Burnt Rapid Road - what can I do now?

[arti040]

Well, i've just burned Rapid Road controller. THIRD in my life. Only by connecting a pendrive. LED does not light up and controller is not recognised by Poseidon stack anymore. This time yet warranty probably already ended. What can do now? Is it "fixable"?

PS Is it really hard to add some electronic "something", don't know, to secure that if it's so sensitive? #angry

[fitzsteve]

I would email Jens and see if he can offer a repair service.

[AmigaHope]

Best thing to do in the future would probably be to never plug into the controller directly and use a hub for plugging/unplugging your devices.

[arti040]

Best thing to do will be to not buy RR anymore Somehow it's the only controller I have (and I got almost ALL Amiga related ones - Thylacine, Subway and Highway) which broke... 3 times.

[dkedrowitsch]

What kind of Amiga do you have it installed in?

If it's in a 1200 I seem to recall Jens mentioning RR is sensitive to power rail transients potentially triggering latch-up conditions (infinite current) in the RR CPLD, permanently damaging it. To prevent this it's important to have a solid 5V throughout the system, which in a 1200 can be difficult with some replacement power supplies that barely provide 5v to the input choke on the motherboard.

[Daedalus]

Yep, Jens seems to blame PSUs, either the original (it's getting old and unreliable) or 3rd party (it's not good enough), yet doesn't really have an alternative suggestion to what boils down to basically just those two choices. Given that it's designed specifically for the A1200 clock port, it seems quite poor that it can't handle typical conditions found in an A1200, especially when there are other USB controllers that don't seem to show such issues.

[clebin]

It’s lucky my Rapid Road didn’t work for years because of driver problems. Who knows what damage I might have done should I have actually plugged something into it?

Great purchase. /s

[cclecle]

its a shame this card id not robust, because apart from that I think its the easyest USB solution for classics.
Can't we hardware fix it by ourselves ? I am thinking about adding a Schottky diode at power input stage...

[crazyc]

Well I have a rapid road in my a1200 checkmate setup however Jens views on this suggest I should be grateful for whatever usage I get out of it:

"We generally recommend to avoid ATX power supplies. Technical data of the PSU in question shows that 5V is not the main regulation rail, which makes is plain unsuitable for an Amiga. Expect trouble, going up to destroying your RapidRoad due to bad regulation of the 5V rail.

While the relatively low ripple of the PSU you've linked to might seem to be good, there's still two more key specifications from Commodore that need to be met. With 5% regulation tolerance, you have at least one spec that's violated, so checking the other ones is of no use any more.

One more time, AVOID, AVOID, AVOID. You're just asking for trouble."

I have concluded that if it burns out then I am not going to replace it, and will live without USB in that Amiga, using network for file transfers. Or maybe even a good old CD

[cclecle]

@crazyc: yes, thats whats I am doing now with my A600, samba share mount over prism2/pcmcia wifi. Its fine but would be great to be able to use usb stick / mouse / cdrom drive etc...

Its very hard to believe that its a normal behavior, considering others peripherals does not have this issue + the cart is connected to the miggy motherboard which have decoupling and load...

[Mathesar]

Originally Posted by dkedrowitsch

What kind of Amiga do you have it installed in?

If it's in a 1200 I seem to recall Jens mentioning RR is sensitive to power rail transients potentially triggering latch-up conditions (infinite current) in the RR CPLD, permanently damaging it. To prevent this it's important to have a solid 5V throughout the system, which in a 1200 can be difficult with some replacement power supplies that barely provide 5v to the input choke on the motherboard.

Does the rapid road use a Xilinx 9500 series CPLD and does it fry the moment you stick the USB device in? I then highly doubt it is power supply transients. It's far more likely a ESD problem.

In my company we once developed a complex 19" rackmount device employing multiple GPS/Glonass/Beidou receivers, 450Mhz radio's, dual core pentium commexpress modules, pcie, ethernet, usb, the works.
Now, the front panel board utilized a 9500XL cpld running at 3.3V with some I/O lines being driven @5V. Exaclty why these CPLD's are still popular. Otherwise we'd all be using MAXii,V or 10's.

Now, on the prototype a software developer managed to blow up the CPLD by just plugging in the USB stick. We couldn't really find the cause but later, during CE compliance testing... Boom, CPLD in latchup during ESD testing. The discharge didn't even had to be close to the USB connector. I fixed it by improving the ground path from the USB shield to earth. I also added series resistors to some 5V I/O lines going into the CPLD because a latchup is triggered by excessive current into an input triggering a parasitic thyristor.

TLDR, if the rapidroad uses a 9500 CPLD and it goes in latchup the moment you insert a usb device it is very likely a ESD problem and not a power supply problem. Does the device has a CE sticker?

[Vypr]

Originally Posted by Mathesar

Does the rapid road use a Xilinx 9500 series CPLD and does it fry the moment you stick the USB device in? I then highly doubt it is power supply transients. It's far more likely a ESD problem.

