PS for 3.7v and 9,216 amps

Thread Starter

okcdev

Joined Aug 24, 2023
3
I have some experience in electronics, have written a few firmware programs and designed a couple of basic PCBs but this project requires expertise! I have a future project coming up that has 4608 (yes over four thousand) modules, and each module needs 3.7v and 2amps. These modules are in a tight array physically. Trying to wrap my head around how I will power these things, and in the most efficient manner given the scale. I figure it will involve coming up with a suitable power supply that powers a couple hundred of modules or something. What's my best plan of attack? Hire an engineer to design a power supply? Thanks!
 

joeyd999

Joined Jun 6, 2011
5,161
I have some experience in electronics, have written a few firmware programs and designed a couple of basic PCBs but this project requires expertise! I have a future project coming up that has 4608 (yes over four thousand) modules, and each module needs 3.7v and 2amps. These modules are in a tight array physically. Trying to wrap my head around how I will power these things, and in the most efficient manner given the scale. I figure it will involve coming up with a suitable power supply that powers a couple hundred of modules or something. What's my best plan of attack? Hire an engineer to design a power supply? Thanks!
Ha, 30kW!

Sounds like you're going to need a high voltage bus (or busses, probably) that is stepped down individually by a switcher on each module. Otherwise, I2R loses are going to be excessive, and overall regulation will be impossible.
 
Last edited:

georgf

Joined Aug 12, 2021
9
Hey okcdev, that sounds like a real fun project, for starters you need to define where all this power is to come from, so you have an idea of source voltage level and other characteristics as AC or DC, then on the module side I would put in a high efficiency step-down regulator for each module with all the needed safety features as short-circuit-proof, over-voltage and under-voltage lock out, over-temperature protection. These step-down converter helps to get up the source voltage and lower the current, like ??-to-48V-to-3.8V, that would reduce the current 48V/3.8V=12.6 times as 9216A/12.6=732A which is still a lot, but may be easier to manage, you may go up to 60V bus bar voltage in order to lower the current more, it all depends on what your source level will be, you are playing with a lot of energy, each level MUST be protected, 4608*2A=9216A, these 4608 modules are better sub-divided into smaller groups, for current heavy distribution use copper bars (may be silver galvanized), measure the current for protection and system analysis, each block must be protected, the step-down for each module already is some protection, also allows each module to be controlled by a system MCU which can handle the control, monitoring, system analysis, alarm generation, indication, manual human intervention. You have some big amount of details to monitor, and you cannot be sure that everything at all times will be working fine, you must prepare for the worst case, 35kW are more than sufficient to start a very nice fireworks, so make a good plan and play it safe.
 

georgf

Joined Aug 12, 2021
9
on the other hand you will need a fool-proof test system for all of this, you cannot simply trust everything is working fine, you need to test the limits and functions
 

joeyd999

Joined Jun 6, 2011
5,161
on the other hand you will need a fool-proof test system for all of this, you cannot simply trust everything is working fine, you need to test the limits and functions
Each individual 8 watt module is no big deal.

Obviously the challenge is safe power production, control, and distribution. And, of course, heat (which I didn't mention previously because I thought that was obvious).
 

dendad

Joined Feb 20, 2016
4,424
More details could help. Can you give us a hint of the module function?
Are these modules packed close together or over a wide area?
If outside, it may well be worth distributing the power as AC as that can help slow down electrolysis if water gets into the system. And it will!
Each module needs a switch mode buck converter and Polyswitch over current protection for a start.
And transient protection too.
It is really a bit hard with the limited details you have supplied.
You may be able to just use a transformer, bridge rectifier and capacitor, but we do not know enough to advise you.
 

nsaspook

Joined Aug 27, 2009
12,726
Each individual 8 watt module is no big deal.

Obviously the challenge is safe power production, control, and distribution. And, of course, heat (which I didn't mention previously because I thought that was obvious).
Heat removal and handling details are usually only obvious to those that deal with compact high power systems in the 10's of KW.
40KW of RF power and DI water cooling.
1692923757802.png
 
Last edited:

strantor

Joined Oct 3, 2010
6,771
I don't know what these top secret modules are, how far apart they are, or what kind of distribution topology will be used, but let's say they're in a straight line and 6" apart. You would want the power supply right in the middle to minimize the distance to the furthest ones. So that's a line 0.44 miles long and the longest distance from supply to any device is 0.22 miles and 4608 amps will need to flow each direction. It will break the laws of physics to supply each one with 3.7000 volts exactly; there must be a range. I will arbitrarily make the range 3-4V.

Looking at one side only, the wire nearest the source will be carrying 4608A and after the first device it will be carrying 4606A and after the next 4604 and so on. The wire between each device will act as a low value resistor and will have a voltage drop, meaning the next device has slightly less voltage and so on. So the resistance of the wire needs to be low enough that the devices nearest the center, on either side of the supply will receive 4.0 volts and out at the very ends will get 3.0V.

I'm sure there is an equation to calculate this but I don't math good enough to know it. Instead I made a spreadsheet to figure it out. According to my spreadsheet you can tolerate no more than 0.0000001884 ohms between each device, and this is the total of the positive wire and the negative wire, so each wire will need to be half that, or 0.0000000940 ohms over a length of 6 inches.

0.0000000940 ohms in 0.5ft = 0.000188 ohms per 1,000ft. Ohms per 1kft is the typical unit of resistivity. Let's consult the table below and see what kind of wire meets this criteria:

Screenshot_20230824-225629_Chrome.jpg

So the largest gauge of wire commonly available is size 0000 with 0.049 ohms per 1k feet, which is too small by 26,000%. We need 0.000188 ohms/1kft, AKA 0.188milliohm/1kft. Let's see what other options there are; how about copper Busbar?

There's a nifty online calculator for this:
Screenshot_20230824-230536_Chrome.jpg


So it looks like you'll be needing two busbars, each 1ft x 6in x 0.44 miles. That's 2304 cubic feet of copper. Copper weighs 559lb/cu ft so that's 1,287,936 lbs of copper, which at the current market price of $3.84/lb would cost over $5M just at the commodity price, not including the manufacture of the busbars.

I wouldn't advise paying anyone to design you a power supply until you figure out the more important issue of distribution.

I think this is not as feasible as I think you think this is.
 
Last edited:

crutschow

Joined Mar 14, 2008
33,962
Is there any way to reduce the power that each module takes?
That's a heck of a lot of power and subsequent heat that has to be dissipated.
 

Thread Starter

okcdev

Joined Aug 24, 2023
3
Thank you everyone for all the amazing information, you've all given me a lot to think about and consider, especially the ambient heat being created, if nothing else for HVAC considerations. I will try to reduce module power needs, but even best case scenario it will be in the ballpark of this. Sorry I was unable to provide more details, just am not able to at this time. Thanks again everyone!!
 
Top