# Cleaning up a noisy signal

Discussion in 'General Electronics Chat' started by Hybird, Jan 5, 2011.

1. ### Hybird Thread Starter New Member

Jan 5, 2011
12
0
Hello everyone, I have a gieger tube circuit which has the following design:

Code ( (Unknown Language)):
1.
2. [FONT=System]HighVoltage ----- R1 -------o------ R2 ------- GMTube --- GND
3.                                        |
4.                                        |
5.                                       C1
6.                                        |
7.                                        |
8.                                     Signal -- Rpd --- GND[/FONT]
Where
R1 = 220KΩ
R2 = 2.2MΩ
C1 = 2200pf Rated for 1000V
HV = 400V
Rpd = ? (internal pull down resistor of my oscilloscope)

I have to read the signal as a pulsed square wave into a PIC micro (so there is a resistor tying the signal to ground (pull down) to read the voltage across). I have two cases I have to sort out:

1.) Protect the micro from really high voltages
2.) Make signal square like and not dirty

My assumptions, GM tube can be modelled as a switch which closes only when radiation hits it, other than that it is just a open circuit. Signal is measured across Rpd (pull down resistor of Micro)

1. When I first turn on the HV source, C1 charges up through R1 and the Pull down resistor of the micro. This is a quick boost in current to get the cap to 400v, the same as the HV source. So I see it as a positive spike on my osciloscope. I need some way to 'not' see this as the micro is not rated for that high of a signal.

2. When the GM tube does fire, it creates a voltage divider between R1 and R2 which drops the voltage over R2 to about 390v. Now the Cap has to drain off 10v to match this change and does so by running current through Rpd in the opposite direction as when the cap was initially charged in 1. So you see a negative voltage drop across Rpd. Next the GM tube stops conducting and we get the voltage back to 400 so the cap must recharge, and does so by a positive voltage.

I need a way to clean this signal to a positive square wave triggered only when the voltage drops anywhere lower than -1v say. I'm going to attemp to draw my signal here ha.
Code ( (Unknown Language)):
1.
2. [FONT=System]   |
3.    |
4.    |           ____
5.    |          /        \_
6. 0v-------------------------------
7.    |     | |
8.    |     | |
9.    |     |/
10.    |     |
11.    |
12.    |[/FONT]
13.
I hope thats readible ha, but anyway, does anyone have any suggestions for signal processing, links? Books? I just need knowlegde I was not really trained in circuit design so much as physical processes (Physicist). I'm leaning toward using diodes some how and a separate 5v power supply, but am a little lost. (2-5v is about the size I want to input into the micro)

Anything you could post would probably help me.

2. ### Hybird Thread Starter New Member

Jan 5, 2011
12
0
I forgot to mention, once I can limit the voltage swings I would then use a comparator or something to get out a nice digital signal. But I first need to figure out how to clean it up a bit.

3. ### Kermit2 AAC Fanatic!

Feb 5, 2010
3,851
967
try adjusting the resistance value to get the current needed by a zener. Then put the zener between the bottom of the resistor and ground. Take your 'output' from the top of the zener, so you always have a (Vz) volt high on the signal.

Another diode in the zener output line would stop any negative voltage spike from getting past it.

After that, schottky buffers could square up the signal.

4. ### Hybird Thread Starter New Member

Jan 5, 2011
12
0
But my signal to count is the negative pulse? I'm not sure I quite get your method.

I am reading the signal across Rpd. Are you saying I place a Zenar in series with Rpd and GND? And then take my signal from the top of the zenar? Assuming the zenar is set to hold the bottom of Rpd at 5v this would be ok for limiting the voltage to 5v while turning on the HV, but when a signal goes through and the current changes direction I still have no control of a negative signal.

I was going to try first placing a regular in4007 diode paralell with Rpd and point it down so on charging there is no current flowing through the Rpd resistor and hence 0 volts is scene. Then when the signal comes in and the current changes directions it can't go through the Diode so it goes through Rpd and we observe a negative unregulated voltage spike as before. But at least that kills off issue 1 of my problem.. if my logic is correct that is.

5. ### Kermit2 AAC Fanatic!

Feb 5, 2010
3,851
967
I misunderstood your explaination. The picture shows a negative spike then a positive square pulse.

I assumed the negative spike was a problem and the small positive pulse is what you wanted.

Same logic applies with the diodes in reverse. That will let a negative voltage through and block a positive one. The zener is to keep the voltage clamped to a low voltage level.

Since I misunderstood the post, perhaps you should try describing what you need again, in a more verbose and technical manner.

6. ### Hybird Thread Starter New Member

Jan 5, 2011
12
0
Sorry, its hard to describe without actual pictures. Ok, so I put a regular diode in the circuit in paralell with Rpd that points to ground, this got rid of all positive signal. Yay! So when current flows from the cap through Rpd to GND it takes the path of the diode and so the voltage across Rpd is 0. When the GM tube conducts, the cap must drop to 390 instead of 400, so current flows from GND through Rpd (because it can't go through the diode) and up to dis-charge the Cap. This results in a negative voltage spike as the cap charges and shows up on my oscilloscope as a negative spike.

So the issue now is how to take that negative spike and turn it into a positive digital signal when it is at a certain height.

Also, can I use a zenar some how to clamp this negative voltage to a maximum level? Then I could have a safe signal to feed into a schottky buffer or what ever (I still have to research this, looked into schmidt triggers as well but that was awhile ago).

Last edited: Jan 5, 2011
7. ### Kermit2 AAC Fanatic!

Feb 5, 2010
3,851
967
If you are feeding a logic gate then use a high value resistor divider network to lower the peak voltage of the spike.

The digital gate won't need any current of consequence to the circuit. Something like a 100k and 20k voltage divider, depending on the magnitude of the pulse voltage, should get it down into the proper range for a schottky buffer to invert and square up

Jul 26, 2010
2,358
201
9. ### timrobbins Active Member

Aug 29, 2009
318
16
What about if you put a 4V7 zener across your micro input (to limit positive input to approx 4.7V, and negative input to approx -0.7V, which should be ok for micro if your C1 is not too large). And also put a high resistance R3 in parallel with your C1, to bias the zener to approx 4V when the tube is idle (you want very little dc current, so zener won't reach 4V7).

When tube discharges, C1 will push uP signal to '0V' for a period dependant on tube and R3.C1. The time spent at 0V should be short to allow next pulse to be 'read' by uP, but not too short that uP doesn't read the input.

Also C1 seems a little big for the job. Maybe you could change the ratio of R1 and R2. Maybe if you sim the circuit and do some parameter iterations to get a feel for the likely changes - that may help in choosing component values for a suitable waveform.

Ciao, Tim