"Of course now I'm going to have people demanding I write a flight simulator" Well... you made a joystick interface. You know the rules.

James, I'll say it again. this has got to be one of the cleanest 8 bit builds I have seen as well as being one of the fastest custom 74 series CPUs. The test on bread boards and transition to surface mount along with the attention to aesthetics makes watching your videos very enjoyable. I look forward to seeing this evolve.

I learned about this way of doing A/D conversion just like you by looking at the "gameport" implementation. In my case it was on an ISA card. I needed someone to explain how this trivial looking bit of electronics achieved analog input and when it clicked it blew my mind. Mainly because it made obvious to me how a change of timescale can make a qualitative difference. There is code, which we easily antropomorphize to be patiently waiting for the cap to discharge. Yet it seems instantaneous to us actual humans.

analogue is much-maligned, but it is as close to life as we can get. This is making it real. One can almost taste the actual electrons watching this stuff being created and made to obey. Thanks James. Nice bit indeed!

I always wondered about that calibration step, and after 30+ years, I've finally seen an explanation! Thanks james!

Wow, only just watched this video some 5 months late! Such a beautiful and efficient Analogue to Digital conversion technique. They really knew how to do things the easy way back in the day! Thanks also for the link to Dave Jones video.

Perfect way to start my weekend

Going through JS video withdrawals, so I asked an AI to write a poem. It's not Keats, but hey: James Sharman, whose videos bring light, His content brings joy, even in the night. His creativity, wit, and charm, Leave us wanting more and more. With every video he does share, We fall in love with his videos, we dare. His work, it's truly a masterpiece, We can't wait for his next release.

I used to use the same technique back in the 90's when I was playing with PIC16F84 MCUs and making line following robots with them. Just sample the pin until it went from high to low, switch it to output to charge the cap, switch back to input and sample over and over until it went low. This gave a poor man's ADC and worked like a champ. No 555 needed.

This is great, and could work the same way we do it with MCUs, by having it interrupt driven. Start discharging the cap with a timer set to zero, then increment the timer via the CPU clock (via a divider I guess, as you probably don't need extreme resolution), then the GPIO generates an int to store the timer value and set a flag indicating the conversion is complete. A lot of MCUs only have one or two actual ADCs, and multiplex them too. So I guess if you multiplexed you could indicate the channel when starting the timer and the elapsed count could be stored in a location related to the channel, so all of this would essentially happen in the background, and the main CPU code only needs to read from those memory locations when it feels like it :)

Excellent stuff !

Thanks for another great video!

Oooohhh something decent to watch! 😊

Love it! Very simple, but surprisingly effective

Loving this series, gets better with every episode. Clear explanations, utterly compulsive viewing

Brilliant, as always!

What a nice way to do that. I learned that in school but forgot it. This is such a clean ADC. Awesome!

I've always considered ADC and DAC to be just "black boxes"... It's quite a revelation watching you reduce it back to it's basics and seeing how simple it is to make a ladder of resistors for one and time the charging of a capacitor for the other. As always, someone pulls back the curtain and you say "oh! Is that all you need to do?!" :)

Thanks James. That was interesting. Since it depends on resitance there is no need for a voltage divider. Very cool. Take care.

This is super cool! The circuit works very similarly to a dual slope ADC, just using an input resistance rather than an input voltage. It’s quite a nice simple way to read the potentiometer with quite good resolution and little error.

very good job

What you have built there is a rudimentary single ramp integrating ADC. If you added a counter and a comparator and a latch to count and store the rise time it would reduce your processor utilisation and turn your circuit into stand alone integrating ADC.

used to do this on pic's using one pin. Set pint to output and set low-- ground the cap Set the pin to a input and time until it sees a 1 Works well for most things.

Good to the last second. Ah the good old days. Before wimpy crowbars on the power line or fussy low voltage circuits. When opening your TV gave you a haircut like Yahoo Serious and when it said danger of death it didn’t mean brain death as you examine parts smaller then your tweezer end. I say you throw in something that uses ultrasound and create the worlds least capable Wii game. Cricket might be nice. I seem to recall the spectrum had cricket with stick figures in some sort of isometric projection. That. Plus a ultrasound sensor. Boom. Wii. Done :)

Given a transistor is a current multiplier and given the voltage on a capacitor linearly increases in time when given a constant current, it would be very easy to implement this with a transistor and the behaviour would be pure linear, which gives you more accuracy.

Smashing stuff, you really want to get a 100uF cap across the power rails near the NE555, they crowbar the supply when discharging and can cause all sorts of nastiness :) you could see it in the rising edge (overshoot) on the scope at one point.

Join us on Discord: discord.gg/jmf6M3z7XS Follow me on Twitter: twitter.com/WeirdBoyJim Support the channel on Patreon: www.patreon.com/JamesSharman On the Discord's Video-Discuss channel I've pined a zoomed in picture of an early pc (ISA) game port card, it's easy to see the circuit is conceptually very similar to what I do here alongside just enough decoding logic to map the IO lines in. discord.gg/jmf6M3z7XS

Very cool! Did you consider extending the circuit with a counter that would just put the number on the bus, so that the CPU could do other things instead of counting until a line goes high? (You don't want to go more into that, but maybe interesting for other A/D conversation builds in the IO module, depending on your plans.)

So awesome - GPIO interrupt feature would be handy...

