Hacker News new | past | comments | ask | show | jobs | submit login
Inside the vintage Xerox Alto's display, a tiny lightbulb keeps it working (righto.com)
206 points by darwhy on Oct 14, 2017 | hide | past | web | favorite | 50 comments

Using various interface-ish components only for their electrical parameters is neat trick that is very common in certain applications. In this case the lightbulb is slightly weird but probably serves to shape the 600V pulses in some way and also to protect the switching transistor should the grid outputs get shorted to ground. Looking on the schematic there are also neon lamps across the outputs to protect the CRT from overvoltage (which you will find in essentially any modern CRT monitor).

Another such neat trick is in most simple photographic flashes: the neon bulb which indicates that flash is ready also provides feedback to the flyback charging supply by increasing it's load when correct voltage is reached.

And then there is typical CRT degaus circuit, which produces the required damped oscillation waveform by running mains voltage throught two thermally coupled thermistors.

Edit: Horizontal deflection circuits of CRT TVs and monitors are usually full of neat tricks: shifted hysteresis inductors, deriving various voltages from the flyback and output multiplier, synchronisation of main SMPS switching frequency to horizontal retrace... One can view CRT VGA monitor as one big switchmode power supply with video amplifier bolted onto it (and many CRT monitors are actually mechanically built this way, with RGB video going directly from the captive VGA cable to amplifier on the neck board)

My old Dodge instruments need 5 volts, so there's a mechanical device to convert 12V to 5V. It involves heating a bimetalic strip that opens and closes a circuit quickly, i.e. buzzing, which produces 5V.

Some people have decided to fix this by using an electronic voltage circuit to get a clean 5V from the noisy 12V supply. But the instruments just wouldn't work right.

It seems the instruments relied on the 5V supply being jittery. With a smooth voltage, the mechanical instruments would get stuck.

Watch a pilot in an older airplane. He'll tap the gauges, too, to get an accurate reading.

Actually the reason is that the instrument itself also uses a bi-metal strip and it relies on the ratio between the two strips to cancel out voltage fluctuations.

It's very elegant, but if you take out one, it goes haywire.

> My old Dodge instruments need 5 volts, so there's a mechanical device to convert 12V to 5V. It involves heating a bimetalic strip that opens and closes a circuit quickly, i.e. buzzing, which produces 5V.

Are you sure that this is not a https://en.wikipedia.org/wiki/Vibrator_(electronic) that would be powered by a solenoid and not just a bimetallic strip?

I think you're right.


You were actually initially right, I searched for that actual part and it is indeed a bimetallic strip along with a heater (that turns off when the strip moves away from a contact) and not a device with a solenoid/relay.

Apparently whatever it supplies is tolerant of that sort of interruptions in its power source -- I imagine that it would operate on a quite lower frequency than a "vibrator"-type power supply where magnetic force moves the contacts (and elasticity returns them).

I read about this probably 10 years ago, my memory was clearly a bit faulty on it. I didn't want to take the device on my Dodge apart just to verify it :-).

The thing is, the instruments require the noisy 5V.

You can buy kits with an electronic 5V source, and also to replace the old guts of the instruments with modern ones. That retains the original look, with modern accuracy and performance.

Nice examples. By the way, those symbols on the schematic aren't neon lamps, but spark gaps - simply pointed PCB traces separated by about a millimeter. Presumably these are to safely discharge any excessive voltage buildup.

In this case that probably suffices as there is single intended video source.

Probably all multisync monitors I've seen have actual neon lamps and I believe that these neon lamps are in fact crucial for the "multisyncness", because they provide time for the control electronics (which given the package and pinout seems to usually be 8051) to settle after mode change without blowing out the CRT.

For a long time I've considered all the warnings of the "wrong video mode may damage your display" kind as nonsense. Then about ten years ago I had blown two 12" POS monochrome CRT monitors by unintentionally running them at 1024x768@75Hz, there was correct image for about a second followed by loud bang.

By the way I was thinking about all the multisync monitors I disassembled and got the impression that in all of them horizontal deflection was driven by class D amplifiers (there was half-bridge with switching MOSFETs, IRF640 and such). Does anyone on HN know whether that is correct assumption?


