wafer.space Community - fpgas-online

@carlfk - Where is the CAD for your little Tiny Tapeout & RPi holder shown here?

2:37 a.m.

@carlfk - And what is the little red board on the top?

Tim 'mithro' Ansell

@carlfk - And what is the little red board on the top?

pi uart to usb to see pi boot messages

@carlfk - I also had Claude generate some stickers to label fpgas.online hardware -- https://github.com/mithro/rpi-hdcp-output/blob/hardware-inventory-audit/docs/labels/README.md

Tim 'mithro' Ansell

@carlfk - Where is the CAD for your little Tiny Tapeout & RPi holder shown here?

I cant find it

@Andrew Wingate help?!! lol

carlfk

I cant find it

@Andrew Wingate help?!! lol

uhhh.... let me look?

3:01 a.m.

Part Studio called cut this and drawing called export me Seems to have worked at least once lol. https://cad.onshape.com/documents/6f9e5a77e8ba5c1cd288800d/w/471fd4c5d9442ef9f289ae02/e/86a97e67fa6a2316e0fa2b1c

export_me.dxf

108.45 KB

Onshape

Sign in to Onshape, the #1 fastest growing CAD system in the world.

@Andrew Wingate - What would you call the type of diagram which is used to show the mechanical parameters of something which is used by people to design cases and such?

Tim 'mithro' Ansell

@Andrew Wingate - What would you call the type of diagram which is used to show the mechanical parameters of something which is used by people to design cases and such?

It would just be a mechanical drawing.

3:52 a.m.

I mean, there are other words you can use, but it's unnecessary. If you're designing for something else to fit in/on/around you can apply GD&T stuff that basically outlines some tolerances on the part they may not have. Other stuff like form/fit/function, but again, a lot of this is just embedded into a mechanical drawing if you desire.

@Tim 'mithro' Ansell are there any online-fpgas with radios attached to them?

storborg

@Tim 'mithro' Ansell are there any online-fpgas with radios attached to them?

Not at the moment -- but I separately have set up a bunch of RPi hardware which have SDR systems connected to them -- some which have FPGAs. I think it was/is a Pluto, Kraden SDR, RTL-SDR and soon a xsdr

Ah, hence the adapter to put the xsdr on a Pi HAT.

5:06 a.m.

Does your Pluto setup intend for the bitstream to be overwritten or is it just IQ receive via Pluto IIO APIs?

storborg

Does your Pluto setup intend for the bitstream to be overwritten or is it just IQ receive via Pluto IIO APIs?

I have no idea!

5:09 a.m.

I think I broke the pluto while working on trying to get my 433mhz to wifi bridges working

Ok I will stay tuned!

@storborg - I'd most certainly appreciate any help with improving fpgas.online

I think there is some basic infrastructure stuff that's broken at the moment? I tried, as advised by the docs (https://github.com/CarlFK/pici/wiki/Getting-Started) to pick an FPGA, e.g. this one: https://welland.fpgas.online/fpgas/pi38.html I built an Arty A7-35T bitstream locally and tried to upload it with the form in the top right. I get a 413 Request Entity Too Large served by nginx, on a 2.1MB .bit file. So then I tried to connect via ssh using the ssh -p 23822 pi@welland.fpgas.online instruction. No dice, doesn't look like anything is listening on that port. Then I tried clicking the "Reset" button, which I presume is supposed to power cycle the PoE port. It did update the websocket messages to indicate the control socket was cycled, but it didn't affect the state of the LEDs in the video feed at all, and ssh login isn't any more successful.

Getting Started

Contribute to CarlFK/pici development by creating an account on GitHub.

storborg

I think there is some basic infrastructure stuff that's broken at the moment? I tried, as advised by the docs (https://github.com/CarlFK/pici/wiki/Getting-Started) to pick an FPGA, e.g. this one: https://welland.fpgas.online/fpgas/pi38.html I built an Arty A7-35T bitstream locally and tried to upload it with the form in the top right. I get a 413 Request Entity Too Large served by nginx, on a 2.1MB .bit file. So then I tried to connect via ssh using the ssh -p 23822 pi@welland.fpgas.online instruction. No dice, doesn't look like anything is listening on that port. Then I tried clicking the "Reset" button, which I presume is supposed to power cycle the PoE port. It did update the websocket messages to indicate the control socket was cycled, but it didn't affect the state of the LEDs in the video feed at all, and ssh login isn't any more successful.

