Restoring an HP 521A Frequency Counter: Part 1
ฝัง
- เผยแพร่เมื่อ 1 ก.ค. 2024
- We’re back into vintage HP stuff baby! In the previous HP video I asked y’all what piece you wanted to see next, and the while the voting was tight, the 521A Frequency Counter won out in the end. Honestly, I’m quite happy it won because it was the one I was most excited about! There’s all sorts of digital and analog magic going on inside of the purely tube instrument, so let’s dig in deep and see if we can’t figure out what’s going on inside!
Here’s my previous episode on the HP AC4 Decade Counter:
• In-depth Look at the H...
If you want to support the channel please hop over to Patreon:
/ usagielectric
Also, we now have some epic shirts for sale!
my-store-11554688.creator-spr...
Come join us on Discord and Twitter!
Discord: / discord
Twitter: / usagielectric
Music:
Artist: Ben Prunty
Title: Love Story
www.benpruntymusic.com
Ben, if you see this, thank you so much for the years and years of excellent music! You’re one of my personal music heroes and to have some of your music in one of my tiny little videos is absolutely mega. Thank you so much!
Intro Music adapted from:
Artist: The Runaway Five
Title: The Shinra Shuffle
ocremix.org/remix/OCR01847
Thanks for watching!
Chapters
0:00 Intro
1:57 Cleaning, painting and recapping
4:27 What I didn’t replace yet
5:43 Theory of operation
8:20 Amplitude discriminator
10:23 Crystal controlled time-base
12:16 AC-4A decade counters
17:55 First start
19:31 There’s going to be a part 2 isn’t there?
21:05 Outro
21:58 Bunny! - วิทยาศาสตร์และเทคโนโลยี
Very nice. I'm always fascinated by these old pieces of gear. Mainly because it seems so different to be an engineer back in the day. Every part, even capacitors and resistors were expensive. So you had to keep the part count as low as possible. You had to figure out a way to handle the huuuge temperature difference when operating and you had to find a creative way of doing even "simple" tasks. Sometimes it feels like it got a lot easier to build stuff, today everything is just a microcontroller and some software. What is fascinating to me about this is that the complexity of a small microcontroller that is worth only cents is orders of magnitude higher than the greatest pieces of engineering from back in the day.
Thank you!
And you're so right, that's what I find most fascinating about all this stuff. You can't just "fix it in code," so everything had to be really well engineered from a pure hardware standpoint. The engineers really knew their stuff and were able to build fascinating things with a minimal amount of components!
I love your D6 mobile paint booth. Should be enough to keep the 521A case from blowing away in the wind.
We are spoiled by the simple ease of using modern digital logic components. Vacuum tube logic takes 10X the engineering to make it work.
Haha, thank you! Though, it's not so mobile at the moment, we need to do some work on the cooling system of it, but after 70 years of brutal service, it's still tough as can be.
It's really amazing how inventive the engineers had to be back in the day. You couldn't just "fix it in code," so you had to really work on getting all the hardware doing just the right thing. I think that's why I find these old HP equipment so beautiful, the engineers really had to get the hardware perfect to get the job done!
HP kinda fudged a bit in order to get all that stuff in a smallish case. Expecting a phantastron to divide by ten for more than a few minutes is a bit optimistic-- the basic timing is an RC ramp, which is a stretch to expect to hold its timing to 10% over time and voltage and temperature. The better counters attempted only to divide by 5 in the phantastron and used a flip-flop to bring that up to ten. So you will need to tweak the adjustment pots to get each divider to divide by 10 instead of 8, 9, 11, or 12. The decade counters you can focus on one at a time, work on the lowest digit first and watch it count up a 1Hz square wave and ensure it hits all counts. Again the critical bits are the RC skip-ahead circuits and the dual triode gains and gassyness levels. The many 100K and up resistors are probably drifted up in value by 15% or more so you may need to send out to DigiKey for a slew of resistors. Fortunately 1% resistors are now cheap, like 3.5 cents each so you can actually improve the working margins over the original resistors. You may have to select dual triodes that are especially good to get the modules to count up reliably, especially at higher speeds. To add to your worries, the neons may be reluctant to light as their Krypton-85 stimulus gas has deteriorated. So you may have to replace some neons too.
