NETUSBee problem on TT

Troubles with your machine? Just want to speak about the latest improvements? This is the place!

Moderators: Mug UK, Zorro 2, Greenious, spiny, Moderator Team

Winner61
Atari User
Atari User
Posts: 30
Joined: Thu Mar 29, 2018 8:32 am

Re: NETUSBee problem on TT

Post by Winner61 »

Thanks all. Would be weird that it sometimes works (half) without power...
CartTT.PNG
You do not have the required permissions to view the files attached to this post.
Winner61
Atari User
Atari User
Posts: 30
Joined: Thu Mar 29, 2018 8:32 am

Re: NETUSBee problem on TT

Post by Winner61 »

I measure 5v on one end of the fuse and 2.5v on the other.... Just like Arne's picture ... Where is this 2.5v come from you could ask....
User avatar
Arne
Atari Super Hero
Atari Super Hero
Posts: 876
Joined: Thu Nov 01, 2007 10:01 am

Re: NETUSBee problem on TT

Post by Arne »

Winner61 wrote: Tue Oct 05, 2021 12:29 pm Where is this 2.5v come from you could ask....
I assume you used a multimeter and not an oscilloscope?

The multimeter averages the voltage over a given time where a scope shows you the real voltage curve i.e. the voltage can be 0V for (let's say) 10ms and then 5V for 10ms. On average it's 2.5V - what your multimeter displays.
I once had a colleague who studied (!) electronics and stumbled over this and I had to explain that to him :roll:
Image
Winner61
Atari User
Atari User
Posts: 30
Joined: Thu Mar 29, 2018 8:32 am

Re: NETUSBee problem on TT

Post by Winner61 »

Thanks for the explanation Arne. Yes, I used a simple multimeter. Quite confident now that that is the culprit. First get a replacement fuse :-) . Still wandering why the fuse blew ... old age maybe, about 30 yrs ...
stormy
Atari God
Atari God
Posts: 1273
Joined: Tue Jan 26, 2016 12:39 pm

Re: NETUSBee problem on TT

Post by stormy »

So, exactly the same issue with the guy on the exxos forum.
czietz
Hardware Guru
Hardware Guru
Posts: 1922
Joined: Tue May 24, 2016 6:47 pm

Re: NETUSBee problem on TT

Post by czietz »

Did you happen to connect any power-hungry USB devices to your NetUSBee in the past? In your schematic it shows as 5.0 A fuse, but other TT schematics have a 0.5 A fuse for the cartridge port. I have no idea if one of the fuse ratings is a typo of if this was changed for certain TT revisions. Officially, USB devices (USB 1 and 2) are allowed to draw 0.5 A, but some need considerably more.
User avatar
krupkaj
Captain Atari
Captain Atari
Posts: 321
Joined: Wed Oct 22, 2003 11:00 pm
Location: Olomouc/Czech Republic
Contact:

Re: NETUSBee problem on TT

Post by krupkaj »

OK, I also checked it and in my TT is glass fuse and it is OK. So this is not the case in my TT. And still it would not probably work with the Mighty Mic.
Winner61
Atari User
Atari User
Posts: 30
Joined: Thu Mar 29, 2018 8:32 am

Re: NETUSBee problem on TT

Post by Winner61 »

I cut the bugger out, this is what was in there. Must be a typo in the schematic.
fuse.PNG
You do not have the required permissions to view the files attached to this post.
czietz
Hardware Guru
Hardware Guru
Posts: 1922
Joined: Tue May 24, 2016 6:47 pm

Re: NETUSBee problem on TT

Post by czietz »

This confirms what I've said before: https://www.atari-forum.com/viewtopic.p ... 66#p402766. Please be very careful when connecting USB devices to your NetUSBee! In some hardware revisions, the NetUSBee's overcurrent protection is inoperative; therefore, it's disabled by default by the driver, afaik. And even if you enable it, you can still draw more current from the USB ports than a 0.5 A fuse can handle.

