Coupled Pulse & square wave generator for electron gun

Discussion in 'The Projects Forum' started by beamcircuit, Apr 24, 2008.

  1. beamcircuit

    Thread Starter New Member

    Apr 23, 2008
    9
    0
    I have been working on this project for a while on a 'when-I-have-time basis' and now it is crunch time. Here's a general description:

    My company obtained a DC electron gun as the first stage of a much larger accelerator. It emits a pulse of electrons with 50kV energy from a heated cathode. There are two transformers in the gun body itself. The gun body is a 'black box' so i don't know the turn ratios. The gun consists of two main circuits: the heater and the pulser.

    Heater:
    The user, being me, is to apply a 1kHz [50% duty cycle] square wave of -12V to 12V to the transformer in the black box. See PDF, square wave attached to connection#4.

    Pulser:
    A circuit design was given for this part, but that doesn't mean I've gotten it to work. See second pdf. Apply +800V, and when the thyristor is triggered, a half-sinusoid of -400V is produced [duration of 4.8microseconds] and is applied to connection#2 of the first pdf. The pulse is stepped up to 50kV by the second black-box transformer and biases the cathode.

    Since the anode is held at ground, the 50 kV pulsed bias on the cathode accelerated the electrons out through a hole in the cathode. The heater creates thermionic electrons.


    So here is my first question...[I almost forgot]...Can I just use a simple transistor multivibrator? Like this one, but with +/- 12V rails.[​IMG]

    I should point out that I am an accelerator physicist, not a electronics guru. This seems like it should work. Any suggestions?
     
  2. thingmaker3

    Retired Moderator

    May 16, 2005
    5,072
    6
    You'll want to select components which can handle your plus or minus 10% frequency specification. Remember that temperature has an effect. Frequency will skew if the transistors get too warm. Will the oscillator be in a controlled environment? Or in one with changing temperature?

    Note also your + and - 12V are maximum values. A pair of 12V Zener diodes across the output & sources might be in order.
     
  3. Wendy

    Moderator

    Mar 24, 2008
    20,766
    2,536
    You could build a simple square wave oscillator and amplify it afterwards, it would be a lot more stable that way.
     
  4. thingmaker3

    Retired Moderator

    May 16, 2005
    5,072
    6
    Bill is correct. A nice, stable, low power oscillator driving Darlingtons or MOSFETS in a push-pull arrangement would work nicely.
     
  5. Wendy

    Moderator

    Mar 24, 2008
    20,766
    2,536
    If you want some help I'm sure this crew could have you a crystal control locked ocsillator with amps, in a relatively simple format. :D We like electron guns and lightning bolts.
     
  6. beamcircuit

    Thread Starter New Member

    Apr 23, 2008
    9
    0
    The 1 kHz square wave is a nominal number and it does not need to be too stable shot-to-shot, but I get what is being said about frequency drift. When in doubt, the frequency should be higher than 1 kHz [lower frequency means that we are driving current in one direction longer and may risk burning out the cathode material when pulsing at full energy].

    That said, I agree that the transistor circuit I built was not ideal. My original plan was to build a 555 oscillator and amplify, but I was unsure of how to make the amp. The heater actually needs to scale in amplitude from 0 to +/-24V [+/- 12 is for steady state, 24V is for activation, and we need everywhere in between because we can't slam the cathode with full heating from the get go]
    .

    All that being said [more of that background], the problem seemed to have been that the pulser was interfering with the square wave oscillator. Will a darlington pair be more resistant to coupling with the pulser? Also, isn't the frequency of a crystal oscillator too high? I know I can do frequency dividing, but isn't that overkill?

    Also, how do I do this? I know what a darlington pair is, but how to make circuits usually takes some time for me to figure out.

    One last thing: a 5V pulse [or, say a TTL pulse] to the thyristor should switch it to 'on,' correct?
     
  7. beamcircuit

    Thread Starter New Member

    Apr 23, 2008
    9
    0
    By the way, the circuits are housed in a rack-mounted box near the gun, in the bunker, so temperature is stable but not controlled.
     
  8. thingmaker3

    Retired Moderator

    May 16, 2005
    5,072
    6
    A 30 kHz and a 32 kHz signal from available crystals could be heterodyned yielding a 2 kHz signal. This 2kHz signal could then be run through a J-K flip-flop to generate two out-of phase 1 kHz signals for control of the Darlingtons.

