Is it normal for an opamp to "rectify" a signal?

[Zerosum]

Forgive my poor understanding of this subject, but I have built a simple white noise generator, which is working. I then fed the signal into a UA741 low-noise opamp. The signal was indeed amplified, as I expected, but I now only get the upper half of the waveform, the opamp seems to be cutting the negative component.

Is this normal behavior? How would I prevent this while still amplifying the signal?

 

[(*steve*)]

It is normal behaviour if your circuit is not correctly designed.

Please post the schematic.

 

[Harald Kapp]

... or it the negative supply voltage is missing.

 

[eman12]

Forgive my poor understanding of this subject, but I have built a simple white noise generator, which is working. I then fed the signal into a UA741 low-noise opamp. The signal was indeed amplified, as I expected, but I now only get the upper half of the waveform, the opamp seems to be cutting the negative component.

Is this normal behavior? How would I prevent this while still amplifying the signal?

Hi Zerosum
Did you know that most of Op-amps need a Dual symmetrical power supply (ie "+ ground -") or a virtual ground instead to normally work? Sometimes you need to put a capacitor between the stages (ie between an output to the next input) as a coupler cap as well?

 

[eman12]

And why did you need to use 741 op-amp in this regard? To amplify the signal?

 

[Zerosum]

Hello,

This is the schematic that I am using:
http://electro-music.com/forum/phpbb-files/whitenoise_969.jpg

I hadn't tried using a split power supply for the op amp, I should have thought of that! However I just tried it, and it didn't seem to make a difference? I was still getting a half wave. Is there something in the circuit that would cause this?

The opamp is just to amplify the signal, yes, it is actually producing noise without the opamp.

I've attached a screengrab of the oscilloscope trace, the yellow (CH1) is the input to the opamp, and the blue (CH2) is what the opamp is outputting. It seems that the wave is being truncated somehow?

Thanks for the help!

 

[Harald Kapp]

Where did you get that schematic from?
This is not the way to use an OpAmp.
Read the basics on operational amplifier circuits e.g. here. If your circuit is meant to be an inverting opamp, you need to connect at least the noninverting input to ground (0V).

What is your frequency range? The circuit will operate as a differentiator with high gain. The 1Meg feedback resistor is probably much too high. A diffenretiator works similar to a high pass filter, whereas DC gain is zero (blocked by the capacitor at the input). Is that what you want?

Btw: The 741 is not a particularly low noise opamp. Nor is it modern. Why do you look for a low-noise amplifier in a noise generating circuit anyway?

 

[duke37]

The output socket is connected to -V. The output cannot go below this.
Th socket should be connected to the centre ground of the power supply.

 

[CDRIVE]

Besides the other odd stuff on the schematic check out the NPN connections.????

BTW., Agreed that the 741 is neither low noise nor modern but we need a low noise amp for a white noise amplifier ..why?

Chris

Edit: I would think that the prime requirement would be a wide band amp.

 

[(*steve*)]

Something like this might be better

Note the THREE power supply connections. Your supply should consist of 2 nine volt batteries in series. The +ve end is the +ve rail, the -ve end is the -ve rail, and the connection between the batteries is the 0v rail (or ground),

R4 and R5 determine the gain. Increasing R5 will increase the gain. You could place a 1M trimpot here and you could adjust the gain from 1 to about 20.

Note that the +ve rail (also the op-amp +ve power supply) is shown at the top and the -ve rail (also the op-amp's negative power rail) is shown at the bottom. This is a convention that serves to make your schematic more readable and lessens the chance of confusion.

edit: CDRIVE -- Are the creases more visible now?

 

[CDRIVE]

Steve, I really am loosing short term memory. It took me nearly 3 minutes to understand that.
Yes, the bar napkin!

Chris

Edit: Steve, where is the noise gen being injected?

 

[(*steve*)]

Steve, I really am loosing short term memory. It took me nearly 3 minutes to understand that.

Damn, I'm not being obscure am I?

Edit: Steve, where is the noise gen being injected?

The reverse biased BE junction generates heaps of noise.

 

[Zerosum]

Hello, wow, thanks for all of the helpful replies!

