Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Help! w/op amps

Status
Not open for further replies.

carbonzit

Active Member
I'm an experienced electronics tinkerer with many gaps in my knowledge of the subject, and am currently trying to educate myself properly. Part of this involves experimenting with actual circuits. I'm hoping to get some help here from folks more knowledgeable than I.

Current project involves fooling around with a simple noise generator. I'm using a reverse-biased NPN transistor, as shown in lots of circuits I've seen. I'm trying to use an op amp to amplify the output. At this point, I'm just trying to get the damn thing to work at all; I've constructed the circuit shown below, with the output connected to a small audio amplifier, in order to hear if it works.

**broken link removed**

Well, it ain't working. The noise generator seems to generate nothing. When I remove the wire connecting it to the op amp noninverting input, I hear no difference in the sound. The op amp definitely works; if I touch the input lead, I get very loud hum. Very frustrating; I've tried a bunch of stuff to no avail. Tried a bunch of different transistors, all tested good. Put in a coupling capacitor between the noise generator and the op amp input. Removed the feedback resistors from the op amp. Put in a load resistor across the op amp input. The only noise I get out of this is the noise of the op amp itself.

So apart from the actual application here (I have some fun things involving random noise generation in mind here), I'm trying to teach myself how to use op amps, and also just basic circuit design principles: computing op amp gain, different op amp configurations and feedback, impedance matching, etc.

My assumption (which could be wrong, of course!) is that since the noise generator produces such a small signal (in the millivolt range), I need a fairly high-impedance input so as not to overload it. But now I'm wondering if there's not enough of a load; should I use a load resistor? Part of my problem, it seems, is an impedance mismatch somewhere.

Also, it seems that the circuit here should give me plenty enough gain to at least hear something through a regular line-level audio input.

Any help will be very much appreciated! Ask me about anything I may have neglected to mention that may be relevant.
 
The opamp is not properly biased. The inverting input is returned to ground, while the non-inverting input is sitting at V+ (because the reverse leakage through the transistor is tiny. This will cause the opamp output to rail positive.

What is the supply voltage?

Even if it worked, the gain of the opamp is only 11, which is no where near high enough. I'm guessing you need a gain of >1000.

The LM324 is not a low-noise opamp. Question is, will you be hearing the opamp noise or the transistor noise?
 
Last edited:
Hey, that was quick. Thanks!

The opamp is not properly biased. The inverting input is returned to ground, while the non-inverting input is sitting at V+ (because the reverse leakage through the transistor is tiny. This will cause the opamp output to rail positive.

Well, in my defense I should say that I lifted this circuit almost verbatim from one I found on the Web that (presumably) works. But I see what you're saying: the noninverting input is 100K ohms away from the positive rail.

But shouldn't this have been fixed by inserting a coupling cap betwixt transistor and op amp? I tried that (used a 45 uf tantalum, I believe) but made no difference.

What is the supply voltage?

~15 volts (single-ended supply).

Even if it worked, the gain of the opamp is only 11, which is no where near high enough. I'm guessing you need a gain of >1000.

The LM324 is not a low-noise opamp. Question is, will you be hearing the opamp noise or the transistor noise?

So one of my questions is, how do I rig this for more gain? I don't really need stablilization here (well, as long as the damn thing doesn't break into spontaneous oscillation). What if I just remove the feedback network entirely? Doesn't this give me the maximum gain possible? (Which could be waaaay too high, I guess.)

Also don't care about low noise here (obviously, since I'm trying to generate it!). I used this op amp since I had it. (I also have some 4136PCs and an ECG727 in my motley collection; should I use one of these instead? (Keep in mind I'm really really cheap and am trying to use what I have here.)
 
Last edited:
Try this connection:

The voltage at the non-inverting input should be the reverse Vbe breakdown voltage of the transistor. Select a transistor that has a reverse breakdown of about 5V. 2N3904 comes to mind.
 

Attachments

  • NoiseSource.png
    NoiseSource.png
    9.8 KB · Views: 413
Last edited:
Hey, thanks, Mike. That works a lot better. I'm obviously getting a lot of gain out of that stage.

So here's what I ended up with:

**broken link removed**

Still some problems. (I should point out that a lot of my trouble is due to the funky construction method; I'm using a prototype board, and it's a snarled-up mess of components and wires.)

I tried cascading two stages like the one you suggested, but got absolutely no output out of the 2nd stage. I tried all the op amps separately and they all work fine individually.

