Alright this looks interesting. I know this is a little early for this, but how will you get the battery to be registered by windows? I was thinking about doing a sort of laptop build in the future (think 2020-2021) but one of the issues I came across while designing is that I didn't know how I would be able to get the batteries recognized by windows.
-
Save 15% on ALL SFF Network merch, until Dec 31st! Use code SFF2024 at checkout. Click here!
This will be my last update until next month. I'm going on a road trip to my wife's hometown and will stay there for the rest of the month.
It actually might be a perfect time to test stuff, but I'm still on the fence about bringing the project along.
Anyway,
I've bought 3 fractal design R3 silent 50x10mm fans...
I swear I've never seen or heard 50 x 10mm fans from Fractal Design before. Is this a new design?
At this point I won't be doing that yet. It's sort of an extended goal once all is running well and met my expectation.
The idea would be to use a BMS with open source programmable microcontroller with usb, and write the code to report SoC and voltage as HID ups/battery class device via usb.
I'm still looking for similar device but with integrated instead of simulated usb, so I don't need to bother with v-usb.
First time I've heard about it is when @Reglar used 2 in tandem with a single slim 92m to cool his 1080ti, also in an S4M-C. He recorded all his noise testing vs 40mm noctuas and I liked the sound profile of the 50mm R3s.
I'd gone for noctuas for the accessories alone (3pin to 2pin, Y cable) but spec wise 2 40x10mm noctuas would just barely provide 10cfm, while also spinning faster @4500rpm compared to the R3s @3500rpm.
This is just spec hunting though, I'll know when I test them out ^^
Thanks for the info. I'm interested because this means tshere's a possibility that I can use them in my MI-6 case.
So I decided to bring along the nitecore UM4 charger and all 18 cells with me through the road trip.
Managed to charge them all here and there within 4 days; only threw away one cell!
Didn't take photos tho sorry...
Here's the breakdown:
My very very old, dark green cells all work! 3 out of 6 are 'poor' cells, according to UM4, but after fully charged only one still shows 'poor'. I took record of internal resistance (IR), averaged over all 4 slots (the reading is hit or miss - and different slot gives different reading - it's just the UM4's weakness apparently) and wrote it on the battery with a sharpie.
One of the purple cells is dead (!), one almost dead (must be the dead cell's pair) but revived successfully. I'm really suspicious of them - all of them registers 120 mOhm IR, across all slots. They also drop down to ~4.14V once out of the charger then put back in again, but fully charged within 15 seconds or so. Very curious.
The lastly procured cells are panasonic's light green cells. All is good, IR reading varies but generally well below 200 mOhm which is good.
I plan to let them stored for a week or two then measure their self discharge level, to weed off two worst cells.
After that, to discharge them all to 3V and then charge them to full to record their capacity.
I did wonder if I should just discharge to 3.2V to test for capacity I will actually use, but to think that after being packed, we can't guarantee that all cells in series will, on load, fall at the same rate anyway, I figured that would be useless to me anyway...
And this is one thing I learned: if you plan on recycling laptop cells, just do yourself a favor and get Opus BT-C3100 or better 'analyzer' Li-ion charger. They can do all that automatically (charge - discharge - charge, record capacity and IR), and in case of the Opus one, it can also balance your cells once they are made into a pack.
Managed to charge them all here and there within 4 days; only threw away one cell!
Didn't take photos tho sorry...
Here's the breakdown:
My very very old, dark green cells all work! 3 out of 6 are 'poor' cells, according to UM4, but after fully charged only one still shows 'poor'. I took record of internal resistance (IR), averaged over all 4 slots (the reading is hit or miss - and different slot gives different reading - it's just the UM4's weakness apparently) and wrote it on the battery with a sharpie.
One of the purple cells is dead (!), one almost dead (must be the dead cell's pair) but revived successfully. I'm really suspicious of them - all of them registers 120 mOhm IR, across all slots. They also drop down to ~4.14V once out of the charger then put back in again, but fully charged within 15 seconds or so. Very curious.
