Page 1 of 1

Anyone got any experience rigging LiFePol4 batteries?

PostPosted: Thu Oct 05, 2017 2:28 pm
by Cheb
Does anyone know if these are as good as advertised? I'm making a DIY laptop this winter, cheap and powerful, using desktop parts only 8) But I totally want minimum 2 hours of playing Fallout 4 on battery / 10 hours of office while desktop parts are hogs :x

6pcs brand lifepo4 3.2v 20Ah battery 95140240 3.2v cell for lifepo4 36v 20ah battery pack diy ev energy solar power bike coach

P.S. I also want it could be used for manslaughter and survive with only paint scratched. :oops: So 18 mm plywood for the prototype, then we'll see :roll:

P.P.S. If I screw up, do I go up in a fiery explosion leaving my fics forever unfinished? Or are these batteries "it only burned straight though my leg" safe?

P.P.S. (GTX-1060 = 10 A * 12V + Ryzen 5 = 5.4 A * 12V) * 1/0.85converter efficiency = 14.5 A at battery's lowest discharge of 15 V or 12 A at reasonable discharge of 18 V - and that's not counting mobo, ssd, monitor & other stuff I have currently no data on. So any battery rated for less that 20A continuous discharge takes a hike. It's better to over-engineer such things. The batteries I listed are rated for 60 A, but I have suspicion these are Chinese Amperes. :x

Re: Anyone got any experience rigging LiFePol4 batteries?

PostPosted: Fri Oct 06, 2017 10:01 am
by Spica75
P.P.S. If I screw up, do I go up in a fiery explosion leaving my fics forever unfinished? Or are these batteries "it only burned straight though my leg" safe?


I believe they tend to be considered "potentially feisty". However, they are much safer than the other common Lithium-ion batteries. They also do NOT normally either melt or explode if overstressed.
Right, here:
https://en.wikipedia.org/wiki/Lithium_i ... ery#Safety

So any battery rated for less that 20A continuous discharge takes a hike. It's better to over-engineer such things.


Very much so, don't forget that batteries tend to degrade sadly fast over time and frequent usage.

The batteries I listed are rated for 60 A, but I have suspicion these are Chinese Amperes.


:mrgreen:

Does anyone know if these are as good as advertised?


I have zero knowledge of the specific one, but generally nowadays, unless you're unlucky or buy ultracheap budget batteries, they at least tend to hold up to their rated specs or close to it.

Re: Anyone got any experience rigging LiFePol4 batteries?

PostPosted: Fri Oct 06, 2017 12:22 pm
by Cheb
unless you're unlucky or buy ultracheap budget batteries,

Hmm, maybe I should buy a pack of 8 when I only need 6. Would be a tad more expensive but can potentially save a lot of grief.

tend to degrade sadly fast over time and frequent usage.

Hmmm... My DIY controller, my rules. I can normally operate them in the range of 3.0..3.45 V per cell and have a tumbler "go all the way" to allow full range. Should allow them to live that much longer.

However, they are much safer than the other common Lithium-ion batteries.

That's what I hoped for :D After all, there would be 3 kilograms of them in the machine :lol:

P.S. I just tried one interesting "what if". Started my gaming rig with external power unplugged from the GTX 1060 (I plan to re-use it in the laptop). :roll:
Got a black screen with an aesthetically centered white line "PLEASE POWER DOWN AND CONNECT THE PCIe POWER CABLE(S) FOR THIS GRAPHICS CARD" :mrgreen:

Will perform the next experiment in a month, when I have basic circuitry. What would GTX 1060 do if I lower external voltage? AFAIR, GTX460 reacted to an inadequate power supply by throttling itself down but still working perfectly.

Also, there's this great idea: you can throttle Ryzen by disabling its cooling fan if the battery gets low. It will then throttle itself when it reaches the temperature limit.

Re: Anyone got any experience rigging LiFePol4 batteries?

PostPosted: Sun Oct 08, 2017 12:20 pm
by Spica75
Also, there's this great idea: you can throttle Ryzen by disabling its cooling fan if the battery gets low. It will then throttle itself when it reaches the temperature limit.