(EDIT: To clarify: The Lightning VME is not affected by this, as it a) is not powered from the cartridge port, and b) has a two-stage overcurrent protection.)
Winner61
Atari User
Atari User
Posts: 30
Joined: Thu Mar 29, 2018 8:32 am

Re: NETUSBee problem on TT

Post by Winner61 »

microfuse.PNG
Put in one of these and all is working network & USB :D
did not bother to take the whole mboard out, soldered it on the wires that were left when cutting the old fuse out. 8)
You do not have the required permissions to view the files attached to this post.
User avatar
ChainQ
Atari maniac
Atari maniac
Posts: 83
Joined: Sat Jan 21, 2017 9:35 am
Location: 1 AU, EU, DE/HU
Contact:

Re: NETUSBee problem on TT

Post by ChainQ »

I decided to resurrect this topic, because I had the exact same problem as the original post (except it wasn't in a TT).

So it turns out, the NetUSBee (at least some revisions, including the Lotharek one I have) has a slightly under-designed RESET circuit, and also a few other issues, that might cause non-operation in several machines, under some conditions. I had this problem in my STE. Exact same problem as the first poster in this topic. USB test passed, network test didn't, all failed. Also UIPTool didn't work.

The gist of it is, that the simplistic RESET circuit the RTL8019AS has on this card lets the chip to attempt to initialize before the power really stabilizes (especially on PSUs with slow ramp up!), and in turn this leads to the chip latching wrong values from its configuration pins, probably because the VREF is too low (despite having internal pull-downs on those pins according to the documentation). Also, the 8/16 bit bus mode selector pin is not connected, which might result the chip sometimes initializing into 16bit mode, while only 8 data pins are connected. And there might be more... I made some bodges, and now the card works fully reliable in my STE, at least with NETUS.TOS test, and also UIPTool. I still wait for parts to rebuild the reset circuit, then do more test, because some of the bodges are really ugly, and I want to remove them. But hey!

The changes I did, all on the RTL8019AS:
- grounded IOS0-IOS3 pins, directly with a solder blob (Yuck. These are pins with internal pull-downs in theory, so a proper/improved RESET circuit would make this bodge unnecessary.)
- grounded IOCS16B pin via a 22K resistor (the chip's manual asks for 27K for 8 bit operation, close enough)
- grounded AUI via a 10K resistor (this is probably not needed, but the chip's manual asks for it, so whatever)

I used this RTL8019AS manual. I also used this Open Source 8 bit ISA card based on the same chip for reference.

Please note this also isn't a silver bullet, at least someone else tried to implement it and it didn't work in his TT (but his card also had no USB parts, so no way to verify against anything), but if the USB part works on your system, and the network isn't (as in the opening post), it's a good chance this might be the/a problem. What pointed me in this direction, that the card in my STE worked when I ran the entire shebang from my bench PSU, but it didn't work from the stock PSU. With nothing else changed. And of course the bench PSU has a really fast power ramp-up, compared to the stock PSU. (On the other hand, some modern aftermarket PSUs might even have a slower(!) ramp-up than the stock one, making this problem worse! But it's not the PSU's fault!)

Proper fix: I'll try to implement a "proper" reset circuit with an MCP101-450 power monitoring chip. Sadly the 8019AS is an ISA chip, and the ISA reset signal is active high, which is somewhat challenging to produce without side effect, just with passive components. Thanks Intel! Also, kudos to Atari for making an expansion slot without a reset signal... Oh well. :) I know it's technically not needed for ROMs...

