Hello all,

I have a photo transistor which produces a small signal (1 kHz) I'd like to further amplify.
Currently I use a transimpedance amplifier like image below, using a TL074.
From the first stage the amplified signal is further amplified by an inverting circuit (100x).

All is setup on a breadboard and picks up a lot of 50Hz noise which I want to remove.
To do this I have a piece of wire which also picks up the 50Hz noise, and use this signal to subtract from the amplified signal (currently using my scope for the subtraction).

This seems to work, but I'm sure there must be a more sophisticated way to get rid of the 50 Hz noise.
I want to avoid using a lot of filtering to eliminate the 50Hz signal. Any ideas?

Thanks,

 

Is the 50Hz being picked up by the phototransistor from the room lighting?
If so then fitering is your only option other than screening the phototransistor from the ambient lighting.

 

It is definitely em noise being picked up. Also when the room is dark

 

Electrostatic shielding is probably required. The overall circuit has a gain of 100 mV per nanoamp, so it really doesn't take much noise coupling for large output. The phototransistor and input amp should have a "tin can" shield connected to circuit common, leaving only enough of a hole for the optical path. If the transistor must be remote from the amp, it will likely be necessary to separately shield each and use coaxial cable between them. Since the shielding required is just electrostatic, any reasonably conductive metal will do. Thin sheet brass is usually a standard item at hobby shops that sell model railroad supplies and it is quite easy to solder (prep with fine abrasive paper).

In the event that there is electromagnetic coupling, then shielding is a much more difficult problem requiring, at a minimum, fairly thick iron or steel for 50 Hz. Exotic materials like mu metal are often used, but it is expensive and difficult to work with (it must be annealed after any sort of forming in order to maintain its magnetic shielding properties).

 

Converting the 50 Hz into CM and rejecting with differential amplifier
as you have done excellent way as long as the signal paths same for
pickup on both legs of the diff amp. Filtering it single ended with a
notch or HP filter handles that problem.

This might help -

http://web.mit.edu/jhawk/tmp/p/EST016_Ground_Loops_handout.pdf

Your circuit may suffer from C coupled noise, and its unbalanced because
your bias current will create an offset due to one side looking like 1M ohm
whereas the other side is looking at 1K ohm. You might try changing the
bias network to 2 M ohm Rs to see if it is C coupled 50 Hz. Normally
I would not advise that as purposely forcing nodes to HiZ is a recipe
for noise pickup/coupling.

Regards, Dana.

 

"... and its unbalanced because your bias current .."

It is a FET input amplifier. Attempting bias current balancing is largely futile and usually just adds more Johnson noise.

The resistors in the biasing network can certainly be increased save supply current. It is important to include a bypass capacitor across the bottom resistor. Raising the resistor values without a capacitor would increase the noise (Johnson noise, again), but it also makes it easier to filter the noise with a fairly low-value capacitor. With 1M resistors, a bypass cap of 100 nF would be quite effective and mean only one resistor value for the circuit. The cap not only acts against the Johnson noise but also any noise or ripple on the amp's power supply.

 

Totally agree ebp, I should not have even mentioned bias current.

Mainly was thinking what happens if we make both side of the diff amp
subject to the same level of CM from stray C coupling and pickup by
elevating both inputs to same impedance (assuming layout symmetric
and same, relatively speaking).

The TL082 only has 8 nA hot bias, so 8 mV generated offset nothing
to complain about. In this application.

Regards, Dana.

 

Balancing impedance might help, but I strongly suspect the first requirement is proper construction. Solderless breadboards (don't know if that's what's in use here) are a long long way from ideal for high-gain circuitry. Many of the current crop aren't mounted on metal plates that would provide at least some shielding if connected to circuit ground. Layouts need to be very compact and lead lengths minimized. Shielding is pretty much essential. Even some of the integrated IR remote control receivers in the little plastic side-view packages have shielding cans around them, and there isn't a whole lot of "antenna" there in the first place.

One thing to be careful of with any multi-amp package, but especially with those with ultra-high input impedance is to be sure any unused amps are properly dealt with so they don't behave like unruly high-gain comparators.

BkraM, can you post a photo of your breadboard?

 

Isn't the photo-transistor C-E junction reverse biased here?

Not a circuit configuration I have seen before.

 

Hello,

I see you got the circuit from stackexchange.
On this page there are some suggestions:
https://electronics.stackexchange.c...totransistor-with-virtual-ground-opamp-design

Bertus

 

there's a lot of detail we don;t see. Put sensor and amp very close to one another, Twist the wires to the sensor.
Twisting reduces EMI. Some OP-amps need a "harder" artificial ground. Use a rail splitter IC.

Eliminate power line and harmonics of light from the general area. RFI is reduced by shielding. So a twisted pair shielded cable does wonders.

 

Thanks for the replies all,

Attached are two images, one of the breadboard and one image capture of my scope.



I'm well aware that I'm building an antenna with using a breadboard, it's a thing that has been troubling me for some time with some of my projects. I have the same trouble (but less) with soldered breadboards with shorter leads. I'm working towards a (DIY) project where I need to amplify very small signals where these noise level will be a problem. So I decided to make a simple breadboard circuit where a small signal is being amplified to see what solutions there are to deal with it.

My initial assumption was that it will be impossible to prevent any 50Hz signal (or oter EM noise) from entering the circuit, although you can do a lot to minimize it (which in this case i didn't do). My thought here is that creating by a second 'matched' antenna and using this signal as a CM in a difference amplifier would, regardless of the EM noise entering the system, eliminate the 50Hz hum.

The scope capture shows the result of the amplified signal (yellow) and the antenna (pink) and the difference signal (white). Although not perfect ,the setup does reduce a lot of the noise in the signal. It seems to be (in principle) a workable solution.

I was wondering if this technique is commonly used and how to approach it in a more controlled manner. Also if there are other techniques to get rid f the noise. From the responses I get that shielding is more used than I would have thought, could the noise levels be reduced enough? for my intended application I would be looking at amplifying a uV signal up to 5V.

Thanks for the help so far.

Kind regards,

 

Hello,

Did you see this image posted by EM FIELDS on the given page?


As you can see the E and C of the photo transistor are reversed compared with your schematic.

Bertus

 

If you are using a pseudo ground, then the NPN phototransistor emitter lead should be connected to ground also.
The output signal comes from the collector.

If you are still getting 50Hz line frequency noise, try turning off all room lights.

 

Hi,

I've tried flipping the Photo-transistor, but it doesn't seem to work in that orientation. Not sure why that is.
(unsure how it is currently wired as there is no indication on the device what the collector or emitter is, but only produces a signal in one orientation)

I've rebuilt the circuit on a soldered breadboard (so less of an antenna), and wrapped it in (connected to ground) aluminum foil as a sort of basic shielding, with only a small gap for the optics.
Does seem to reduce the noise more than I initially expected. Attached is the signal with for te "wrapped" circuit.


Conclusion as mentioned in the responses; small circuit and shielding does most of the trick...

Thanks for the help

 

From the photo, I think the device is likely a photodiode, not a phototransistor.

 

From the photo, I think the device is likely a photodiode, not a phototransistor.

If the photodiode is black then it is likely to be an infra-red photodiode, i.e. one that is sensitive only to wavelengths in the infra-red region.

 

Is the 1 kHz signal source near the photo transistor amplifier? If you you can make a lock-in amplifier. Easy to do with op amps and transmission gates.

 

If the part really is a QSD122, then it really is a phototransistor. The 122 is in black 2-lead T-1 3/4 package.

 

Looks more like the "noise" was an uncoupled ground clip on the scope probe. Check to see if the wire is broken in the insulation.