Whew. Had to replace one of the 4TB drives in the main NAS. That was a success. I then look at the removed drive... last known bad sector count while installed in the main NAS just before removal was 21 - from 9, 8, 2. Even just before replacement, it did not show any pending sectors, either.

After now looking at it on my laptop, there are now 9 pending sectors after that. Glad it was replaced then...

--

The 'bad' drive is my near five year old abused SATA SSD (as an OS drive). I plan on upgrading it to a 1TB drive from the current 512GB anyway.

Organizing and planning my network for the new place. Lots of things to consider. Like whether to move or change ISP (AT&T vs. Charter), whether I want to have a 800watt X2 Server chugging away in the attached garage for plex/retro filesharing/internal-ftp/maybe a home intranet page - vs just using the same FTP Share I have on my Linksys ReadyShare currently (which is what I use for the vintage machines more often).

Trying to decide if I want to keep the T5400 I have, or even consolidate the "Server" stuff to that - or keep using the T610 - and maybe use it as a dual Workstation/Server/Linux Gaming setup.

Then there's the whole thing of getting internet down to the garage, since it's a rental, I'll probably be going the Wireless Extender/Wireless Bridge route.

If the garages are on the same phase as your place, maybe Powerline would get over there. Though you maybe wanna try old/cheap/slow thrift store units before buying a top spec pair.

Or get a HP Z420 or Z440 workstation and do what other poster did for storage duty and low power CPU, low end GPU like Quadro 600 or K600? The power usage is low due to PSU being 90% plus efficient.

Cheers,

"mod" phase of 1060 rehabilitation beginning, the one I got here Re: Bought this (Modern) hardware today

I have decided that I do want to slim it down a bit, as it is a 3 slot card with that chonker of a fan on. So pondering between two options...
i) two very close to stock evercool eighty something mm fans with four wires (but not both spliced to a harness)
ii) slimline 90mm fans that I've got with 4 wires.

Sanity check me here, if I splice two 4 wire fans together to plug into one connector, I just leave off the tach feedback of one of them right?

i) are quite probably the same fan model original to the 1060, but I don't have the plastics or anything and I'd have to reconstruct some sort of shield, mount slash airflow guide.
ii) should be good, and as normal "case fan" type, they have their own frame and could just be ziptied or otherwise affixed with minimal fuss.

It's maybe gonna look bodgy either way, I was only leaning to i) when I was thinking that was my sole option with parts on hand, but then I realised ii) was on the table, and had the advantage of reserving the evercool fans for direct replacement of other GPU fans, and keeping those GPU nice and stock.

Edit: gah, I am a dirty rotten liar, the 90mms turn out to be three wire. Will have to see how loud they are, was hoping for fan control back, with PWM header that happens at the fan right? Not varied from the board with tach feedback.

editII: so I'm working on a tiny gizmo "fanbus", basically slap a mosfet in there and control 3 wire fans from PWM.. theory-O-hectically ... Also special harnesses for the m73 that's actually an m83, dunno why I've been saying m73 a lot lately, I think that's an object in Orion or something, as well as a lenovo... they have 4 pins on the board for the sata drive power, gotta dedicate one to 6 pin PCIe power and the other needs to have 3 drive connectors on it.

I prefer blower type GPUs and remove fan. Make shroud duct and use a larger blower. This will generate more pressure with less rpm and less noise. I'll find out once I have all the goodies in the future.

Note, not all blower fans are equal, majority of these on ebay and others are low pressure even it is high CFM, The ones that is high rpm to generate decent pressure are noisy and loud. There are very few designs that works well. I need to go into detail about these.

Axial fans has their place but not all are equal either. I did experiments including 3 in series, even counter-rotating fans. Doesn't generate enough pressure and bulky, noisy. But push-pull works great for some heatsinks.

Just please buy Noctua and Bequiet, and any with sufficiently broad blades fans is my strongest advice. They are quiet, motor low-impulse design to cut down on vibration. Look at Microsoft Xbox One (all) and Xbox series X used 5 blade fans. Quiet because they also designed around the heatsink too. Works well.