Looks like I have broken welland.fpgas.online -- try ps1.fpgas.online?

9:37 p.m.

@storborg - Thanks for testing things!

Sweet, yeah, ps1.fpgas.online works, or at least, the one I randomly picked does.

7:03 p.m.

Ex xylinx dev

Can't quite read

"Why Care? Aren't FPGAs a niche thing?" is the only bit I can read....

FPGAs allow for true hardware replication rather than emulation

??

??

??
Allows the community to preserve truly functional hardware forever

@carlfk - I've been trying to find a better way to mount the tt boards -- https://github.com/fpgas-online/fpgas.online-mechanical/blob/main/tinytapeout/output/tinytapeout-sheets.pdf

fpgas.online-mechanical/tinytapeout/output/tinytapeout-sheets.pdf a...

2D orthographic mechanical drawings of the hardware fpgas.online is built from: boards, adapters, accessories, and the plates and drill templates to mount them - fpgas-online/fpgas.online-mechanical

11:45 a.m.

@carlfk - https://github.com/fpgas-online/fpgas.online-mechanical/blob/main/tinytapeout/mounting_plate/output/tt-generic-mounting-plate-drill-template.pdf

fpgas.online-mechanical/tinytapeout/mounting_plate/output/tt-generi...

2D orthographic mechanical drawings of the hardware fpgas.online is built from: boards, adapters, accessories, and the plates and drill templates to mount them - fpgas-online/fpgas.online-mechanical

storborg

I think there is some basic infrastructure stuff that's broken at the moment? I tried, as advised by the docs (https://github.com/CarlFK/pici/wiki/Getting-Started) to pick an FPGA, e.g. this one: https://welland.fpgas.online/fpgas/pi38.html I built an Arty A7-35T bitstream locally and tried to upload it with the form in the top right. I get a 413 Request Entity Too Large served by nginx, on a 2.1MB .bit file. So then I tried to connect via ssh using the ssh -p 23822 pi@welland.fpgas.online instruction. No dice, doesn't look like anything is listening on that port. Then I tried clicking the "Reset" button, which I presume is supposed to power cycle the PoE port. It did update the websocket messages to indicate the control socket was cycled, but it didn't affect the state of the LEDs in the video feed at all, and ssh login isn't any more successful.

Claude is claiming that it has fixed a bunch of the problems you identified. In theory I have an end-to-end test which shows it works now (by using the system like a human) but I haven't had a chance to test it myself manually.

It's broken in new and different ways now

3:10 a.m.

ssh to the individual pis doesn't seem to be listening, or maybe the ports are wrong. Uploading a bitstream directly with the web form sometimes throws a 500 and sometimes succeeds, but never seems to actually do anything.

storborg

ssh to the individual pis doesn't seem to be listening, or maybe the ports are wrong. Uploading a bitstream directly with the web form sometimes throws a 500 and sometimes succeeds, but never seems to actually do anything.

Got a test bitstream?

storborg

It's broken in new and different ways now

I haven't fixed the direct ssh connection yet -- does the webssh terminal work for you?

Here's an Arty 35 bitstream that should make the LEDs do something noticeable and distinct on the video stream

blinky.bit

2.09 MB

7:54 p.m.

The webssh terminal sort of works, although it seems to sometimes not load. I guess I can curl from somewhere else via webssh.

7:55 p.m.

If I do that and flash the board manually via openFPGALoader, it completes successfully, but I don't see anything on the video feed.

7:57 p.m.

I tried on a different board (https://welland.fpgas.online/fpgas/pi37.html) and it seems to be working, although the view of the LEDs is really blurry.

7:58 p.m.

It would be kind of cool if the Pi could just read LED or Pmod state directly and show it on the web interface. I presume there is some wiring between the Pmods on the boards and the Pis that I could use to do that.

7:59 p.m.

I assume the reason this is more reliable on ps1 and less reliable on welland is that ps1 doesn't use the per-port VLAN configuration?

