How to design the pulse generator correctly...

[Folnia]

Hi,
I'm working on a pulse generator with 2N3904.
I've gone through the AN-47 and I made one almost same as 
[url]http://www.kerrywong.com/2013/05/18/avalanche-pulse-generator-build-using-2n3904/[url]
with only power supply changed.
The BJT is 2N3904TFR manufactured by ONSEMI. I found 120V cannot work so I changed the voltage to around 134V then the pulse generated.

Attached are the scheme and captured output.
My questions:
1) Is there any way to know calculate/simulate the minimal voltage to generated the pulse based on the spec of the BJT?
2) How to calculate the theoretical pulse amplitude ?  Can I use the E =  1/2 CV^2 and give an estimation of the power consumed on the 50ohm resistor?
3) From the captured waveform I can see some oscillation. I used a 50ohm coaxial connected to the scope BNC directly and the BNC input resistant is 1Mohm, so there might be a negative reflection. But why there's also oscillation before the pulse?
4) If the pulse amplitude is too large, is there a clever way to do an attenuation to match the 50ohm 5Vmax scope BNC connector?

Thanks in advance!

[T3sl4co1l]

Simulate -- no, this mode of operation is rarely if ever modeled.  (Possibly the Diodes avalanche parts have models, I forget.)

Breakdown is always between Vces and Vcbo(br).  This can be quite a bit higher than the datasheet minimums (a mere 40V!) but you don't really know how much more until actual testing.

Adjusting the base resistor is a good way to find the sweet spot, both in terms of what Vce and Ic is needed for trigger, and over what range stable triggering will occur.  I found 4.7k, and 100-120V supply, effective for the Fairchild 2N3904s I have on hand, but YMMV.

Yes, a simple resistor attenuator affords any lower output amplitude.

Tim

[tggzzz]

Attached are the scheme and captured output.
My questions:
1) Is there any way to know calculate/simulate the minimal voltage to generated the pulse based on the spec of the BJT?
2) How to calculate the theoretical pulse amplitude ?  Can I use the E =  1/2 CV^2 and give an estimation of the power consumed on the 50ohm resistor?
3) From the captured waveform I can see some oscillation. I used a 50ohm coaxial connected to the scope BNC directly and the BNC input resistant is 1Mohm, so there might be a negative reflection. But why there's also oscillation before the pulse?
4) If the pulse amplitude is too large, is there a clever way to do an attenuation to match the 50ohm 5Vmax scope BNC connector?

1) not that I know of, because avalanche operation is outside the specification and outside the standard Ebers-Moll model. There are a few (?Zetex?) transistors which do have avalanche specifications.

2) No.

3) This might be a measurement artefact, either due to lead lengths (1mm of wire is 1nH of inductance), or the interpolation functions inside your scope (use an analogue scope or RTFM ).

4) The pulse amplitude is poorly defined. A very narrow 1kV pulse could be "integrated" by a scope's bandwidth so that it appears the same as a 10V wider pulse. In addition, while "matching" and "attenuation" can  go together, they are different.

If a 250ps-300ps 2.5V step into 50ohms would suit your purposes, a 74LVC1G* generator would give more predictable results. FFI see the "show us your square wave thread".