Converting a voltage/resistance into a PWM signal is clever for this kind of applications - the CPU can naturally measure time with high precision and you save on bus logic because the "analog" signal is basically treated like serial data rather than parallel data with a real ADC. The only question is, how linear is the pulse width from the 555 with the resistance? I'd guess the nonlinearity is negligible for gaming, or you can implement a very simple LUT for linearization (which is what the "calibration" screen in a videogame does).

2:30 The circuit you're describing is somewhat similar to how capacitive touch sensors work as well. That's kind of cool.

Since you've mentioned that you're considering another ADC video, I'll throw out a two-bit circuit I've been thinking of trying. You run your input voltage to the bases of two "identical" NPNs (Q1 & Q2). Those are (independently from each other) tied to ground through a resistor (R3 and R4 in my schematic), a diode (for the voltage drop), and another resistor. One bit is pulled from the points between the diode and the ground-most resistor (through buffers, to avoid problems), resulting in two bits. Those same points each control a PNP (Q5 & Q6 in my schematic) transistor, which is connected to the _opposite_ branch's R3 or R4 & diode through a resistor (R9 or R10) of it's own (the transistor is meant to operate in saturation, allowing the current flow to be controlled directly by those extra resistors). The concept is that Q1 & Q2 produce basically identical amplifications of the input signal, which will turn on the buffers if the input voltage was high enough, and also will turn on Q5 & Q6 in that case. Meanwhile, if Q5 & Q6 are turned on, then they will steal a _limited_ amount of current from the opposite branch through R9 or R10, which in turn will reduce the voltage in that other bank, possibly turning off whichever of Q5 or Q6 is controlled by it. By itself it sounds like this should be just an oscillator, but if you give R9 and R10 sufficiently different values then one will decisively be the "low" bit and the other the "high", with the bits "quickly" settling, and oscillations only if you're presenting an intermediate voltage. I assume that these could be chained at the low-bit current-theft points, but I'd worry about stability with each additional bit.

I saw that the Atari 2600 paddle uses this method. Very simple solution to an otherwise complicated problem. It works even better on the 2600 because you can measure if the cap is charged on each scanline to position things like paddles in pong.

Analogue is great - my prime-dean (the head of university) made a thesis over the analogue processors :) - i mean using simply the opamps and so instead of digital maths in an electronic machine :) - the SSDs though digital are really analogue physical devices - you need to adjust the voltage sometimes to be able to read a page :)

Super cool! Also not the approach I'd guessed you'd take; I'd expected you'd go right to implementing a binary-search ADC -- that is, an R-2R ladder DAC plus a comparator to tell you if the DAC output is higher or lower than the analog input.

Very nice design. Another IC, i have used in this kind of designs is 74HC123. It got 2 monostable inside and it will be less parts. No real difference, the results will be the same though.

Nice one. Even this is a sidebuiild only, one might expect you to run a demo using the joystick. For the record, I did not say doom.

You should write a flight simulator! (Sorry 🙂 )

Best of luck with the flight simulator. You have some videos that you write code in C/C++. Do you have a video of where you did the compiler port?

When I saw what you planed to do, I knew you were going for the timing method, but my mind went in another direction: could you use a 8-bit (or so) counter IC, maybe a 74xx590 with output register. Connect it up to the system clock and have the reset/latch driven by the capacitor/potentiometer circuit? sure, this is more involved, but performance wise it could read the value without hogging the CPU

First thought. Add a “I is a dumbwit, did I watch part one?” Link in the description at the top. “Part one, in case you missed it”

Superb, I didn't know how creative they were back in the day but I guess they had to be. Does this technique have a proper name I would love to learn more ?...cheers.

Only just started the video and have to admit, even though electronics is my main hobby, I had never considered or read about analog input without adc. My guess would be forming an RC oscillator with the potentiometer and count CPU clock per RC oscillator cycle Edit: OK, maybe I took "no adc" a bit too literally. I figured discrete single slope adc were ruled out

a newbie here. maybe you can replace the two 555s with a single 556? excellent videos! keep it up!

Heh, as soon as I saw "analogue", I was thinking "He's gonna pull out a 555 and a cap"... and sho 'nuff :D It's remarkably straightforward and useful. This is the way (although in a slightly more slick fashion) they used to do it in the Atari 8-bit line's paddle inputs. It was effective enough that, with a little glue electronics, you could digitize audio (at a ridiculously low bitrate, but it worked).

Could you do both axes with just one channel by interleaving them? Using a couple of transistors on the analog side to switch which channel is being read? Edit: aha, I jumped the gun on that comment, I didn’t see the part about how you wanted to read them simultaneously. It’s a really cool ADC though, so simple and yet perfectly effective for a low precision input.

well, maybe not a flight simulator, but a racing simulator? like the thing you did for scrolling demonstration

Using a 556 timer you can ditch one timer, for one less component.

were can you find those old joy sticks. There must have been millions of them but try and find one now is impossible.

I demand you build a flight simulator.

Shouldn't it behave more like an ADC peripheral in a microcontroller, i.e. start the sampling process with a register, and be notified when it finishes? Because currently, you are wasting precious CPU cycles just with waiting for the capacitor to charge. Wouldn't take much more than a counter and a few more gates.

i was always wondering how ADCs are made

TTL overflow reset counter and 555 timer for counter capture register. no code neded

soooooo... flight sim?

atari 2600 does it this way

I mean this is what a DAC does too...