The lastly procured cells are panasonic's light green cells. All is good, IR reading varies but generally well below 200 mOhm which is good.
I plan to let them stored for a week or two then measure their self discharge level, to weed off two worst cells.
After that, to discharge them all to 3V and then charge them to full to record their capacity.
I did wonder if I should just discharge to 3.2V to test for capacity I will actually use, but to think that after being packed, we can't guarantee that all cells in series will, on load, fall at the same rate anyway, I figured that would be useless to me anyway...
And this is one thing I learned: if you plan on recycling laptop cells, just do yourself a favor and get Opus BT-C3100 or better 'analyzer' Li-ion charger. They can do all that automatically (charge - discharge - charge, record capacity and IR), and in case of the Opus one, it can also balance your cells once they are made into a pack.
After 2 weeks or so I test the cells again for self discharge: another purple cell kicked the bucket; showed up less than 2V, and wouldn't charge. All other purples were at abysmal 4.03V.
Green cells are all fairly healthy, most stayed at 4.11V while one dropped to 4.09V. Light green cells are quite random with one cell each at 4.03V and 4.05V, three cells at 4.07V, and one at 4.13V.
For guidelines, good 4.2V cells drop to 4.15V within 24 hrs, then drop again 5% in capacity within a month. That 5% usually means ~0.2V or so.
So this results are not too good, but it is to be expected of such old and used cells.
And I guess sony does use good cells back then. Pretty sure they're low capacity tho.
I've got myself this discharger/capacity tester:
But mine came with 8.2KiloOhm instead of 8.2Ohm (ideally 2.2Ohm or less) so it was discharging too slowly (~0.5mA) that I gave up. I'll get a proper resistor when I come back home in couple days.
In the mean time I designed and sent out the enclosure to be lasercut on 2mm acryllic:
5 cross sectional parts, that long one is the 'back' panel, one long L shaped 'top' panel, 2 piece 'front' panels, and 2 side panels with fans on them.
No bottom panel since the bottom was meant to serve as ac cable run.
It's not perfect - the twin fan side panel is only fixed to the back panel - simply too tight for tabs. Thinking just superglue it to adjacent panels.
I spotted some design mistakes but it's been sent out already. Not that it's un-fixable after cutting.
Should arrive today ^^ exciting!
Green cells are all fairly healthy, most stayed at 4.11V while one dropped to 4.09V. Light green cells are quite random with one cell each at 4.03V and 4.05V, three cells at 4.07V, and one at 4.13V.
For guidelines, good 4.2V cells drop to 4.15V within 24 hrs, then drop again 5% in capacity within a month. That 5% usually means ~0.2V or so.
So this results are not too good, but it is to be expected of such old and used cells.
And I guess sony does use good cells back then. Pretty sure they're low capacity tho.
I've got myself this discharger/capacity tester:
But mine came with 8.2KiloOhm instead of 8.2Ohm (ideally 2.2Ohm or less) so it was discharging too slowly (~0.5mA) that I gave up. I'll get a proper resistor when I come back home in couple days.
In the mean time I designed and sent out the enclosure to be lasercut on 2mm acryllic:
5 cross sectional parts, that long one is the 'back' panel, one long L shaped 'top' panel, 2 piece 'front' panels, and 2 side panels with fans on them.
No bottom panel since the bottom was meant to serve as ac cable run.
It's not perfect - the twin fan side panel is only fixed to the back panel - simply too tight for tabs. Thinking just superglue it to adjacent panels.
I spotted some design mistakes but it's been sent out already. Not that it's un-fixable after cutting.
Should arrive today ^^ exciting!
I'll most likely do a second version once fitting and testing is done - want to match the holes to skyslots for airflow.
This enclosure is very tight, I had to remove 2 extra cross section panels since it's too long to fit.
Well we'll see when we get there ^^
This enclosure is very tight, I had to remove 2 extra cross section panels since it's too long to fit.
