Unknown noise/flickering in circuit [images from scope attached]

Hi there!

The posted circuit is a low noise charge amplifer and a shaping filter for measuring very tiny pulses appearing on the input of the charge amplifier. The first oscilloscope image is from our PSU bias voltage, and as you can see it is very clean (had some problems but this post fixed it. The second image is on the output of the charge and shaping amplifier, and we get this wierd periodic flickering. Does anyone have any idea where this might be coming from?

Any help would be very appreciated! Thank you!

As your amplifier is very high gain, and it includes the low frequency response, it may be a pickup from the AC power line. There are switching transients from fluorescent tubes, dimmers, etc.

I would first try to connect the input of your amplifier to ground. This prevents any stray signal from getting there. Then check grounding blocks of your DC power supplies; add 10 ...100 uF capacitors to + Vcc and - Vcc lines against ground, and see the result.

If this does not help, check shielding of your amplifier. Sensitive amplifiers like yours should be enclosed in a metal box, with connectors for input and output, and capacitive feedthroughs for DC +/- lines. Also try to replace the feedback resistor in your preamplifier with a trimmer; reducing gain will show if the pulses are generated there or elsewhere.

Your amplifier is quite similar to those used with PIR detectors in alarms. Maybe you can check one of these; they are usually quite stable but their frequency response is limited to ~ 100 Hz.

And preventing light to reach semiconductor devices is also a good idea. Opamps in metal cases with glass insulators allow flicker to enter. One can wonder what a sensitive amplifier can "feel".

---------- Post added at 18:22 ---------- Previous post was at 18:18 ----------

On your oscillogram I cannot see the time scale. If the period is not corresponding to 50 or 100 Hz, it may be a relaxation oscillation, probably due to the first stage with the FET and opamp. Why there is no "series" resistor before the gate? That resistor and 10 MegaOhms in feedback loop determines circuit gain.

The time scale is at the top. 2ms/line segment.

I belive we do not have a series resistor befor the gain as we are using the pre-amplfier as a charge sensing devise. The pre-amplifer receives a charge from a diod detector when an alpha particle hits the detector. This charge is then placed over Cf and the shaping filter is used to amplify the signal for further analysis.

We are using a referance design for our cicuit.

I've been looking at your post and the oscillograms and I'm thinking along pretty much the same lines as jiripolivka, so I won't repeat everything he said.

From the second oscillogram, the period of those mystery pulses is 10msec, strongly implying that they are somehow related to 50Hz mains frequency. In the other thread, we concentrated on the possibility of the rectified 100Hz ripple from the diode bias supply somehow leaking into the signal path. Now I'm thinking that it's also possible that a 100Hz ripple from the power supply of another mains-powered product might be responsible. This includes the fluorescent tubes, etc brought up by jiripolivka. It shouldn't be difficult to check for photo-interference by temporarily enclosing everything, including the sensor diode, in a light-tight cover and see if the pulses disappear or not.

We have tried tunring the lights off, but the problems still persists.

Adjusting and optimizing such sensitive device like your charge amplifier always requires a patience.
Maybe you can try to replace one or both DC power supplies with batteries; two 9V in series, with a 15 V regulator, could work. And this way you will see which way the stray pulses arrive.

If you use 7815, 7915 or similar regulators, they sometimes become unstable if their INPUT capacitors have a bit higher internal resistance. Paralleling 1 uF and 0.1 uF often helps. If you can use LM317T adjustable regulators, I never had a stability problem with them.

Good luck!

sinnadyr said:

We have tried tunring the lights off, but the problems still persists.

Click to expand...

OK, that apparently eliminates one possible source. Have you checked the power supplies for any 100Hz ripple? Do you provide additional filter caps near the amplifier +/- power supply pins and near Ra?

Yes we have checked for riple and provided extra filter caps. The weird thing is that when connecting the 200V in series with a signal generator we see nothing of this noise, but when connecting the detector it suddenly shows up. Also, on test setup with signal generator nothing happens when we knock on the table, but with the sensor every little movement shows on the oscilloscope.

Even our teacher can not figure out this, but they seem to have less experience than you guys, so thank you!

I've read old Bob Pease articles where he talked about the
piezo effects of cabling on low current / high impedance
measurements. There could be microphonics in the sensor
or the blocking capacitors that respond to any acoustic
input as well.

Running the whole thing off two 9V batteries instead of
the power supply, inside a Faraday box, would be the way
to test whether this is ambient or internally generated
noise.

All right, that is definitely something to consider, but how do I step up 18V or +/- 9V to 200V?

From what you have gradually excluded, it seems to me that somewhere around you there may be a power-line powered source of alpha-rays that your detector responds to.
Please enclose the detector element in a good metal case, and try again. You can also use a heavy flat piece of metal (steel, aluminum, lead) to find out the direction of the radiation being detected.

Generating -200 V from 9 V is not difficult as your detector needs only a small current, a couple of mA. Use a 555-keyed transistor, and find a 1:50 to 1:100 transformer. Use a rectifier diode (or the part of your 200 V supply), and you will get the required -200 V. Filtering the noise from ~ 500 Hz or 1 kHz will be another problem; I think a good screening will help.

Good luck!

What is e 555-keyed transistor?

Thanks again!

Oh. my! I expected you are experienced electronics experts!

A 555 is a timer, integrated circuit most used as a pulse generator. You can use any multivibrator, I often use CD4011 . The pulse output at TTL level is fed to transistor base via ~3-5 kOhm resistor, and the transformer primary is connected in collector circuit. Across the secondary you get 50x the primary voltage pulses, ready to be rectified and regulated like you did with AC, 50 Hz transformer.

Haha, an expert at the age of 24 would be quite cool, but unfortunately I'm not. I know a little bit about 555 timers, but the expression "555- keyed" was new. I usually generate a pwm from any microcontroller when I need a square wave pulse, or just a signal generator.

So you kind of simulate AC with chopped up DC then?

Will try it tomorrow, its time for å Friday snack here in Norway

Thank you!

Yes, the DC/DC converters all operate like that: like with 50 Hz AC, you generate low-voltage pulses, get several times more voltage in a transformer, and rectify it.
There is one more option for a low-current higher voltage: diode voltage-multiplier chain. It looks like a zigzag -connected diodes in series, with capacitors between neighboring joints. Each AC voltage cycle charges the first, second, etc. capacitor. After X cycles, all capacitors are charged, and as they are connected in series, the full voltage is N times higher than the AC peak input. You need N diodes and 2 N capacitors, no transformer.

Thank you!

This has really helped in the troubleshooting, but unfortunately we do not have time to develope the circuit further, so we just have to use this information in our report. But since my quriosity around this has no end, I just stumbled over this article:
Peak Repetitive Reverse Voltage (VRRM)
where V(RRM) of the rectifier diode is said to be at least (in our case) 200V. We have a little over 200V in our ciruit, about 210V or so. Is it possible that this is the reason for the yellow lines 100Hz negative drops on the oscilloscope on the first post of this thread?

I seem to have missed the point, what is the yellow trace in the second CRO screen shot. Its not ripple as I know it unless there is something taking pulses of current that is synchronised with the mains. - Which seem to be inducing the noise spikes on the signal (green trace).
Frank

The yellow one is output from the charge sensitive amplifier, look at the schematic to see better where the probe is. The green one is output from the shaping filter,also described on the circuit with their respective names