If regular CRT televisions behave the same way, that means at some point in time it would have been possible to create a "jamming signal" that transmitted a signal with the wrong refresh rate and blew up all the TVs in the area.

Regular CRT televisions weren't multisync - at best they'd be able to deal with both 50Hz and 60Hz inputs, but they wouldn't attempt to sync to anything else.

Multisync means the CRT will succeed at syncing to multiple specified frequencies.

There is nothing stopping non-multisync monitors (including TVs) from trying to sync to an out of range signal and depending on the design they might damage themselves trying.

I never cease to be amazed at the sheer complexity and precision required to make analog electronics work. There are so many parts to a CRT, and about half of them seem like magic (directing electron beams with electromagnets!), and the first was built in 1911. LCDs seem downright simplistic in contrast.

It's kinda the same as the move from internal combustion engines to electric. The discrete parts count drops to a tiny fraction of the former. (Though, I guess if we were to count transistors in an IC as individual parts, LCD trumps CRT.)

Then again, one could build a CRT without a billion dollar fabrication facility. The same can't be said of LCDs (AFAIK).

Regardless, I really enjoy these posts. I don't understand half of it (though I understood more of this one than usual, as my dad owned a TV shop when I was a kid and I got some hands-on experience with CRTs early in life), but they're very satisfying. I wish I could play with cranky old computers more often.

It's all pretty simple when you know how. It's getting there that's the hard bit. It's arguably more challenging than software but once you've learned the abstractions and techniques then the knowledge is useful forever so you don't have to wipe the slate clean again when the emperor's new clothes are delivered.

The pinnacle of analogue electronics and precision is the line of Tektronix analogue oscilloscopes. They are marvelous from just about every angle. Each box is a blend of physics, mathematics, mechanical and electrical engineering that is simply unsurpassed. The engineering is reminiscent of the space race. No expense spared. Everything covered in gold and full of precision machined parts.

My current favorite Tektronix unit (I have a few!), a 475, is 42 years old, works perfectly and is still classified as relatively high performance compared to modern equipment. And the knowledge gained from looking after it and understanding it is STILL valid and STILL 100% usable in 2017.

You don't get that value from a Javascript framework!

I'm hoping my dad kept his old scope. I haven't gone through his tools (he passed away a few years ago, and I didn't have the heart at the time, since fixing stuff was how I'd always bonded with my dad and every tool has associated memories), but plan to visit my mom in a few weeks. I'll see if it's still around. He owned a TV shop through the 70s and early 80s. His equipment was usually top notch stuff from the peak of the analog electronics era, but he got rid of some of it over the years, especially during moves.

Someday, I'll have more time to play with electronics. It's something I enjoy, and I have most of the tools for it (except a scope), but I always seem to have too many other things requiring my time and attention.

I worked with high-end 90s-era Tektronix scopes when I worked in television. They definitely weren't pure analog by that time, as they had memory and a variety of modes providing both digital and analog data, but they were still pretty cool.

Yeah they’ve come a long way since the analogue units. I’ve got a new Chinese digital unit as well (Rigol) and it’s got Ethernet and USB and stuff and you can drag data off it with python trivially for example. Colour wide screen and measures just about every parameter for you automatically.

But it feels like it’s lacking a soul.

Playing around with Electronics is my disconnect from reality for a bit. Works pretty well :)

If You find any of your father’s stuff, spend some time with it. It helps things believe it or not.

Those older Tektronics are works of art. The boards were meant to be repaired. I still remember the feeling I got when I took the case off mine. Yes, complicated, but so much thought was put into the layout.

I guess those days are over?

You can buy used ones from army surplus stores at incredibly low prices. I got mine that way and the only thing that it needed to be in spec again was a minor recalibration. Super nice gear.

Indeed. Not if you buy old oscilloscopes they’re not!

They’re still perfectly useful instruments to be honest.

Reminds me of an electric organ a friend had. It was totally analog and dated back to the 60s or 70s.

The volume control knob abruptly stopped working; the organ was stuck at maximum volume.

He opens it up to discover the volume control simply controls the brightness of a light bulb, which had burned out.

The light bulb shone on a photo resistor, which governed the output volume.

Faders in DJ equipment are often optical -- apparently a potentiometer would be noisy.