I actually just got this circuit to work, I think I was connecting the two oscilloscope ground leads incorrectly? I have one lead connected to the input, and one to the output of the op amp, and when I changed their position I get a full wave output, but it's pretty clear from this thread that I've done things wrong, so I will certainly take those suggestions on board!

The reason I am using a low noise amp is because I got that circuit from a youtube video (I know, I know ) and that's the opamp that the video used. I have a general purpose opamp which I can try, and I have built a circuit which splits a power supply into +ve, 0, and -ve, so I should be able to use that instead of 9v batteries.

@steve, thanks a lot, I'll definitely try out that circuit! I'm pretty sure I have all the parts I need.

Another quick question: I have some 5.1v zener diodes, would they be better for this application than a transistor with one pin unconnected? I understand the basics of "reverse breakdown" and "avalanche noise" but could the unconnected pin of the transistor act as a simple antenna, and pick up signals from the environment, which might affect the noise generated, whereeas a zener diode wouldn't?

Thanks again for the help!

 

[CDRIVE]

Damn, I'm not being obscure am I?

No, you're not. It's me!

The reverse biased BE junction generates heaps of noise.

He said he built a noise gen that I thought was not shown on the print. I thought the transistor was a mis-wired buffer amp... Duh!!

It's been a while but I remember Tunnel Diodes being a favorite for noise gens.

Chris

 

[(*steve*)]

I actually just got this circuit to work, I think I was connecting the two oscilloscope ground leads incorrectly?

That would do it! Both of the ground leads are connected together. Perhaps you were shorting out half of your power supply.

Fortunately it also sounds like you didn't destroy either your meter or your power supply.

Beware of this trap in the future. It is very possible (not to mention potentially very destructive) to short a high energy source through your oscilloscope.

The reason I am using a low noise amp is because I got that circuit from a youtube video (I know, I know ) and that's the opamp that the video used.

If you're using a 741, then this is just a general purpose op-amp. It's not low noise. I can't see much purpose to use a low noise amplifier here, and in any case the amplitude of the signal from the transistor is reasonably large anyway (I think).

I have a general purpose opamp which I can try, and I have built a circuit which splits a power supply into +ve, 0, and -ve, so I should be able to use that instead of 9v batteries.

The breakdown voltage of the transistor is probably around 7V. This will limit the life of the battery because if the terminal voltage falls too far the white noise will cease before the battery is exhausted. A +/- 9V power supply will keep it running longer.

Another quick question: I have some 5.1v zener diodes, would they be better for this application than a transistor with one pin unconnected?

The best suggestion is to try them and find out.

I suspect the transistor will be the noisiest source, and that is for one major reason. Zener diodes are made to be operated in this region and reducing noise is one design criteria. The transistor is not expected to operate in this region and thus the lack of noise is not an issue.

We did some tests on transistors some time ago and discovered (perhaps found is the better word) the reason why this kills transistors. It possibly also goes some way to describing why they're so noisy. "Hot carrier injection" are the magic words.

I understand the basics of "reverse breakdown" and "avalanche noise" but could the unconnected pin of the transistor act as a simple antenna, and pick up signals from the environment, which might affect the noise generated, whereeas a zener diode wouldn't?

I am reminded that any piece of wire can be an antenna. In this case, the wire is short, it's likely to only be effective for high frequencies, far higher than your op-amp can amplify. In addition, a small voltage on the collector is unlikely to affect the base.

You could experiment by connecting the collector and emitter together. I'm pretty sure that would have no effect on the noise characteristics, but would go a long way to preventing the collector lead being an effective antenna.

 

[Zerosum]

Uh oh, it's possible to destroy a scope that easily? Wouldn't it have some kind of internal overvoltage protection? I have a cheap multimeter that has a removable fuse, surely a relatively expensive scope would have something similiar? In any case, i'll certainly be more careful in future!

I am getting a pretty strong source without the amp, but with the amp I can connect the circuit to a speaker and I get a high-volume sound out, which is useful.
One note about low-noise vs regular opamps. My understanding (admittedly limited) is that the low-noise amp is designed to reduce the noise generated by the amplification process, and NOT the noise in the signal itself, is this not correct? So, if the intention is to produce clean white noise, (which I want to feed into a computer to make a true random number generator) using a low noise amp would prevent "regular" noise getting into the noise source, which could create a predictable pattern in the supposedly white noise. Is this assumption correct at all?

