• Save 15% on ALL SFF Network merch, until Dec 31st! Use code SFF2024 at checkout. Click here!

Log Project šŸ¦RoadRunner (Battery Powered Portable PC, in S4M-C #524)

Vlad502

Airflow Optimizer
Nov 4, 2017
258
211
Yes, that is correct. Not completed and a bit slow to get all parts
Plan changed back to 11.6" tablet pc, but powered with QD188 PD, because OpenUPS is too expensive. Found later power bank with UPS mode
I will use type c 5A 100W cable with PD converters type c to 7.4x5.0 and 5.5x2.5 to power Thin ITX MB, but still not decided Thin ITX Ryzen APU or CPU 35W with GPU on B360I-P-ITX (need 4pin kycon to barrel)
 
Last edited:
  • Love
  • Like
Reactions: Dapuma and Choidebu

Vlad502

Airflow Optimizer
Nov 4, 2017
258
211
6pin is 12V power output for GPU (i.e.GTX1060)
Does that powerbank support passthrough charging? Couldn't find it mentioned anywhere in the details. Would be nigh perfect if it does.
Unfortunately, no. Tried barrel DC plug with 12V adapter, but no charging - only USB-C charging works, but there is only one port
 
Last edited:
  • Like
Reactions: Choidebu

Choidebu

"Banned"
Original poster
Aug 16, 2017
1,199
1,205
Whoa that is neat. Yep I saw it now.

PCIE-Eę’ę§½1Ɨ16X ę’ę§½ć€1Ɨ6PIN 12V供ē”µč¾“å‡ŗęŽ„å£ļ¼ˆ120Wļ¼‰

Looking forward to a build log!
 

Choidebu

"Banned"
Original poster
Aug 16, 2017
1,199
1,205
Hit the jackpot while sorting through old stuff today. This must be my sister's old vaio's battery pack.



All of then registered ~3.2V, shone bright on bulb test (means good current). Will charge now and capacity test tonight.

Edit: found a datasheet online. It was only rated for 2.2A max discharge. Should be fine eh...
 

Choidebu

"Banned"
Original poster
Aug 16, 2017
1,199
1,205
Lucked out on this one. Discharged 2 cells, 1800 and 1900mAh. Voltage sag was bad that they went 3.3V on 1.4A load, but that just indicates higher IR from age. I'm getting IR readings close to 200mOhm. Nevertheless, very grateful for it.

Edit: With next one post my title would be 'SFF Guru'! So I'm saving it up for a big update!
 

Choidebu

"Banned"
Original poster
Aug 16, 2017
1,199
1,205
So.. that update I promised is here!

I finished capacity testing the new batteries, they _are_ amazing. Well, a bit better than my dark green ones. I'm not listing the capacities here, take a look at the pic below. Half of em 1700mAh, the other 1900mAh. IR seemed to correct itself after refreshed, all register 120mOhm.



After I finished soldering the parallel groups. Not using the purple cells.
Please don't mind my soldering skills - these batteries are NOT meant to be soldered. Had to first sand the poles, flux, tin, then solder the salvaged nickel strips (also fluxed). It gets warm, but nothing dangerous.



Then on to series them.. Again, don't judge me. I put the weakest, light green cells at the outermost in the series so that I can easily remove them if they prove to be a bottleneck.

Note the cardboard and the rubber band to secure them while I solder.



Stacked together with cardboard in between. After this I just wrap them in cellophane to secure the pack a bit.



Voltage's all good ?

After that, I tackled the wiring. Rewired @Thehack 's meanwell cables to my dc-ups. I just happen to shorten my dynamo's eps cables few days ago, so I double down them cables on each output pin on the meanwell. Made life harder because the screw connector is relatively small, but made it work somehow.



For the dc-ups's output I used the cables that theHack sent me as-is since it's bulky enough. That goes straight to dynamo's input.

Andd.....



IT LIVES!



Pack's load voltage in the first minute I unplugged the power cable.