That would make DJ gear the exception, then. There are slide pots for that purpose which do quite well with audio across them, they’re just in bigger mixers for sound reinforcement and radio gear. Digital obviously negated needing them to work the same way — it actually rearchitected how an entire mixer works — but analog pots got very good toward the end of their “run”. Lots of quality radio gear out there with beautiful 10Ks, much still in use. I wouldn’t call it “old” like that article does. Then again, DJs also beat the shit out of faders when they work, so it might be a good idea from a reliability perspective.

Coming from the reinforcement side, a lot of DJ gear baffles me. They look at audio much differently, which I suppose is understandable when you’re mixing two decks with an occasional bandpass versus a 50+ channel metal band with subgroups and racks of outboard effects.

Same reason some guitar wah pedals use optical control. There are ones with mechanical pots but you have to keep replacing the pots every couple of years when they become "scratchy" as you move the pot back and forth quite a lot in use

Yep. Definitely maintenance involved and a culture of care and feeding of faders on high-end gear.

> Digital obviously negated needing them to work the same way

So, how do faders and knobs work nowadays?

Oh, are they perhaps still potentiometers, with the difference that you can now low-pass filter them before reading off and digitising their values, since you don't have to pass the audio signal through them?

Some knobs use rotary encoders. A nice side-effect is that they are "infinite" in either direction, unlike helical or single-turn pots.

Yeah, I have vague recollections of taking apart an analog electric organ (like https://en.wikipedia.org/wiki/Hammond_organ) and seeing an "optocoupler" with an incandescent bulb and an LDR in a little enclosure.

This is a very common little device. The V-I characteristics are non-ohmic i.e. non-linear which make it very useful for feedback. Some notable uses I've seen it for floating around:

1. One of my DMM's, a Fluke 8600A, has a light bulb in it as part of the charge circuit: https://i.imgur.com/TMvRmKF.jpg

2. My Tektronix DM44 strapped to the top of one of my scopes has one as part of the ohmmeter input circuit.

3. Also used for amplitude stabilisation in feedback networks in the negative feedback path.

Very common tricks and very useful little things.

Only problem is they get hot, eat a lot of current and snuff it occasionally. They were replaced by JFETs for typical feedback applications.

The famous HP 200A oscillator, which was their first product, used a light bulb in the feedback loop of a Wien bridge oscillator to stabilize its amplitude. Not only was it non-ohmic, but it had a nice dynamic response: Fast enough to keep up with minor amplitude drift, but not so fast as to cause harmonic distortion.


Indeed. Works better with tubes. I built a slightly more modern version with an op amp and a small 28v bulb in the feedback circuit and it didn’t quite perform as well as I’d hoped. Was interesting however.

Wow, if you could get your hands on a Dorado that would be something. On one of my visits to PARC there was an office with the door closed and the name plate "E. Dorado" on it. The offices on either side had displays and keyboards but the machines for those offices were in this office. The display and keyboard cables being fed through the walls. Not a 'machine room' per se but a separate office.

I recently came across a couple big boxes of ECL boards at Xerox PARC that I think are from a Dorado. But without the backplane, it would be pretty hard to make a Dorado out of the boards.

Some photos: https://photos.app.goo.gl/cn1CV9ts5sINLlwa2 You can see the 10181 ALU chips - these are ECL versions of the popular 74181 ALU.

Wow, they appear to be about the right size. And yes, it would be impossible to assemble without a backplane, and fabricating one would be challenging even with documentation.

The dorados and (some of the dolphins) were in machine rooms, not offices, as the ECL machines ran so hot. There were not a lot of dorados so you could connect to an available machine. One reason I used work at night was that a dorado would usually be available for me to connect to from my office.

Surprisingly, despite being exposed to 600 volt pulses, the problem with this 28 volt bulb wasn't that it had burnt out, but that one of its tiny leads had broken.

That's because the bulb doesn't have 600V across it. In a normal application as an actual glowing bulb, it's rated to have 28V across it, pass a certain current, and emit light at a certain brightness.

In this application, it's being used as a current limiter, and the high voltage supply to CRT grids is very low current (several milliamps at most) --- thus, the current through the bulb and the voltage across it will also be very low.

It would have 600v if whatever else it was in series with shorted. It's not clear how much resistance the rest of the circuit has, so it's hard to say whether the pulses briefly exceed 28v across the bulb or not.