Anyhow, I went ahead and constructed the low-pass filter stage as shown. It works, but at a greatly reduced output. This should be a unity-gain stage, shouldn't it? But obviously there's a loss instead of any gain. It does, however, seem to be doing its job; the high-frequency component of the sound is greatly reduced. (I got this circuit from my current textbook, Malvino's Electronic Principles). The formula he gives for the cutoff frequency is

Code:
f(c) = 1 / 2ΠRC√2

which should be about 34 Hz with these values.

Questions:
1. When I connect the filter stage's output to the inverting input as shown, it goes dead. It works (sort of) without it. What gives?

2. The whole idea of this exercise, besides trying to learn some stuff, is to generate subsonic noise. Unfortunately, that doesn't seem possible with this method. I suspect that the spectrum of the noise generator doesn't include much below, say, 20 Hz. (What I'm hoping for is a low-frequency flicker.) Is this because of the noise generator? or are the lower frequencies disappearing somewhere else here?

3. In your suggested circuit, what's the function of the capacitor ground for the inverting input? I've seen lots of circuits without this. Is this to establish an AC ground? or to block DC?

By the way, I went through my parts box of transistors; some were obviously much louder (= more noise output) than others. The best one was some old TO-92 marked "7128" (god knows where that came from ...).

By the way, here's a link to another similar circuit I found on the web: **broken link removed**
 
I'm an experienced electronics tinkerer with many gaps in my knowledge of the subject, and am currently trying to educate myself properly. Part of this involves experimenting with actual circuits. I'm hoping to get some help here from folks more knowledgeable than I.

Current project involves fooling around with a simple noise generator. I'm using a reverse-biased NPN transistor, as shown in lots of circuits I've seen. I'm trying to use an op amp to amplify the output. At this point, I'm just trying to get the damn thing to work at all; I've constructed the circuit shown below, with the output connected to a small audio amplifier, in order to hear if it works.

**broken link removed**

Well, it ain't working. The noise generator seems to generate nothing. When I remove the wire connecting it to the op amp noninverting input, I hear no difference in the sound. The op amp definitely works; if I touch the input lead, I get very loud hum. Very frustrating; I've tried a bunch of stuff to no avail. Tried a bunch of different transistors, all tested good. Put in a coupling capacitor between the noise generator and the op amp input. Removed the feedback resistors from the op amp. Put in a load resistor across the op amp input. The only noise I get out of this is the noise of the op amp itself.

So apart from the actual application here (I have some fun things involving random noise generation in mind here), I'm trying to teach myself how to use op amps, and also just basic circuit design principles: computing op amp gain, different op amp configurations and feedback, impedance matching, etc.

My assumption (which could be wrong, of course!) is that since the noise generator produces such a small signal (in the millivolt range), I need a fairly high-impedance input so as not to overload it. But now I'm wondering if there's not enough of a load; should I use a load resistor? Part of my problem, it seems, is an impedance mismatch somewhere.

Also, it seems that the circuit here should give me plenty enough gain to at least hear something through a regular line-level audio input.

Any help will be very much appreciated! Ask me about anything I may have neglected to mention that may be relevant.


Hello there,

You didnt mention how high your V+ supply was. You may have to go as high as +20 volts to get it going.
You can also try heating the transistor a little to see if it will ever work at all.

Zener noise sources may be a little better. If you use zeners of say 12v you should get a nice flat band of white noise (lower voltage zeners tend more toward tunneling rather than avalanche) as long as the op amps dont filter the noise and create their own. I dont know how good your noise has to be (pure white or anything goes) as you can also use logic gates biased into linear operation (lots of gain) if you are not too picky about the generated spectrum.
 
Hello there,

You didnt mention how high your V+ supply was. You may have to go as high as +20 volts to get it going.
You can also try heating the transistor a little to see if it will ever work at all.

Zener noise sources may be a little better. If you use zeners of say 12v you should get a nice flat band of white noise (lower voltage zeners tend more toward tunneling rather than avalanche) as long as the op amps dont filter the noise and create their own. I dont know how good your noise has to be (pure white or anything goes) as you can also use logic gates biased into linear operation (lots of gain) if you are not too picky about the generated spectrum.
But, as Mike pointed out, the op amp will saturate at the positive rail unless a cap is added in series with the 10k resistor.
 
Hello Ron,

I didnt mean to imply that the same circuit could be used for a zener. You're right though that i should have mentioned that another circuit biasing the zener properly would have to be used.
 