Honestly, I kinda feel like you could forego the camera feed entirely and just show a bunch of outputs of different types that could be driven directly from the FPGA. I could imagine the 'thumbnail' of the board instance showing whether or not it's running a bitstream / has a user interacting with the pi, the filename of the bitstream that's running, and a country flag to indicate the remote user that loaded the bitstream.

storborg

Honestly, I kinda feel like you could forego the camera feed entirely and just show a bunch of outputs of different types that could be driven directly from the FPGA. I could imagine the 'thumbnail' of the board instance showing whether or not it's running a bitstream / has a user interacting with the pi, the filename of the bitstream that's running, and a country flag to indicate the remote user that loaded the bitstream.

I somewhat agree -- but I don't think you should underestimate the draw / appeal of being able to see a camera feed of a real hardware device.

7:06 a.m.

Also the camera feed is what makes it different from just running a simulation

Yeah. You're right of course. But then it needs to be less blurry.

I was thinking that radio would be a great way to make it feel more "real", especially because there are many ultra-cheap SDR FPGA boards these days. But maybe a cool thing on video would be robotics? Some kind of motion control? Tim, what if we replace the guts of all of your household appliances with internet-connected FPGA boards? Random strangers could improve the control algorithms of your dishwasher and washing machine. Nothing could possibly go wrong.

storborg

Yeah. You're right of course. But then it needs to be less blurry.

Yes, that will be fixed sometime when @Tisham Dhar (CSIRO EASI) get a chance to do that

storborg

I was thinking that radio would be a great way to make it feel more "real", especially because there are many ultra-cheap SDR FPGA boards these days. But maybe a cool thing on video would be robotics? Some kind of motion control? Tim, what if we replace the guts of all of your household appliances with internet-connected FPGA boards? Random strangers could improve the control algorithms of your dishwasher and washing machine. Nothing could possibly go wrong.

Well, I have the sdr radios separately -- but generally the government doesn't like you running uncontrolled radio transmitters

Just receivers is ok too! That's still an interaction with the real world.

@Tim 'mithro' Ansell I will be re-building the ps1 fpga box in an hour ish. do you have anything to PR into my ansible?

bah. [ERROR]: Task failed: Module failed: non-zero return code

4:10 a.m.

Upgrading: 0, Installing: 0, Removing: 0, Not Upgrading: 9", " 1 not fully installed or removed.", " Space needed: 0 B / 103 GB available", "", "Setting up initramfs-tools (0.148.4+rpt2) ...", "update-initramfs: deferring update (trigger activated)", "Processing triggers for initramfs-tools (0.148.4+rpt2) ...", "update-initramfs: Generating /boot/initrd.img-6.18.34+rpt-rpi-v8"]}

hmm.. the date... 9932 -rw-r--r-- 1 root root 10168774 Jun 9 14:55 /boot/vmlinuz-6.18.34+rpt-rpi-v8

@carlfk - At some point I would like to rebuild your gateway box at ps1 onto the stuff under https://github.com/fpgas-online/

fpgas.online

fpgas.online has 22 repositories available. Follow their code on GitHub.

4:24 a.m.

@carlfk - But I'm not in a super hurry to do that because it hasn't quite been working as reliable as I would like yet.

I know that feeling

@carlfk - Did you see the new mechanical diagrams and the Tiny Tapeout adapter plate?

yes, is it the collection of all the TT moutning holes?

@carlfk - I would also like to get the compute blade devices at ps1 wiring finished.

4:27 a.m.

@carlfk - It's a plate which has all the different TT mounting holes and a set of other holes

4:28 a.m.

@carlfk - And it makes sure that the pmod connectors are all in identical locations

4:28 a.m.

@carlfk - https://github.com/fpgas-online/fpgas.online-mechanical/blob/main/tinytapeout/mounting_plate/README.md

fpgas.online-mechanical/tinytapeout/mounting_plate/README.md at mai...

2D orthographic mechanical drawings of the hardware fpgas.online is built from: boards, adapters, accessories, and the plates and drill templates to mount them - fpgas-online/fpgas.online-mechanical

in your fpga.online deployment - are you able to "apt install nfs-common" into the pi's fs from the x86 cpu? know what I am talking about?

