Jim Williams avalanche pulse generator build--modern choice for L1

[0culus]

I've decided I'm going to try my hand at this circuit as presented in App Note 47. I have some of the parts on hand (scored an armful of new in package, old stock 2N2369s at my local surplus store, for instance), but I'm putting together a BOM on digi-key to get the stuff I don't have.

For L1, he specifically calls out a Toko 262-LYF-0095K, which is discontinued by Murata it appears. There's such a dazzling array of fixed inductors available that my head is spinning trying to sort them all out. Any particular recommendations for a substitute?



Thanks!

[bd139]

Two options:

1. FT37-43 ferrite core with 21 turns of wire on it.
2. Bourns radial 150uH inductors or the cheap chinese clones of them.

Pretty non critical for this one.

From the TEA thread yesterday you can see the corner of option 1 in my failed attempt to build the same circuit due to the leg falling off the LT1073



Edit: then the bodged up non LT1073 replacement as I'm not paying £7.50 for another IC!



Also beware you might need more than 90v out of it get the transistor to avalanche so be prepared to futz with the divider a bit. This isn't one of those circuits that just works.

[HighVoltage]

We had a discussion about this great pulse generator...

here:
https://www.eevblog.com/forum/blog/eevblog-306-jim-williams-pulse-generator/

and here:
https://www.eevblog.com/forum/projects/new-version-pulse-generator-(based-jim-williams-pulse-gen)/

[MagicSmoker]

L1 isn't terribly critical given the tiny average current required by the pulse generator. I would look for a drum or toroidal core type with a DC saturation current of at least 200mA. E.g. - any from this pre-filtered list except for the axial types that somehow qualified as "shielded":

https://www.mouser.com/Passive-Components/Inductors-Chokes-Coils/Fixed-Inductors/_/N-wpczZ1yzvvqx?P=1z0wreaZ1z0t1rp&Rl=wpczZer9dZ1z0wrdmZ1z0wpl5SGT

However, you'll make your life a bit easier if you use a 3.6V lithium cell - or even a 9V battery - instead of a single 1.5V alkaline cell, as this will dramatically reduce the peak current needed from the cell - and which the boost inductor must tolerate without saturating; after all, stepping up 1.5V to 90V requires a corresponding 60x increase in current drawn from the source - and also allow you to use a wider range of switcher ICs than the comparatively few that will run off a single 1.5V cell.

[JackJones]

Also beware you might need more than 90v out of it get the transistor to avalanche so be prepared to futz with the divider a bit. This isn't one of those circuits that just works.

The switch is rated at 50V, so there isn't a huge amount of playroom. Don't futz it too far!

It uses a voltage doubler so the switch is seeing only about half of the output voltage. Going much closer to 100V might cause more problems too.

[bd139]

Good point! My build above used an IRF620 which has 200V Vds(max)

[0culus]

Thanks for the advice...I do have a bag of random inductors...no idea what the core material is though.

[MagicSmoker]

The switch is rated at 50V, so there isn't a huge amount of playroom. Don't futz it too far!

It uses a voltage doubler so the switch is seeing only about half of the output voltage. Going much closer to 100V might cause more problems too.

Yeah, good point. In light of this I would recommend going with either a tapped boost or non-isolated flyback instead of a boost with voltage multiplier, especially since it might take more than 90V to reliably avalanche the BJT.

[David Hess]

Another change you might want to make is to replace C1 with a coaxial charge line so you get a flat topped output pulse more suitable for rise time testing.  Later schematics show this configuration.

Yeah, good point. In light of this I would recommend going with either a tapped boost or non-isolated flyback instead of a boost with voltage multiplier, especially since it might take more than 90V to reliably avalanche the BJT.

The advantage of the capacitive multiplier with a boost converter is the lack of a leakage inductance spike which is why little voltage derating is necessary.

I would change it to a tripler or add a high voltage cascode transistor before I used a tapped or flyback configuration as long as the current is low enough which it always is in this case.

[0culus]

So, after building a couple of prototypes of the power supply, I feel like I may have made a bad assumption about the assumed units on the caps. Are they supposed to be 0.47 uF?

[bd139]

First one of them I built I used 0.47uF and it worked. I used film caps.

[pgo]

There are a few design on the web - search will give you several.

For reference here is the one I did a while ago.

https://github.com/podonoghue/Jim_Williams_Pulse_Generator

You can check the BOM for part numbers from Element-14

[0culus]

OK, so I think I have everything correct...but it's giving me ~8-900 mV instead of of the nominal 90 V.

I noted that the SET and A0 pins on the LT1073 are unused. I tied those to ground but it didn't make much of a difference. I managed to get some crappy Chinesium radial inductors and got a 200 uH and a 100 uH one out of the box after testing a lot of them (they weren't marked at all, even to show what value should be in what compartment  )

[0culus]

I figured it out...I did a rookie mistake and didn't notice that the datasheet for the LT1073 specifies a maximum soldering temp of 300 C...I had my station set at 370.  So I cooked that one good. Fortunately I have more than one on hand. With the 200 uH inductor, it's producing about 96 V output now.