I'm not too worried about battery duration, I'm more concerned with proof of concept, but I take your point. What would be done in a commercial application? I've never seen products which require 2 9v batteries, would they use some kind of boost converter and power splitter like I am using?

I did try the zener's, they didn't work, I didn't think about the fact that they would be designed for low noise operation, that makes sense... That's probably why the transistors worked much better. When you say "kills transistors" do you mean the transistors will eventually burn out?

What you said about the antenna makes sense, it just seemed plausible to me that it could pick something up. Since I'm trying to build an RNG from this, I want the noise to be a "ideal" as possible, other wise I am going to get predictable patterns in the data.

Thanks a lot for your help, I appreciate it!

 

[(*steve*)]

Uh oh, it's possible to destroy a scope that easily? Wouldn't it have some kind of internal overvoltage protection? I have a cheap multimeter that has a removable fuse, surely a relatively expensive scope would have something similiar? In any case, i'll certainly be more careful in future!

The ground of the probe is connected to the metal outside of the BNC plug (which is most often used) at the scope end. These are typically connected directly onto the chassis which is earthed. There is no fuse here.

If you clip the ground lead to something that has a voltage relative to mains earth, current will flow through the shield of the probe, to the scope, then through your mains cable to ground. Either the current will be limited by something, or something will melt. It may be a fuse in the equipment under test, it may be wires, or it may be your scope lead. It could also trigger an earth leakage breaker (which would be a good thing).

If you have something with an isolated supply, this won't happen, but if you connect 2 earths (which are connected together at the scope) to 2 points with different potential, then current will flow through one probe lead to the scope, through the chassis, and back through the other probe lead. Again, something will happen. If the device is powered by small batteries, then you might short one out (it might get warm) and nothing much will happen. For a higher power supply, see above (except you won't blow an earth leakage breaker).

This is one of the hidden traps of using a scope.

I am getting a pretty strong source without the amp, but with the amp I can connect the circuit to a speaker and I get a high-volume sound out, which is useful.

You should probably check that you're not getting clipping of your output.

One note about low-noise vs regular opamps. My understanding (admittedly limited) is that the low-noise amp is designed to reduce the noise generated by the amplification process, and NOT the noise in the signal itself, is this not correct? So, if the intention is to produce clean white noise, (which I want to feed into a computer to make a true random number generator) using a low noise amp would prevent "regular" noise getting into the noise source, which could create a predictable pattern in the supposedly white noise. Is this assumption correct at all?

In general, noise is noise. The noise you want has a constant power from DC to daylight. As long as all sources of noise are similar, you're OK. It *might* be a problem if you were trying to generate pink noise, but even then, all other sources of noise will be tiny compared to what the transistor is creating.

I'm not too worried about battery duration, I'm more concerned with proof of concept, but I take your point. What would be done in a commercial application? I've never seen products which require 2 9v batteries, would they use some kind of boost converter and power splitter like I am using?

They might. Or they might employ a circuit to generate the -ve rail. 2 batteries is probably simpler, but probably not as commercially acceptable.

I did try the zener's, they didn't work, I didn't think about the fact that they would be designed for low noise operation, that makes sense... That's probably why the transistors worked much better. When you say "kills transistors" do you mean the transistors will eventually burn out?

That's a very good question.

The transistor will have steadily deteriorating gain the longer you use it in this application. (If you're interested, you can read this thread https://www.electronicspoint.com/emitter-base-breakdown-t257654.html)

Will that affect the noise spectrum or the amplitude of the noise? I really don't know. However it would be really interesting to find out.

What you said about the antenna makes sense, it just seemed plausible to me that it could pick something up. Since I'm trying to build an RNG from this, I want the noise to be a "ideal" as possible, other wise I am going to get predictable patterns in the data.

Let us know how you get on with that.

Thanks a lot for your help, I appreciate it!

No problems, it's a pleasure.

 

[Zerosum]

Hello,

Sorry, I meant to reply to this sooner!

