Control of 1000 servo motors simultaneously

[ClydeCrashKop]

How about a 32 X 32 array of photo transistors stuck to a CRT monitor screen.
You could control 1024 solenoids on or off with video.
Maybe with video into the computer of the tennis ball’s position, your program could move the ball in patterns. Circles, squares, play pong, etc…

 

[Pommie]

I don't understand how a matrix of servos could be controlled either. You would at least need an AND gate at each node. I suppose you could use diode logic with pullups but that still requires 32 simultaneous pulses which would be an interesting circuit in itself.

Mike.

 

[dougy83]

I don't understand how a matrix of servos could be controlled either. You would at least need an AND gate at each node. I suppose you could use diode logic with pullups but that still requires 32 simultaneous pulses which would be an interesting circuit in itself.

If you're referring to my post, the AND gate (a transistor) is shown, and yes it is required for each servo. The simultaneous pulses for an addressed row is easy to accomplish either using 32 outputs from a uC, or 32 outputs from a SIPO shift register. The fact that the resolution of the pulse width required for this application is coarse makes the pulse production trivial. If you have a 5MHz SPI clock, then you can have 156 positions between 0° and 180° for 32 outputs.

 

[ronsimpson]

https://www.amazon.com/TowerPro-SG9...=UTF8&qid=1411568791&sr=8-7&keywords=rc+servo

Here is a $3.00US servo. That eats up $3000 of your budget.

 

[Wade Hassler]

I earlier said that an FPGA would work fine......

Because there was one of these right in front of me:
https://www.aliexpress.com/item/EP4...-Board-Core-Kit-free-shipping/1675716755.html
and I was learning about Verilog 'generate' syntax, I typed up a controller for manymany servo-motors.

- Receives data from from USB in funky-but-straightforward format using one of these
https://www.sparkfun.com/products/7841
- Controls as many servos as the FPGA has available pins (less the USB, clock etc inputs)
- Synchronous-serial output to other FPGAs for I/O expansion

The board above has enough I/O for 56 outputs, and logic resources are 21% used. Adding something like this :
**broken link removed**
... would allow adding 90+(?) outputs for about 25¢ US each.
(Note that the latest version 14.0 of Altera tools has dropped support for Cyclone II: you'd need version 13.0 or earlier.)

All in all, this would make your hardware-only cost about 12 boards@$30 = $360
Wiring, carrier boards and I/O protection (a single level-translator 3.3-5V on each output) would add at least $150
(..... says the guy with no money at risk.)


This is only one way to do it, but the low hardware-outlay and wiring simplicity deserve consideration.

 

[dougy83]

Using a microcontroller will be simpler for most people, and it's generally easier to get help with programming in C vs Verilog or VHDL. Using a microcontroller development board, the following costs apply (postage included):
1x Arduino (ATMEGA) or Tiva C Launchpad (ARM): ~$15 or ~$13 (Launchpad is $17 from au.element14.com)
10x serial in parallel out shift registers: $1.70 (au.element14.com)
1000x transistor matrix decoder: $26 (surface mount SMD), or $42 (leaded/through-hole TH) (au.element14.com)
Expander PCBs: $14 (https://dirtypcbs.com/) or DIY
Transistor decoder PCBs: $28 (https://dirtypcbs.com/) or DIY or protoboard
1000x SG90 servo: $1845 (**broken link removed**)
Wiring: depends on layout.

 

[Nathaniell]

Still absorbing all the input.. Thanks all!

 

[Pommie]

PCBs: $14 (https://dirtypcbs.com/) or DIY

Thanks for that, very useful to know.

mike.

 

[koolguy]

Hello Nath.,
What is the need of using
Control of 1000 servo motors

 

[MrAl]

Hi,

Something else to think about...
The ARM SAM 3 will give you more than 70 outputs per board.
That means you'd need 14 such boards for the slave controllers, one more for the master.
The master takes the human input, sends control data to the slaves, the slaves drive whichever actuators are used.

 

[Pommie]

The other things to think about,
If you read the thread you will have your answer.
It's over 1 year old.

Mike.

 

[Andrei_D12]

Hi!
From what I have read, you are new to electronics and you want to begin with a circuit that controls 1000 servomotors!!! To be sincerely, this project is both complex and expensive.
Servomotors will eat a lot from your money. And the circuitry is pretty complex.
However I have an idea.
Solenoids are a better choice because these are much simplier. With lots of wire, magnets (neodymium are the best choice) and patience it can be done. But why complicate things with 0 to 180 degrees???
My idea is just like in the diagram. Both methods use solenoids, but each in a diferrent way.
About the price of a solenoid, it is around 2-3 dollars and it's still too much if you want to buy a thousand.
The cheapest way is to make them, and it wouldn't be too hard because they are simple, yet effective for what you want. As a core, you can use iron, or magnets, better neodymium because they are small but strong (and very fragile)!
This is what I'm talking about.

As for the ball, you can use the 2nd method with a metallic ball, or magnetic...
And don't forget, if you want to make them, make sure you have enough copper wire!!!

 

[Andrei_D12]

One more thing...
To check if the application is working, you can use LEDs, laid on a matrix board, like this one:

Then you can use solenoids and, with buffer transistors, it will be done!
You will need a lot of...

 

[Les Jones]

How about using 64 outputs driving 32 x 32 matrix (1024 points.) Each point of the matrix would consist of a 2 input and gate driving the clock input of a flipflop (D type latch) The data input of all the latches would be fed from outut 65 from the micro (With suitable buffering.) If servos were used then the output of each latch would select between two pulse generates. One would have a pulse duration for the zero position. The other would have a pulse duration for the 180 degree position.

Les.

 

[kubeek]

Like I said in another thread with similar fate, look at how old the thread is, which makes me ask, shouldn´t there be some kind of moderation in place to review posts that are replying to old threads before they get shown?

 

[Andrei_D12]

Lots of ideas and suggestions... But this project is HUGE stuff even for someone which has financial support. My
idea was a cheaper modality to do this project, however 1000 servomotors is ridicullously complex. Even 100 is a lot and if you stay and think a little on how many components will be necessary, you will realise that with the money you spend in the project you can buy a new car. 1000 AND gates, the same number of flip flops and actuators... will lead to this!!!
https://encrypted-tbn3.gstatic.com/...NtTaMGJ2ZO8_MzrNcJ2KrflX3ImBcVtXNpXQ2umVMeQZw
I suggest you to make it by using only 25 actuators, see if it works, then extend it!

 

[kubeek]

Does ANYONE actually read the thread before they decide to make a post? It is OVER A YEAR OLD and the author of the original post has never been back since. And the fact it is very old has been said twice already.