EEVblog #306 - Jim Williams Pulse Generator

[rolfdegen]

Next step is an HF case from https://www.meilhaus.de/tbrfh.htm
I want to install the circuit in the HF case. I hope it fits... 

[rolfdegen]

I ordered ten 2n2369 transistors here in Germany. Two transistors didn't work in Avalanche mode. Now I've ordered ten BFR505s from Poland.
Today I'll expand the 2n2369 circuit with a pulse generator based on a small ATtiny404 processor and test it. The rise time of ATtiny404 is 1.5ns at 5 VDD. Better than a 74HC14 with 6ns.
I'm excited...   

ATtiny404 Chip

[rolfdegen]

I found a smaller 20MHz ATtiny412 in my chip box. It only has 8 pins but the same electrical properties.
I programmed a pulse frequency of 10 kHz with a pulse width of 45ns on the ATtiny412. Is this OK for driving the 2N2369 avalanche transistor 

My planned circuit..

[rolfdegen]

Result on a breadboard. It's not perfect. But is a liitle bit faster... 
The supply voltage for the transistor was +50 volts. The Scope probe was 10x.
I'm going to solder it onto a small circuit board and measure it again.

New Measurement 660ps


Old Measurement 720ps


My breadboard

[David Hess]

Result on a breadboard. It's not perfect. But is a liitle bit faster... 

Breadboard construction is going to trash the rise time.  This is the type of circuit which should be built over a ground plane.

[rolfdegen]

Yes you are right  Tomorrow, I'll solder the circuit onto a small circuit board and connect it with a BNC connector. Until then, good night, my friend

[gf]

I assume that this generator is supposed to generate very narrow impulses? Then the "New Measurement" looks a bit weird. OTOH, the "Old Measurement" actually looks more like it could be the impulse response of the scope's frontend, which is expected to show up on the screen under the assumption that the true stimulus impulse is a very narrow one. The true shape of the stimulus impuse is likely not measurable with this scope anyway.

Risetime is a property of a step response. For an impulse response it is not really meaningful, IMO. You must integrate the impulse response to get the step response, if you want to measure the rise time.

[tggzzz]

Yes you are right  Tomorrow, I'll solder the circuit onto a small circuit board and connect it with a BNC connector. Until then, good night, my friend

Keep the inductors as small as possible. Rule of thumb is that 1mm is ~1nH, in wires or in components.

If you want to understand and measure a scope's risetime, a step wave is more useful than a badly characterised impulse. You can use a short transmission line to lengthen the impulse so that it can be considered a step.

[rolfdegen]

Yes, a breadboard is not particularly suitable for high frequency measurements. 
I will solder it tomorrow on board with SMD components.

[tggzzz]

Your scope trace appears to show an amplitude of 3.4Vpeak. Is that 10%, 50% or 90% of the peak voltage output by the transistor?

Another approach to the topic is to consider what peak amplitude would be seen on a 100kHz scope or 10GHz scope.

[David Hess]

I would replace C1 with a charge line.  Otherwise with limited bandwidth, the amplitude measurement and rise time are meaningless.

[tggzzz]

I would replace C1 with a charge line.  Otherwise with limited bandwidth, the amplitude measurement and rise time are meaningless.

Precisely.

Mentioned earlier, but he didn't pick up the point.

Once the signal has been convolved with the scope's response, the scope's response cannot be determined by deconvolution unless the signal is known.

[rolfdegen]

We're getting closer to the matter...

The square-wave generator with the ATtiny barely had any effect. I reused my old circuit (without the square-wave generator) and soldered a 35cm coaxial cable parallel to capacitor C1.
This produced a square-wave pulse. But it's still not optimal







Is there any way to still buy the circuit board? Thanks 

[rolfdegen]

Next step a liitle bit better
I have now soldered the coaxial cable to the back parallel to C1.

[rolfdegen]

Where can I still buy circuit boards for the Jim Williams Pulse Generator ?

[rolfdegen]

I ordered the Fast Edge Pulse Generator 350 Picoseconds Rising Edge from AliExpress for comparison and testing. It costs €40.
I'm curious 

[6SN7WGTB]

All interesting work, but presumably you’re not expecting to better 700ps or so on this ‘scope?

[rolfdegen]

I think so too. According to the formula 0.4/700ps, my MSO5104 (350MHz hack) would have a cutoff frequency of 571MHz. 700ps would be a very good value. I'm curious to see if the Fast Edge Pulse Generator I ordered from AliExpress will produce better results. Unfortunately, my setup isn't optimal.

[rolfdegen]

Wow... what a price. I'm not a millionaire 

[rolfdegen]

Low cost and simple and makes 510ps:-+




Rise time of 510 ps


Link: https://www.edn.com/creating-a-very-fast-edge-rate-generator-for-testing-or-taking-the-pulse-of-your-scope/

[tggzzz]

Low cost and simple and makes 510ps:-+

Low cost and simple makes ~250ps 2.5V into 50ohms. Need to be very careful with the PCB layout, lead length, and decoupling at those ^di/_dt

https://www.eevblog.com/forum/testgear/show-us-your-square-wave/msg1902941/#msg1902941

[tautech]

Low cost and simple and makes 510ps:-+

Uncharacterized, therefore 510ps is a product of the scopes rise time plus the pulser rise time.


I had one drawn up in Altium then twigged I had nothing to test it properly so instead splashed 50 Quid on Leo Bodnar's V1 pulser with a guaranteed RT of 40ps or better.
It arrived with a printed scope screenshot showing a 28ps RT....