In my company we once developed a complex 19" rackmount device employing multiple GPS/Glonass/Beidou receivers, 450Mhz radio's, dual core pentium commexpress modules, pcie, ethernet, usb, the works.
Now, the front panel board utilized a 9500XL cpld running at 3.3V with some I/O lines being driven @5V. Exaclty why these CPLD's are still popular. Otherwise we'd all be using MAXii,V or 10's.

Now, on the prototype a software developer managed to blow up the CPLD by just plugging in the USB stick. We couldn't really find the cause but later, during CE compliance testing... Boom, CPLD in latchup during ESD testing. The discharge didn't even had to be close to the USB connector. I fixed it by improving the ground path from the USB shield to earth. I also added series resistors to some 5V I/O lines going into the CPLD because a latchup is triggered by excessive current into an input triggering a parasitic thyristor.

TLDR, if the rapidroad uses a 9500 CPLD and it goes in latchup the moment you insert a usb device it is very likely a ESD problem and not a power supply problem. Does the device has a CE sticker?

It appears to use a Xilinx 9572, Maybe soldering a wire to the USB Shield and connecting it to Ground would help?

[Mathesar]

Originally Posted by Vypr

It appears to use a Xilinx 9572, Maybe soldering a wire to the USB Shield and connecting it to Ground would help?

That would help yes. I don't now the rapidroad so well but I assume that in a A1200, the USB connectors themselves are in a plastic bracket. so, any ESD current surge will have to go over the cable to the rapidroad itself. The surge needs to be diverted around the rapidroad. so the best way to solve it is to solder a wire (not too whimsy) from the USB connector shield to the A1200 shield.

[S0ulA55a551n]

Seems like typical Jens to blame everything and everyone except his own product

[Mathesar]

Originally Posted by S0ulA55a551n

Seems like typical Jens to blame everything and everyone except his own product

I am not trying to dismiss Jens's judgement. Power supply *can* be an issue.
However, this particular fault condition is so very similar to something I have seen happening before that I think in this case an ESD event is far more likely.

[whaka]

Originally Posted by Mathesar

That would help yes. I don't now the rapidroad so well but I assume that in a A1200, the USB connectors themselves are in a plastic bracket. so, any ESD current surge will have to go over the cable to the rapidroad itself. The surge needs to be diverted around the rapidroad. so the best way to solve it is to solder a wire (not too whimsy) from the USB connector shield to the A1200 shield.

but the sheild need to be effectively connected to earth.
so, be sure to use a earthed wall plug. otherwise, nothing will really change.

i don't know if it's still the case, but some RR was delivered with a earth cable.

[Mathesar]

Originally Posted by whaka

but the sheild need to be effectively connected to earth.
so, be sure to use a earthed wall plug. otherwise, nothing will really change.

i don't know if it's still the case, but some RR was delivered with a earth cable.

Yes, an earthed wall plug is best. However, even if the case is not earthed it will help a bit as the electrostatic discharge will charge up the shield (and eventually the whole computer). What's important is that the high current discharge path is kept away from any sensitive electronice like the CPLD.

[ljmarent]

iComp's position on power supplies is really a two edged sword.

He (Jens) knows full well the original supplies are old and dying, needing recap at the very least.
He 'should' know they didn't all meet (from new) the OEM design spec. which he likes to site.
Some had remote sense, some didn't. I've opened my A500 PSU up and it doesn't have load based voltage compensation or remote sense, it's just a cheap 90's line side regulator PSU. Finally, if you use replacement power supplies like Mean Well he says those are unsuitable, for a slew of reasons, with varying levels of actual severity.
Like some multi-voltage supplies do in fact follow the 12V rail for regulation, meaning the 5V rail is often way out, so that point is fair enough, but concerns of ripple, much less important.

Okay.. Simple enough.. There are nearly no Amiga's in service today with good, clean, reliable power that meet the original Commodore spec. and that's not about to change.
And the proof of this can be found on the iComp support site
Then the only reasonable conclusion for an engineer and businesssman who knows this; is to design your products for a poor power environment.
That's the back edge of the sword.

Don't whine about linear regulators being too close to dropout and oscillating, potentially causing SCR latch-up; just use a buck-boost regulator topology.
And, while we're on the topic, make sure you don't have caps on your I\O pins and perhaps put some current limiting devices (resistors) on pins that are likely to be in an SCR latch-up situation. In all my years designing for automotive, I saw a handful of parts come back with SCR latch-up damage. I just can't explain how often I read about this condition on iComp's support forum.

Oh..and the RapidRoad ESD protection...
I've read on the iComp forum that they are proud of it's 2kV human body model ESD protection. Just..yikes!
Many people do not know there has been an ESD event in the 2kV - 4kV range.
15kV air and 15kV contact protection is about as low as you should design for stuff in the home.
In automotive we use 25kV for internal and 35kV for anything that can be reached from outside the vehicle.
So, 2kV protection is basically no protection in my opinion.

iComp spent a lot of time and money, I'm sure of it, on getting RR to work and now the clockport version has been discontinued because of returns. I really hope they toughen up the regulator and add some some ESD protection more suitable for use outside an ESD safe room. I'd gladly buy the updated version.