03:24 <d1b2> <sharebrained> Hopefully simple question about the Glasgow memory-floppy applet: I tried it out last night, and found a lot of 0xFFs in the raw stream files I took from a 720K 3.5” (90mm) disk. As I understand the code, 0xFE indicates the end of a record, with the LSB set if an overflow occurred, so 0xFF. There were approximately 600-ish of each of 0xFE and 0xFF for a read of 80 tracks and both heads. The pulse interval statistics I viewed for various heads

03:24 <d1b2> and tracks looked correct. I take the frequent 0xFFs to mean the software on the host wasn’t emptying the USB FIFO on the Glasgow quickly enough. And yet the benchmark applet tells me I’m doing 42MB/second. And my experience streaming USB 2.0 HS from HackRFs has been reliable at 20Msps (40MB/s). Any thoughts on if this is an actual problem?

03:27 <d1b2> <sharebrained> I’m curious if this pending issue could have anything to do with it: https://github.com/GlasgowEmbedded/glasgow/issues/113 If so, I’ll try to make sense of the issue and update memory-floppy accordingly. But having looked at the “finished” applets in that issue, I haven’t yet grasped what changes I’d need to make. But I’m good at staring at code, so it’ll eventually click... if working on this issue would be fruitful.

04:30 <whitequark> sharebrained: if you hit a FIFO overflow IIRC you will get an error or a hang, you will *not* get silent data corruption

04:31 <whitequark> it is completely unrelated to #113

04:31 <whitequark> I would need to recall how the applet works exactly to debug your issue.

04:39 <d1b2> <sharebrained> @whitequark Thanks. I'm happy to look into it myself. I just needed to validate that I wasn't doing something fundamentally wrong or misunderstanding how it should work.

12:47 <_whitenotifier> [glasgow] electroniceel commented on issue #220: Qualification tests for revC2: pre INA233 - https://git.io/JLQDV

12:57 <electronic_eel> anyone already got a glasgow board and wants to help developing?

12:57 <electronic_eel> here is an issue that doesn't look too complicated to me, so it should work well if you want to get into glasgow development: https://github.com/GlasgowEmbedded/glasgow/issues/252

12:58 <electronic_eel> no fpga or nmigen skills required, this should be pure python

13:28 <_whitenotifier> [glasgow] electroniceel opened issue #253: Add documentation of ics used on Glasgow to the archive - https://git.io/JLQyQ

13:29 <electronic_eel> and one more issue where help is wanted: https://github.com/GlasgowEmbedded/glasgow/issues/253

13:29 <electronic_eel> for helping with this you don't need a Glasgow board at all

19:20 <_whitenotifier> [glasgow] marcan commented on issue #220: Qualification tests for revC2: pre INA233 - https://git.io/JLQh7

19:25 <_whitenotifier> [glasgow] electroniceel commented on issue #220: Qualification tests for revC2: pre INA233 - https://git.io/JLQjv

19:34 <_whitenotifier> [glasgow] marcan commented on issue #220: Qualification tests for revC2: pre INA233 - https://git.io/JLQj3

19:41 <_whitenotifier> [glasgow] electroniceel commented on issue #220: Qualification tests for revC2: pre INA233 - https://git.io/JLQj6

19:47 <eddyb> I just realized the reason those "mining" risers have their own 12V power is because an x16 card is allowed to pull more power and when you have more than a few on one motherboard, it's really easy to overload it

19:48 <eddyb> anyway I should pull the flash already. I guess I can go looking for the chip puller thing I might have, just in case it is here somewhere

21:13 <gruetzkopf> you simply cant push 75W (or even 25W) through the usb cable

21:13 <gruetzkopf> anyways, they abuse V+ for CLKREQ iirc

21:15 <tnt> V+ or GND can't remember which ... and the other is PERST.

21:16 <tnt> Ah no, ~PEWAKE is VBUS and ~PREST is GND.

21:17 <d1b2> <Attie> @tnt ... they don't use GND for GND?

21:18 <tnt> Nope.

21:19 <tnt> Only the DRAIN (i.e. the gnd for the sstx/ssrx) is used for ground.

21:19 <tnt> the GND of "usb2" is used for signal.

21:19 <mwk> they what!?!?

21:21 <d1b2> <Attie> eesh

21:21 <tnt> https://i.imgur.com/q92ZqIb.png

21:22 <tnt> That's the pinout I had to use when making my own M2 -> USB3 compatible with those miner breakout things.

21:40 <eddyb> it makes sense if you think about it :P

21:40 <eddyb> they get 3 twisted pairs to work with

21:40 <eddyb> and not much else

21:44 <tnt> sure, I mean it's truly differential so there should be no current on DRAIN, just need the common reference.

21:45 <d1b2> <Attie> interesting, thanks! I couldn't find a pinout / schematic with a quick search the other day...

21:56 <tnt> Unrelated but potentially interesting for glasgow users : https://intona.eu/en/products/buy/7054/

21:56 <tnt> First high-speed usb isolators that's reasonably priced I ever see ...

22:15 <electronic_eel> tnt: see here for my tests of isolators: https://www.eevblog.com/forum/chat/high-speed-usb-isolator/msg2677443/#msg2677443

22:16 <electronic_eel> 335 EUR for a usb 2 hs isolator with 2.5 kV isn't particularly cheap

22:17 <electronic_eel> the exys one i tested has better specs and is cheaper

22:20 <tnt> oh ok, I guess my info was outdated, I think it's been like 5y since I last searched for one.