Hi all, following up on my earlier thread about running the 12MP synchronised (stereo) hat on a Pi Zero. I’ve finally managed to get the stereo board to run intermittently, but it’s extremely jumpy - even when I’m just in live streaming mode, not capturing. It’s not a clean dropped-frame thing, the whole stream stutters and occasionally freezes for a moment before coming back.
I’m fairly confident at this point that the problem is the stereo board itself. Could someone tell me what the minimum power requirements are for the 12MP synchronised stereo hat? I’d like to figure out whether my supply is the limiting factor here.
Hi, welcome back. To narrow this down I need a bit more about your setup:
What exactly is powering the Pi (and the hat)?
Does the same rig behave normally with a different camera?
Any kernel messages / logs when it freezes or stutters?
Any code or config you can share (resolutions, framerates, gstreamer pipeline, etc.)?
The “jumpy even in live streaming mode” detail is interesting - if it’s happening without capture, that usually points at power delivery or bandwidth rather than the capture path.
Setup is: Raspberry Pi powered by a LiPo, which is in turn charged by a suite of D-cells, via a PiJuice RTC board. So the PiJuice sits in the middle between the battery pack and the Pi.
I’ve got a 12MP synchronised (stereo) hat on it. The key thing is that the same setup runs flawlessly when I run a non-stereo HQ camera instead. Same power chain, same Pi, no stutter. Swap to the stereo board and it goes jumpy.
No logs to speak of - I don’t get errors, it just stutters. I don’t have any code to share either, it’s just the stock live streaming view.
Should add: I’ve already tried reseating the ribbon cables and swapping in a fresh Pi Zero, and it made no difference to the stuttering. The HQ camera still works perfectly on the exact same board and power chain, which is why I keep coming back to the stereo hat.
Thanks, that’s helpful - the fact the HQ camera is rock solid on the same Pi and same PiJuice/battery chain is a strong clue. It means the power source can carry a normal camera load fine, but something about the stereo board load isn’t being met.
The 12MP synchronised hat drives two sensors at once plus the sync logic, so its current draw is meaningfully higher than a single HQ camera - and the PiJuice + D-cell setup may not be able to sustain that under peak draw, which would show up exactly as you describe (jumpy in live view, no explicit errors).
Can you tell me whether you can measure the supply voltage at the board while it’s running and stuttering? And do you have any way to check the current draw during streaming?
I don’t have a meter on me right now to measure it properly, unfortunately. I was really hoping there was a published minimum spec for the stereo board so I could compare against what my PiJuice chain is actually delivering.
Is there a number for the minimum current/voltage the synchronised hat needs? If it’s higher than what the HQ camera pulls by a wide margin, that would explain a lot.
I don’t want to quote a figure off the top of my head and have you chase it - I’d rather confirm whether the board itself is the variable first.
Quick question: do you have a Raspberry Pi 3 available anywhere? If so, could you put the stereo camera on the Pi 3 and run the same live streaming test? That tells us something useful either way:
If the same jumpy behaviour happens on the Pi 3 as well as the Pi Zero, it points to the stereo board’s power requirement / draw rather than the host platform.
If the Pi 3 is clean, then it’s more likely something specific to how the Pi Zero is being fed in your setup.
Same test with the HQ camera on the Pi 3 would give us a clean comparison too.
OK, I can get hold of a Pi 3 and test it. So the plan is: run the stereo 12MP hat on the Pi 3 exactly as I have it on the Pi Zero, stream live, and see whether it still stutters? And then repeat with the HQ camera for comparison?
If it reproduces on both boards then we’re looking at a power-delivery limitation of the stereo hat rather than my battery chain, right?
That’s exactly it. To summarise where we’ve got to on this so far:
The 12MP synchronised stereo hat runs only intermittently and is very jumpy, including in live streaming mode with no capture involved.
The identical setup (Pi powered by LiPo, charged from a bank of D-cells through a PiJuice RTC board) runs flawlessly with a non-stereo HQ camera, and reseating ribbons / swapping the Pi Zero changed nothing.
That isolates the variable down to the stereo board and suggests a power requirement / power delivery limitation of the hat rather than the power source itself - it’s a suspicion at this point, not confirmed.
Next steps:
Confirm whether you have a Raspberry Pi 3 available.
If yes, test the stereo camera on the Pi 3, streaming live the same way.
Compare that against the Pi Zero result, and repeat with the HQ camera on the Pi 3 for a baseline.
Once we know whether the behaviour follows the board or the Pi Zero, we can pin down the minimum power the hat needs and whether your PiJuice chain can cover it. Please post your Pi 3 numbers when you have them and we’ll take it from there.