Closer look at how everything is wired. The dangling yellow-black cable doesn't connect to anything, I'm not taking out the hdplex just yet.

It's been 45 mins, and I'm down from 23.5V to 21.5V. The weakest cell groups are still at 3.6V.
Load is writing this up and youtube playing in the background on autoplay.

But it's 5.0AM here, I _really_ need to sleep!
 

Valantar

Shrink Ray Wielder
Jan 20, 2018
2,201
2,225
That looks absolutely awesome! Nice job. I've probably forgotten this from earlier in the thread, but does your UPS board communicate with the PC at all, so that it knows it's being powered by a battery? I remember seeing boards capable of that somewhere (likely some tablet project here on the forums), but I can't see anything except power cabling on yours.
 

Choidebu

"Banned"
Original poster
Aug 16, 2017
1,199
1,205
but does your UPS board communicate with the PC at all, so that it knows it's being powered by a battery? I remember seeing boards capable of that somewhere (likely some tablet project here on the forums), but I can't see anything except power cabling on yours.

No the dc-ups board doesn't do that. OpenUPS by minibox.com does that but it's too expensive for me.

My dc-ups does power path management, charge the battery when needed with proper cc-cv curve and does some minimal protection against under voltage and overcharging.

I'm planning to have the BMS do the battery reporting to the PC.

For this round I opted not to use one, because my earlier concept uses 5S which I bought a BMS for, and it doesn't support 6S.

Originally I was going for Tindie bms with atmega, that is open source, so I can write custom code to have it report stuff over usb.

Aliexpress got some rather expensive boards that can do reporting via uart and bluetooth, but don't think I'll go that route...
 
  • Like
Reactions: Valantar

Dapuma

Average Stuffer
May 18, 2019
72
64
No the dc-ups board doesn't do that. OpenUPS by minibox.com does that but it's too expensive for me.

My dc-ups does power path management, charge the battery when needed with proper cc-cv curve and does some minimal protection against under voltage and overcharging.

I'm planning to have the BMS do the battery reporting to the PC.

For this round I opted not to use one, because my earlier concept uses 5S which I bought a BMS for, and it doesn't support 6S.

Originally I was going for Tindie bms with atmega, that is open source, so I can write custom code to have it report stuff over usb.

Aliexpress got some rather expensive boards that can do reporting via uart and bluetooth, but don't think I'll go that route...
if you ever do get it to report over usb with that, will you upload the source code for it?
 

Choidebu

"Banned"
Original poster
Aug 16, 2017
1,199
1,205
Of course. I'm not much of an embedded programmer myself, so I'd prefer an 80% there approach to it, with libraries and stacks etc.

There might be another way, which is to separate BMS and battery reporting 'module'.

If we think about it the PC only needs the info when it is on, so...
Say we make a device, usb powered, which measures
  • battery voltage
  • battery current flow (in/out)
  • Ac presence
We might get away with simple model of voltage - SoC adjusted with current. A rough approximation. And with usb mcu such as arduino micro or pic16f1455...
 

Choidebu

"Banned"
Original poster
Aug 16, 2017
1,199
1,205
The weekend is here! I'd like to devote tonight and tomorrow to do some testing and benchmarking.

My current gpu is only a placeholder though, R7 240 isn't exactly power hungry at 30W tdp, 50W real draw.

I'm thinking video playback with a 2 hr movie. Then cinebench if it still holds after that.

Anyone have more ideas what sort of test to conduct? I'm not much of a gamer, I've only got diablo 3 and League of Legends installed. And like I said, given my target of 75W gpu like the 1650, if it would matter currently.
 
  • Like
Reactions: fergunat000

Choidebu

"Banned"
Original poster
Aug 16, 2017
1,199
1,205
Did some redesign of the enclosure. I figured it'd be too much hassle and awkward to put 9 cells in 2 acryllic cross sections, _then_ solder/spot weld them. Wouldn't be removable too since the cross sections then pretty much become part of the pack. So..



