There are times when you make the effort to do a superlative job in the construction of an electronic project. You select the components carefully, design the perfect printed circuit board, and wait for all the pieces to come together as they come in the mail one by one. You then build it with tender care and attention, printing solder paste and placing components by hand with a fastidious attention to detail. There follows an anxious wait by the reflow oven as mysterious clouds of smoke waft towards the smoke detector, before you remove your batch of perfect boards and wait for them to cool.
Alternatively, there are other times when you want the device but you’re too impatient to wait, and anyway you’ve only got half of the components and a pile of junk. So you hack something a bit nasty together on the copper groundplane of a surplus prototype PCB in an evening with ‘scope and soldering iron. It’s not in any way pretty but it works, so you use it and get on with your life.
When you are a Hackaday writer with some oscilloscope bandwidths to measure, you need a picosecond avalanche pulse generator, and you need one fast. Fortunately they’re a very simple circuit with only one 2N3904 transistor, but the snag is they need a high voltage power supply well over 100 V. So the challenge isn’t making the pulse generator, but making its power supply.
For our pulse generator we lacked the handy Linear Technologies switcher used by the avalanche pulse generator project we were copying. It was time for a bit of back-to-basics flyback supply creation, robbing a surplus ATX PSU for its base drive transformer, high voltage diode and capacitor, and driving it through a CRT line output transistor fed by a two-transistor astable multivibrator. Astoundingly it worked, and with the output voltage adjusted to just over 150V the pulse generator started oscillating as it should.
We’ve looked at avalanche pulse generators once before here at Hackaday, and very recently we featured one used to measure the speed of light. We’ll be using this one tomorrow for a ‘scope comparison.
WTF? Alredy April? This thing looks ugly.
Probably works because it is fugly. Nanosecond a foot. How many picos do you want to waste on neat board routing?
Only a few components are involved in the high-speed part of the circuit – the components directly connected to T4, and T4 itself. The rest are all just a medium voltage power supply, which WOULD benefit from a little bit of component spacing and neat layout. Sure it works today (and congratulations on that), but want to make any bets about when you drag it out of the drawer next month?
Being honest, I’ll probably not be dragging it out of the drawer next month. After all, how often do you need to measure your ‘scope’s bandwidth?
If I need one again and it doesn’t work, I’ll probably take the time to do a PCB and order the LT switcher.
Yes, there is a difference between doing something once, and “I’ll treasure it always”.
Currently my entire bench looks like your project. I even had to set up a card table annex for the overflow.
It is built from recycled parts.
I forsee plenty of exploding transistors
Can we have a picture of this in high res to make a poster out of. I need it for inspiration.
I love this. It just looks like a huge pile of garbage soldered together.
The PCB looks better than the schematic…
The schematic has a mixture of crossing lines, with and without connecting dots, and even a 4 way junction!
+1. I’ve been bitten by not placing junctions in eagle, when the board was designed some nets were not connected. Took me hours to figure out what’s going on. It’s very easy to add junctions. Tip: assign ctrl+j to the junction command.
At school I’ve learned not to use the dot as a junction. A junction can only be three way, two crossing lines are not connected. It does make it simpler to look at but I now prefer using labels. Anything with the same label is connected. There’s no need to draw the actual line messing up the schematic.
I like the Elektor way of crossing base wires: diagonally at 45 degrees. It’s easier on the eyes and doesn’t turn your schematic into a labyrinth. I’ve noticed when moving parts onto wires, the junctions are created automatically.
I like the rats nest Jenny, reminds me how I build some of my circuits from scavenged parts in my teens. I’m looking forward how the slope will look like on the scopes to be tested, maybe versus a really fast scope that can out perform the avalanche :)
Ah, that’s EAGLE being a PITA.
Eagle let you draw good schematic as well as bad ones, but the blame is on the user.
I have seen far worse, but I still think that someone that publish schematic should at least try to set a better example for others.
That’s right. Here’s a tip for Jenny and others: Use the NET command and not the WIRE command. It will place the junctions automagically for you. Run an ERC afterwards and it will tell you what else you might have missed ;).
Which is pretty much what it is :)
I swear I’ve seen that picture recently…
Oh it’s already on page 3. Then it’s ok to post it again.
was thinking the same thing. Apparently this is a comment generating pulse generator…. it reposts itself every few days to generate more comments….