    Optionally, a 32 kHz signal could be fed through a divide-by-16 counter prior to clocking the J-K. One less crystal and no passive components.

    Or maybe one of these could be used: http://www.hightechips.com/Download/htc555.pdf
     
  9. beamcircuit

    Thread Starter New Member

    Apr 23, 2008
    9
    0
    I'm going to go with a 555, set to 2kHz @ 55% duty cycle for now. This will clock a JK flip flop [actually, a Toggle flip flop]. Q channel goes to the base of a NPN switching [3904] a second power transistor [TIP31]. Q' goes to the base of a PNP switching and then a power transistor. Rails are at +/- 12 [through 24V]. The emiiters of both power transistors are tied to the outputs. I'll also put in [100 ohm?] resistors in between the rails and the bases of the switching bjts.

    Any comments? I still need to add current limiting resistors to the Q/Q' channels, as well as the clock.
     
  10. Wendy

    Moderator

    Mar 24, 2008
    20,766
    2,536
    A little late, but here is a circuit I came up with and then saw used later elsewhere long before I thought of it.

    [​IMG]
     
  11. beamcircuit

    Thread Starter New Member

    Apr 23, 2008
    9
    0
    Well thank you Bill.

    I will need to look at this in more detail. I can't see right away how this will make a bipolar square wave, but I'll try to figure it out tomorrow with a beer on the beach.

    -Marcos
     
  12. Wendy

    Moderator

    Mar 24, 2008
    20,766
    2,536
    OK, I redrew it to make it more obvious. The capacitor charges to 1/3 and 2/3 of the Vcc level, and discharges through R to the output of the 555, which is going rail to rail on the output. The 555 configuration is that of a Schmitt trigger. I threw in a blinker showing another use for this circuit. I've been known to put a pot for R to make a fully adjustable output frequency wise. Pin 4 is a digital on/off input for the 555, the first pulse is a bit long since the capacitor starts fully discharged, but settles down after that. The transistors are rated for 8 Amps, if you need more other darlingtons will work.

    [​IMG]
     
  13. Wendy

    Moderator

    Mar 24, 2008
    20,766
    2,536
    One other thing, due to the base emitter drops across each of the transistors the output voltage will be around 9.2 volts P-P, there are ways to boost this, but I figured you'd need to know. There are probably better ways to work around this using a MOSFET, but I'm not too comfortable around them yet.
     
  14. beamcircuit

    Thread Starter New Member

    Apr 23, 2008
    9
    0
    OK I think I see it somewhat. I'll be using a variable supply, so the base-emitter drop is no problem. The 8 amps is not enough, so I'll put more darlingtons in.
     
  15. Wendy

    Moderator

    Mar 24, 2008
    20,766
    2,536
    You'll probably need to get some transistors rated for the higher current, or parallel them. How much do you figure is enough amps?
     
  16. beamcircuit

    Thread Starter New Member

    Apr 23, 2008
    9
    0
    I was planning on running in parallel for lower cost and ease of expanding.

    I need about 10 amps for now, possibly more later. I'll make this Monday or Tuesday. I'll run the gun on Friday and post pictures.
     
  17. thingmaker3

    Retired Moderator

    May 16, 2005
    5,072
    6
    When paralleling transistors, choose MOSFETs instead of BJTs.

    If one BJT in a parallel group heats up, it conducts more than its neighbors. This makes it heat up more. We call this "thermal runaway." Once the first one self-destructs, the others follow like lemmings.

    MOSFETs have a negative temperature coefficient, and so will conduct less when they get warmer.

    Note that many BJT transistors are rated for 10 Amps or more. One could, for example, choose these: http://www.st.com/stonline/products/literature/ds/4148/bdw93c.pdf
    or these: http://www.onsemi.com/pub_link/Collateral/BDW42-D.PDF
     
  18. Wendy

    Moderator

    Mar 24, 2008
    20,766
    2,536
    How would you connect MOSFETs for this design, complimentary mode? Like I said, I'm not comfortable with designing with them yet, but they seem like they would be ideal for this use.
     
Loading...