Your explanation of the potential to damage a scope makes sense, I will certainly pay more attention to that in future! Am I to infer then, that connecting just one scope earth at a time would be safe?

This circuit is more of a concept test than anything else, so I am not too worried about short-lived transistors or odd power arrangements, as long as it works I'm satisfied enough.

I wrote up my results on my blog, if you'd like to read it:
http://phoenixgamedevelopment.com/blogs/index.php?s=rng

There are a few interesting posts there. Basically, it seemed to "work", but it didn't pass the Die-Hard tests for randomness. I suspect it had something to do with sampling the data with the microcontroller?

Thanks for your time!

 

[(*steve*)]

The problem occurs when you connect the ground lead of the grounded oscilloscope to something with a potential different to ground and where current can flow.

Bad things

e.g. connecting the ground to your mains live wire <-- this is very bad

e.g. connecting your ground to your mains neutral <-- shouldn't be bad, but an earth fault or swapped live/neutral can ruin your day.

e.g. connecting ground lead to power supply rail where another rail is connected to mains earth.

OK things

e.g. Connecting your ground lead to a rail of a power supply that floats with respect to mains earth.

e.g. connecting the ground lead to mains earth

e.g. connecting your lead to a power supply rail that is grounded.

Another consideration is when your scope is not grounded. In most cases this means you have a portable battery operated scope. In this case you won't get nasty currents if you connect ground to something with a non-ground potential, but you might make its chassis or external connectors live.

In this case, if the equipment has 2 earth connections, then connecting them to points which have a potential difference wrt each other will ruin your day.

 

[CDRIVE]

The problem occurs when you connect the ground lead of the grounded oscilloscope to something with a potential different to ground and where current can flow.

Bad things


e.g. connecting your ground to your mains neutral <-- shouldn't be bad, but an earth fault or swapped live/neutral can ruin your day.

Ever have one of those issues stuck in your head and anxiously waited for just the right moment to share it? Well I have and I can't think of a better time to post this. The fact is you don't need a swapped Hot/Neutral to have a very bad day.

What I'm about to describe was discovered while I was contracting for a hospital. A new wing was being built and I had a contract to produce and install a SS control that was located in a 12 x 12 x 6 electrical box located in all the patient room ceilings. This was a patient control that accessed over bed lighting and TV control via their pillow speaker. We called them hand sets.

On that day the electricians had completed installing the boxes, mains power and lines from the fluorescent lamps. I was anxious to pull my LV control cable and install my boards because the ceiling was still wide open and much easier to work in.

The first thing I wanted to check was mains power. I usually did this with an inductive probe but I had left it on my bench in the Bio-Med shop but I had my trusty Fluke DMM in my cart. Anyway, while testing the voltage between neutral and hot the neutral grazed the metal box (GND) and threw a healthy spark. I backed down the ladder to ponder the incident. Went back up and measured the voltage between Neutral and Ground. Eek I measured about 9VAC! Back down the ladder to think a bit longer and sketch a schematic. The 9V difference didn't concern me as much as the healthy arc that I saw. So I measured the current between GND and Neutral. It was substantial but I don't recall the exact measurement. Since the DMM is 10A max it had to be less than 10A.

After further investigation I found that this room was the last room in the string. The electricians had hard wired all the lighting (incandescent and fluorescent) in all the rooms to give them working light. Further down the line was a healthy compressor load located a MED Gas closet. All the lights plus the compressor produced voltage drop on the Hot and the Neutral. The Ground was 1" hard line conduit plus a separate copper (12AWG) conductor.

If this all seems confusing your not alone but it all makes sense. When I grazed the ground I reduced the total return resistance to the panel substantially because I was effectively shunting the Neutral leg with the Ground leg. That's why it produced a substantial current between the two.

What all this has to do with scope ground may also seem irrelevant or maybe fuzzy at best but I'm not done here. Take a look at the attached simulation. What you will see in this sim is 21 peak amps flowing through the scope's ground test lead when connected to the Neutral at 100mS during a transient characteristics sim. This happening because of the 4 Ohm load on the end of this line. All three of the mains conductors form a voltage divider causing very high current to flow through the scope probe's ground lead.

Chris