Depending on the setup, if you want to really try your hand at tinkering, you could make a heatpipe assembly that connects the cpu and gpu to external, large and passive heatsinks.

There's also some heatsinks that already are capable of running CPUs passively, though they're usually of the HUGE sort.
My previous system used a Scythe Samurai(designed to be capable of functioning without a fan) with the Core2 E8400, and even 20% overclocked it ran very fine with the fan off, even if the heat went up about 20C.
And then, the graphics card in that system, a AMD 6770, which uses almost as much wattage as a regular GTX 1060 BTW, had a huge 1+kg passive heatsink that covered all of one side and a bit out the sides, funny thing is that it actually ran cooler than the same card with active cooling. Best graphics card i ever bought.

In short, the only limit is how much work and cash you want to throw at it.

P.S. I just tried one interesting "what if". Started my gaming rig with external power unplugged from the GTX 1060 (I plan to re-use it in the laptop). :roll:
Got a black screen with an aesthetically centered white line "PLEASE POWER DOWN AND CONNECT THE PCIe POWER CABLE(S) FOR THIS GRAPHICS CARD" :mrgreen:


Yeah, even when they don't need the extra power, often they will simply refuse to run without it plugged in.
IIRC, the PCIe port can supply 75W, and as a 1060 tops out around 120W, it's probably not going to be easy to get it running without the extra cable.
The 1050 tops out at 75W, so might be easier to work with one of those if you can't get the 1060 working.

Re: Anyone got any experience rigging LiFePol4 batteries?

PostPosted: Sun Oct 08, 2017 4:04 pm
by Cheb
In short, the only limit is how much work and cash you want to throw at it.

I was actually planning to use water cooling to make it environmentally sealed. To the point you can submerge it and only external ports and maybe cooling fans would fail.

That's a lot more work, though. And lots more points of failure, namely the pump. I'd want to have a reserve pump for hot-swapping, then Peltie-based heat exchanger to run everything cooler when running from mains power, and so on...
In the end I looked at that overcomplicated plan and said: just no. The classic wind tunnel will have to do, it is also much cheaper.

My best idea, I think, was a heat exchanger aped from steam locomotive boiler: a long tube of hot water with tightly packed copper pipes running through it, a fan pumping air through these. With flows in opposing directions for better exchange.
Would be a female dog to make, though, and need to be the length of the entile computer, meaning I'd have to know the finished design to start making it.
Finding relevant formulas and making proper calculations without stupid mistakes would be hard as well. Would it be able to handle 250W max?

The 1050 tops out at 75W, so might be easier to work with one of those if you can't get the 1060 working.

I already have GTX 1060 and wish to save on that expense.
Any available car PC PSUs aren't powerful enough anyway, I have to make my own buck-down converter slaved to it, based on the same Arduino that will be running everything.
My current master plan is like this
1. AC-DC 24V 20A (rated 600W to have sizable reserve, it's a good idea to operate things at 50%, live much longer)
2. Battery (20V nominal) with a safety/balancing board and a charge/discharge controlling MOSFET on top (PWM to stabilize current when charging, fully open when discharging)
3. car ATX PSU rated for 8..28V input
4. Two separate buck-down converters 15..24 to 12V rated 10A, with 5A nominal usage. One for CPU (that extra cable to the motherboard), another one for the external power to videocard. Both slaved to the PSU: they are turned on by presence of its 12V output, plus Arduino only lets the Power Good signal through when these two reach nominal voltage. Arduino cal also lower voltage on these if needed (reqires thta experiment I talked about)

had a huge 1+kg passive heatsink

That seems like a good idea until you remember about hits and vibrations. Even with special supports for heatsinks.

Re: Anyone got any experience rigging LiFePol4 batteries?

PostPosted: Mon Oct 09, 2017 8:03 am
by Spica75
That seems like a good idea until you remember about hits and vibrations. Even with special supports for heatsinks.


Nah that one was a PCIe card and superduper well designed so it was VERY secure. Since it was mounted along the WHOLE gfx card, it actually made the card itself much sturdier as well. You would basically have to bend 2mm of solid aluminium as well as the PCB for that card to break, and that's before accounting for how the fin pattern was laid out both horizontally and vertically, adding even more strength.

