Colorlight 5A-75B: A dive into a popular low-cost ECP5 (FPGA) development board

Posted by gregsadetsky 4 days ago

Counter47Comment14OpenOriginal

Comments

Comment by tverbeure 1 day ago

I’ve used this board for an LED cube, but in general I don’t recommend it as a cheap general purpose FPGA board due to the lack of inputs, the different versions, and the uncertainty about getting access to the JTAG pins on later versions.

The Colorlight I5 is a much better choice for that. It’s also a high volume and this cheap HUB75 controller board but with a plug-in configuration so they all pins can be used as input or output.

See my old blog post about it: https://tomverbeure.github.io/2021/01/22/The-Colorlight-i5-a...

Comment by zoobab 21 hours ago

"getting access to the JTAG pins"

I have seen some of the boards being reflashed with DirtyJTAG and openfpgaloader.

JTAG pins might dissappear from the board?

There is a video here "Pico-DirtyJTAG: Cheap JTAG Programmer! Raspberry Pi Pico for Lattice ECP5 + Icestudio":

https://www.youtube.com/watch?v=Yj1X4PIoQGg

Comment by tverbeure 13 hours ago

I have some HUB75 boards where the JTAG pins are covered with solder mask or only available by scratching solder mask from non-descript traces.

These boards are programmed in production with SPI pins or assembled with preprogrammed PROMs.

Comment by sho_hn 1 day ago

I use this one a lot: https://store.mesanet.com/index.php?route=product/product&pr...

Last month I reverse-engineered a 25 years old embedded system that controls a ground-penetrating radar, which isn't commercially available anymore. It uses ECP parallel port to talk to the radar's Rx/Tx fibre optic control unit, so I used the Mesa board to implement that part of it, hooked it up to a Raspberry Pi and put the touchscreen HMI on that.

Comment by monster_truck 16 hours ago

Have you written about this at all?

Curious if you've played around with a MRMS-esque (Multi-Radar Multi-Sensor) approach to ingesting multiple passes on the same area to "get more resolution for free"

Comment by sho_hn 8 hours ago

> Have you written about this at all?

I'll be giving a (short, somewhat shallow) talk about this and some other things at this year's Akademy, about "Taking KDE to the Pyramids".

> Curious if you've played around with a MRMS-esque (Multi-Radar Multi-Sensor) approach to ingesting multiple passes on the same area to "get more resolution for free"

I haven't! We do have some interesting bespoke highly-stacking radar hardware, and do somewhat unusual things like complex transilluminations, but I'm really more a nuts and bolts software engineer than the actual sensor expert :-)

Comment by monster_truck 4 hours ago

Oh I'm glad that's soon, I don't usually use a calendar but I've set a reminder to check.

Ask chat about it! I have a pretty limited understanding of any of this shit but: I wasn't really able to find a satisfactory explanation for why we do not currently do this with doppler radar[1]. A great deal of CONUS has 3+ overlapping local/national radars from ideal angles & elevations, the smallest cell on the mosaic being ~1km felt "like bullshit/not very 2026". I don't care if the radars are nearly twice my age, they're still quite good.

With Codex 5.2 (and some preprints wrt MRMS from countries that don't even really get tornadoes) I got to the point where this 5800X3D (and 128gb of ram) could attenuate and plot all of conus about as fast as new sweeps are published, but it makes my room very hot. Now that fall is coming and the models are so much better I will revisit it. Will email you if anything good happens. Here's an old mp4 in the meantime https://i.imgur.com/EZjM46W.mp4

I'm certain if you threw Astra at it on high with a pile of slightly different scans it would do something useful.

1) This isn't something they would do half-cocked, I learned weather nerds are very serious about only presenting exactly what can be measured with known error bounds and detest anything resembling speculation

Comment by darknavi 1 day ago

Mostly unrelated, but I (mostly Copilot) spent some time re-authoring a software stack of HUB75 displays in our office with rotating pixel art. A previous team set them up on a overly-complex stack that relied on a bunch of services that haven't existed for years, getting them stuck in place for the last few years.

While I found the hardware pretty simple, HUB75 displays seem _very_ picky about over-volting. The setup is basically a simple controller (such as ESP32-Trinity) plugged directly into one HUB75 (potentially chained to a few more).

When running off of USB power it pulls the voltage down to ~4.7v and the image looks great. When running off our junky Amazon 5v barrel connectors the image gets all sorts of artifacting. Quite frustrating!

Anyhow, I can't recommend HUB75 displays enough. Toss some retro pixel art on there and it is such a nice little picture frame sized treat!

Comment by GeorgeHahn 1 day ago

I suspect you can fix that by adding a level shifter between the display and the ESP32. Input voltage thresholds are relative to the voltage rail, which is why reducing the display's voltage relative to the ESP32's (3.3v) outputs make it able to read the data.

Comment by q3k 16 hours ago

This really doesn't sound right. I've ran these from hundreds-of-watts industrial PSUs without any voltage sag and they did great. I suspect the clanker is making some shit up here, as always.

Comment by darknavi 8 hours ago

I am also not using anything near an industrial PSU. I am using cheap ($20) "5V" DC barrel chargers from Amazon (similar to a shabby laptop charger).

Are you sure you aren't a clanker? You seemed to have missed some critical details from my short post.

Comment by peterus 1 day ago

Another low cost fpga board I've found is the EBAZ4205, which has a Zynq7010 SoC (fpga and dual core arm CPU). I got one a few years ago for 40 AUD. They were originally Bitcoin miner control boards. https://theokelo.co.ke/getting-starting-with-ebaz4205-zynq-7...

Comment by gregsadetsky 23 hours ago

Also another recent FPGA thread: https://news.ycombinator.com/item?id=49531525

Comment by q3k 16 hours ago

More accurate credit for the RE effort would be 'q3k took scans done by someone else for the first board and turned them into a netlist, then other people did the rest'. It's really just hosted under my GH user at this point, nothing more.

Comment by basil_io 1 day ago

[dead]