Mr. Carlson has gotten a creampuff unit to operate without working on it, I recall. Kr 85 in neon bulbs? With a half life of 10 years, that's not going to be very long.
@@SeekingTheLoveThatGodMeans7648 Thinking about it, the neons don't have to ionize to count correctly, they are just for display, so it's okay if it takes a few milliseconds for them to strike. You only need fresh radioactivity if you need the neon to be a quick and reliable striker, like in a sawtooth oscillator.
It is surprisingly compact for the amount of stuff it has to accurately do.
Thank you for the info on the phantastrons, I'm learning a lot more about them and starting to learn which components to really look out for.
The Decade counters fortunately seem to be mostly alright. One isn't resetting to zero and another one has a few bad neons, but since the neons are just an illumination source, they shouldn't be affecting the count. Fortunately, I have a few spare neons salvaged from some old gear (and a full spare decade counter unit with working neons I can use for parts), so I should be able to get the decade counters back up and going well.
The bigger issue is that it's counting a ridiculously high amount. It should be counting 60 pulses when set to "Check" and on 1 second, but it's instead counting about 4,000 pulses, which might be a little too much, haha.
I just got some new caps to replace the four black beauty caps I didn't replace in this episode, and then I'll go through and test each tube and go from there!
@@UsagiElectric Yep, the Phantastron divide ratio goes up *exponentially* with leakage or resistor drift, so I would not be surprised if they are each dividing by 42 or worse instead of 10. Personally I would be tempted to up-engineer the division chain to be 5-2-5-2-5-2-5-2. The 2's could possibly be done with neon or CMOS dividers. But that's just me, a real purist wouldn't massacre the design like that.
@@UsagiElectric Yeah, the critical RC time constants are the 39, 390, 3900, 39,000, 390,000 pF capacitors and the 1% resistors going to the adjustment pots. You're probably just a few parts away from getting this thing back on track!
Great video...and that color choice for the case...sheer genius!
Thank you!
The color choice is actually a funny coincidence. A a bit over a year ago when I redid the case for my 200CD, I didn't want to run up to the store to go paint shopping, so I looked at what we had floating around in the shop, and that same olive drab green was all we had. I gave it a coat and thought it looked great, so all my future HP stuff is getting the same paint!
Oh no... Another super amazing work to watch!
Thank you!
This one should be a lot of fun to get really deep into troubleshooting!
Those dividers on the timebase work by counting pulses, and having the tube switch on hard, when the pulses charge a capacitor to a high enough voltage. Thus you need to have the RC circuit able to charge up to that trigger voltage with the right number of pulses, but also have to have all pulses be the same width, so the charge transfer each input pulse is the same. Analogue, very prone to drift, and also touchy with age and leakage in the capacitors, so probably needs to be checked and adjusted, because likely a lot of those resistors have drifted in value, which makes it erratic or count short.
Very interesting but I guess this makes total sense. Another common divider technique that came around with BJTs was to use a current source to get a linear charging curve on a capacitor with the current being gated by the input signal. For the output you just need a comparator at the desired voltage. This makes it much more stable since you don't need to perfectly filter single pulses but have a solid voltage instaead.
@@gammaleader96 Yes and was very common because implementing a dividor in digital logic needs a lot of transistors and diodes, but with analogue you only need a half dozen or so, and if you relax your standards you can get it down to even fewer, especially if you use unijunction transistors to do the actual comparison, as all you need is the one transistor to switch in current, and the UJT does the rest.
Thank you so much for the explanation!
Seriously, when I read your comment, it was like a light bulb went off and it clicked! Tough, I can definitely see how any imprecision in the resistors or capacitors can definitely make them count wonky.
@@UsagiElectric fast counting is capacitors low in value normally, less charge to get to trigger voltage, though drifty resistors and leaky capacitors make it not count at all.
@@UsagiElectric Also you can do it yourself with your existing long tailed pairs, just use a reference voltage on the one side, and another tube running in saturation to discharge the capacitor when you need it, no anode resistor needed, though it will probably need a grid resistor, or use a diode tube on the output, to both draw charge out of the capacitor and determine your pulse width.
Those secede counters are AMAZING!
Gotta go find your other video.
Can we talk about the Caterpillar? I was not expecting that, but wow.
