Thats my suspicion, I think that on certain chipsets, the ASI is USB Hungry and takes almost all bandwidth on usb bus even if you
reduce the bandwith in the INDI settings.Or at least it tries to download the frames faster the the bus can handle.
The compatible firmware changes the way usb transfers are made but sometimes even that is not sufficient to solve the problem. Initially I had to try 3 different powered hubs to make mine work ok. Maybe some chipsets are more "sensible" than others.

I don't see any crashing of my RPi when I disconnect all other USB devices and use the ASI directly in the RPi USB ports. It still does not work entirely as I like, but it does not crash the whole system which is a major step forward for me. There have also been reports of bad USB connectivity on the RPi with the upgrade from wheezy to jessie, but I guess they should be fixed by now. Anyway, my ASIs are the only devices not working correctly for me which is a shame...
As a side-note: I have also tried to use SharpCap on windows, and it works great for a few minutes but then the video becomes very laggy. FireCapture seems to work fine in all situations.

I've done some more testing with this as I was back at the observatory yesterday. Found spomething interesting (impatient readers can skip to tests 7-9)
I'm not using an RPI, my ASI is plugged into a HP elite desktop PC.

Test 1) Tried to take exposures before changing anything
Result: looping

Test 2) Re-arranged all USB plugs so that just the ASI was plugged into the front of the PC (4xUSB2.0 ports at front). Everything else plugged in at the back of the PC
Result: looping

Test 3) Removed all USB devices except the ASI and a wireless keyboard
Result: looping.

Test 4) Re-flashed the ASI with the 'compatible' firmware (v3). Repeated test 3 above
Result: looping.

Test 5) Repeat same test above with a different cable.
Result: looping.

Test 6) Tried a windows laptop and firecapture
Result: no looping.

All above tests were done at night in the observatory. It was a clear night, about 2degrees c. Ground frost and high humidity.
I decided to take the PC and ASI home with me back to where I live and hack around the next day.

Next day:
Test 7)Setup in living room, Same as test 3 above.
Result: No looping. I can take continuous 2sec exposures no problem. Left running for 30min, still working fine. I've changed nothing here!! Except bringing the gear for a 1hr drive and into a warmer environment........

Test 8 ) Put the ASI into a plastic bag and into the freezer for 20min. Retry same test again
Result: Looping. Camera reporting -8.00 deg c. Left running until temp reached 10, still looping. Stop-start indi. Still looping.

Test 9) Warm the camera up again to room temp.
Result: No Looping!! Camera reporting 19deg c. All working fine again

UPDATE

Test 10) Waited a couple of hours. Warm the camera up again to room temp.
Result: Looping.

It looks like the temperature is a red-herring.

This temperature sensitivity has me rather concerned as I was planning on buying an ASI120MM very soon. And the comment regarding better Linux compatibility makes me wonder if I should I get the USB3 model instead.

Another update, the temperature sensitivity problem is not conclusive..... I just tried again when the camera was +20degc and its looping again.
Going to do some more experiments with some code now....
Derek

After making a small code change to the driver (to default set the bandwidth to auto on connect) it seemed to have really helped this problem when testing in my living room. I could not reproduce the looping no matter how much messing I did, connecting, disconnecting, putting camera in the freezer etc etc....
I was optimistic.

However I setup everything again in the observatory today and I still have a problem. I have narrowed it down though. I can take exposures UNTIL I connect my celestron driver..... This is really strange but the instant click connect on the celestron driver, the continuous exposures fail. As soon as I disconnect the celestron driver, the exposures can be taken again with the ASI....
Its some weird interaction with my specific hardware/PC/environment. Anyone have any ideas?

That's exactly the behaviour I faced a few days ago. Check this post .

Could it be something with udev rules? (desperate guessing here )

I would rather blame some kind of usb connection locking by a mount. In my case it's eqmod.
Low level diagnostics seems necessary here. I will not have time for it in the upcoming days but if somebody is willing to contribute I point out at wireshark usb traffic capturing. This might give us some answers.

Hello guys,
As I had many problems with my QHY5L-II color for autoguiding I decided to sold it and buy an ASI 120MM-S.
I tried it yesterday for the first time. And I have the symptoms you are describing. Whatever the camera is connected to a hub or to an ODROID µdevice.
It happens with EKOS or PHD2, so it is not a question of guiding software but in my opinion of driver.
I have a 178MM which works really fine apart some resolution problems recentely solved. So why the 120MM-S doesn't works fine too is weird..
I begin to feel desperate. When I think that a problem is fixed an other one appeared.
And I don't want to go back to windows to get a full functional set-up. That would be a treason.
I have post a message on the ZWO forum, as you can do to make the ZWO guys aware of the problem.
zwoug.org/viewtopic.php?f=21&t=1688&p=11646#p11646

@Kaczorek
This is not a question of interaction with other drivers. The camera connected alone give same result.

Well, in my case there is direct correlation with Eqmod driver - connected looping, disconnected not looping. Moreover, while looping I click disconnect on eqmod driver tab and it leaves the loop instantly.

Exact same here with me using the celestron driver. Click disconnect on the celestron driver and I can take exposures again.
I'm going to re-build indi-asi with the previous zwo sdk (4.?) and see if that works.
I'll also try wireshark but I've only ever used wireshark for network traffic and I have to do this remotely using teamviewer so I'm not sure how productive I'll be....
Derek