I have a very low voltage (about +/-20mv) wave. What I want to do is split it into two different waves, one reflecting the negative half and one for the positive. Then whenever a wave reaches the 10mv mark in their respective sign, spit out +5v, otherwise nil. I received some advice before for the positive half, I'm just using a voltage comparator and it seems to be working. Now how can I go about taking the negative half and turning it positive so I can do the same thing to it? Also, I'm somewhat worried about using the voltage comparator as a solution as I have a dual yet independent one (the LM393) and when I wire up both comparators to measure the exact same thing and I hook up both channels of my oscope to both outputs the results do not line up and seem sporadic. Is there a reason for this I'm not aware of? It's rather irritating as it seems like there won't be a reasonable solution for my application. I figured I'd come back on here and ask you guys, I search around for solutions to no avail.

Thanks a lot for any info, I'd really appreciate it.

 

How about a simple transistor or op amp invertor, or am I misunderstanding the problem.

 

I have a very low voltage (about +/-20mv) wave. What I want to do is split it into two different waves, one reflecting the negative half and one for the positive. Then whenever a wave reaches the 10mv mark in their respective sign, spit out +5v, otherwise nil. I received some advice before for the positive half, I'm just using a voltage comparator and it seems to be working. Now how can I go about taking the negative half and turning it positive so I can do the same thing to it? Also, I'm somewhat worried about using the voltage comparator as a solution as I have a dual yet independent one (the LM393) and when I wire up both comparators to measure the exact same thing and I hook up both channels of my oscope to both outputs the results do not line up and seem sporadic. Is there a reason for this I'm not aware of? It's rather irritating as it seems like they won't be a reasonable solution for my application. I figured I'd come back on here and ask you guys, I search around for solutions to no avail.
Thanks a lot for any info, I'd really appreciate it.

What frequency range are you dealing with? What is your input waveform?

 

I'd say about 2-3khz. As for the wave form.. I suppose a very narrow triangle form is how I would describe it. Spikes similar to what you would see on a heart monitor.

 

I'd say about 2-3khz. As for the wave form.. I suppose a very narrow triangle form is how I would describe it. Spikes similar to what you would see on a heart monitor.

Are the spikes symmetrical, i.e., positive and negative? How about posting a typical waveform?

 

No, it's manchester/phase encoding. And I can't post a waveform since it's a an old analog scope.

 

No, it's manchester/phase encoding. And I can't post a waveform since it's a an old analog scope.

The Manchester code I'm familiar with looks like this. It doesn't have short spikes.
You don't have to post a scope photo. A sketch will do.

 

No you don't understand, I'm reading in a crappy signal that I will be cleaning up to become a readable manchester code.
_
__/\ / \ __ at +/- 20mv peak.
\/ \/

convert it to:
_ __
__| |_| |___ for + at +5v and:
_ _
____| |__| |__ for - at +5v

I will then be using both signals with a microcontroller to deduce the binary of the manchester code along with the bit rate. (which will be very easy once I can figure out how to get these two waveforms.

 

Damn forums won't allow whitespace.

http://profile.imageshack.us/user/asdasdfasdfafs/images/detail/#232/picia8.png

 

You should be able to AC couple the signal into an amplifier with a gain of somewhere between 10 and 100, then use two comparators to slice the amplified output above and below the baseline. This should eliminate the problem you were having with the comparators. If you need a schematic, I could probably provide one. You would need to tell me your supply voltage.

 

I sort of question the representation of the waveform - the o'scope could be set on AC coupling. Manchester is usually 0 to some positive level.

 

I sort of question the representation of the waveform - the o'scope could be set on AC coupling. Manchester is usually 0 to some positive level.

A signal that's as "crappy" as our OP indicated is probably coming from a transmission channel or some sort of recording device. All DC information is probably lost, but one of the advantages of the Manchester code is that it the DC level carries no information, and is therefore not needed.

