![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
muwlgrOne unpleasant effect I noticed when running my home PC setup from LR1106, was its instability in powering SATA HDDs.
I tuned my input 12V level so that BIOS reported 12V + a small fraction, but 5V level was barely observed and often fell down to 4.92 or 4.96 V.
May be this, or may be some unwanted variance of its voltages at higher frequencies, caused my SATA HDDs to behave unstably.
One particularly old WD20EARX, bought somewhere in Autumn 2011 (and so, produced most probably before Thailand floods which had affected HDD production facilities), started reporting "SATA link slow to respond/reducing link speed" and similar messages in Linux dmesg, audially this was heard as its motor spinning down and then back up, and one newer WD40EFRX sometimes reported numerous SMART errors collecting in its log, which disappeared from smartctl printout after its full power-cycle (that is, with powering down the whole ATX PSU).
So I put back my older but full-scale FSP ATX400PNF, and SATA HDD behavior restored back to its normal stability.
Also I ordered a different DC ATX PSU from AliExpress, marked RGEEK 300L or LD-A300W, with stated ATX output power of 300W.
Also I finally found a time to attach 2*12V car batteries, each of 62A*h stated capacity, in parallel to my existing 2*14A*h internal UPS batteries.
Had to drill 2 openings in the side wall of the UPS case, and run through them 2*6mm2 cables to connect external batts to the internal ones.
Then found some time to test the runtime of the resulting setup, and got 19h36m which approximately corresponds to ~5* capacity growth (14+62=76A*h at 2*12V, that is up to 1824 W*h DC), correcting this by higher power draw of FSP ATX400PNF, and by the fact that I re-powered my network gear back from their bundled AC-DC PSUs, since I relocated the whole setup into a different corner of the room and put them higher and closer to the room's ceiling.
To get fully charged back, the resulting battery setup took >17 hours, but APC SmartUPS 1000 completed this process without complains.
The resulting voltage settled at 27.58V (well balanced as 13.78..13.79V on individual batts) and did not grow further.
apcaccess reported its usual number 28.1V (who knows how it got this number, but it was made in 1998 after all).
After some time I received my LD-A300W from the post office, and several days later my FSP ATX-400PNF started triggering power surge protection on my ASUS B85M-G motherboard (which is in its 9th year of almost continuous 24*365 operation as my home server).
So I decided that the time has come to put LD-A300W into its production operation (did not even test it preliminarily), and put it into my box instead of FSP.
External 12V DC I took from the same S-120W-12V mentioned at https://muwlgr.dreamwidth.org/49507.html .
Did not even adjust its output voltage, and the BIOS happily reported both +12V and +5V with small added fractions.
The motherboard, HDDs and additional PCIe and USB adapters worked expectedly stable after system boot, and SATA errors observed with LR1106 did not happen again.
Load% reported by apcaccess reduced from 6.3% to 5.2% (a step back from what I achieved at https://muwlgr.dreamwidth.org/50053.html, just need to improve packing of my PSUs and one more battery to more confidently put them under the ceiling).
Battery runtime with the current setup will test in some future.
I tuned my input 12V level so that BIOS reported 12V + a small fraction, but 5V level was barely observed and often fell down to 4.92 or 4.96 V.
May be this, or may be some unwanted variance of its voltages at higher frequencies, caused my SATA HDDs to behave unstably.
One particularly old WD20EARX, bought somewhere in Autumn 2011 (and so, produced most probably before Thailand floods which had affected HDD production facilities), started reporting "SATA link slow to respond/reducing link speed" and similar messages in Linux dmesg, audially this was heard as its motor spinning down and then back up, and one newer WD40EFRX sometimes reported numerous SMART errors collecting in its log, which disappeared from smartctl printout after its full power-cycle (that is, with powering down the whole ATX PSU).
So I put back my older but full-scale FSP ATX400PNF, and SATA HDD behavior restored back to its normal stability.
Also I ordered a different DC ATX PSU from AliExpress, marked RGEEK 300L or LD-A300W, with stated ATX output power of 300W.
Also I finally found a time to attach 2*12V car batteries, each of 62A*h stated capacity, in parallel to my existing 2*14A*h internal UPS batteries.
Had to drill 2 openings in the side wall of the UPS case, and run through them 2*6mm2 cables to connect external batts to the internal ones.
Then found some time to test the runtime of the resulting setup, and got 19h36m which approximately corresponds to ~5* capacity growth (14+62=76A*h at 2*12V, that is up to 1824 W*h DC), correcting this by higher power draw of FSP ATX400PNF, and by the fact that I re-powered my network gear back from their bundled AC-DC PSUs, since I relocated the whole setup into a different corner of the room and put them higher and closer to the room's ceiling.
To get fully charged back, the resulting battery setup took >17 hours, but APC SmartUPS 1000 completed this process without complains.
The resulting voltage settled at 27.58V (well balanced as 13.78..13.79V on individual batts) and did not grow further.
apcaccess reported its usual number 28.1V (who knows how it got this number, but it was made in 1998 after all).
After some time I received my LD-A300W from the post office, and several days later my FSP ATX-400PNF started triggering power surge protection on my ASUS B85M-G motherboard (which is in its 9th year of almost continuous 24*365 operation as my home server).
So I decided that the time has come to put LD-A300W into its production operation (did not even test it preliminarily), and put it into my box instead of FSP.
External 12V DC I took from the same S-120W-12V mentioned at https://muwlgr.dreamwidth.org/49507.html .
Did not even adjust its output voltage, and the BIOS happily reported both +12V and +5V with small added fractions.
The motherboard, HDDs and additional PCIe and USB adapters worked expectedly stable after system boot, and SATA errors observed with LR1106 did not happen again.
Load% reported by apcaccess reduced from 6.3% to 5.2% (a step back from what I achieved at https://muwlgr.dreamwidth.org/50053.html, just need to improve packing of my PSUs and one more battery to more confidently put them under the ceiling).
Battery runtime with the current setup will test in some future.
