ADVICE: Circuit Software Needed

Thread Starter

strokebow

Joined Jan 26, 2008
14
Hi All,


I need to simulate a network of resistors.

The network will have approximately 4,000,000 resistors in it. Within the network I will want to find the equivalent resistance at various nodes.

Can anyone please recommend a piece of software that can
(a) do this simulation (i.e. can meet memory requirements)
(b) can do it fairly fast (i.e. Within a day or 2)

I have tried this with LTSpice but it runs out of memory. It complains that cannot allocate enough bytes in to one contiguous block.

Can anyone recommend any alternative software that can complete this task? Are there any alternative versions of spice of other circuit sim software that could possibly handle this?

cheers
 

thatoneguy

Joined Feb 19, 2009
6,359
Doubtful, unless you are on a 64 bit machine and get a 64 bit simulator.

I've just got to ask, why? I am reminded of the xkcd "geek sniping" resistor network.
 

Thread Starter

strokebow

Joined Jan 26, 2008
14
Yes, you are correct - It is very similar indeed.

I have a wire interconnect of about 2000x2000 wires. So I wanted to simulate the response modelling each interconnecting piece of wire as an individual resistor.

So, is there any hope for me?
 

thatoneguy

Joined Feb 19, 2009
6,359
There is a way to calculate the resistance between two points on an infinite grid (the xkcd 1Ω snipe one), between adjacent points, knights cross, etc.

Is that what you are looking for? There are many solutions on google, that cartoon created quite the stir with... well, us geeks... so many people went so far as to make web calculators for for resistances a distance apart, though I don't think they go to 4x10e6. There are the bare formulas posted in many places as well. Google will probably help you out a lot, as one may be closer to what you have in mind other than the nerd sniping grid.

Sorry I am using the "try google" for answer, but not enough information on which data you are looking for makes it hard. Especially when it is a somewhat well known problem thanks to the cartoon.
 

Thread Starter

strokebow

Joined Jan 26, 2008
14
Hi,

Thanks for your response.

Yes, I've seen the solutions on google and I have the bare bones equations for the infinite grid.

However, I was having some issues matching the theoretical/infinite values with experimental ones. Particularly close to the boundaries.

The data I am looking for:
IDEALLY... I would like to have the equiv resistance from a given reference node to every other node within the approx. 2000 x 2000 interconnect.
What I want is to simulate this with circuit simulation software to see if it is matching up with my experimental values.
 

Thread Starter

strokebow

Joined Jan 26, 2008
14
I was thinking and I may possibly be able to get access to a high performance computer.

Would it be feasible then?

If so, which circuit sim software would people recommend and why? Also, what would be the approx amount of memory required?

cheers
 

thatoneguy

Joined Feb 19, 2009
6,359
You would need a 64 bit OS, like Win 7, with a SPICE program compiled and built to work with that number of nodes, I'm not sure if one exists as I haven't looked. There usually aren't 16 million connect point in real world systems.

MATLAB may be able to do the math solution for you, which is another option, if you can write the equations for MATLAB.
 
Top