 

Are you sure about manchester being 0 to some positive voltage? According to wikipedia it definitely does contain a negative

http://en.wikipedia.org/wiki/Manchester_coding

Not to mention, if you examine how manchester/phase encoding works, it would make sense for there to be a negative. If the stream started with a 0 for instance, and you only listened to the positive, you would have to assume the first bit was a 1 and continue from there only to have some distorted bitstream that would fail a checksum (assuming there is one) and then your software would have to auto adjust and take into account the possibility that the code started 1/2 a bit earlier than you were able to detect. With the negative half of the wave along with it, you can easily see when a stream starts and stops.

Ron H: I would love a schematic. My supply voltage is 5v. Thank you very much for your help.

I was thinking there are opamps that don't need the - rail right? I could just ground it? And then like you said amplify the signal by ~100 gain. But do it twice, one inverted and the other normal. And then do a voltage comparison or something from there on both signals. I don't know, I'm interested to see your schematized solution.

 

Are you sure about manchester being 0 to some positive voltage? According to wikipedia it definitely does contain a negative

http://en.wikipedia.org/wiki/Manchester_coding

Not to mention, if you examine how manchester/phase encoding works, it would make sense for their to be a negative. If the stream started with a 0 for instance, and you only listened to the positive, you would have to assume the first bit was a 1 and continue from there only to have some distorted bitstream that would fail a checksum (assuming there is one) and then your software would have to auto adjust and take into account the possibility that the code started 1/2 a bit earlier than you were able to detect. With the negative half of the wave along with it, you can easily see when a stream starts and stops.

Ron H: I would love a schematic. My supply voltage is 5v. Thank you very much for your help.

I was thinking there are opamps that don't need the - rail right? I could just ground it? And then like you said amplify the signal by ~100 gain. But do it twice, one inverted and the other normal. And then do a voltage comparison or something from there on both signals. I don't know, I'm interested to see your schematized solution.

Do you have good access to parts, or are you stuck with a supplier like Radio Shack? I have to pick an op amp, and it might be nice if I pick one that you can actually get.

 

Actually this is perfect timing. I'm ordering a bunch of parts off of jameco.com today. So anything available there, which is a pretty decent collection. I'm purchasing a linear grab bag from there as well, I'm tired of not having a random assortment of these things and having to salvage them off old boards.

 

The grab bag you'll get will probably have at least a couple of LF351 and/or LF353's in it (single and dual FET-input opamps). They're cheap and ubiquitous.

 

It's a bag of 300 randomly assorted linear series IC's of these varieties (LM300's, 500's, 700's, 7800's and 7900's).

But he can pick anything he wants from jameco and I'll just add them to my order. I wish there was a place that would review these IC's. When you are starting out and there are thousands to select from it's really a pain. Which is why I've opted for the grab bag in which I can explore the most common. Get a feel for it and move on from there.

 

No, it's manchester/phase encoding. And I can't post a waveform since it's a an old analog scope.

The Polaroid is your friend.

 

Here's a circuit that should work. You might want to change some resistor values. You could also add a little hysteresis to the comparators, but you shouldn't need it unless your signal is noisy.
The TLC2272 is available from Jameco. It isn't cheap, but it has rail-to-rail output, which will give you a little more output range. You can use an LM324, which is cheap and ubiquitous, but it's a fairly crappy op amp, and you can only get about 3.5V p-p (which is actually plenty) out of it.
I biased the TLC2272 inputs at approximately the center of the common mode range.

 

Thank you very much, this looks good. I'm going to order the parts today. I'll be keeping you posted to let you know how it worked out. What software did you use to draw up that schematic? And also can it emulate the circuit? I need to see about purchasing some software that can because I believe it would greatly shorten my learning curve.

Edit: Aren't both of these amps doing the same thing on this circuit? I mean shouldn't one of them be reversed in order to spit out an inverted yet amplified wave? I could be completely wrong but I'm just curious.