I thought everyone had a 70 year old bulldozer hanging out in the backyard 😅
Replying a bit late but the Phantastron was an unexpected surprise. It's pretty easy to find descriptions of how analog dividers were used to generate the TV master signals used in a studio for the cameras and resulting in the video signal. What's not easy is finding examples of those circuits. I'd suspected monostable multivibrator setups were used with very asymmetric duty cycles. Instead of binary counter type circuits these effectively do division by 'not listening' for the time period appropriate to drop that number of pulses and only gating a fraction of the input through. The trick is finding a circuit with good enough time stability over a time range for a useful divisor and have a fast enough recovery period to fit in the off duty cycle of the pulse train being divided.
The horizontal sweep circuits for oscilloscopes also need such circuits. Here the recovery period is the flyback time where the beam current is cut off and the deflection plate potential is reset to draw another line. This also suggests the real source of the name for the flyback transformer of the CRT TV. Here the flyback pulse was a great source to create the high voltage supplies and reasonably cheaply.
Another surprise is that a lot of this work was channeled into WW2 radar work (plan position indicator displays, radar pulse generation, etc.) Reading one article had me thinking I bet A D Blumlein was involved and I was right!
Here's a great story about him: www.inventricity.com/a-touch-of-magic-alan-dower-blunlein
Lastly, back to the video signal, for NTSC most sources indicate the master clock is 2x the horizontal line frequency. Though the vertical sweep is much slower it uses 6 pre-equalizing pulses, 6 broad vertical sync pulses and 6 post equalizing pulses during a 9 horizontal line time interval. That's 2 pulses / line or 2x the horizontal rate. The divider chain is then: master divided by 2 = horizontal, horizontal divided by 3 for vertical pulse trains, divided by 5 and 5 again for a 210-line clock and divided by 7 for the 30Hz frame rate clock. Phantastron certainly has the range for that job.
Thanks for sharing!
Edit to add: a 555 timer circuit could do this sort of job and how I first found your channel was when Hackaday covered your vacuum tube 555 project!
So nice that youtube finally recommend this video to me today. Only 2 weeks late 😒 and yeah I have notifications on.
You may have a look at CuriousMarc's video "HP 5245L Repair-a-thon - Part 2". Looks like they recycled the basic design with the discharge lamps for the bcd couting. It uses trans instead of tubes, but the schematic looks very similar (at least to me).
I wonder why they use those contraptions for counting. When I was at school in physics class (around 1990) we still had an old tube based frequency counter. It was really amazing and also quite large. But it used Dekatron tubes for counting which have a dot that lights up and when you send an impulse the dot jumps to the next position, quite a simple but ingenious design. Maybe they were already unavailable when HP built this.
Phantastron works something like, using a single tube as a differential pair. The trick is, whereas we normally use pentodes etc. with the screen grid at fixed voltage and ignore its current flow, this parameter can also be made useful.
The fundamental at play is, when plate voltage is low (saturation region), cathode current remains constant (emission is determined by total electrode voltages, weighted by their respective mu factors: think of a virtual diode whose plate voltage is Vg1 + Vg2 / mu_g2g1 + Vg3 / (mu_g3g2 * mu_g2g1) + ... ), and current is instead absorbed by the screen. The same is true if current is diverted by a grid (as in mixer tubes: dual-control pentodes, heptodes, etc.). Put another way, the screen and plate act like the outputs of a differential pair, the currents depending on either their relative values, or of the grids controlling them. By strapping one around, a negative resistance characteristic can be had.
Though, even that is just a start, and the negative resistance path is still non-obvious in this circuit.
it's that, I think, at least given the curves they showed -- when the cathode pulls up harder, Vg1k doesn't change much (hardly at all), but Vg3k pulls negative, cutting off the plate and forcing cathode current into g2. This condition is bistable, I think, maintained by g1 positive bias. There happens to be just enough downward tug from the input diode, coupled through C1, to turn off g1 and thus Ig2, and screen voltage rises, cathode voltage falls and so g3 goes forward and plate current ramps up (along a steady ramp set by C1-2.2M).
