uip-tool

[sqward]

FTP support! Coming soon.

Uiptool early FTP support demo

[jonord]

I'm looking for a like-button or thumbs up-emoticon, but, alas, none of that exists. Anyway. Yay!

[neanderthal]

Cool,have to dig in the sources when its finished and maybe do a local port
And while talking of fixing stuff more than ftp support,anyone else noticed the odd thing when exiting the atari side server thing (that is quitting uip-tool on the atari) your browser tab/window will begin to draw quite a lot of CPU time?
In my case about 30% extra or so with firefox under ubuntu mate on the core its running on.
Not a big issue or anything,I just happened to notice since cpu fan would not go down to idling speed.
Btw, a page reload fixes it since active script is flushed and no server connection

[Bikerbob]

A couple of days ago I was using the same NETusbee on the same TT, no problems.. yesterday and today.. now I get DHCP IP: and flashing cursor.

I know there is the uIP.cfg you have somewhere.. this is generated and supposed to be edited in a text editor. Couple of days ago I did not even know about the cfg it just worked.

What has changed? How do I fix it.

[susher]

Knowing the failure modes of the NetUSBee on the TT, the "write using read" is failing and so the ethernet interface isn't doing anything. The other failure mode I've seen s the system not being able to read the MAC address and getting all zeros.

NetUSBees are extremely finicky and will stop working on specific machines for the most minor or reasons.

e.g. on my STE, it will work with the stock power supply but not with the Exxos power supply which has less noise and better voltage control. i.e. the hardware only just about works under the best of conditions.

[Galvez]

Now that we know that Lotharek made changes into the original design [1] I wonder if the origin of your NetUSBee has an impact too in the problems you're having.


[1] http://www.atari-forum.com/viewtopic.ph ... 36#p400230

[sqward]

I never had any issues with Netusbee with my machines, though, anecdotally, TT's romport was never very compatible. Most if not all issues I hear about are on TT.

@Bikerbob: I don't know what has changed in your environment or your Atari computer. If other machines in the network can do DHCP then it's unlikely using static IP config would help. Can you use USB or Sting/Mintnet on that system?

[neanderthal]

e.g. on my STE, it will work with the stock power supply but not with the Exxos power supply which has less noise and better voltage control. i.e. the hardware only just about works under the best of conditions.

Wow,,that sounds really bizarre,does the NetUSB have some coils or something inside?,pretty much the only thing that comes to mind,Exxos uses a old school low freqeuncy mains-transformer in his design.The usual RF-shielding cant filter that out.

[Bikerbob]

This is a Lotharek design. Yes.
The TT has lightning VME working perfect.
The netusbee lights are powered everything seems to be good except nothing. I will try static for argument. I know this worked 2 days ago I transfered a lot of stuff through the network.. in fact. Maybe this is interesting the router had not even released the lease on the IP address .. I had to reboot the router. I can try this on a megaST I have here. I guess lóok into one of these usb2net adapter for vme

[sqward]

Bikerbob: you can press F1 key to toggle between manual na DHCP.

[susher]

e.g. on my STE, it will work with the stock power supply but not with the Exxos power supply which has less noise and better voltage control. i.e. the hardware only just about works under the best of conditions.

Wow,,that sounds really bizarre,does the NetUSB have some coils or something inside?,pretty much the only thing that comes to mind,Exxos uses a old school low freqeuncy mains-transformer in his design.The usual RF-shielding cant filter that out.

I notice that Lotherek isn't using TTL 74LS chips and is using CMOS HCT ones instead. That can introduce power line noise issues. Maybe a capacitor between the Vcc and GND pins of the 74xxxxx chips would help? I've tried pull-ups and that made no difference, and anyway, the TT's address and data signals are cleaner when looked at on an oscilloscope.

[Bikerbob]

Well this is a kick in the balls, I take that Netusbee and another I have to my workbench with my megaST4 - Using the NETUSB05 driver set.. Neither of them can be found on the MegaST4 - USB drivers.. Fn messed up.

James

[Bikerbob]

Well I can confirm the Netusbee seems to pick what machine it runs on.

Hooked it up to a megaSTE I took down off the shelf, everything works, USB Uiptool. all good

Hmmmm

James

[Perdrix24]

Well this is a kick in the balls, I take that Netusbee and another I have to my workbench with my megaST4 - Using the NETUSB05 driver set.. Neither of them can be found on the MegaST4 - USB drivers.. Fn messed up.

James

On my MegaST with TOS 1.04 I have to run POOLFIX4.PRG prior to the USB drivers.

[Bikerbob]

This megaST that I have was Pak68/3 with 3.06 in it.

James

[sqward]

Netusbee doesn't work in my MegaST with Pak68/3.

[Bikerbob]

AHHHH.. there ya go.. I was just about to post that sqward.. It is a megaST with a Pak68/3 and my TT030.. So nothing with 68030

Took the Pak out.. and it works. SO it must be a CPU issue then if people with TT030s (there have been several here that posted) OR speed issue? Could you slow the system down with a autofolder program.. maybe its the running at 30mhz that's doing it?

frank has a fully tricked out Pak68/3 not sure on a TT.. frank have you run a netusbee?

James

[sqward]

I think that PAK changes some details about signal timings when accessing the romport. Netusbee works perfectly in Falcon with 030 and even better with 060 at 100Mhz.