4:31 a.m.

this worked before, but now I need an apt upgrade which is failing

@carlfk - Yes, I have / had both qemu userspace chroots and full qemu virtual RPi

4:38 a.m.

@carlfk - Now I've kind of moved to having the RPi nfsroot just built on github using the native ARM github action runners -- it takes between 5 to 20 minutes to build the complete image.

@Tim 'mithro' Ansell that is compelling

@carlfk - It is basically building an "ARM docker container" (not actually docker - podman or something) that it just then extracts the contents from.

4:41 a.m.

The ARM github action runners are some type of big beefy server ARM systems too, they run a lot faster than a real RPi

for your qemu, do you have a script like https://github.com/CarlFK/pici/blob/main/ansible/roles/fixpi/files/scripts/chroot-mount-pi-fs.bash

pici/ansible/roles/fixpi/files/scripts/chroot-mount-pi-fs.bash at m...

Contribute to CarlFK/pici development by creating an account on GitHub.

4:42 a.m.

the apt update errord with Error: Can not write log (Is /dev/pts mounted?) - posix_openpt (19: No such device)

4:42 a.m.

which makes me think I need: (very much guessing) mount -t devpts pts "${target}/dev/pts"

4:48 a.m.

nope. mount: /tmp/pi/dev/pts: mount point does not exist.

carlfk

which makes me think I need: (very much guessing) mount -t devpts pts "${target}/dev/pts"

I think you need something like mount -o bind /dev/pts "${target}/dev/pts"

Tim 'mithro' Ansell

I think you need something like mount -o bind /dev/pts "${target}/dev/pts"

+ mount -o bind /dev/pts /tmp/pi/dev/pts mount: /tmp/pi/dev/pts: mount point does not exist.

4:54 a.m.

I was hoping you fixed the script before you moved to the arm github thing

I may have a whole back, but its moved quite a bit further

Error: Can not write log (Is /dev/pts mounted?) - posix_openpt (19: No such device)

4:56 a.m.

woppsp, wrong window

5:00 a.m.

can you grep your files for "dev/pts"

@carlfk - https://github.com/fpgas-online/fpgas.online-netboot-pi/blob/main/qemu/chroot-mount-devices.bash.sh maybe?

fpgas.online-netboot-pi/qemu/chroot-mount-devices.bash.sh at main ...

Netboot filesystem preparation tools for fpgas.online Raspberry Pi boards - fpgas-online/fpgas.online-netboot-pi

5:01 a.m.

@carlfk - Or maybe - https://github.com/fpgas-online/fpgas.online-netboot-pi/blob/main/scripts/chroot-mount-pi-fs.bash ?

fpgas.online-netboot-pi/scripts/chroot-mount-pi-fs.bash at main · ...

Netboot filesystem preparation tools for fpgas.online Raspberry Pi boards - fpgas-online/fpgas.online-netboot-pi

@Tim 'mithro' Ansell neither of those have pts

5:04 a.m.

it is possible I can ignore the lack of pts error, but it is the only line that stands out as a problem that needs fixing, and something needs fixing

Do you have a proper log?

Tim 'mithro' Ansell

Do you have a proper log?

https://paste.debian.net/hidden/6dbe9275

@carlfk - It still needs a lot of work, but I'm trying to put together https://docs.fpgas.online/

@carlfk - https://github.com/CarlFK/pici/pull/56

chroot-mount-pi-fs: mount a private devpts in the chroot by mithro ...

Fixes this error when running apt inside chroot-mount-pi-fs.bash (from https://paste.debian.net/hidden/6dbe9275): Error: Can not write log (Is /dev/pts mounted?) - posix_openpt (19: No such device)...

Pull request successfully merged and closed

5:29 a.m.

it got rid of the error, but there is still soemthing wrong...

5:30 a.m.

@Tim 'mithro' Ansell https://paste.debian.net/hidden/2bb89792

5:32 a.m.

root@base:/# grep -r MODULES /etc/initramfs-tools /etc/initramfs-tools/initramfs.conf:# MODULES: [ most | netboot | dep | list ] /etc/initramfs-tools/initramfs.conf:MODULES=dep

@carlfk - Claude sent you a few more.