It's not uncommon to exceed forward constant max current specs for an LED with pulses (PWM), for example.

Really cool pictures!

Light bulbs are sometimes used for stabilizing oscillators. I think it works like an (audio) volume "compressor" would.

For example the Thorens TD125 turntable uses a light bulb as part of the oscillator circuit that generates the AC signal that powers the motor.

A bulb was also a neat hack used in the 200A, the first product made by a small and relatively obscure garage startup in Palo Alto: https://en.m.wikipedia.org/wiki/HP200A

I like it how someone that really knows their subject can explain things so fluidly. I have a good idea of how CRTs work but in the many diagrams and lessons that I have endured I have not read such a fluid, easy to understand overview of how a CRT works.

Not everyone has this gift of being able to explain stuff in layman's terms even if they are expert on the subject. Carl Sagan was best at it, on the other extreme you have those that use jargon beyond the audience's knowledge, pretending to know but not understanding it really. This whole Alto series has had some great writing in it.

Our boat has a Hitachi alternator on the engine. From reading the schematics one of the leads connects to a warning light. This warning light is to let you know the alternator isn't charging.

That warning light also has a second, and even more important, function. It induces a load on the alternator to energize the charging coils. A few hours spent removing what I thought was a defective alternator and $100 having it rebuilt. Only to find out it was a $1 bulb.

In this digital world analog circuits are like witchcraft to me. How you think things work is not at all how it works.

The bulb limits the current draw from the battery. When the alternator is starting up, the field coil current comes from the battery. Simply connecting the battery across that coil with no additional resistance would draw to much current from the battery.

Alternators have a circuit such that when the battery current is no longer needed (the alternator has spun up and is putting out more voltage than the battery) an arrangement of diodes cuts off the battery field coil circuit and then the light goes out.

When the alternator is self-energizing, there is a feedback mechanism then at play which controls the field current, keeping the alternator voltage level: without it, the voltage would rise uncontrollably: more voltage from alternator -> more current through field coil -> greater induction of voltage. Before the alternator is self-energizing, what regulates the field current is that bulb.

So that is to say, if you see the light on all the time when the alternator is spinning, that warns you that it's not actually working. It is not it's not putting out enough voltage (perhaps none at all) to overpower the battery circuit and become self-energizing. This could be because the rotor coils are burned out, or other problem: wiring or whatever.

Since that bulb is off most of the time, we don't expect to have to change it very often.

If an alternator bulb has burned out, and the unit is not very old, that could indicate a problem.

Right on the money! We actually did have alternator problems as this was a 12 year old boat. The bulb was on constantly, had built up a layer of soot in the glass, and couldn’t be seen in the sun.

Going back and forth with alternator mechanic he asked me to pull the instrument panel and inspect the bulb.

As part of my electronics training many years ago we did a subject on TV repair as a practical introduction to circuit design, at the same time we studied transistors from a more theoretical perspective at college. I will always remember the instructor saying 'there is the design on how you are meant to use transistors and there is how they are used in practice in a TV'. For a consumer TV there was just as much focus on optimising costs as theory. We studied many amazing individual tricks like where three transistors can be used to replace four by putting two backwards. Just assuming a junction voltage of .6V would make you assume everything is broken in a working TV. Many of the tricks rely on non-linear characteristics of the components, such as a light bulb filament.

Lightbulbs can function as a PTC, as the bulb heats up the resistance goes up, and current drops until it stabilizes. This allows you to use a lightbulb as a way to stabilize a voltage or a current.

In the model train world, we use car brake light bulbs as current limiters for a reason similar to why they used one in this monitor. http://cs.trains.com/mrr/f/744/t/215408.aspx

Nice little touch there on measuring the 17kV signal from a distance of two inches, without touching it at all, thereby avoiding lots of safety issues.

The classic example was the light globe used in a polaroid camera.

The globe was powered from a mercury cell, and because it's voltage was so constant, the light stayed at constant brightness.

Then a simple plastic lens overlayed alternate stripes from the image and the globe. The user adjusted the aperture until the stripes vanished, which guaranteed that the exposure was correct.

Cool. That reminded me of a very simple tube amp design I had seen that used a light bulb as a current source.


Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact