This is a demonstration of a basic analog computer that I built. I solve a simple second-order linear differential equation and walk through the schematic.
Dunno a whole lot about this but all i know is that its very cool! And you could be doing very well for yourself if you continue working with stuff like this! Analog computers may have a uprising here!
Hello dude, so awesome explanation, you are so smart bastard, thanks for sharing your knowledge. I have a question... what electrónic component (or thing) is needed for put in your board so that the computing turns into digital??
To get something like this to interface with the digital realm, we would need most of all a DAC (Digital to Analog Converter) and perhaps some kind of USB interface or microcontroller to connect to a PC. Ideally, the DAC should have enough resolution to accurately capture and render the various waveforms as raw data.
This is excellent work. I'm trying to build something similar myself. I believe the quick explanation of the multiplier circuit is it computes via exp(ln(x) + ln(y)) = x*y. Have you gotten good results with the differentiator? My understanding is the bandwidth is too high so people would combine it with an integration topology on the same op amp.
I've read about standalone differentiators being not accurate enough to be usable. I think I see why now. The one I made seems to work okay at low frequencies (< 100Hz) but has some weird problems at higher frequencies; the waveforms don't look right. I made R adjustable with the DIP switches but it might be better if C is adjustable or both. Basically, it does work but it seems to be very situation specific. The multiplier didn't work as good as I thought it would. It multiplies, but only half wave. Anything above zero volts gets cut off. I wonder if "full wave" logarithmic amps would be better. In fact, I bet a Gilbert Cell circuit or something might be more linear.
Hey A. S. (?), I'm in contact with a company looking to build an analog computer on a chip. However they plan to build an analog computer for students first. Are you in the located in the continental EU? What is the largest analog computer you could plug together with the components you have? In traditional analog computers the use of differentiator was frowned upon because one just could not implement them as the transfer function of an differentiator is not implementable in physics. Any circuit i wanted to use an differentiator for i could usually reformulate it using an integrator instead. Have you looked into crossbar switches or to turn your analog into a hybrid computer but adding some D/A A/D converters?
Interesting, I didn't think anyone really cared enough about these to develop them any further. I am not located in the EU, I'm in the USA. This prototype I made was meant to inform me of which parts of the computer were useful and how they should be implemented. I made a differentiator to test its viability and soon realized why they didn't use them back in the day. I think a multiplier would be super useful but implementing it cheaply and accurately may not be easy. Expo and log amps might be useful too. Some oscillators or PLLs would be something I'd like to add. Making the mixers have voltage controlled gain would be great too. I ultimately plan to make a massive computer with multiple circuit boards. Instead of a patchbay, each section will be connected with analog switches controlled by some kind of microcontroller like an Arduino Due. The mixers will have digipots for gain control and CMOS switches to connect everything. You would then program the Arduino for the equation and be able to change parameters as easily as the code can implement changes. There would perhaps be some kind of DAC with and SD card so the computer would have "storage" or "memory" like a digital computer. It could then recall certain functions back into the circuit as the code directs. I would also include extra scaling amplifiers, instrumentation amplifiers (for sensors) and motor drivers for interfacing with the outside world. Similar to an Arduino, it would effectively be an analog Swiss Army knife. As far as the size of computer, the sky is the limit. The trick is keeping it reasonable. This little prototype was a lot to handle. Implementing the solid-state switching might get complicated.
@@thetuberoaster8321 Cross bar switches, also called Switch matrices are an have 2 groups of n inputs and can many any input to any output. They also have some digital input used to program the mapping in real time. The form factor usually quiet small compared to patch panels and there are non expensive ones available. If you are not aware of them they might be neat for your applications. You probably can reach out to Dr. Bernd Ulmann from Analog Paradigm (there is an acm.org email out there google can find) if you are interested in getting in touch with an experienced analog computer builder/historian who has actual analog computers in his basement.
I opted to use the DIP switches because I could do any parallel combination I want. For the integrators for instance, I sometimes need a 1M resistor around the feedback capacitor(s), regardless of which ones are selected.
Go to About on the channel page and under details click view email address. I can't message you personally on RUclips and I don't want spammers finding my personal email in the comments so just use the business inquiry one for now.
Really cool project! Thank you for sharing this with the world! 🙏
Thank you for making this. I found it super clear to follow being someone who REALLY doesn't know what they are doing. 😜
Dunno a whole lot about this but all i know is that its very cool! And you could be doing very well for yourself if you continue working with stuff like this! Analog computers may have a uprising here!
I think you're right
Impressive work, really cool
Where are the calculus part (the Best part ) ?
extraordinary work thanks
This is wonderful, thanks for sharing this
this needs more viewsss.
Superb work I’m also developing an analog computer trainer but I’m having issues with the inverting and multiplier circuit
Thanks, I was also having trouble getting the multiplier to be linear.
Hello dude, so awesome explanation, you are so smart bastard, thanks for sharing your knowledge.
I have a question... what electrónic component (or thing) is needed for put in your board so that the computing turns into digital??
To get something like this to interface with the digital realm, we would need most of all a DAC (Digital to Analog Converter) and perhaps some kind of USB interface or microcontroller to connect to a PC. Ideally, the DAC should have enough resolution to accurately capture and render the various waveforms as raw data.
Very cool.
Can you please comment on how the disks were rotated with sinusoidal frequencies in the mechanical integrators?
Very cool man
I don’t think it’s Turing complete. It doesn’t seem to have a JumpIf instruction
Can I ask you if you can share the project schematics?
This is excellent work. I'm trying to build something similar myself. I believe the quick explanation of the multiplier circuit is it computes via exp(ln(x) + ln(y)) = x*y.
Have you gotten good results with the differentiator? My understanding is the bandwidth is too high so people would combine it with an integration topology on the same op amp.
I've read about standalone differentiators being not accurate enough to be usable. I think I see why now. The one I made seems to work okay at low frequencies (< 100Hz) but has some weird problems at higher frequencies; the waveforms don't look right. I made R adjustable with the DIP switches but it might be better if C is adjustable or both. Basically, it does work but it seems to be very situation specific.
The multiplier didn't work as good as I thought it would. It multiplies, but only half wave. Anything above zero volts gets cut off. I wonder if "full wave" logarithmic amps would be better. In fact, I bet a Gilbert Cell circuit or something might be more linear.
Hey A. S. (?),
I'm in contact with a company looking to build an analog computer on a chip. However they plan to build an analog computer for students first.
Are you in the located in the continental EU?
What is the largest analog computer you could plug together with the components you have?
In traditional analog computers the use of differentiator was frowned upon because one just could not implement them as the transfer function of an differentiator is not implementable in physics. Any circuit i wanted to use an differentiator for i could usually reformulate it using an integrator instead.
Have you looked into crossbar switches or to turn your analog into a hybrid computer but adding some D/A A/D converters?
Interesting, I didn't think anyone really cared enough about these to develop them any further. I am not located in the EU, I'm in the USA. This prototype I made was meant to inform me of which parts of the computer were useful and how they should be implemented. I made a differentiator to test its viability and soon realized why they didn't use them back in the day. I think a multiplier would be super useful but implementing it cheaply and accurately may not be easy. Expo and log amps might be useful too. Some oscillators or PLLs would be something I'd like to add. Making the mixers have voltage controlled gain would be great too.
I ultimately plan to make a massive computer with multiple circuit boards. Instead of a patchbay, each section will be connected with analog switches controlled by some kind of microcontroller like an Arduino Due. The mixers will have digipots for gain control and CMOS switches to connect everything. You would then program the Arduino for the equation and be able to change parameters as easily as the code can implement changes. There would perhaps be some kind of DAC with and SD card so the computer would have "storage" or "memory" like a digital computer. It could then recall certain functions back into the circuit as the code directs. I would also include extra scaling amplifiers, instrumentation amplifiers (for sensors) and motor drivers for interfacing with the outside world. Similar to an Arduino, it would effectively be an analog Swiss Army knife.
As far as the size of computer, the sky is the limit. The trick is keeping it reasonable. This little prototype was a lot to handle. Implementing the solid-state switching might get complicated.
@@thetuberoaster8321 Cross bar switches, also called Switch matrices are an have 2 groups of n inputs and can many any input to any output. They also have some digital input used to program the mapping in real time. The form factor usually quiet small compared to patch panels and there are non expensive ones available.
If you are not aware of them they might be neat for your applications.
You probably can reach out to Dr. Bernd Ulmann from Analog Paradigm (there is an acm.org email out there google can find) if you are interested in getting in touch with an experienced analog computer builder/historian who has actual analog computers in his basement.
To bad you didn't decide to use a rotary switch, rather then those mini-switches.
I opted to use the DIP switches because I could do any parallel combination I want. For the integrators for instance, I sometimes need a 1M resistor around the feedback capacitor(s), regardless of which ones are selected.
Next up a programmable analog computer
Yes, exactly
I'm just waiting for that to happen one of thies days
@@Scudmaster11 It's in the works but I've been really busy lately.
If it's a quantim computer that's not really analog... it's more like a collapsing analog value into binary "not 0 and not 1"
Can I have your email
Go to About on the channel page and under details click view email address. I can't message you personally on RUclips and I don't want spammers finding my personal email in the comments so just use the business inquiry one for now.