Second, What heatsink makers needs to do is increase fin pitch a bit and must double thickness of the metal fins. This way you can have low pressure resistance and thicker fins can dissipate the heat better in larger area. The noctua is best in that regard almost. But the one I like most is their Noctua's passive heatsink.

Heatsinks makers do not paint the fins any color, the paint increases heat resistance badly. Anodizing is best or leave the metal natural.

PS: The high speed fans that scream at 10k rpm at 3 amps something, uses high impulse (hard magnet cogging feel), motor is the source of the noise more clearly heard over the roar of the air and is what annoys the most is high pitch shriek.

Cheers,

I think I'm gonna do this...
https://www.techpowerup.com/forums/threads/so … pin-fan.115752/

I had halfass brained it out with a 7805 for the pullup, but I started feeling worried about linear reg wasting watts from the GPU supply when I don't know exactly what it's good for, so as long as I can find a 5.1V zener, I'm going with that... though I suspect it may work with a 4.7 or so. .. but anyway, sticking close to "someone else risked their hardware to test it" sounds like a win.

Edit: 5.1 zener found, random headers, bit of stripboard.... only thing bugging me is that I only know where I've got "good" 10 amp MOSFFTs... seems a bit of a waste, swear I had some 2A which would be fine for this, but nooooo they're hiding.

Meatball wrote on 2023-06-20, 12:30:

No, I checked dimensions before buying; plus, this is this is the second time buying the case, so I had an idea of what I can st […]
Show full quote

BitWrangler wrote on 2023-06-20, 03:46:

Did you try moslo ? 🤣

Looks like a neat little cube, did it get you worried whether everything would fit?

No, I checked dimensions before buying; plus, this is this is the second time buying the case, so I had an idea of what I can stuff inside. The bigger issues were concerning this teeny, tiny, ITX motherboard:

A) Impossible to reset the BIOS during tuning with the Graphics card; installed it blocks the jumpers to short and the battery is buried under the rear fan.
B) I could not install any NVMe drives because if any of them ever fail or I want to make a change, I would have to take the entire system apart and remove from the case, and even the tower cooler. So, 2 SSDs and 1 HDD were used to spread I/O. Plus, there's a 5TB HDD USB3 drive attached for backups.

I did not think about these when I bought it, and I didn't have these issues on the last ITX because A - I used an AIO keeping upside NVMe accessible, and B there wasn't a gigantic plate covering the entire board preventing removal of the NVMe on the backside like this board has. Lesson learned for next time.

You can always wire a small front panel switch to the clr cmos header.

After doing some figures, I got locked into a "Goldilocks" hunt. The brand shiny new MOSFETs are too powerful, too high a transition voltage, meaning during half the pulse the MOSFET is dissipating practically all the power as heat, and I am not sure the GPU fan header can stand it. Theoretically, I could just pull the 12 and Ground off a drive connector, just use the PWM and tach on the card and not worry about it, but I wanna keep it all self contained on the card, bad enough already without having a "cat's cradle" (kids string game) of extra power wiring all round the case. Anyway, 400V 10A MOSFETs at one end Papa Bear, and 10V 500mA mosfets at the other end, baby bear, and no momma bear middle stuff... well after hunting all last night and on and off today, I finally found a donor, half picked out MSI OEM board, gonna give me some Niko-Semi P3057LD and if those ain't right, then screw it, wiring fans off the PSU or putting up with 3 wire full speed. Won't get much done tomorrow though, family stuff.

Drive harness stuff for the M83, there's two 4 pin motherboard power headers, the PSU just plugs straight to the board, no drive harness on the PSU. Two by two SATA on them. I need one to be three SATA power and one to be GPU 6 pin dedicated. Scared up all the stuff for that easy enough. 3 drives to be ODD, spindle and SSD. Actual cable I'm putting the 4 pin end on has a regular old 5.25 drive plug on it too, just gonna leave that on in case I need another fan or something.