Well we'll see when we get there ^^
The panels arrived!
Here's with sticker backing intact...
Here's without...
I overestimated panel width - can go tighter like .2mm on some holes/notches.
As expected the top panel needs tweaking - the part that hits the side fan panels would need to go.
The cross section panels that hold 6 cells need a notch at the edge.
Wouldn't be hard to fix with some pliers and cutters d^__^b
Here's with sticker backing intact...
Here's without...
I overestimated panel width - can go tighter like .2mm on some holes/notches.
As expected the top panel needs tweaking - the part that hits the side fan panels would need to go.
The cross section panels that hold 6 cells need a notch at the edge.
Wouldn't be hard to fix with some pliers and cutters d^__^b
There's a bit of a bad news however..
I actually received some proper 2.2Ohm 10W resistors to capacity-test my cells today.
Which I did...
On full, starting current was 1.4A something, after 5 mins or so I noticed cell voltage has dropped to 3.3V while it only discharged around 150mAh.
My fear was proven - at 3V cutoff voltage I set - the thing stopped. It was only expending ~500mAh.
But the voltage settled at 3.7V. So I set lower cutoff voltage at 2.7V, and start discharging again.
It stopped again at ~350mAh expended.
Set to 2.5V, stopped at another ~150mAh but settled at 2.7V.
So what's happening here?
It's not the capacity - I knew that ~1000-1500mAh of usable capacity is to be expected.
I forgot that battery voltage drops on load. The bigger the load the lower the drop.
What's the implication? Well, my target lower output voltage is 16V which is 3.2V individual. This meant on load I'd probably only get 1 minute then the whole pack would get below 16V, shutting down dynamo. This'll happen even though the pack still has juice in it.
The solution is to go back to drawing board and go 6S (2.67V - 4V) and 18 cells (6S3P). I do have the room for it, and not like it's any less air cabin legal.
There is a problem with going 6S though, the charging circuit in my dc-ups is a buck circuit which needs 1V higher input voltage than pack's maximum charging voltage.
EPP-200-24 can actually be set to output anywhere from 22.8V to 25.2V, so say at 25.2V it can charge the pack at 24.2V or 4.03V per cell. No problem there, but the dc-ups doesn't regulate output so in the state where mains power is used, 25.2V goes to dynamo directly.
But restricting to 24V means 23V charging, 3.83 individual cell voltage --> roughly 60% SoC. Now that is bull.
Guess I'm gonna throw larry an email to ask whether dynamo mini can handle that high of a voltage.
I actually received some proper 2.2Ohm 10W resistors to capacity-test my cells today.
Which I did...
On full, starting current was 1.4A something, after 5 mins or so I noticed cell voltage has dropped to 3.3V while it only discharged around 150mAh.
My fear was proven - at 3V cutoff voltage I set - the thing stopped. It was only expending ~500mAh.
But the voltage settled at 3.7V. So I set lower cutoff voltage at 2.7V, and start discharging again.
It stopped again at ~350mAh expended.
Set to 2.5V, stopped at another ~150mAh but settled at 2.7V.
So what's happening here?
It's not the capacity - I knew that ~1000-1500mAh of usable capacity is to be expected.
I forgot that battery voltage drops on load. The bigger the load the lower the drop.
What's the implication? Well, my target lower output voltage is 16V which is 3.2V individual. This meant on load I'd probably only get 1 minute then the whole pack would get below 16V, shutting down dynamo. This'll happen even though the pack still has juice in it.
The solution is to go back to drawing board and go 6S (2.67V - 4V) and 18 cells (6S3P). I do have the room for it, and not like it's any less air cabin legal.
There is a problem with going 6S though, the charging circuit in my dc-ups is a buck circuit which needs 1V higher input voltage than pack's maximum charging voltage.
EPP-200-24 can actually be set to output anywhere from 22.8V to 25.2V, so say at 25.2V it can charge the pack at 24.2V or 4.03V per cell. No problem there, but the dc-ups doesn't regulate output so in the state where mains power is used, 25.2V goes to dynamo directly.
