HP 8082A Pulse Generator Repair Part 3: Ersatz ECL Fabrication and Pulser Repair 

Aka enterprise software

I'm mot ready to talk about my relationship with Maximo.

Good luck, Paolo!

rjy8960 Thanks mate!

Knowledge is everything

Good luck. This helps me stay motivated when online class stinks

I think it’s been repaired the ONLY way that resembles the hp way :)

But is attrocious like the said Eagle. Use different tools, such as the Saturn PCB Toolkit.

Sixta16 It is at least very simple code to change. That wizard can be improved in an afternoon and I do suggest you have a go at it first. Otherwise I’ll have to get a pulled request ready. It’s fairly fun code to work with.

reckon that must have been a decent fraction of a million dollars worth of kit when new on that cart lol

These old Tek scopes were truly marvelous creations, and pleasant to use.

@@zyeborm The 7854 was $10,500 in 1981

@@taunusmechanics3121 That's 25k in today dollars.

The transistors you got surely could power those strips but if they were overloaded they would get hot and eventually emit the magic smoke perhaps. But if the transistors did nothing then they were not used in a proper circuit, and I bet if you shared with us what part the transistors were, me or someone else could tell you how to hook them up to make them work. When it comes to an RGB LED strip, most common transistors can control them and it doesn’t matter whether it’s a mosfet or a bipolar :)

It's fun when you need an x-ray machine to see what they are doing with filters and other components BETWEEN layers on a board.

@@projectartichoke I think, I once saw someone build a simple radar gun (that doppler type for speed traps) with the RF path with no components, just cleverly formed traces on the board. You can do some fun stuff if you understand Maxwell...

That is definitely a more specialized piece of gear. I believe the main market for this was test and characterization of digital ICs, also radar work. My interest in it is for working with core memory, as you can see in this episode: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-AwsInQLmjXc.html.

@@CuriousMarc thanks so much for taking the time to respond. I've really been enjoying going through your videos. I had been introduced to the concept of core memory in a Smarter Every Day video a while back and found it fascinating. Your video gives more insight on the subject and I greatly enjoyed it. Thanks again!

Weren’t they great? I have a paper from my Bell Labs days from research on laser diode high speed effects that were only visible in sufficient detail to be studied on that scope.

Back in the early 90s, I had a pirate version of Autodesk Animator Pro, which was wonderful and a pirate version of Autodesk 3D studio, which was just far too complicated - especially compared to my friend's pirated copy of Lightwave.... thank heavens for KiCad and it's friends now i can "pirate" things without the guilt.

Hard to say. Month and a half start to finish, but I was doing three projects in parallel, and there was lots of waiting for parts and PCB fab. Plus filming slows me down considerably. Maybe two weeks of work?

With KiCad’s wonderful progress there’s no sane reason to use Eagle anymore. Thankfully.

"Ersatz" found it's way into English many years ago now.

My. If they ever see that one I’ll have to hide in a corner. I fear that they are going to put it on the big screen at one of our strategic technical meetings to embarrass me, and everyone is going to roll on the floor laughing, “look he made a 2mm 50 Ohms line at 1 GHz”, mwahahahaha!” and I’ll have to pay for all the beers. These guys make 110 Gb/s go through 3 feet of cable and two 100 pin backplane connectors for breakfast. This is not real RF engineering, it’s just RU-vid Electro-Monkeying-Around. That said, you can do it with amateur means, and it demonstrates the basic principles so it is very educative. At least that will be my line of defense...

CuriousMarc they have the benefits of simulations with bells and whistles though. Get them to do it with a slide ruler and a pencil, then they can laugh at you!

@@CuriousMarc I mean it works :) so there should be nothing to complain about haha

@@CuriousMarc Hi Marc. Do not be so tough with yourself...you did a good job considering that it is your first time in the RF world. I have been a microwave designer and test engineer for the last 32 years so I know what I am saying! You used only amateur tools and facilities. No fancy microwave substrate with low loss (ohmic and dielectric), no em or microwave circuit simulator...and, basically it worked at the first turn on!!!Believe me: your collegues will offer to pay 10 beers to you!!! Regards from Italy and again, bravo!!

@@CuriousMarc It is the 3 feet vs. an inch or so. Been there, done that with ECL at 600 MHz. Interestingly enough a 30 AWG wire wrap wire flat on a ground plane makes a really nice microstrip line at 50 ohm impedance. A 12" distance at 600 MHz on a breadboard - no problem.

Thanks! But I have a second one (or was it the first one?) already that worked out the gate, so I think I am good for that.

Wow! Gold epoxy!

I only realized it was slightly slower when I put the video together. I was really shooting for the 1ns rise time. Looks like it affects only the ND pulse for some reason, which you can see has a min of 3 ns. But not the “regular” delayed pulse. Not sure how it is in the original instrument. I might swap in the faster circuit once I get the correct resistor for it.

No you can't. The semiconductors used in LNB are HEMT which are similar to FET transistors but for microwave frequencies. The semiconductor is GaAs I.e. gallium arsenide. ECL uses BJT semiconductors. They are on silicon substrate. Regards from Italy.