This new cross section can just be put on afterwards, with spacers for airflow (over-engineer, I know).

Also made protected slots to route cable/nickel strip between left and right pack.

All of these changes have made very narrow parts which I seriously worried whether lasercut, will, pun intended, cut it.
 

ExplodingWaffle101

SFF Lingo Aficionado
May 11, 2018
129
82
This project seems very interesting and cool: if you could get the USB HID Power Device stuff working I'm sure there might even be a market for it (if it were cheaper and less specialised than the OpenUPS I suppose).
 

Choidebu

"Banned"
Original poster
Aug 16, 2017
1,199
1,205
Curiosity got the best of me... tried to cram everything in the case without waiting for lasercut acrylic I just sent out the design yesterday.

First I extend the power cable and mod the new C6 inlet to fit, just like what I did before on the original build, plus putting a brass insert into the inlet with soldering iron.



Insulated the bottom of the epp with some vinyl paper (actually a polarizer sheet from my broken lcd). Tried mounting the fans with zip ties (later removed)



Here's how it look like in the case with the battery. Shouldn't have put that spacer...



To its side, the dc-ups board being held only by mass and tension of cables coming in and out of it.



The money shot.

Drive bracket no longer fit, so I temporarily mount my hdd like so:



Only one screw hole can be fastened; that other one just sort of 'grips' the steel frame edges. Not too bad actually.

With the drive installed:



Look, ma! Only one cable!

 

Choidebu

"Banned"
Original poster
Aug 16, 2017
1,199
1,205
On my todo-list:
  • Battery life test.
  • Figure out how to mount the epp.
  • Figure out how to mount dc-ups, keeping in mind I'd be stacking more boards on top of it like an arduino micro for battery reporting.
  • Order a 40 mm fan.
  • Order a 6S BMS, something tiny.
  • Order an arduino micro, an INA219 breakout board, and one pair of XT60.
if you could get the USB HID Power Device stuff working I'm sure there might even be a market for it (if it were cheaper and less specialised than the OpenUPS I suppose).

That's my plan, to stick with USB hid battery/ups class so that I don't need to write custom libusb driver/software for it.

Determining a battery's State of Charge (SoC), however, is not an easy task to do accurately; I'll stick to current/voltage monitoring, and approximate SoC with a simple model.
 

Choidebu

"Banned"
Original poster
Aug 16, 2017
1,199
1,205
I charged the pack last night standalone with the epp, measured 24.2V open circuit. edit: yeah I messes that up - see next post.

But after I finished cramming everything in, I was going to test it but too tired and fell asleep, as a result the rig idled (and likely put itself to sleep) for like an hour before I woke up and shut it down.

Just started at 3.45 pm with movie playback (where's that 3 hour Return of the King when you need it?), settled in Invictus at 2 hours 10 mins.

After that, I'll check battery voltage and continue with sketchup and youtube audio. Got something to design for the rig.

[16:32] opened up the case to check voltage. Shocked to see 16V :confused: It's only been 45 mins playback - maybe that 1 hour + idle did take toll.

[16:37] shit 13.6V. The dynamo's holding up.. light blue cells are 3.4V, dark green cells are 3V but...

[16:40] pulled the plug on the test and put mains in. Because I found the light green cells shorting out. It was 0.6V. It was getting warm.

Well that's unfortunate.
At least we now know the dynamo still doesn't shut down on 13.6V.

Quite possibly I made a mistake of measuring charging voltage, not open circuit voltage last night. I haven't got around to change the default charging current on the dc-ups which is about 0.2A. That would take ages to charge this big of a pack, ideally it should be 1.5A for about 0.25C. Even then, It's still gonna take like 6 hours at that current. And I only plugged it in for like 4 hours or so, after it has been running for 1.5 hrs on my first test.

A BMS would've taken the whole pack offline when a cell dipped below the cutoff voltage.
 
Last edited: