Unexpected sign reversal

Discussion in 'The Projects Forum' started by jety89, Oct 7, 2014.

  1. jety89

    Thread Starter New Member

    Oct 7, 2014
    5
    0
    Hi guys, I have three soviet nuclear signal preamplifiers to fix. They are used to 4x -- 40x amplify short, 2 microsecond impulses. I came to call them PR1, PR2 and PR3. I have been testing them with a digital oscilloscope and a square signal generator for over a month now.
    https://drive.google.com/file/d/0B0sjs6PVblg1V3RTZ0JBTEQ4N2s/view?usp=sharing
    Only PR2 is working close to properly. It amplifies a +50mV 1 microsecond square signal to something like -1 volts. Have I mentioned that I've been trying to fix them for over a month now?

    The FETs have been replaced a few times before, but I can confirm that all of them are working fine. PR3 has two soviet FETs in parallel, while PR1 and PR2 have BF245C-s installed. Everything else is the original. R2 and C2 of the smaller plate are not required and are not present in any of them.

    The bias supply on input x2 gives a steady -1800V, and the germanium crystal on x1 behaves like a reverse biased diode with the other end grounded on the aluminum casing. Every time a gamma-particle hits the detector a small amount of charge is collected, which becomes an exponentially decaying 2-50 microsecond impulse, depending on the initial charge. What gets then measured is the amlitude of the output signal.

    The input to the preamplifier is a short positive burst, which is turned into a short negative burst on the field-effect transistor VT1*, which then gets further processed and amplified. The way these preamplifiers work, is that the incoming signal gets integrated on the feedback capacitor C5. I don't get it completely, but that's what the literature says.

    There are two outputs called x6 and x4. The part of the signal path I am interested in is VT1* VT2 VT3 VT6 VT7.

    This is my main question:
    Yesterday I've noticed, using a digital oscilloscope, that PR3 does something weird: it reverses the sign of the signal twice! I am not sure which one of the abovementioned transistors does it, or WHY... how can I fix this?

    Over the course of the month I became fairly familiar with the description of these things, probed every leg of every transistor with a voltmeter and an oscilloscope, and turned all of the five available knobs. The most important is R1. It controls the phase of the feedback signal. If R1 is not set properly, the circuit can amlify itself instead of the signal. It also is the main control for the amount of amplification.
    https://drive.google.com/file/d/0B0sjs6PVblg1Rjd5NnQzYVo0cXM/view?usp=sharing
    https://drive.google.com/file/d/0B0sjs6PVblg1ZXdjXy15UE9aamM/view?usp=sharing

    I've tried everything, and read every available document, including the IEEE standard on preamplifier testing, amongst others. I've come a long way, an yet I am not mouch closer to fixing these things. They take forever to warm up, they can't hold their amplification steady for 24 hours, and/or they behave in funny ways.

    This is my second question:
    In PR1 the main problem is: VT7 corrupts the signal: on the input leg the signal amplitude changes properly as I play with the square signal generator, but on the output leg only the length of the signal is modified. The voltages on the legs are: Collector +18V; Base +2.2V; Emitter +1.7V. What could be the problem.

    PR2 works fine after a 2-3 hour warmup, but I never quite know whet it will change its amplification slightly, only to mess up my measurements completely. What can be done about that?
     
  2. jety89

    Thread Starter New Member

    Oct 7, 2014
    5
    0
    Oops, I was meant to say *signal* reversal
     
  3. MrChips

    Moderator

    Oct 2, 2009
    12,449
    3,363
    This is all Greek to me.:D

    Nuclear radiation preamps are designed to work with detectors that produce a unipolar pulse, positive or negative going. Typical input signals have a very short leading edge (less than 10ns) and a long decay (greater than 1us). The inputs and outputs of the preamps are typically AC coupled.

    If you use a square wave signal generator to test the preamp the result will be the derivative of the square wave, that is, a pulse at the leading edge and a signal reversal at the trailing edge.

    You really need a proper pulse signal generator that mimics the signal that you would obtain from the radiation detector.
     
  4. wayneh

    Expert

    Sep 9, 2010
    12,145
    3,056
    I hope these are not LAUNCH signals? :eek:
     
    jety89 likes this.
  5. jety89

    Thread Starter New Member

    Oct 7, 2014
    5
    0
    Thanks for the answers, yeah, I think I've really oversalted this. MrChips, I guess, You're right. How can I Make one?
     
  6. jety89

    Thread Starter New Member

    Oct 7, 2014
    5
    0
    Last edited: Oct 8, 2014
  7. jety89

    Thread Starter New Member

    Oct 7, 2014
    5
    0
    It isn't such a bad choice though. According to the IEEE standards document,
    "
    3.2.1 Pulse Generator, Preamplifier Tests
    It is recommended that the pulse generator produce a flat-topped pulse, although a tail pulse may be used for all tests
    except examination of the preamplifier's exponential recovery to the baseline. Here, a rectangular pulse shall be used
    for the test because the decay of a tail-pulse signal will be confused with that of the preamplifier (see A3.2.1 and A4.2).
    "

    Still, I agree with You. In my situation it will be helpful to use a proper shape pulse generator.
     
Loading...