz but I need faster reaction diodes because my input signal has 12.5 or 100 kHz frequency.

Err, no you don't need faster rectifiers; they will do perfectly well.

Clearly your circuit is wrong. Until we see your circuit however, we will not know why. It is in fact, an "XY Problem"

IvanRocha:
Is there any diode which I can purchase to do the job like, maybe, an ultra fast one? Or perhaps an IC that would make it work for me?

Yes, I figure 1N914s. The thing is, they are much faster than most common op-amps.

Cute pictures, and illustrate a point.

Trouble is, no voltages specified, so we are unable to suggest an appropriate or more specific circuit than CrossRoads', which has the potential problem with speed of the op-amps.

XY Problem ?
(ie: why is the OP trying to do this ?)

Usually it is to generate a clock that can then be used to decode the signal.

Usually it is to generate a clock that can then be used to decode the signal.

hence my comment here:

If you are TRYING to extract a clock signal from an RZ signal then Crossroads circuit would certainly do that.

Honestly, I am not sure what the correct term is;

A- "generate"
B- "extract"
C- "reconstruct"

Many fast op amps are available that could handle this 100KHz signal, such as this 100 MHz bandwidth designed for video applications

That's exactly what I'm trying to do, to extract a clock signal from an RZ signal. I don't know what a crossroad circuit is, but I'm now trying to think about those comparators for a solution :smiley:

IvanRocha:
That's exactly what I'm trying to do, to extract a clock signal from an RZ signal. I don't know what a crossroad circuit is, but I'm now trying to think about those comparators for a solution :smiley:

lol, it is the circuit CrossRoads gave you in the 1st answer :wink:

If you are TRYING to extract a clock signal from an RZ signal then Crossroads circuit would certainly do that.

Oh, my bad , I forgot the apostrophe (" ' " )("Crossroad's")..

I think that would be
CrossRoads' answer
not
CrossRoad's answer

but the wife says I also like to argue ...

CrossRoads:
Can use a circuit like this.
May a have to AC couple your pulse stream into the op-amp inputs & use resistor divider to get into the same voltage range.

Use fast op amp and high speed logic family.

I think you need a better scope. Those waveforms look awfully wiggly and rounded off! ;D

I for one, am still waiting for an answer regarding what the signal actually is; what voltages and source impedances are involved. As always, the actual circuit.

Raschemmel's question is not adequately addressed.

raschemmel:
XY Problem ?
(ie: why is the OP trying to do this ?)

And it would indeed be - CrossRoads' answer.

If you are TRYING to extract a clock signal from an RZ signal then Crossroads circuit would certainly do that.

That's exactly what I'm trying to do, to extract a clock signal from an RZ signal.

And it would indeed be - CrossRoads' answer.

And (the solution) would indeed be the circuit in - CrossRoads' answer

The word might be ‘derive’