BitWrangler wrote on 2023-07-16, 04:18:

After doing some figures, I got locked into a "Goldilocks" hunt. The brand shiny new MOSFETs are too powerful, too high a transition voltage, meaning during half the pulse the MOSFET is dissipating practically all the power as heat, and I am not sure the GPU fan header can stand it. Theoretically, I could just pull the 12 and Ground off a drive connector, just use the PWM and tach on the card and not worry about it, but I wanna keep it all self contained on the card, bad enough already without having a "cat's cradle" (kids string game) of extra power wiring all round the case. Anyway, 400V 10A MOSFETs at one end Papa Bear, and 10V 500mA mosfets at the other end, baby bear, and no momma bear middle stuff... well after hunting all last night and on and off today, I finally found a donor, half picked out MSI OEM board, gonna give me some Niko-Semi P3057LD and if those ain't right, then screw it, wiring fans off the PSU or putting up with 3 wire full speed. Won't get much done tomorrow though, family stuff.

Drive harness stuff for the M83, there's two 4 pin motherboard power headers, the PSU just plugs straight to the board, no drive harness on the PSU. Two by two SATA on them. I need one to be three SATA power and one to be GPU 6 pin dedicated. Scared up all the stuff for that easy enough. 3 drives to be ODD, spindle and SSD. Actual cable I'm putting the 4 pin end on has a regular old 5.25 drive plug on it too, just gonna leave that on in case I need another fan or something.

Linear regulator is efficient at delivering power directly across regulator to the fan, than resistor and zener combination beacuse zener and resistor is taking about half of power and is limited by resistor wattage on capable of driving a fan. No way I would do that. Also Linear regulators is most easiest way to build and cheap. I drove 3 fans with one regulator in TO-220 on a plate heatsink. Practically no heat.

The most efficient step up from this is buck switching regulator (called DC-DC converter) if you have no choice to drive these 3 or 2 wire fans. But the most best way ever is use fan or blower via PWM pin because you are regulating the magnetic density direct to the windings for the fan. Much like the DC-DC converter does.

PWM fans, The power input is replaced with a transistor feeding the two coils in the fan. Other side of windings are alternating grounded via pair of transistors and is long periods of on and off, very low frequency. While these two transistors switching the coils causing the rotor to spin, the PWM transistor controls magnetic flux density by rapidly switching the coils off and on in on/off ratio at 25KHz.

PWM fan is very very smart design indeed.

In large high HP applications, especially hybrid cars and battery powered cars, The controller uses large dc-dc converter to boost voltage to around 300V to 600V from 100V input from there feeds power to the 3 phase power mosfets cooled by liquid cooling system. The signal to switch off and on these mosfets is also PWM to limit current and adjust power (HP) directly to the coils causing BLDC rotor to spin. To make power impulses smoother and boost efficiency, the BLDC rotor is spun up to 10,000 rpm max and drive the reduction gears, remember it is direct driven to the wheels, not shifting at all. The gear ratio set is fixed around 9 to 10 to one.
Generator/motor one and generator/motor two is coupled via a differential gear, between two drives the gear set. Either one or both. This allows the car to be at standstill while engine is generating power for motor/generator one while the second BLDC is held still magnetically. Final drive is regular differential gear driving the wheels much like gas powered differential is.

Also there is no starter for the engine. The generator/motor starts the engine directly.

BLDC rotors dimensions is about 30 cm diameter or slightly smaller, and length of the rotor can be anywhere from 15cm to 40cm. BLDC rotors is constructed purely of laminated soft iron with holes to squeeze magnetic fields harder through the rotor, increasing torque.

Windings of these motors/generators in electric cars is about 5 turns per pole. Exactly similar to DC-DC converter's inductor number of turns, in some cases only a one turn on high end VRMs.

Cheers,

No internet for at least 2 days at the old apartment - Spectrum did not ask moving date when we planned to close their service and move to our new place with AT&T fiber. Guess that's motivation to get more moving stuff done. Good news is we could end up with the equipment on day 1 at our new place....so that'll be nice. Anxious to get in there so I can start work on my new computing area.

Good luck with the move.