5:37 a.m.

@carlfk - Any of those fix your problems above?

Tim 'mithro' Ansell

@carlfk - Any of those fix your problems above?

mount pts fixed Error: Can not write log (Is /dev/pts mounted?) - posix_openpt (19: No such device)

@carlfk See 57, 58, & 59

MODULES=most seems to fixed something as the update-initramfs is working instead of erroring

5:44 a.m.

the fixes seem elegant - I'm amazed.

ah - 56 and 59 fix both copies of the same script. guesisng the tools one was from before I put it in ansible

5:54 a.m.

im ignoring the cmd one for now - my pi stuff seems to be working

6:01 a.m.

phooy. cant' do lets-encrypt certbot when the server isn't live on a public IP

6:04 a.m.

also I am installing unicorn, gunicorn, asgi and who knows what. pretty sure I only need one of those.

6:05 a.m.

and I need to rename some things to have a 2nd set so veyepar.nextdayvideo.com can live on the same box

6:06 a.m.

on the pluss side, ansible ran to the end, no errors

6:07 a.m.

01:00 - good stooping point.

6:07 a.m.

carl@notp:~/src/tv/pib/pici$ git status |grep modified | wc 24 48 1311

6:09 a.m.

huh. I thought I was better at committing things

6:09 a.m.

tomorrow problem

@storborg - Maybe I'll be able to put something like https://essenceia.github.io/projects/alibaba_cloud_fpga/ into fpgas.online after I have finished getting the netv2 and acorn PCIe stuff validated.

Alibaba cloud FPGA: the 200$ Kintex UltraScale+ [Updated]

Using a decommissioned Alibaba cloud accelerator card as an FPGA dev board

That would be cool. If you do one of those you put a loopback cable between the SFP28 cages, or better yet give them a fast switch port on the same VLAN.

@Tim 'mithro' Ansell how do you mange what hardware is plugged into what port? like what file or spreadsheeto or db or where does this info live?

10:44 p.m.

like I edit https://github.com/CarlFK/pici/blob/main/ansible/inventory/host_vars/ps1.fpgas.online.yml#L64 - { port: 2, mac: "b8:27:eb:2f:5d:08", sn: 042f5d08, model: "Raspberry_Pi_3_Model_B_Rev_1",

pici/ansible/inventory/host_vars/ps1.fpgas.online.yml at main · Ca...

Contribute to CarlFK/pici development by creating an account on GitHub.

carlfk

@Tim 'mithro' Ansell how do you mange what hardware is plugged into what port? like what file or spreadsheeto or db or where does this info live?

I've moved to a scheme where the Pi & FPGA details are autodetected by scripts in the nfsroot and then each port is on it's own isolated vlan and thus gets a specific IP address & name automagically. IE 10.X.1.2 is "switch 1, port 2" and 10.X.1.55 is "switch 1, port 55" -- also stops each of the RPi traffic from interfering with each other and makes the RPi3b netbooting a lot more reliable.

storborg

That would be cool. If you do one of those you put a loopback cable between the SFP28 cages, or better yet give them a fast switch port on the same VLAN.

100g switches are surpisingly affordable.

Tim 'mithro' Ansell

I've moved to a scheme where the Pi & FPGA details are autodetected by scripts in the nfsroot and then each port is on it's own isolated vlan and thus gets a specific IP address & name automagically. IE 10.X.1.2 is "switch 1, port 2" and 10.X.1.55 is "switch 1, port 55" -- also stops each of the RPi traffic from interfering with each other and makes the RPi3b netbooting a lot more reliable.

I like the sound of that

It would be neat to make a board in the form factor of e.g. a Mini ITX motherboard, which just breaks out a PCIe edge connector onto a ton of SFP28/QSFP28 cages and puts a low-jitter programmable oscillator into the PCIe refclk pads. Then you could turn those cheap surplus accelerator boards into optical switch platforms.

@carlfk - Claude claims to have solved the video delay problem. I'm just having it merge and deploy the functionality so I can check if it really has.

"interesting"

@carlfk - It claims to have gotten to network-latency + 2 video frames

2:57 a.m.

@carlfk - I'll believe it when I actually see it, which should hopefully be within an hour or two.