Well, since you mentioned zeners, can you tell me what the difference is between zener noise and reverse-biased transistor (E-B) noise, in terms of the noise spectrum? I'm guessing if there is any difference that it's pretty subtle. By the way, I am now able to get plenty of usable noise signal out of the noise generator + amplifier stage.

By the way, I did mention what the supply voltage is (~15 V), which turns out to be plenty for both the noise generator and op amp.
 
Last edited:
Hello again,

I believe the zener produces more noise output so less amplification is needed, but you still need plenty of amplification anyway. The frequency band is supposed to be flat too.
Did you ever figure out what voltage the base emitter started to avalanche?
 
No, but I don't really need that information now, do I? As I said, the current setup produces plenty of noise, at ~15 volts. Since I now have an op amp stage with a gain of about 1000, there's plenty of signal.

I wish someone would address my other questions asked above.
 
Well, since you mentioned zeners, can you tell me what the difference is between zener noise and reverse-biased transistor (E-B) noise, in terms of the noise spectrum? I'm guessing if there is any difference that it's pretty subtle.
The noise from both those sources is likely fairly white with a similar spectrum since they are both generated by a similar breakdown process.
 
I'm guessing that your filter is just too aggressive (The Sallen-Key configuration needs the voltage-follower connected OpAmp), and that the reverse-breakdown transistor just doesn't have much spectral output below 30Hz.
 
The noise from both those sources is likely fairly white with a similar spectrum since they are both generated by a similar breakdown process.

OK, thanks for that helpful reply. Now, can you tell me if it's possible for me to extract any usable amount of subsonic (say, < 10 Hz) components from such a signal, using one or more low-pass filters? Or am I just going on a fool's errand here?
 
OK, thanks for that helpful reply. Now, can you tell me if it's possible for me to extract any usable amount of subsonic (say, < 10 Hz) components from such a signal, using one or more low-pass filters? Or am I just going on a fool's errand here?
I would think the noise likely continues to a very low frequency. Make sure the output capacitor resistive load does not roll-off the frequencies you want. (For example a 10kΩ load will give a low frequency roll-off of anything below 0.6Hz with the 25µF output cap.)

Edit: You may have to add another amp after the LP filter to get a reasonable output level.
 
Last edited:
Your cutoff frequency is about 3.2Hz. You probably slipped a decimal point in your calculations.
The rolloff is not 40dB/octave. It's 40dB/decade. I'm guessing that was a typo.
 
Your cutoff frequency is about 3.2Hz. You probably slipped a decimal point in your calculations.
The rolloff is not 40dB/octave. It's 40dB/decade. I'm guessing that was a typo.

Not a typo, a brain fart. Thanks. Yes, it's 40 dB/decade.

So I'm wondering if adding another low-pass filter stage like this one (probably after a gain stage) would make any difference. Or do I need a higher-order (that is, steeper-rolloff) filter? (Warning to self: getting into deep waters here, considering my lack of formal training in such matters.)
 
Not a typo, a brain fart. Thanks. Yes, it's 40 dB/decade.

So I'm wondering if adding another low-pass filter stage like this one (probably after a gain stage) would make any difference. Or do I need a higher-order (that is, steeper-rolloff) filter? (Warning to self: getting into deep waters here, considering my lack of formal training in such matters.)
Are you currently getting noise below 3.2Hz?
 
Are you currently getting noise below 3.2Hz?

All I can say is "A little?" You gotta realize I'm flying blind here. No test equipment to speak of other than a couple of multimeters.

I can feel the cone of the woofer twitching at what seems like <5-10 Hz, thus the basis for my guess. However, there's still quite a bit of audible hiss above that, which I'd like to eliminate.

Since I have very little feel for what these kinds of numbers mean, could you give me a guess as to what kind of dB reduction/rolloff might be required here? 40 dB/decade sounds like a lot, but maybe I need a lot more.

Sorry if this has deviated away from original topic (op amps) ...
 
What I did

If you look here https://www.electro-tech-online.com...ignal-to-test-a-fir-filter.116825/#post958241 I found the way to get enough noise.

I am about to start soldering that in minutes.

I selected the noisiest transistor from a batch of 10.

Keep in mind that your +V should be higher than the eventual zener voltage. In my case is little less than 8V.

If you use a single supply, make sure the + input is biased to +V/2. Look into the sticky threads (op amps) to make sure you know how.

Emitter resistor is 8K2.

edit

It seems that you know all this. Sorry.

/edit
 

Attachments

  • Noise source.jpg
    Noise source.jpg
    16.2 KB · Views: 238
Last edited:
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top