Still plugging away at my 1060 build/mod things are coming together. Pulled it from the AM2A2X2 box and stuck a GTX440 in there that had been kicking around for ages, and I finally figured out how to fix fan on that a couple of months back, but hadn't tested it. So that runs. Hilarity ensued as I tested it on some of the same stuff as the 1060, 3fps on the covid bench 🤣 7ish on the 3D.Benchmark.OK... yay beat a 450, but was under a 710, boo... though they have same shaders, not a 610 relabel. Then the unigine ones I couldn't get to run still didn't run so software environment fubar somewhere. Still seemed plenty decent for desktop, browsing and old stuff (Hundreds of fps in GPUcap <OGL 2.xx ) though.... not sure it's going to find a "seat" when everything is put together, might have to see if it's PhysX score is good enough to partner a Radeon box. (I should think it might be, a 9400GT got about 4 times the bench of a high Athlon X2 in CPU mode, so it would probably need a real brute of a CPU, like modern Ryzen, intel i7 to better it. Even then your CPU is doing other stuff at same time so unloading it still helps)

Anyway, back to the 1060 stuff, gotta do some marking and drilling, ( I know right, don't drill through the black plastic things, they hold the magic smoke)

Bought an NVMe, but goes up to 64°C and stays there.
Guess I'll have to buy a heatsink. Too bad the slot is on the back of the board (ITX), no airflow there.
It was cheap and I went for it. Guess this is the future if you want some speed 😀

How it started ....

How it's going ....

Mechanics are about sorted, still have to build the PWM to 3 pin fan driver. This is what you might call a "rat rod" build in that I just used stuff I could find around the "garage" as it were.

Heatsink for the NVMe arrived. 20°C difference.
From 64 to 43/44 under load. That was unexpected 😀

Edit: spoke too soon, 53°C. Probably it had to stay on more than 5 minutes.
Anyway, better than before.

Edit2: minimal airflow putting a piece of cardboard to deflect in the direction of the drive: dropped 5°C.

Modern activity?

I finally am getting around to testing out pine rosin powder (dissolved in isopropanol) as a bed adhesive for printing with nylon. So, basically playing with the 3D printer today.

Eww, tree juice, that's just tacky... hopefully. Well it's kinda good for holding parts in place when you're soldering, more so than most fluxes. I guess you aren't going to burn it off though, so I guess it's gonna take some more alcohol to clean up.

It seems to be working well so far.

Nylon prints at a very high temperature (253C, in my case. My printer alarms higher than 255C), which is above the melting temp of the rosin. I put the mix in a spray bottle and misted the bed with it, and then heated the bed up to 80C for a few minutes to dry the rosin, then let it cool back down to room temperature. This left a "smooth" and non-tacky surface on the build plate.

When the hot nylon hits it, it melts and sticks, then cools again very fast (because glass has a high thermal capacity). This grabs the nylon pretty good. When I want it to let go, I will just heat the bed back up and then peel it off.

Got an hour on the 1060 proj... MOSFETs successfully graverobbed and tested, 3 pin headers ditto, board layout revised for 3rd time, all parts present and correct, think I can put it together tomorrow.

I might be further ahead but I took about a half day messing around with that engine turning/jewelling effect. 3 methods didn't work, guess the grit wasn't gritty enough or something in various substances. Possibly that stuff is a harder duralumin rather than soft aluminum. Anyhoo, ended up making a custom steel wool brush for it, then being my first rodeo it came out a little uneven still. I had read up on it and thought I'd get it done in an hour 🤣

I built a NAS that will house 8 x 12TB Seagate IronWolf Pro drives. This will host my Retro Driver collection that has exceeded the space on my old QNAP.

Specs:
AUDHEID 8-Bay NAS Chassis
ASUS TUF GAMING B550M-PLUS WiFi II Motherboard
AMD Ryzen 7 5700G 8-Core/16-Thread CPU (w/ Thermalright AXP90-X47 CPU Cooler)
Patriot Viper Steel DDR4 128GB (4 x 32GB) 3200MHz
2TB Silicon Power UD90 NVMe 4.0 Gen4 PCIe M.2 SSD NVME (@ PCIe 3.0)
240GB Kingston HyperX Savage SSD
9207-8i PCIE3.0 6Gbps HBA LSI FW:P20 IT Mode
Fujitsu Dual Port 10Gb SFP+ Intel x540 Fibre Card
8x 12TB Seagate IronWolf Pro Enterprise NAS HDDs
700W Flex Power Supply

Running TrueNAS