1
\$\begingroup\$

I have built a scintillation detector. The detector outputs 0,1 ms 50-200 mV negative pulses from a photomultiplier (DC). The pulse shape is shown in the picture below.

before and after shaping

To detect the pulses I need to change their properties (still DC however). Their new shape has to be pretty much identical, but more than 500 mV positive voltage (their duration can be 0,1 ms to 0,5 ms, but preferably 0,3 ms to 0,4 ms). The shape is, again, illustrated in the picture above.

Does anyone with more experience than me have a small form-factor schematic that will work? If I need to, I have a 5 V rail that I can power something with.

Thanks!


Update: The op amp that I chose (TLE2081) has sufficient properties. However, I can't get my schematic to work. See the picture for my schematic.

The output I am getting is my Vcc+, which is odd. I am not getting the pulses that I was looking for. I think I did something wrong, does anybody have any ideas?

schematic diagram

Thanks!

\$\endgroup\$
6
  • \$\begingroup\$ Welcome to EE.SE. Most any quality op-amp in inverting format with a gain of -2 will work. It will be a mirror but inverted image of the input pulse. A high-pass filter can shorten the pulse width but needs to be hand adjustable. \$\endgroup\$
    – user105652
    Mar 30 '18 at 20:03
  • \$\begingroup\$ What is the rise time? 10~90% and why do you need a decay time of 0.3~0.4ms? \$\endgroup\$ Mar 30 '18 at 20:04
  • \$\begingroup\$ Also what is source impedance roughly \$\endgroup\$ Mar 30 '18 at 20:12
  • 1
    \$\begingroup\$ Why can't you process the signal in the form it currently is in? The context provided by an answer to this may help some. \$\endgroup\$
    – jonk
    Mar 30 '18 at 20:13
  • \$\begingroup\$ Is the timing of the pulses important? You might want to look into a constant fraction discriminator circuit. \$\endgroup\$
    – D Duck
    Mar 30 '18 at 20:36
2
\$\begingroup\$

You say that you have tried this circuit:

but that the output is stuck at Vcc. You didn't say what opamp you are using, but most opamps should be able to handle this. They certainly should not produce Vcc with nothing connected.

One possibility is that you have the inputs flipped. Start by debugging the case where the input is tied to ground. The output should be very close to ground. Until you get that, there is no point trying to feed pulses to it.

You also need to consider the impedance of the source. Your inverting amplifies has a 10 kΩ input impedance. If the source impedance is not significantly lower than that, then the signal will get attenuated. It doesn't seem like you need unusually high speed, so you should be able to use about 10x higher resistors. In other words, 100 kΩ and 800 kΩ if you still want a gain of -8.

Added

Many opamps can do what you asked, but not the one you picked. Now that you have supplied a link to the opamp datasheet, the problem is clear. The common mode input voltage range needs over 4 V headroom from the negative supply. You are trying to operate the amp with the inputs at the negative supply. That's clearly well out of spec.

To fix this, use a more appropriate opamp, or give it a negative supply of -5 V or lower. This is, of course, after properly reading the datasheet.

\$\endgroup\$
0
0
\$\begingroup\$

schematic

simulate this circuit – Schematic created using CircuitLab

Gain of -2 with Rail to rail output. Is roughly what you asked for.

But it depends on how fussy you want to discriminate.

Include Specs: Vp threshold, PW50 min ( 50% Vth pulse width) Integral of pulse (I) , Integral squared ?. Threshold fixed or variable depending on amplitude? Threshold controlled by derivative slope ? or controlled by % of light energy (V^2)

\$\endgroup\$
0
\$\begingroup\$

Something like this

enter image description here

In this circuit (from a canned demonstration of how a low-pass-filter fails to reject the high frequencies), I've made R1 be 750K, Rn be 1 ohm, Rf be 1.5Megohm. And C1 is disabled, otherwise the circuit likely will oscillate because the massive phaseshift at the virtual-gnd node.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.