This one is my second one. I wanted to have a second one for when I work with core memory, and need two precisely controlled rise and fall time pulses with adjustable delay for the X and Y address wires. And even a third one for the read gate pulse. But really it's a great instrument to have around the lab.

If you switch to KiCAD, watch these series of quick video by DigiKey and Shawn Hymel: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-vaCVh2SAZY4.html . They got me going in no time. The best UI I have seen in a free program is DesignSpark.

"Stockholm Syndrome" 😀 😀 😀

Thanks for the tip!

@CuriousMarc 0805 or 0603 will work with those traces and be solderable by hand with regular iron. No need for stencil and paste. You do not have to worry about coupling or insufficient feedback, that is only an issue with fast but high impedance opAmps. Do only worry about inductance wherever there is AC current. Draw the AC loops into the schematic with a marker, then keep them tight in the layout.

Hahaha! Bench-Marc’ing. I like that. I might re-use it!

👍👍

Looks like MAME has a HP9k driver, so getting them dumped and running on any modern computer might be feasible.

That era of HP9000 700 machines could use HP-HIL dongles (or flex LM license server.) So, depends on the software. Getting the right dongle is .... a pita but yes if so the protection can probably be cracked.

With smallest surface mount transistors available today the area taken by transistors would be smaller than the area of the chip packages. Marc used leaded packages. There are unleaded BGA packaged transistors that are ~2x2mm in size. You could replicate the entire IC in the area of the outline of its socket :)

Yes I have a lot of work to do on the documenting. I need to revise my PCB and schematics too with the new values and the correct landing pads for the transistors before I publish it. Will be available on my website (www.curiousmarc.com) eventually.

It’s mostly eBay sourced, some local auctions too. The scope was tossed from the lab at work as being hopelessly outdated and slow many years ago. It is the instrument that started the collection.

Oh my, these were from my box of spare parts. The pins I am pretty sure are DB-25 connector male pins. The female connectors I don't remember exactly. They might also have been from DB connector pins, or might have come from another pre-assembled thing that I took apart.

They are called SMC. Older miniature RF connector.

It's really the rise time I was after. I think it just elongated the minimum width of the non-delayed pulse, in the video it looks like it's 3ns instead of 2ns. Maybe I'll go back and try with the faster circuit once I have the correct resistor in it. But it would be quite rare that I need a 2 ns pulse.

Cool . I’d love to see that . Your RU-vid’s are great, a beacon of intellectual inspiration in this tic tok world . Keep them coming!

@@best1brit So in the pursuit of perfection I put the circuit with the faster transistor in. It gave shorter pulses, but first pulse still longer by 1ns than second pulse. Turns out there is a “factory selected” resistor to balance them, and another trim pot to adjust min width. Made my own “factory” balance adjustments, tweaked the width, now I have two picture perfect 2ns pulses at the min setting. Bill and Dave should be pleased.

You are first! You win a free 50 Ohm SMD resistor that flew away from my tweezers when I tried to place it. Whenever I find it again that is.

@@CuriousMarc Haha Thank you for your reply. I am afraid my vacuum is picking up resistors every time I do some cleaning to my room. Still, high precision temperature stable resistor in voltage standards could be worth a lot.

Yea, at my hackerspace in the Netherlands “de jonge onderzoekers” that’s common and we just grab another resistor

I can't agree that the UI on Eagle is anywhere close to "atrocious" btw. lol I tried many many PCB CAD programs before Eagle, including Cadence, Protel, PADS, P-CAD, Proteus, and most others that begin with a "P". I found Eagle to be far more intuitive to pick up than most, but I get that it's a personal taste thing as well. I am hoping KiCAD will continue to improve though, as I definitely agree that the subscription model for software like Eagle isn't ideal. What I can say is that they started moving the toolbar icons around from about Eagle 8 onwards, after they were in the same places for over ten YEARS. Which was incredibly annoying. I still stick to Eagle 7, mainly for that reason. (and they removed the colour from most of the new icons, which makes them much harder to see on a quick glance. sigh) If KiCAD can fix the problems with mousewheel zooming in Windows, and make a few other UI improvements (like making it more obvious where the button is for tying a new footprint to a symbol library), then it could end up being the de-facto PCB software. KiCAD does seem to be a bit more reliant on learning a few keyboard shortcuts, but it's not too bad otherwise. I found it very frustrating to use at first, but I guess it's a case of sticking with it. For now, I'll probably continue to use Eagle, or use Altium for the most complex projects involving BGAs / FPGAs. The auto pin swapping in Altium is awesome, and the Interactive Routing tools, but I do find it's library editor confusing. I'm interested in trying more RF stuff at some point, but it's still a lot of voodoo to me. lol The closest I got to "RF" is using Saturn PCB Calc to work out the trace width to get closer to 50 Ohms impedance for RDRAM traces (250 MHz / 500 MHz DDR)

But you can’t, because it’s the wrong logic levels.