But restricting to 24V means 23V charging, 3.83 individual cell voltage --> roughly 60% SoC. Now that is bull.
Guess I'm gonna throw larry an email to ask whether dynamo mini can handle that high of a voltage.
Larry mailed back within minutes - f**kin awesome guy. I asked what's the absolute maximum voltage the unit can take.
AND..... 30 V !!!!!
I'm cranking that EPP-200-24 to the max!
@Choidebu you can use the openups from minibox to make the batteries detectable by windows. Here's the link: http://www.mini-box.com/OpenUPS
I did mention OpenUPS up there; basically an expensive all-in-one solution, but limited to 120W.
is there something else like that but a bit cheaper? That is really interesting and enticing, ive been looking for something like that for a while now
Last edited:
Finished testing the green cells; there's some sort of fuckery going on with either the tester's or the nitecore's coulomb counting - I discharge more than I charge. I redid the test on couple cells and the result is fairly consistent, so I'll dismiss it as measurement differences.
With that I'm grouping closest cells together as parallel: 2,5,6 and 1,3,4.
All tests are done by fully charging a cell to 4.2V, then discharging and coulomb counting with 2.2Ohm resistor (starts out 1.4A down to about 1A) to cutoff voltage of 2.5V, then charging and coulomb counting with the nitecore. Charging itself is CC-CV with about ~800mA on CC phase, just because that's what my supply to the nitrcore is capable of.
I've started tests on light green cells and it doesn't look too good. They all have good IR but the voltage drop under load is crazy high. This is just 1.4A - I planned for peak current of 2.4A across all cells.
So far, first couple of light green cells only discharges below 1000mAh before they cutoff at 2.5V. Had to let them settle first and restarting the discharge to squeeze out additional 200mAh or so.
Alright for testing with normal load I guess.
Gonna have to look for some more used 18650s, reckon I'll need 5 more.
Next up, fitting the battery enclosure in the case, marking up the parts to be adjusted. I didn't take into account that there's a 120mm bracket on the back which means the flange extends alongside of it.
Last time I underestimated the lasercut acrylic's tolerances though, so I'm fairly confident I can tighten up a mm and a half.
Ooh, meanwell hookup cable package from @Thehack arrived! Gonna have to rewire them though, too short -_-!
| Cell # | Discharge (mAh) | Charge (mAh) | IR (mOhm) |
| 1 | 1622 | 1446 | 136 |
| 2 | 1705 | 1557 | 182 |
| 3 | 1593 | 1549 | 173 |
| 4 | 1519 | 1380 | 191 |
| 5 | 1955 | 1817 | 162 |
| 6 | 1858 | 1667 | 215 |
With that I'm grouping closest cells together as parallel: 2,5,6 and 1,3,4.
All tests are done by fully charging a cell to 4.2V, then discharging and coulomb counting with 2.2Ohm resistor (starts out 1.4A down to about 1A) to cutoff voltage of 2.5V, then charging and coulomb counting with the nitecore. Charging itself is CC-CV with about ~800mA on CC phase, just because that's what my supply to the nitrcore is capable of.
I've started tests on light green cells and it doesn't look too good. They all have good IR but the voltage drop under load is crazy high. This is just 1.4A - I planned for peak current of 2.4A across all cells.
So far, first couple of light green cells only discharges below 1000mAh before they cutoff at 2.5V. Had to let them settle first and restarting the discharge to squeeze out additional 200mAh or so.
Alright for testing with normal load I guess.
Gonna have to look for some more used 18650s, reckon I'll need 5 more.
Next up, fitting the battery enclosure in the case, marking up the parts to be adjusted. I didn't take into account that there's a 120mm bracket on the back which means the flange extends alongside of it.
Last time I underestimated the lasercut acrylic's tolerances though, so I'm fairly confident I can tighten up a mm and a half.