I posted some pictures in this tweet. I'll update when I manage to update my card with the better reset circuit. Also, I'm not a hardware guy by any stretch of the imagination, just a software guy with a soldering iron (YOLO!), so please correct me if I got something very wrong somewhere. But hey, I made it work, at least for me!
Free Pascal Compiler for m68k - an Amiga fan outside his comfort zone
czietz
Hardware Guru
Hardware Guru
Posts: 1922
Joined: Tue May 24, 2016 6:47 pm

Re: NETUSBee problem on TT

Post by czietz »

A very good find and a plausible explanation. By the way: the PSU used in the TT can take up to 5 seconds to stabilize (according to its specs; I didn't measure it). That might explain why these NetUSBee issues are more common when used with a TT.
User avatar
ChainQ
Atari maniac
Atari maniac
Posts: 83
Joined: Sat Jan 21, 2017 9:35 am
Location: 1 AU, EU, DE/HU
Contact:

Re: NETUSBee problem on TT

Post by ChainQ »

I have some more evidence that point towards the reset/latching of wrong config values, other than just a working bodge. For example, first I wired up a resistor, to keep the card in reset as long as the machine really started to boot (I mean, TOS boot screen appears), and then bring it out of reset by hand. This worked first try, and in fact this was the first time ever the card worked fully with the stock PSU. However, when I used this "circuit" later to reset the card after going into a non-working state during power-on, it didn't work. And the manual writes, RTL8019AS doesn't re-latch its configuration after power-on, not even on a hardware reset...

So I theorized, what if the chip actually starts, just with the wrong config, and tries to answer on the wrong IO address? Ironically, this could have been tested with some software hack, scanning for the chip on various addresses, but I was too lazy to write one, so I just did the solder-blob first, as the iron was hot already. And then it worked, first try. So yeah.

I just didn't want to slam this all in my post, it was already long enough.
Free Pascal Compiler for m68k - an Amiga fan outside his comfort zone
User avatar
sporniket
Captain Atari
Captain Atari
Posts: 174
Joined: Fri Feb 16, 2018 5:39 pm

Re: NETUSBee problem on TT

Post by sporniket »

czietz wrote: Wed Jun 15, 2022 6:31 am By the way: the PSU used in the TT can take up to 5 seconds to stabilize (according to its specs; I didn't measure it). That might explain why these NetUSBee issues are more common when used with a TT.
There is a "Power Good" signal coming from the PSU if I understand correctly. If it is suitable, the TT version of the reworked reset circuit of the USB could use it, maybe ?
czietz
Hardware Guru
Hardware Guru
Posts: 1922
Joined: Tue May 24, 2016 6:47 pm

Re: NETUSBee problem on TT

Post by czietz »

sporniket wrote: Wed Jun 15, 2022 12:19 pm There is a "Power Good" signal coming from the PSU if I understand correctly. If it is suitable, the TT version of the reworked reset circuit of the USB could use it, maybe ?
But that signal is not available at the cartridge port and therefore not available to the NetUSBee. Neither is the reset signal (as ChainQ had already mentioned). The TT itself and any internal (i.e., VME) add-on cards are of course held in reset until power good becomes asserted.
User avatar
Cyprian
10 GOTO 10
10 GOTO 10
Posts: 2579
Joined: Fri Oct 04, 2002 11:23 am
Location: Warsaw, Poland

Re: NETUSBee problem on TT

Post by Cyprian »

what about adding the reset button to the NetUSBee?
Lynx I / Mega ST 1 / 7800 / Portfolio / Lynx II / Jaguar / TT030 / Mega STe / 800 XL / 1040 STe / Falcon030 / 65 XE / 520 STm / SM124 / SC1435
DDD HDD / AT Speed C16 / TF536 / SDrive / PAK68/3 / Lynx Multi Card / LDW Super 2000 / XCA12 / SkunkBoard / CosmosEx / SatanDisk / UltraSatan / USB Floppy Drive Emulator / Eiffel / SIO2PC / Crazy Dots / PAM Net
Hatari / Steem SSE / Aranym / Saint
http://260ste.atari.org
User avatar
ChainQ
Atari maniac
Atari maniac
Posts: 83
Joined: Sat Jan 21, 2017 9:35 am
Location: 1 AU, EU, DE/HU
Contact:

Re: NETUSBee problem on TT

Post by ChainQ »

Cyprian wrote: Wed Jun 15, 2022 12:26 pm what about adding the reset button to the NetUSBee?
As I wrote (and my experiments kinda confirmed, also the chip's manual confirms), retroactively resetting the RTL8019AS will not reset any wrong configuration it might have picked up earlier. So going down this path you'd end up with a reset SWITCH, which you have to change to reset position before every power up, and then manually bring the card out of reset. Not something very nice and practical. There are better ways.