Thanks very much for this – I need a pulser with lots and lots of UHF harmonics for a spectrum analyzer calibrator. I was going to try to hunt down a tunnel diode for this, since that’s what we used at Tektronix back in the day for ultra-fast-rise pulses. Tunnel diodes also need a high voltage pulse to do their thing (seems like they were around 70 V), but are getting hard to find. What I see on eBay is mostly Russian surplus parts. Now that I know about this circuit, I’ve got something in my arsenal that behaves like a tunnel diode but actually exists. In fact, exists for about a penny. My troubles are over – almost – I still need to drive it with a crystal-controlled frequency, which means amplifying a wider pulse to around 100 V (should be easy with a pulse transformer), and just shaping it with this circuit.
I’m still very fond of the way Tektronics did it way back when using a magnetic reed switch carefully magnetically biased just short of activation then driving a multivibrator signal into opposing coils. Smoking rise times! Just a major PITA to get properly adjusted.
Well, I calibrated a few Tek 485s using that monstrosity, and I can tell you I was really happy that we used tunnel diode pulsers for all of the other models I worked on. Trying to even SEE a 1 ns edge on a 60 Hz waveform was a pain.
I was looking for harmonic comb generator circuit this morning and I found something you might find useful: generator that goes up to 2GHz, with some spectrum analyzer project attached to it. I hope this helps:
http://lea.hamradio.si/~s53mv/spectana/sa.html
Thanks!
How disappointing. I came here expecting a tale of a MacGuyver-type hacker running from his arch-nemeses through snow-covered mountain passes…
Okay, so you’re McGuyver being pursued through the Alps by bloodthirsty Nazis from a secret werewolf conclave you discovered (Obviously they bred, because if 90 year olds were chasing you you’d just wait until they started breaking hips. Plus it doesn’t matter if the good guys kill nazis, so nazis.) You have a paperclip, two matches, packets of powdered soup, a scrap of tinfoil and a garbage bag, what do you do?
…
…
˙ɹǝʌʎbɔɯ ǝsnɐɔǝq ‘sʞɹoʍ sıɥʇ ןןɐ puɐ ‘ǝoɟ buınsɹnd ɹnoʎ ʇno buıdıʍ ǝɥɔuɐןɐʌɐ uɐ sɹǝbbıɹʇ ǝsıou ɟo ǝsןnd ɥɔıɥʍ ‘buɐq pnoן ɐ ɥʇıʍ uoısoןdxǝ ɹıɐ ןǝnɟ ɐ sɯɹoɟ buıuɹnq ןןıʇs uı buıʇʇıɥ ɥɔʇɐɯ ǝɥʇ ʎq pǝʇıubı uǝɥʍ dnos ǝɥʇ uı ɟɟnʇs pǝɹǝpʍod ɹǝɥʇo puɐ ɥɔɹɐʇsuɹoɔ ǝɥʇ ‘bɐq ǝbɐqɹɐb ǝɥʇ ʇɐ buıɯıɐ ‘sıɥʇ pǝʇıubı noʎ ɥɔʇɐɯ ɹǝɥʇo ǝɥʇ buıʞıɹʇs ‘ɥɔʇɐɯ ǝɥʇ ɥɔunɐן oʇ ‘dıןɔɹǝdɐd ǝɥʇ ɥʇıʍ ןıoɟuıʇ ǝɥʇ ʇɐ pןǝɥ ‘ןıoɟuıʇ uı pǝddɐɹʍ ɥɔʇɐɯ ǝuo ‘ʇǝʞɔoɹ ɥɔʇɐɯ ɐ bıɹ puɐ noʎ puıɥǝq ʇı ssoʇ noʎ ˙ʎןsnoɹobıʌ ʇı ǝʞɐɥs puɐ ʞɔǝu ǝɥʇ buıʎʇ ‘ʇı ǝʇɐןɟuı ‘ʇı uı dnos pǝɹǝpʍod ǝɥʇ dɯnp ‘bɐq ǝbɐqɹɐb ǝɥʇ ǝʞɐʇ noʎ
I just saw the build photo and see that you meant the other kind of avalanche…
Just what I needed for my future radar transmitter :)
I know, right? Since a bipolar transistor in avalanche mode seems to work much like a tunnel diode, I wonder if you could use one transistor for the pulse generator feeding another mounted in a resonant cavity as an oscillator.
Today, hacking means doing something cool with an ARM-based system on a chip. A couple years ago, it meant doing something with an AVR, PIC or other 8-bit microcontroller. A few years before that, it meant throwing a 555 at any challenge, and a few years before THAT the go-to do-everything component was the 2N3904. It wasn’t the first transistor by a long shot, it wasn’t even the best transistor available in its day, by any measure. It was just the first general-purpose transistor you could get a bag of 100 for just a couple bucks. In my book this qualifies it for the hacking hall of fame. You probably COULD make a device to trigger avalanches with it.