So yeah, it's weird. :D
Vacuum tube frequency counter... actually counting, not just an LC bridge... This reminds me of the (not that bad) shock I had in my late teens when I found out there used tp be full vacuum tubes color TVs when I was still watching a BW fully integrated set...
meow thats a cool thing to have! i have got older Tesla counter with digitrons, but its on its way to become a chasis for audio amplifier, its 21st century u know 😀
You could leave the black beauties in but out of the circuit too, that’s what I do with ones I can’t bring myself to replace
The neon scheme is almost charlieplexing
At least it sort of works, I've bought stuff that would cost more than a new one to fix. :)
Can't wait for part 2.
Thank you!
This one may end up costing more to fix than buying a nice one would be, but I'm hoping it's only a few capacitors and maybe a bad resistor that's holding me up. Either way, I'm excited to be working on it because it's just so beautiful inside!
4:36 Mr. Carlson wouldn't be too particularly happy with you leaving the black beauties in there lol... could just leave them in the mounts as a display and detach it from circuit and put modern caps in place of them.
I agree but he goes his own way,(without me)
Looks like a 4 slot toaster on it's back! Great video!
It would probably make toast with that many tubes.
Hahaha, that is now one of those things that I will always see everytime I look at it!
Phantastron? I totally thought you made that up for a minute... ;-)
Phantatron is a cool name, have only seen them controlling the klystron in a old radar there also had a thyratron.
The engineers back in the day definitely had a keen eye for awesome names! Phantastrons, Klystrons, Thyratrons, Dekatrons, Ignitrons, etc., they all sound awesome!
One question. I have a Tempo 521a Irrigation valve locator ( with wand). It has a setting where you can dial transmitter power. Even if I do the low setting it goes allthe way high on the transmitter not the wand. What could the problem be?
3:49: That's a very interesting moiré effect on the plaster.
Also, I remember being told leaving fingerprints on bulbs that get hot could damage them or shorten their lifespan - does that apply to vacuum tubes? How hot do they get in normal operation? I suppose at the 24V level, they shouldn't get hot and the odd fingerprint should be okay, right?
Gold
Thank you!
Ayo is working
It's almost there! I'm hoping it'll be much closer in the next episode!
Where do you get good quality capacitors at?
cool! is failing correctly!
golden pins... sounds logical, golden reference
are the neon part of the dividing/counting circuit?
Thank you!
As far as I can tell, the neons are really just being used as a display and aren't being used to influence the dividing/counting circuit. One of the decade counters actually has a few bad neons, but despite that it does still seem to count correctly.
Is that a yellow AZ-1 sitting in your drive? I've loved those since Tom's Turbo Garage bought one, but I'd never fit into it!
Good eye!
Though it's actually a Honda Beat. Interestingly, I thought about spending the extra to get an AZ-1 when I was shopping for the Beat while I was living in Japan. But I sat in an AZ-1 and didn't even come close to fitting. My knees were in the dash, my head was above the window line and gullwings couldn't be shut, and it was just way too tight. The Beat on the other hand is quite spacious actually!
Back in the day, I saw a Suzuki Cappuccino at an auto show here in Melbourne. I decided, for a laugh, to try and shoehorn myself into it. I was very surprised to find that it was rather roomy for its diminutive size. Of course, this was with the top down. Japan, as you know, doesn't do height. I can't remember the number of times I had to duck through a doorway when I was there. Haha
super intressant
Thank you!
With no input signal applied it shouldn't count at all, and all the decades should remain at zero when the gate opens. Once a large enough amplitude signal (I forget the spec) is applied at the input it should count, and hold the count until the next gate cycle when it counts again.
Yup! Though, with the Input Sensitivity control switch set to "Check" it's actually using the 60Hz from the main transformer as the input, which is why we see it counting here without an input connected. But, it should only be counting about 60 pulses and is instead counting around 4,000 pulses, which is a bit too many!
Как я понимаю,это частотомер на электровакуумных приборах.
Дисплей на неоновых лампах.
You got it! That's exactly what it is!
Ты получил это! Это именно то, что есть!
you should check out mr carlsons lab on YT I learned a lot from him regarding valves :) i believe he restored some HP equipment as well , is there another episode on the mini computer upcomming ?
I've been watching Mr. Carlson's stuff for a while, he's super smart!
There is indeed another episode on the Minicomputer in the works, so hopefully this month! It got put on hold due family stuff, but things are starting to fall into proper place again and we're going to get to work on it again soon!
Actually
Watch the transitions on this non-binary display! :)
It's beautiful isn't it!
Like some other non-binary's I know!