Chasing a camera fault
17 to 23 October. Indoors and out, in a back garden in Northern Ireland.
The camera is an Altair Hypercam 183C, made in June 2017 and still on its first firmware. It has no memory of its own to hold a frame, so every row goes straight down the USB lead as it’s read. That turned out to matter.
Stripes in the cold
On the cold nights of 17 and 18 October, near freezing, faint green and magenta stripes ran across every frame, and long exposures started going missing while short ones still arrived. I opened a support ticket with Altair on the 19th and sent the camera’s own diagnostic file.
Warm on the bench that afternoon, cap on, it behaved: 116 dark frames of 16 seconds over an hour and a half, every one arrived, and the stripes were at their ordinary level.
Altair’s first reply said the diagnostic file showed the camera running at USB 2, and suggested a good USB 3 lead of 1.8 m or less. The USB 2 was only on the PC where I made the file; on the laptop it always linked at USB 3. So I sent back something more useful. Claude measured the stripes in every saved session and set each night beside the temperature. They were about 50% stronger on the two nights near freezing and normal at 6 °C or above, a correlation of −0.90 with the air temperature.
The morning it broke
After that the camera was fine again for two nights. It kept 16 and 32 second frames for hours, and through the night of the 19th it took 992 frames of a planet transit.
Then at about five in the morning on the 20th, at the end of a cold night with cloud coming in, it hung four times in half an hour. Since then it hasn’t kept a single frame of 2 seconds or more at USB 3, whatever we’ve tried. One-second frames still arrive.
Ruling things out
Altair’s second reply suggested Windows, since that’s what they test on, a ground loop through the laptop’s charger, other mains sockets and other computers. We went through the lot, cap on, indoors:
- A shorter lead. A 1 m USB 3 lead fails the same as the 3 m one.
- Mains. The kitchen socket with and without the extension lead, the laptop on battery, and a socket upstairs on another circuit. 2-second frames were lost every time.
- Windows. Claude helped me set the laptop up to dual boot Windows 11 and ran the camera through Altair’s own software there. Same cut-off: every 1-second frame kept, every 2 and 16-second frame lost.
- Another computer. On my desktop PC it was the same again, on both leads.
- The socket. I photographed the camera’s USB socket and it looked fine, then cleaned it with isopropyl anyway. Holding the lead up, then down, while a test ran changed nothing either.
Two clues came out of it. In USB 2 mode on the desktop, long frames arrived. And the readout speed decides it:
| Readout speed | 2 s frames | 16 s frames |
|---|---|---|
| 2, the fastest, which it always used | lost | lost |
| 1 | kept | lost |
| 0, the slowest | kept | kept, 20 of 20 |
The faster the camera sends its data, the shorter the exposure at which it starts losing frames. That looks like a USB 3 connection inside the camera that has gone marginal: fine when it’s asked for less, failing when pushed.
A friend wondered whether an older driver would help. The laptop’s own records answer that: its driver hadn’t changed since early October, it worked with the camera until the 20th, and it fails the same way as Altair’s newest one. The driver didn’t change; the camera did.
Other people’s cameras
Claude also read up on what others have found. Another Hypercam 183C owner in 2019 lost frames on three different computers, sometimes imaging for hours and sometimes stopping after three frames. Others’ cameras with dropped frames became stable on USB 2. One owner of a different camera watched it go from now and then to almost always, on every cable, port and laptop. Altair’s own help pages point at power saving on the computer, but that wouldn’t explain a change on one morning seen on three computers.
Where it stands
For now the camera runs at its slowest readout speed, which keeps 16-second frames at USB 3 for about a second extra per frame. Claude made a command that tries each speed from the fastest down and keeps the fastest one that loses nothing, so it can be checked again any time. The exposure timing at that speed still has to be checked against star trails on the next clear night.
That’s a workaround, not a fix. I’ve sent Altair the full set of results, with three questions: is this a known fault on the early 183C, is there newer firmware that would help, and what would a repair involve. No reply to that one yet. If it can’t be repaired, the likely replacement is a camera that holds the whole frame in its own memory before sending it, so a hiccup on the lead can’t lose it.