OTOH, my current system uses a Noctua U14S, which is 935g with the 15cm fan, and it's pretty well mounted. Of course, it's a stationary PC, so not nearly as much movement and risk of jostling. Vibrations however comes mostly from fans, so once you remove the fans, no worries about that at all. Next most is from HDDs, and you can use dampening mounts for those(and generally should as it improves reliability for the HDD as well).

I was actually planning to use water cooling to make it environmentally sealed. To the point you can submerge it and only external ports and maybe cooling fans would fail.


:shock:

Wow, yeah that would be a nice piece of work.

That's a lot more work, though. And lots more points of failure, namely the pump. I'd want to have a reserve pump for hot-swapping, then Peltie-based heat exchanger to run everything cooler when running from mains power, and so on...
In the end I looked at that overcomplicated plan and said: just no. The classic wind tunnel will have to do, it is also much cheaper.


Good choice. SANE choice! :mrgreen:

My best idea, I think, was a heat exchanger aped from steam locomotive boiler: a long tube of hot water with tightly packed copper pipes running through it, a fan pumping air through these. With flows in opposing directions for better exchange.
Would be a female dog to make, though, and need to be the length of the entile computer, meaning I'd have to know the finished design to start making it.
Finding relevant formulas and making proper calculations without stupid mistakes would be hard as well. Would it be able to handle 250W max?


Dont forget that copper has poor ability to transfer heat to air. Also, blowing air through pipes works well if they're big enough, but the smaller the internal diameter, the more pressure you will need to push air through effectively.
I also think the "opposing directions" would add much less in effectiveness than they would add troublesome complexity.
Would probably be better off having a fan(s) blow from the underside and up through shorter such pipes.
There's also the lesser issue that the air in the pipes would be very poor at sucking up the heat radiated as IR.

Also, have you accounted for how the water density will change as temperature does? It's just 1-2% from 20C to 60C, but water is highly uncompressible, and as unlike a steam boiler you will need to have it all filled up, there will be a fair amount of strain on the design.

Actually, how about you just make a 2kg worth of aluminium to make a sort of classical style heatsink on the outside(edge?) of the laptop, and connect it with heatpipes? Vertically aligned fins to allow automatic airflow.
Not as radical, but definitely easier to make and less risk of missing something important.

And i think you can certainly make it handle 250W, just not sure how hard it will be to get that far. The "oversized classical aluminium" is definitely doable for that. 250W is twice that of my 6770 card, and it ran at 110W, with the GPU rarely going above 70C, meaning that the heatsink has to be cooler than that. When i tried putting fans blowing at it, i dropped it down around 50C, so even basic airflow on a big enough heatsink can get rid of a lot of heat.

I already have GTX 1060 and wish to save on that expense.


Of course. What i meant was that if it refuses to work without the extra power connection, full stop no way to change, you may be forced to consider it.

My current master plan is like this


Well that looks workable, although it's too long since i played around with electrics to give any guarantees. :P
Fun project either way! :D

Re: Anyone got any experience rigging LiFePol4 batteries?

PostPosted: Mon Oct 09, 2017 10:04 am
by Cheb
I also think the "opposing directions" would add much less in effectiveness than they would add troublesome complexity.

I meant "air flows left to right, water right to left" so hottest water first meets hottest air and coolest water matches coolest air directly from the intake. This should have maximized temperature transfer, otherwise you couldn't get outflowing water cooler than the hot air that already absorbed part of the heat.

and as unlike a steam boiler you will need to have it all filled up,

Was already accounted for, the heat exchanger to serve as expansion tank as well, with some air left in it and the inflow nozzle placed in such way (along the central axis, some distance from the wall) that it always stays submerged.

classical style heatsink on the outside(edge?) of the laptop,

LOL, that was exactly what I was thinking, a thin metal plate with two heatsink rows, one outside, encased in a wind tunnel, and one inside.
But even *that* won't do because heatpipes only work above certain temperatures, meaning that the inside will be constantly cooked at 40+C.

I will simply shape the air intake like that no-spill inkwell of yore and hope it holds against any reasonable amounts of splashing. And choose fans for pressure, not CFM.