Tim 'mithro' Ansell

@carlfk - I'll believe it when I actually see it, which should hopefully be within an hour or two.

that fix for modules=something... I already had that fix, but it wasn't being done in the right order. so I had 2 copies of the same thing that didn't work

3:11 a.m.

neat that it figured out the same thing I did, including the not quite right

@carlfk - I now have a hopefully decent diagram of what wiring needs to be finished on the compute blade devices at ps1 - https://docs.fpgas.online/en/latest/boards/acorn/wiring.html#compute-blade

Tim 'mithro' Ansell

@carlfk - I now have a hopefully decent diagram of what wiring needs to be finished on the compute blade devices at ps1 - https://docs.fpgas.online/en/latest/boards/acorn/wiring.html#compute-blade

Do you know if the one I did was done right?

@carlfk - I think so, but I should have a webpage which declares if it is true shortly.

@carlfk - https://github.com/fpgas-online/apt/pull/11

Register the Acorn debs from fpgas.online-test-designs by mithro ·...

Adds the two Acorn packages that fpgas.online-test-designs now publishes to its v0.0 release (test-designs #34, merged as 37d2407): Package Version Contents fpgas-online-acorn-bitstreams 202...

is there a .deb I can wget?

@carlfk - In theory, yes -- let me check. It should be linked from https://github.com/fpgas-online/fpgas.online-test-designs but does not currently seem to be.

GitHub - fpgas-online/fpgas.online-test-designs: Designs use to ver...

Designs use to verify the functionality of the fpgas.online hardware is working correctly on RPi boot. - fpgas-online/fpgas.online-test-designs

current ps1.fpgas.online box is about to be replaced - new box is ready, just need to spend an hour moving stuff around

1:39 a.m.

so if you want to use my box to test stuff, now is a ok time to break it

@carlfk - Okay, I'm just currently juggling making sure that Run #3 is going to actually make money and potentially trying to open up Run #4 due to selling out of 1x1 slots.

Tim 'mithro' Ansell

@carlfk - Okay, I'm just currently juggling making sure that Run #3 is going to actually make money and potentially trying to open up Run #4 due to selling out of 1x1 slots.

yay for selling out slots - that's exciting

@carlfk - While I hoped it would happen with Run #3, I think I didn't really believe it would happen, so not as prepared as I would like.

2

Tim 'mithro' Ansell

@carlfk - While I hoped it would happen with Run #3, I think I didn't really believe it would happen, so not as prepared as I would like.

@Tim 'mithro' Ansell I just missed to save a 1x1 slot for Run #3, do you have already a rough schedule for Run #4?

gonsolo

@Tim 'mithro' Ansell I just missed to save a 1x1 slot for Run #3, do you have already a rough schedule for Run #4?

Hoping to have something next week.

1

@carlfk

FWIW, I've been primarily focused on two things connected to fpgas.online

(1) Getting to a position where I'm confident that we can create changes and be confident they don't break things.

(2) Getting the RPi's to do automated self testing of their FPGA board connectivity on boot going. Kind of following your idea of the GPIO test for the Arty.

@carlfk - I'm getting close to having a debian package you can install which has both bitstreams and scripts for checking an FPGA board is wired correctly.

Tim 'mithro' Ansell

@carlfk - I'm getting close to having a debian package you can install which has both bitstreams and scripts for checking an FPGA board is wired correctly.

close in hours minutes or days?

@carlfk - Within 24 hours, maybe within the next 8 hours...

1:46 a.m.

Depends on how distracted I get

3:05 a.m.

that is if I power it with a usb ps... POE dongle.. it gets the lights on but no hdmi / boot

3:05 a.m.

I have been told some of these poe dongle things are crap.

https://www.amazon.com/PoE-Converter-Ethernet-Charging-PT-PGC-AF/dp/B0CP296VXN

Gigabit PoE to USB C Converter, USB-C to Ethernet Adapter with 10W ...

PT-PGC-AF is an indoor PoE to USB-C adapter. It allows power and data obtained from an IEEE802.3af PoE injector or PoE switch to be provided to USB-C PD and network connectivity through the adapter. PT-PGC-AF features a white plastic housing design, making it portable for on-the-go use or for fix...

