2022-10-28 19:36 from Nurb432
ARM is dead. The fat lady is warming up to sing
ARM is dead -Nurb
Nurb is dead -ARM
2022-11-01 07:13 from Nurb432
Learned something yesterday abut USBC PD power. I'm sure everyone
knows and im late to the game, but i never needed to know this until
I guess when a device turns on, the power supply gives it a trickle
of current, with the expectation that the device will actively
'negotiate' what it wants to fully power up. And if it does not in
x microseconds, the power supply powers down, causing the device to
reboot and start the process over again.
Most devices do this early on in the boot process,perhaps the very
first thing, ( its inside UBoot if its ARM ), and if you are stupid
and put this into the kernel, by the time you get there the power
supply has already given up.
I think this is inaccurate.
USB-C defaults to 5V power supply, same as USB-A. Consumers can negotiate higher.
I believe that all the negotiation logic is provided by chipsets (or "IP") that you can buy from Texas Instruments and the like. Probably in quantities of 1000+
Direct from their FAQ ( and a lot of others talking about the same thing of why its currently f-ed )
ROCK 5B supports USB PD power negotiation to higher voltage such as 9V, 12V, 15V, 20V to meet the total system power load requirements(around 30W with WiFi, SSD, USB peripherals etc). Currently the PD negotiation is implemented in the kernel driver, we need to boot to the kernel to start negotiation, however some Power Supply can not wait the ROCK 5B to negotiate and cut the power off when timeout, which cause the infinite boot loop. We provide the following suggestions:
Seems like that particular implementation might be an outlier.
The way they are talking, its not, just they f-ed up the implementation and we have to wait for code fixes ( or the workaround of getting a stupid PD or make your own and hardwire it to the GPIO pins ) But either way, at least there will be a fix. just frustrating as i waited almost a year to get this thing.. now it dont work.
and yes i did get out of buying these things as i dont do any projects that need SBCs anymore, but that was after i pre-ordered back January. ( but this should be mid-range desktop level stuff with built in AI TPU, once it works )
Tue Nov 01 2022 07:41:59 PM EDT from LoanShark
Seems like that particular implementation might be an outlier.
cool ill read it after bit
I have one more PS to try before i just build one.
USB-C defaults to 5V power supply, same as USB-A. Consumers can
LS is correct. USB ports have always delivered a constant +5 volts DC, and a USB-C port is no different, which is why you can plug USB 1.0 and 2.0 devices into a USB-C port with the right cable and they will work fine.
"USB Power Delivery" (USB-PD) requires USB-C and then the devices on each end of the cable negotiate for higher voltage and current, which requires some smarts. This can be implemented either in the device's USB interface or in a separate chip (which, as was pointed out, TI can provide). So yes, it's completely feasible that a device which wants USB-PD but isn't getting the volts and amps it's asking for might just keep reinitializing until the user gives up and pulls it out.
Incidentally ... power over ethernet (802.3af and 802.3bt standards, not the "borrowed pair" hack) operates on the same principle of negotiated power delivery, and can go up to 73 watts.
i suspect in the ARM sbc world everyone does it via software in Uboot, and not a chip. its 'just the way'.
.....and these jokers just dropped the ball and should know better.
Right, with a little care to not assume the source has gone about 5V before it actually has done so, power should remain uninterrupted.
Tried my last PD supply today, same results. ( apple OEM i had at work, people were saying some models did work. )
So i either wait for the to fix the code ( im reading people tried moving the stuff to uboot, but its still broke ) or buy a regular supply and hardwire it, since i dont think i have any laying around now. I got rid of nearly everything like that this year. Even had to borrow a resistor to test a sensor in the jeep this summer. i had nothing left.
Broke down and ordered one of their supplies, and should be here in 2 weeks. Id even have ordered one with the board if they had told us they dropped the ball... I'm annoyed. Its not the cost, ( shipping was more than the supply ) its just that there really isn't an excuse for this from an established board maker.
We will see now if they get u-boot code fixed, or if the PSU arrives first. The race is on!
The last one I built used an Arduino and a WWVB receiver. It's been chugging away in my bedroom for several years and I love it. But I wanted the next one I built to be an NTP server as well. That requires a network connection and a computer more powerful than an Arduino so it can handle NTP protocol.
I have another WWVB receiver, so that would work even though I'd have to rewrite the decoder to run on Linux. No big deal, really. Or I could get a GPS puck and things would work more or less out of the box.
Part of me is even thinking, why does it even have to be a Stratum 0 at all?
Just grab the time off the network and display it.
Might look at a higher end ESP32.
Hmm or a used tablet...
built in network, display, battery....
The problem with a tablet is that it looks like a tablet. There is zero coolness factor in that.
Here's a photo of my WWVB clock. I deliberately chose a photo frame as its "case" because the point is to show off the electronics.
Anyone can slap a used tablet on the wall. I want custom electronics with a dedicated LED display. Something that not only looks cool but says "I made this."
Then you better get out the 3D printer and discrete components and build your logic from scratch.. Buying a ESP/Arduino ( or whatever, even FPGA ) and slapping some code in it is for amateurs. :)
Oh, and blow your own glass to make nixie tubes from scratch too. ( kidding aside, i have seen people do this, its pretty cool )
Tue Nov 15 2022 06:11:18 PM EST from IGnatius T Foobar
Something that not only looks cool but says "I made this."
Recently i came in to possession of a few 7200 rpm server class drives. Put them in my 'big' box in my cluster to put the storage up to ludicrous amounts for testing stuff and added them as shared storage to the farm. That machine normally is turned off. its just for when i need real horsepower for something, like AI stuff or video format conversions ),
I went to only solid state drives here a while ago. I had forgot **HOW FREAKING SLOW** spinny disks are. geesh. i'm about to rip them back out and toss them in the lake. just a simple clone takes forever...
funny part was during a clone last night " what is that odd noise. something is up, that is a new noise " Thinking it might be the furnace or fish tank or something...went in the hall and living room looking and listening.. no its from the computer room.. "oh, dumb ass, those are the drives" between it being so long and the case is muffled from the 5 fans on the 2 video cards + cpu + power supply + 2 case fans.. damned thing is a helicopter
there goes your electric bill. gretta is giving you frownies, you don't want gretta unhappy with you
Even before the drives she would be mad at me, pointing her finger and yelling 'how dare you'. this machine is a beast.
Might be fun actually.