Ooh, meanwell hookup cable package from @Thehack arrived! Gonna have to rewire them though, too short -_-!
Last edited:
More bad news: not related to the battery project, but my revered laptop lcd is unusable. The board to drive it arrived but turns out my ccfl is dying - one second on then it's off. In spur of moment I ordered an led replacement kit, not expensive at 8$, but then I did my homework and dissassemble the lcd to see how the panel shows under light.
It wasn't good - bubbles in the panel, blackening at the edges, discoloration throughout. Time (and bad storage condition) has taken its toll.
I'm thinking of making a glorified led lamp light once the backlight kit arrived, just with the frame and light reflection + diffuser layer.
It wasn't good - bubbles in the panel, blackening at the edges, discoloration throughout. Time (and bad storage condition) has taken its toll.
I'm thinking of making a glorified led lamp light once the backlight kit arrived, just with the frame and light reflection + diffuser layer.
I've got some time today and actually finished testing the light green, Panasonic CGR18650CG cells. The reason that makes it so quick, is not a good one. Here it is:
These should be 2250mAh cells. Out of factory IR should be around 65 mOhm.
They charge and discharge in prob half the time of my dark green ones.
They're in bad shape, but usable for testing. I think I'll do 1,5,6 and 2,3,4 parallel groups.
Now I also had time to do the purple cells - in which I am at loss of words.
They drop to 3V immediately on 1.2A load, then goes to 2.5V in seconds. I had to drop the cutoff voltage to 2V to even test them properly. Even then, they only register 60-90mAh. I tried overcharging them to 4.3V (who knows? they might be those rare cells), which charges them for an additional 5 or 10 mAh.
Not to mention one of them gets quite hot on discharge - a red flag.
So right now, I can do 5S3P with one questionable parallel group, or 6S2P with risk of drawing peak load of almost 4A across all cells.
Thoughts? I might be able to do both...
My prediction is at first load, 5S3P will drop to 17.2V (3 + 2 * 3.3 + 2 * 3.8) immediately, then to 16V in 5 mins. Then in another min or so either it's dead or the dynamo can cope that low (why didn't I ask its absolute lower boundary to Larry?).
With 6S2P, it'll go well, but nothing should crank it up above 110W mark or I risk overheating them cells. A youtube test might work..
The only decision I made is, this time around I won't be using a bms since it'll likely only worsen the pack condition because it is mismatched from the start.
| Cell # | Discharge (mAh) | Charge (mAh) | IR (mOhm) |
| 1 | 560 | 682 | 182 |
| 2 | 696 | 631 | 161 |
| 3 | 881 | 801 | 147 |
| 4 | 744 | 777 | 164 |
| 5 | 760 | 631 | 190 |
| 6 | 455 | 593 | 120 |
These should be 2250mAh cells. Out of factory IR should be around 65 mOhm.
They charge and discharge in prob half the time of my dark green ones.
They're in bad shape, but usable for testing. I think I'll do 1,5,6 and 2,3,4 parallel groups.
Now I also had time to do the purple cells - in which I am at loss of words.
They drop to 3V immediately on 1.2A load, then goes to 2.5V in seconds. I had to drop the cutoff voltage to 2V to even test them properly. Even then, they only register 60-90mAh. I tried overcharging them to 4.3V (who knows? they might be those rare cells), which charges them for an additional 5 or 10 mAh.
Not to mention one of them gets quite hot on discharge - a red flag.
So right now, I can do 5S3P with one questionable parallel group, or 6S2P with risk of drawing peak load of almost 4A across all cells.
Thoughts? I might be able to do both...
My prediction is at first load, 5S3P will drop to 17.2V (3 + 2 * 3.3 + 2 * 3.8) immediately, then to 16V in 5 mins. Then in another min or so either it's dead or the dynamo can cope that low (why didn't I ask its absolute lower boundary to Larry?).
With 6S2P, it'll go well, but nothing should crank it up above 110W mark or I risk overheating them cells. A youtube test might work..