Again, I'm working on an improved reset circuit for the ethernet chip using a real power monitoring chip (that might sound complicated, but it's still a really simple part with only 3 pins), which could be then hopefully retrofitted to existing cards, and would solve this problem for good. But I'm waiting for parts, before I can do the experiment, then post the results.

(An alternative idea I had would be to use some sort of flip-flop logic, which would keep the chip in reset, until say, the ROM3/4 lines are first asserted by the machine (this is first happening when TOS is looking for expansion ROMs, I assume?). This is basically the equivalent of a reset switch I described above, but it would be fully automatic. But I don't think it will be necessary, but maybe it's an useful idea for new designs, if there's a few unused NAND gates at hand.)
Free Pascal Compiler for m68k - an Amiga fan outside his comfort zone
User avatar
sporniket
Captain Atari
Captain Atari
Posts: 174
Joined: Fri Feb 16, 2018 5:39 pm

Re: NETUSBee problem on TT

Post by sporniket »

czietz wrote: Wed Jun 15, 2022 12:21 pm
sporniket wrote: Wed Jun 15, 2022 12:19 pm There is a "Power Good" signal coming from the PSU if I understand correctly. If it is suitable, the TT version of the reworked reset circuit of the USB could use it, maybe ?
But that signal is not available at the cartridge port and therefore not available to the NetUSBee. Neither is the reset signal (as ChainQ had already mentioned). The TT itself and any internal (i.e., VME) add-on cards are of course held in reset until power good becomes asserted.
Ha, I forgot we were out of the casing.
susher
Obsessive compulsive Atari behavior
Obsessive compulsive Atari behavior
Posts: 141
Joined: Sat Oct 03, 2015 7:06 pm

Re: NETUSBee problem on TT

Post by susher »

You could use a zenner diode to trigger the power-on of the board via a MOSFET. In that way the board would only be enabled after the +5V rail was fully up to spec.

On my TT there are other issues (and this has been since new). No "write using read" devices have ever worked, e.g. Spectre GCR.

Looking at the oscilloscope trace on any of the address lines probably shows why, they ring like bells! The transition from 0V -> 5V is more 0V -> 3V -> 1V -> 4V -> 2.5V -> 5.5V -> 4V -> 5.25V -> 4.5V etc. Often it only stabilises just before transitioning back down to 0V via another set of oscillations. It's amazing that it can actually read the contents of ROMs at all.
User avatar
frank.lukas
Hardware Guru
Hardware Guru
Posts: 2210
Joined: Tue Jan 29, 2008 5:33 pm
Location: Germany

Re: NETUSBee problem on TT

Post by frank.lukas »

I send Lotharek a mail if he can mod the design ...
fancy Atari Musik anDA Dance "Agare Hinu Harukana" 1998 ATARI http://www.youtube.com/watch?v=JX10fxb5eYE
mikro
Hardware Guru
Hardware Guru
Posts: 2769
Joined: Sat Sep 10, 2005 11:11 am
Location: Kosice, Slovakia
Contact:

Re: NETUSBee problem on TT

Post by mikro »

While this is not related to the ongoing discussion with the reset signal (I can't observe such behaviour on neither of my machines with Lotharek's NetUSBee -- MegaSTE, TT, Falcon) it is related to the TT itself. After quite a few tests and discussions I have discovered that my TT simply doesn't supply enough current via ROM port. As soon as I connected my keyboards via USB hub, all problems were gone.

The symptoms were pretty misleading, one keyboard responded much slower than the driver expected, another one didn't respond at all.

If I find time, I'll try to measure the exact values, maybe even compare it with MegaSTE and the same TT without NOVA (as I suspect this eats most power) but for everyone else in the meantime -- this could be worth trying on your Atari as well.
Post Reply

Return to “Hardware”