PHD only managing 1-ish image per second? Regardles of exposure time.

[pipnina]

At first I thought I may have been running against limitations of the raspberry PI 4's processing or memory, but I recently switched to using my laptop (I7-6700HQ, 16GB ram, nvme SSD) and realised PHD wasn't *quite* performing as I had anticipated.

I'm using the ASI 120MM mini as a guide cam, which claims 30+ fps at its max resolution, although I anticipate that's set to JPEG compression and 8-bit mode. I have run my guiding on 8-bit raw mode by mistake before, which didn't actually improve the frame rate beyond the (slightly below) 1fps. I seem to be getting this 1fps update interval regardless of whether the exposure is set to 1s, 0.5 etc. Could it be an issue with my laptop, with a setting in PHD2, or maybe the camera really can't handle RAW frames at all above that framerate? The guide SNR went waaaay down in 8-bit raw mode so I can't imagine how poor it would be in jpeg-compressed mode!

My typical setup is to run PHD2 on the laptop, which is plugged in to the hardware directly, and then PHD2 reports back to Kstars indoors which is controlling the setup as a whole. In theory this should be ideal as guiding is not being performed over the network but the lag remains. I have tried to run guiding directly at the laptop with kstars running on the same machine but I still couldn't get 2fps guiding, and occasionally I get skipped frames!

Has anyone had this issue before?

[licho52]

Enable subsampling in the camera settings of PHD and see if that enables quicker fps.

[pipnina]

5 minutes ago, licho52 said:

Enable subsampling in the camera settings of PHD and see if that enables quicker fps.

Is that the subframe tickbox in the bottom right? I'll give it a go! Thanks!

[Paul M]

I use a 120MM camera for guiding but on a W10 machine, I just looked at my default PHD2 settings and see I have "noise reduction" set to 3x3 wich I guess will speed up frame rates if I was struggling. 

My question would be why you would want sub second guiding? I use 2 or 3 seconds but a bit longer is not a problem. Too short a guide period can cause more problems than it solves; guide pulses can be greater than the required correction ammount - overshoot and subsequent over correction and then saw-tooth guide graph!

I only use sub second frame rate when using the polar alignment routine in PHD2, which is often done in late twilight and frame rates over 1 sec result in white-out due to sky glow.

[kens]

PHD2 operates the ZWO cameras in single shot mode rather than in video mode. The single shot mode has a latency per shot of a few hundred milliseconds (up to 500 from memory). This is a function of the hardware and ZWO driver. So if you choose a 1s exposure then the time between shots is going to be 1.5 seconds. On top of that you add the time taken to make corrections as you dont want to take a guide image while the mount is moving (and probably why single shot mode is used) and any download time. In my setup I can just get a 1s sampling rate if I choose an exposure time of 0.5 seonds.

[wimvb]

5 hours ago, kens said:

On top of that you add the time taken to make corrections as you dont want to take a guide image while the mount is moving

Quite correct. Guiding corrections can be up to 2 500 ms or 2.5 s, the default maximum pulse length. Besides, why would one use sub seconds exposures with a guide cam anyway? You'd be chasing the seeing.

[licho52]

My former mount worked best with 0.5-0.7s exposures.  There are some disagreements about the "chasing the seeing" theory which sounds plausible but then there are practical results and for my old mount the short exposures and corrections worked the best.  This is likely also be the reality for the new ZWO AM5 owner who will need to guide with below 1s exposures on account of the huge oscillation amplitude of harmonic motors.

That said my current EQ6-r is more stable and likes longer exposures and does fine with 1-2s times.  But it has about 1/3 of the PE of the old one.

[michael8554]

If you must use sub1 second exposures, use the PHD2 MultiStar mode, which compensates for "Chasing the Seeing".

THough I've not seen any posts where HEQ5-PRO users have to guide sub-second.

Michael

 

 

[pipnina]

11 hours ago, wimvb said:

Quite correct. Guiding corrections can be up to 2 500 ms or 2.5 s, the default maximum pulse length. Besides, why would one use sub seconds exposures with a guide cam anyway? You'd be chasing the seeing.

Ah, this might make sense. I was wondering why even at 1s I was sometimes getting big delays between frames of almost 2s.

I mainly wondered if shorter exposures would help PHD correct for the oscilations better, I guess probably not.

6 hours ago, michael8554 said:

If you must use sub1 second exposures, use the PHD2 MultiStar mode, which compensates for "Chasing the Seeing".

THough I've not seen any posts where HEQ5-PRO users have to guide sub-second.

Michael

I have been using multi star mode til now, it does seem to work quite well.

 

20 hours ago, licho52 said:

Enable subsampling in the camera settings of PHD and see if that enables quicker fps.

Sadly with this turned on the calibration fails, so I think the feature doesn't work well when operated remotely. Thanks for letting me know it exists though!

[wimvb]

35 minutes ago, pipnina said:

mainly wondered if shorter exposures would help PHD correct for the oscilations better

Imo, it’s better to find the cause of the oscillations and try to fix it. High frequency oscillations are usually caused by the gear close to the motor. I had a 10 s oscillation that was caused by a loose belt. Tightening the belt fixed it. Guiding can remove slow tracking errors, but not fast ones.

[kens]

8 hours ago, michael8554 said:

If you must use sub1 second exposures, use the PHD2 MultiStar mode, which compensates for "Chasing the Seeing".

And/or use the Z-filter algorithm which lets you adjust the guiding response as a multiple of the exposure time (or, more correctly, a multiple of the sampling rate). But it is best suited to mounts that can make small, precise corrections.