3:19 a.m.

recomened by another PS1er, but provides ethernet over usb-c so no go for pi

@Tim 'mithro' Ansell did you find some robust or reliable poe dongles?

@carlfk - Yes, I found a bunch of different things.

3:59 a.m.

@carlfk - This is the current "trusted" one - https://www.waveshare.com/poe-splitter-25w-type-c.htm

Industrial Gigabit PoE Splitter, options for 5V 5A Type-C / DC Powe...

Industrial Gigabit PoE Splitter, options for 5V 5A Type-C / DC Power Output Port, Onboard MPS Control Chip, Safer And More Stable | POE-SPLITTER-25W-TYPE-C

https://www.amazon.com/Waveshare-Industrial-Gigabit-Splitter-Onboard/dp/B0D7955R6N

Waveshare Industrial Gigabit PoE Splitter, 5V 5A Type-C Output Port...

Waveshare Industrial Gigabit PoE Splitter, 5V 5A Type-C Output Port, Onboard MPS Control Chip

4:24 a.m.

$20 vs $32

4:28 a.m.

im confused - whats the diff betten that at

4:28 a.m.

https://www.amazon.com/waveshare-Industrial-Protection-at-Compliant-Auto-Negotiation/dp/B09MC4QDYG

Industrial Gigabit PoE Splitter with Metal Case Protection, 5V 2.5A...

Features Incorporates original chip Si3404, high-efficiency, high-safety, stable-performance 10/100/1000Mbps auto-negotiation Ethernet port 802.3af/at-compliant PoE (Power over Ethernet) standard Isolated SMPS (Switching Mode Power Supply), effectively protecting the powered device 5V DC ou...

4:29 a.m.

ah.. 5a vs 2.5a

Using a hat might help with assembling the kit

4:41 a.m.

https://www.amazon.com/Ethernet-POE-Raspberry-802-3af-Compliant-Connecting/dp/B0BLT3MZB9/

Waveshare Power Over Ethernet HAT (E) for Raspberry Pi 4B/3B+, IEEE...

Overview The PoE HAT (E) is a mini 802.3af-compliant Power Over Ethernet accessory for the Raspberry Pi 3B+/4B. By using this small HAT and some proper 802.3af-compliant power sourcing equipments, it is possible to provide both network connection and power supply for your Raspberry Pi in only ...

4:42 a.m.

https://www.amazon.com/waveshare-Raspberry-802-3af-Compliant-Connecting-Compatible/dp/B0BKKCDQ6B/

Waveshare Power Over Ethernet POE HAT for Raspberry Pi 4B/3B+, IEEE...

Overview This is a mini IEEE 802.3af-compliant PoE HAT for Raspberry Pi 3B+/4B. If you already have an IEEE 802.3af-compliant PoE router or switch, with this HAT you can provide both network connection and power supply for your Raspberry Pi in one cable. Feature Suitable for the Raspberry Pi ...

4:45 a.m.

What is the difference between those 2 hats ?

doesnt matter. poe hat needs to be on the bottom to grab power from the poe headers on the pi

5:35 a.m.

so a poe hat needs to pass the 40 pins up to the pmod hat

look at this hat!!!

8:58 p.m.

https://digilent.com/shop/pmod-hat-adapter-pmod-expansion-for-raspberry-pi/

Pmod HAT Adapter: Pmod Expansion for Raspberry Pi

Shop now for USB test and measurement devices, FPGA development boards, programming solutions and educational products.

8:58 p.m.

er, not that hat, this poe hat:

8:58 p.m.

https://www.waveshare.com/product/raspberry-pi/hats/interface-power/poe-hat-b.htm

Power Over Ethernet HAT (Type B) For Raspberry Pi 3B+/4B, 802.3af P...

Power Over Ethernet HAT (Type B) For Raspberry Pi 3B+/4B, 802.3af Power-Sourcing Equipment Required

9:00 p.m.

$20. cheapest I can find that will work with the pmod hat, which will cover up the 0.91" OLED display

https://www.waveshare.com/poe-splitter-25w-type-c.htm is $20 USD direct from China?

Industrial Gigabit PoE Splitter, options for 5V 5A Type-C / DC Powe...