The only decision I made is, this time around I won't be using a bms since it'll likely only worsen the pack condition because it is mismatched from the start.
Last edited:
Did fitment of cells on the enclosure..
Dark green fits perfectly, purples are too tight. Didn't test light greens. Reckon I'll enlarge the holes' diameter by .25mm
Before chassis fitment the pieces are superglued to held them together, and fans are ziptied to the respective holes.
Man that was tough. I really thought I had more clearance. That 120mm builtin bracket got in the way but at least the acryllic was flexible enough to be shoved in. Crack a few pieces - that's what prototype is for I guess...
From the cpu side..
The psu got like 2mm to the frame and 3mm or so to the fans. Also, might not easily spotted here, but the single fan that was supposed to cool the cells didn't make it. Clearance between AC inlet and the gpu side was only 46mm.
Here we can see how tight it was - 1mm!
Need a special sequence to put them all in, involving removing the pcie bracket and installing it first.
The absolute permissible gpu card height is 165mm - bummed me out a bit since that ruled out Asus GTX 1650 LP OC at 182mm. Even without fan and acryllic piece, the max height then becomes 180mm - it'll practically hit the psu.
After that I marked the spots to be adjusted.. here's the side with the 120mm bracket:
And the other side plus the intake fan..
You can see where I broke off the top part of the fan section to fit the gpu. The line below marked where a fan wouldn't hit the AC inlet.
Speaking of that single fan...
I have 2 choice :
1. Move the inlet more to the bottom of the case. It's possible, but some fat wall power cable would hit the wrap around bezel. My current one is fine though, up to 3mm outwards. Does involve more cutting of the frame.
2. Just give up 5mm fan and get a 40x10 noctua. Lose about ~1-2cfm.
While at it I also tested out running the rig with the epp, went okay.
Dark green fits perfectly, purples are too tight. Didn't test light greens. Reckon I'll enlarge the holes' diameter by .25mm
Before chassis fitment the pieces are superglued to held them together, and fans are ziptied to the respective holes.
Man that was tough. I really thought I had more clearance. That 120mm builtin bracket got in the way but at least the acryllic was flexible enough to be shoved in. Crack a few pieces - that's what prototype is for I guess...
From the cpu side..
The psu got like 2mm to the frame and 3mm or so to the fans. Also, might not easily spotted here, but the single fan that was supposed to cool the cells didn't make it. Clearance between AC inlet and the gpu side was only 46mm.
Here we can see how tight it was - 1mm!
Need a special sequence to put them all in, involving removing the pcie bracket and installing it first.
The absolute permissible gpu card height is 165mm - bummed me out a bit since that ruled out Asus GTX 1650 LP OC at 182mm. Even without fan and acryllic piece, the max height then becomes 180mm - it'll practically hit the psu.
After that I marked the spots to be adjusted.. here's the side with the 120mm bracket:
And the other side plus the intake fan..
You can see where I broke off the top part of the fan section to fit the gpu. The line below marked where a fan wouldn't hit the AC inlet.
Speaking of that single fan...
I have 2 choice :
1. Move the inlet more to the bottom of the case. It's possible, but some fat wall power cable would hit the wrap around bezel. My current one is fine though, up to 3mm outwards. Does involve more cutting of the frame.
2. Just give up 5mm fan and get a 40x10 noctua. Lose about ~1-2cfm.
While at it I also tested out running the rig with the epp, went okay.
Adjusted the new enclosure design.
- As seen here I opted for 40 mm fans instead of the 50 mm.
- Side panel is enlarged all the way, to act as thermal barrier against the GPU.
- I've had an idea to make cell compartment slide-in and removable, but alas no more space.
- After enlarging battery holder holes and accommodating those bracket decision to push the limits and try ~1 mm clearance in 2-3 spots has to be made. Should be fine considering that the 50 mm fan grill already went to 1 mm in some places.
- Tried incorporating skyslots for airflow, not all the way for the sake of structural integrity
Last edited:
Similar threads
