Sounds odd. From the hw designer point of view you can change logical resolution whenever and however you like, provided the physical scanline rate remains the same, which should be the case for CRT, as it is unable to change video timings at all.
But CRT may be more forgiving about vertical synchronization and small fluctuations in in the video geometry (basically CRT is totally ignorant). And LCD needs to recognize the geometry of the incoming video in order to resample it properly, and it may back off if it finds inconsistencies in sync signals. I'd guess Amiga does not maintain sync structure particularly strict/aligned, and this upsets the LCD.

Here's the famous Amiga boing ball demo showing what I'm talking about.
The Workbench screen is a different resolution than the boing demo, but you can pull it up and down.
I can display anything from my Amiga 3K on a vga monitor... but the multi-resolution is what seems to break...

https://www.youtube.com/watch?v=UyNr3yaUQA8&a … channel=amigang

Would be great if you could capture the sync signals with some logic probe.
Otherwise we can only make guesses, is it Amiga not maintaining proper timings, or its 'Built-in "display enhancer" (scan-doubler and de-interlacer) for use with VGA monitor' not doing its job properly in some cases, leaking certain details of interlaced video timings to VGA and scaring the monitor.

I remember doing that on my A1200 but not thinking too hard about it at the time. Would have the standard res workbench in 256 color then have an 18bit 720x512 interlaced Ham-8 screen from TAD (The Art Department) you could pull down. This was on standard PAL monitor.

Might be it's switching to a 31khz signal?

I've seen that trick on Amstrad CPC, but...

The CPC monitor was a fixed frequency monitor (in fact, it could be described as a stripped down TV) . The "changes" in resolution were limited to doubling pixels and or lines (but the "modes" had differente colour capabilities), so sync timings were the same. If you threw that signal into a LCD screen... nothing strange happened. You could get a "full" colour 160x200 screen on the first two thirds of the screen and change to a "mid res" 320x200 on the fly.

When you change sync signals (timing and/or polarity), drop the signal output or throw a vertical blank out of time, you'll probably trigger the mode detection logic of every LCD screen (and some "digital" CRTs). So, if you want to change resolution "seamless", I guess you're limited to "Amstrad" tricks... doubling lines and/or pixels.

Zup wrote on 2025-01-07, 07:56:

So, if you want to change resolution "seamless", I guess you're limited to "Amstrad" tricks... doubling lines and/or pixels.

The idea is the Amiga already does this, or supposed to do with that "enhancer" of theirs on VGA port, yet LCD backs off, so something is not quite right there.
Many LCDs won't support interlaced VGA, I believe this should be the culprit - the deinterlacer inside the Amiga likely does not handle the switch in logical resolution gracefully.

The physical scanline rates never changed, no CRT would ever handle switching between 15 and 31kHz mid-frame. Amiga outputs two fields of 256 lines each, and whether you see 256 or 512 lines on the screen - depends on how Amiga would drive the vertical sync pulse, would it tick the vertical sync in the middle of the scanline in every second field to draw interlaced video or not. "Switching" the vertical resolution midframe means Amiga already needs to output interlaced video for high-res (512 lines) to be possible. Changing horizontal resolution does not need any changes in video timings at all.

So the only real question is - how exactly that scandoubler and deinterlacer inside the A3000 behaves, and whether it really produces valid progressive video for VGA port, or it leaks timings related to interlaced video which won't be supported on many LCDs.

cool! Thanks for all the technical info!