It works in some TT's. I wonder if this is about the 16Mhz and 32Mhz versions?

[Mark_G]

I use Lotharek Netusbee on my TT, without any problem (so far).
It works fine with UIPtool under Magic 6.20.

However, I got one strange behaviour :
I used a Microsoft optical wheel mouse (PS2/USB) attached on the USB port of Lotharek Netusbee.
With the new NETUSB05 driver set this mouse works flawlessly.

When I use the old drivers (umouse) the TT stalls after a while (couple of minutes or more).
In the past a got an original Netusbee and the mouse worked perfect with the umouse drivers.

Sadly, the original Netusbee fried from a lightning stroke I got some years ago. So I'm not in the possibility to test him with the new drivers.
Mark

[Galvez]

I use Lotharek Netusbee on my TT, without any problem (so far).
It works fine with UIPtool under Magic 6.20.

However, I got one strange behaviour :
I used a Microsoft optical wheel mouse (PS2/USB) attached on the USB port of Lotharek Netusbee.
With the new NETUSB05 driver set this mouse works flawlessly.

When I use the old drivers (umouse) the TT stalls after a while (couple of minutes or more).
In the past a got an original Netusbee and the mouse worked perfect with the umouse drivers.

Sadly, the original Netusbee fried from a lightning stroke I got some years ago. So I'm not in the possibility to test him with the new drivers.
Mark

We're detecting more differences between NetUSBees depending on who made them, just yesterday we discover a new one. According to the NetUSBee documentation to write data (LSB) you must read some address (0x00FA8000), but reading that address doesn't work well with NetUSBees produced by Lyndon and Tuxie.

https://github.com/freemint/freemint/co ... 305a50186d

[frank.lukas]

Netusbee doesn't work in my MegaST with Pak68/3.

For me too. It works not with my PAK68/3 Mega ST and not with my normal Atari TT with SMD CPU. But I have Lightning ST and Lightning VME for an alternatives ...

[susher]

I think that PAK changes some details about signal timings when accessing the romport. Netusbee works perfectly in Falcon with 030 and even better with 060 at 100Mhz.

It works in some TT's. I wonder if this is about the 16Mhz and 32Mhz versions?

Having leant a lot this week about decoupling caps and then looking at the Lotharek NetUSBee those caps look an awfully long way away from the chips. Also the 74 series chips look to be HCT technology, which would be more sensitive. I wonder if that's the issue?

I've a bunch of decoupling caps arriving tomorrow for another project. I may tack a one on the Vcc and GND legs of each of the chips and see what happens.

[neanderthal]

I think that PAK changes some details about signal timings when accessing the romport. Netusbee works perfectly in Falcon with 030 and even better with 060 at 100Mhz.

It works in some TT's. I wonder if this is about the 16Mhz and 32Mhz versions?

Having leant a lot this week about decoupling caps and then looking at the Lotharek NetUSBee those caps look an awfully long way away from the chips. Also the 74 series chips look to be HCT technology, which would be more sensitive. I wonder if that's the issue?

I've a bunch of decoupling caps arriving tomorrow for another project. I may tack a one on the Vcc and GND legs of each of the chips and see what happens.

Interesting issue,Then again I dont have a NetUSBee,I compiled a own version of Sqward:s version of uip-tool for my internal ISA-version of a network-card on the falcy. But do know that could not get the PRO-24 cartridge to run on the falcon,even with all trix to get it behave 'ST-ish'.
Then again those key-carts are a bit of 'black-boxes'.

But afaik about LS and HCT,they do have some level trigger stuff that are slightly different..I did use some HCT chippys in my ISA-adapter design,,but there I used the standard bus timing and so on.The thing with HC could be the over/undershoot tho?(each chip had its usual 100nF decouple tho in my design)
A bit depending how the level triggering and all that is done in the device,I would quess that there might be some slight timing issues or something?
I've never even checked the ROM-port for timings since almost always use it for some key-dongle or such.
In theory there should be no problems reading the thing,it is after all a 'ROM'-port,,no r/w line here,just a chip select and timing.
Then again,,the 'write' ,since we only have the Address lines as output?
And there is some differences on 000 versus 030 bus timings aswell.

[susher]

I'm thinking that it has to be electrical given the behaviour with my STE and different power supplies. It works the Atari one (which overshoots on the +5V line a bit) but not with the Exxos one which provides a very accurate (and less noisy) +5V.

Maybe the board just needs more that 5V. I know my TT's +5V is very accurately 5V too.

[susher]

I'm thinking that it has to be electrical given the behaviour with my STE and different power supplies. It works the Atari one (which overshoots on the +5V line a bit) but not with the Exxos one which provides a very accurate (and less noisy) +5V.

Maybe the board just needs more that 5V. I know my TT's +5V is very accurately 5V too.

OK, I've looked again. The decoupling capacitors are OK. The only issue I can see then is that they are 74HCT parts, which are generally slower than 74LS parts, though get faster as the Vcc voltage increases towards 6V. Given that changing the Vcc voltage on the STE caused the NetUSBee to fail, empirically this seems to be the most logical possibility. i.e. the cumulative propagation delay through the chips is just too high if the voltage is precisely 5V but sneaks in if slightly higher.