Industrial Gigabit PoE Splitter, options for 5V 5A Type-C / DC Power Output Port, Onboard MPS Control Chip, Safer And More Stable | POE-SPLITTER-25W-TYPE-C

here is shipping on 4 items:

5:13 a.m.

ordering from waveshare.com Registered Post Air Mail (ETD: 12-28days) $13.60 UPS (ETD: 12-15days) $59.85 UPS-DDP (ETD: 12-15days) $118.55 DHL (ETD: 9-12days) $40.07

5:13 a.m.

$100 and still over a week?!

5:14 a.m.

I think I will pick the $13 option.

Most products are shipped from Shenzhen, China. https://www.waveshare.com/shopping_q_a

Shopping Q&A

Waveshare

The 100$ option also includes taxes (DDP)

https://www.robotshop.com/products/waveshare-power-over-ethernet-hat-b-for-raspberry-pi-3b-4b-8023af-poe-network?qd=3e9ad6bd28dc429b90a6c584078d9fcc

5:44 a.m.

US distibuter.. so maybe I can get it sooner? No.

5:45 a.m.

This is an On Demand item and is not normally stocked. We will order it for you along with our next purchase from the supplier.

woot - pishop.us/checkout USPS Priority Mail (Delivery in 2-4 business days, not guaranteed)$13.18

@Tim 'mithro' Ansell you should get a few too https://www.waveshare.com/product/raspberry-pi/hats/interface-power/poe-hat-b.htm

Power Over Ethernet HAT (Type B) For Raspberry Pi 3B+/4B, 802.3af P...

Power Over Ethernet HAT (Type B) For Raspberry Pi 3B+/4B, 802.3af Power-Sourcing Equipment Required

@carlfk - I have a bunch of those.

Tim 'mithro' Ansell

@carlfk - I have a bunch of those.

will the pmod hat work on top of them?

@carlfk - It is kinda awkward

8:07 a.m.

@carlfk - I'm also about to get some of @psychogenic's pmod hats which are much nicer than the digilent ones.

1

Tim 'mithro' Ansell

@carlfk - I'm also about to get some of @psychogenic's pmod hats which are much nicer than the digilent ones.

Let me know how they work out in the fpgas.online setup, and if you'd need any mods.

@psychogenic - I'm just still trying to check if they will work mechanically for what I want to do but been distracted with other things.

1

Tim 'mithro' Ansell

@psychogenic - I'm just still trying to check if they will work mechanically for what I want to do but been distracted with other things.

I know the feeling. Lots of travelling on my side in the near future, but I'll keep an eye on @mentions, whenever they should appear.

psychogenic

I know the feeling. Lots of travelling on my side in the near future, but I'll keep an eye on @mentions, whenever they should appear.

can I have kicad files? mostly to see/add holes for the PoE header

2:28 p.m.

also reset - which have many places - see https://github.com/CarlFK/pj

GitHub - CarlFK/pj: Pi JTAG hat

Pi JTAG hat. Contribute to CarlFK/pj development by creating an account on GitHub.

2:29 p.m.

(which don't have Poe thing, yet :p

@carlfk - https://github.com/mithro/tinytapeout-demoboard-to-raspi

GitHub - mithro/tinytapeout-demoboard-to-raspi: Connects all I/O fr...

Connects all I/O from TT demoboard to RPi ribbon cable while still allowing the PMODs to be used with external modules. - mithro/tinytapeout-demoboard-to-raspi

Impossible!

9:45 p.m.

Probably possible. I'm curious how this works

@carlfk - It looks I must have dreamt I had already made the modifications....

@carlfk - The RPi side would use a connector more like https://www.adafruit.com/product/2187

SMT GPIO Header for Raspberry Pi HAT - 2x20 Short Female Header

For super-slim SMT designs, this custom-made, 2x20 short SMT female header can be used to plug a Raspberry Pi HAT onto any modern Raspberry Pi computer with 2x20 headers It's harder ...

Tim 'mithro' Ansell

@carlfk - The RPi side would use a connector more like https://www.adafruit.com/product/2187

are you talking about the POE support ?>

10:43 p.m.

support/pass though

10:45 p.m.

oh wait. this isn't a hat

@carlfk -