exxos's Atari, Amiga & retro forum

I find the netusbee lockup on one of the H4's completely strange. The other cartridges I have work fine but not the netusbee. The only difference with the H4's is the SMT GLUE and MMU on the one that locks up. @PhilC , you built a H4 with SMT GLUE and MMU, if you have a netusbee give it a try if time permits. Its an easy test, just plug it in and see if it boots. If it hangs then maybe there's a difference with the SMT chips. If not I have no idea. Nothing jumps out at me but I suspect one would need to connect a logic analyzer to the cartridge port and see what's in the wrong state.

@PaulJ I will try and give it a go later on.

PaulJ wrote: 11 Apr 2020 14:53 I find the netusbee lockup on one of the H4's completely strange. The other cartridges I have work fine but not the netusbee. The only difference with the H4's is the SMT GLUE and MMU on the one that locks up. @PhilC , you built a H4 with SMT GLUE and MMU, if you have a netusbee give it a try if time permits. Its an easy test, just plug it in and see if it boots. If it hangs then maybe there's a difference with the SMT chips. If not I have no idea. Nothing jumps out at me but I suspect one would need to connect a logic analyzer to the cartridge port and see what's in the wrong state.

What would be really nifty would be a cartridge port break-out adapter so as to easily connect LAs and oscilloscopes etc. Tricky to do otherwise as the connector is two-sided.

stephen_usher wrote: 11 Apr 2020 15:50 What would be really nifty would be a cartridge port break-out adapter so as to easily connect LAs and oscilloscopes etc. Tricky to do otherwise as the connector is two-sided.

Like this - https://www.exxosforum.co.uk/forum/viewt ... =74&t=1528

Not quite... one with a through connector so you can plug a cart into it to diagnose the signals going to/from it.

@stephen_usher - could be done but getting hold of a cartridge port connector to make the pass through would mean probably scavenging off an old ST. They are like gold dust :)

I think I may have found the issue with the Lotherek version of the NetUSBee... He's using 74HCTxxx parts.

Looking at the product specs the timing on them is very sensitive to the Vcc voltage, with the speed increasing greatly the higher the voltage, with a 20% change in speed between 4.5V and 6V. (At its lowest supported voltage of 2V it's glacial.)

Now, I've noticed that the NetUSBee works best with the standard ST power supply and that seems to give more than 5V typically. My TT and the Exxos power supplies are both very accurately 5V and it fails on them. So, I'm wondering if the timing is marginal and if the NetUSBee gets > 5V its reaction time is just good enough to work.

It would be a tricky one that... The higher the voltages, technically the faster the chip runs.. BUT.. as rise and fall times also increase the propagation time is actually longer.. This also depends on the logic thresholds of the chip as well..

But even so, the speed at which the ROM port is going at isn't really that fast anyway, so a few ns here or there shouldn't matter. Mostly this boils down to noise issues.. Where using slower chips is actually beneficial because it squashes out more noise. Its why I generally recommend LS series chips in ST's. switching for anything faster like "Fast" series is generally asking for trouble.

You could do a "Atari fix" and place something like a 4001 diode in series with the 5V to the cartridge port... That should be easy as I think there is a fuse on that somewhere anyway..

For me the netusbee isn't worth the effort to really dig in. I purchased it for the falcon and it works fine on that platform. Its a little disconcerting that it locks up a mega st and an H4 by just being plugged in but hey.. the lightning ST works good in all the the DIP64 machines. If I need additional networking hardware my preference will be additional lightning cards but I would get another netusbee if the situation dictates but I would test with the current netusbee first. :) :) :)