24 randomly switching transistors

[jpoopdog]

hi,

I wish to create a visual effect using a canvas painted in thermochromic pigment, which will require 24 heated pads randomly switching on and off, to turn the thermochromic pigment, clear or black respectively.
is there any IC which can randomly output and do this? it doesnt need to be all 24 going randomly, it could be 12 with half having simple delay in place to turn the second or even thrid transistor on later, at least giving the illusion of it being random.

alternatively, two out of phase sequencers that run through each transistor at two different frequencies so they never match up, again, giving the illusion that they are occuring randomly.
i considered just having simple flasher circuits on each transistor, but that wont be as effective as i want there to only be 1 or 2 pads active at a time, and even so, its too basic and i want to needlessly challenge myself.

so please, if anyone knows of some method or IC that will randomly switch one output on at a time (in which case it doesnt matter wether or not the time on/off is random), or a 12/24 pin sequencer IC, please let me know. Prefferbly though ide like the random output instead of the sequencer.

thanks

 

[MikeMl]

24 bit shift register and an XOR gate.
Use **broken link removed** that have parallel outputs.

 

[jpoopdog]

i looked into it and it seems perfect!
though i havent a clue how it would operate (how i would apply it and how i would operate it), also , using 3 in parallel is in order to get 3 outputs right?

thanks though!

 

[MikeMl]

No, use three 4015 8-bit SRs in series to get 24 randomly switching outputs.

Did you follow the link to **broken link removed**?

 

[jpoopdog]

no that link didnt appear until now. ok i think i get it now. in the setup with the schmidt trigger, the register goes through each of its pins, and for every additional register i add on it will go through those also.
but they will go through the outputs one at a time, one by one in a repeating sequence. or will it actually be pseudo random?

 

[MikeMl]

no that link didnt appear until now. ok i think i get it now. in the setup with the schmidt trigger, the register goes through each of its pins, and for every additional register i add on it will go through those also.
but they will go through the outputs one at a time, one by one in a repeating sequence. or will it actually be pseudo random?

The Schmitt trig in the Doctronics circuit is used as a free-running clock. Some of the 24 bits will change on every clock edge. Once the sequence is well established, any given bit of the 24 outputs could turn on, turn off or stay the same on every clock edge. One the sequence gets going, any given bit of the 24 could be on or off with roughly a 50% probability...

I have some questions for you. How frequently do you want a new bit? Once per second? 1000 times per sec?
What is the characteristics of the thermochromic pixel? Ohms resistance, current to operate, voltage drop to operate?

I have found this chip which has integrated 20mA open-drain drivers which might simplify your circuit.

 

[4pyros]

What is the current and voltage needed for the "heated pads"?

 

[jpoopdog]

well im going to be building the pads myself, so they use extremely minimal amounts of power, they will be 2-3cm2 and will be designed to get temps of at most, 40 degrees. i think ill be limiting the current going into the pads at 1-2A

the kind of output im looking for is adjustable depending on the ambeint temp, for instance, when air temp is 30 degrees, it can take several minutes for the pigment to turn black again, whilst at 25 it happens within a few seconds ideally ill be wanting the pads to switch at 10 second intervals. the voltage theyll be using is either 7 or 12v, depending on wether a 12v nimh 3000ma battery pack or 7.2v li-ion battery of unknown capacity, is able to supply enough current to quickly heat the pads up, and reliably.
eventlually though, i might be making another which will operate with peltier pads, which ill be running using mosfets.

 

[MikeMl]

So how many of the 24 pads are you expecting to turn on at the same time? The psuedo-random bit generator I pointed you to will at some time turn almost all the pads at the same time. Where are you going to get ~40A to turn on that many pads? You will need a power supply capable of ~1/4kW!

 

[ccurtis]

With power consumption like that, I would consider multiplexing the array so that only one pad is activated at any instant, then use a single-bit random bit generator.

 

[jpoopdog]

ok so how do i go about doing that then?

 

[MikeMl]

ok so how do i go about doing that then?

Start by redefining the problem. Now you want 24 outputs, each capable of supplying 12V @ 2A, only one of which is turned on at any given period, which switch randomly to the next output, such that the switching sequence from output to output appears random.

Remaining questions: How long to dwell on each output? Should that dwell period be equal for all outputs? Should the dwell period be random? If so, what is the average dwell? Min dwell? Max dwell?

 

[jpoopdog]

no it would be the same for each output. ideally ide like to be able to adjust the period though to around 10 seconds. once heat is applied it will retain some heat and stay active for a while. at any time, depending on how hot it gets, there will be one or more spots getting clearer (the latter all loosing momentum as the temp begins to drop below their activation tempurature), and all the rest of the screen should therefore be constantly varying in darkness depending on how the heat sinks off them and other factors.

anyway, power supply is super simple, all i need to worry about right now is making sure only one can be active at a time, for a set static on/off time, ideally ide like to be able to adjust the duty cycle with a potentiometer so i can establish a dial or something to change the settings depending on what the air tempurature is, but otherwise to make sure i dont have to replace parts to change that frequency. though if the duty cycle is locked, thats fine, then ill just have to adjust the overal frequency or raise it so much that each pad must be activated like 5 times each second before it can fully go clear.

ANything is possible, but i still need to limit it to one at a time.

 

[MikeMl]

Ok, I have an idea on how to do this. It still utilizes the linear feedback shift register (LFSR) consisting of an 8 bit cmos shifter and an adjustable clock oscillator. By selecting 5 of the 8 available output bits, connecting them to a 5-to-32 line decoder, that will give you 32 mutually-exclusive outputs, of which you can (randomly) wire 24 to your pixel drivers...

I'll work on this later today.

 

[jpoopdog]

cool thanks.