Solved Weird monitor issue I have never seen before - any ideas on what causes this?

I have a new Mini PC based upon an Intel i7- 14650HX CPU. This PC has three video output: HDMI, DisplayPort, and USB-C. I am attaching 2 identical 28" 4K monitors (Acer V287K), connecting one with HDMI and the other with DisplayPort. The weird thing is that when I connect to the DisplayPort, everything works flawlessly, but when I connect using HDMI I get a weird problem. It's as if the timing is off because the image on the monitor connected with HDMI will be a little too big. It is like the image extends a little bit off of the left, right, top, and bottom of the screen.

If I switch from 60Hz output to 30Hz then the image is just fine, and sometimes it will stay fine when I switch back to 60Hz but only temporarily.

Now, I have found a weird fix for the problem, but I'm curious to know if anyone has a better solution: If I connect one of those dummy EDID plugs with passthrough capability to the HDMI port, then the monitor works 100% perfectly every single time. The EDID plug I am using defaults to 1920 x 1080 but if I tell Windows that the monitor is 4k @ 60Hz it will work flawlessly forever. And yes, the EDID plug fully supports 4K @ 60Hz as well as 120Hz and 144Hz.

Note two that I have three of these monitors (I was only trying to connect two). All three exhibit the same issue. I have also tried three different PCs and have the same problem. I do have a 13" 4k portable monitor and do not have this problem on that monitor from any PC so the problem appears to be confined to all three of these Acer monitors.

That "timing" issue you say completely sounds me to analog technology, the same as a "magnetic read/write head" could sound me to an 80's cassette reader/recorder or to the most advanced HDD, but definitely no to SSD "let alone" RAM.

The "let alone" is a trick: could any SSD have any "magnetic read/write head"? Between an SSD and a RAM stick, is any of them more "prone" to have a magnetic thing inside? The same happens to me with timings (through the monitor cable) in this context. I don't know the concrete format of frames/equivalent, but it has to transmit the R, G, B and the other channels if they exist (*) digitally. Idk if it's through square waves or a carrier/modulation thing, but the outcome for the receiver (the monitor) is the same: one or more square waves, one clock or several clocks (possibility if there're several signals) and a format/rules to extract the R, G and B values for every subpixel.

Once extracted, the monitor sends them to D/A converters (I deduce there uses to be one per line and color channel instead of one per subpixel, but isn't this completely inside the monitor? can/should the gfx card/thing do this work? at what -analog- voltage for example? how do you transmit analog waves through the 100% digital channels (I'm sure about this) called "HDMI", "DVI-D", "DP" etc?) so the "twisting" in TN or the "switching" (bad term) in IPS can be regulated for every pixel. Obviously the number of pixels per frame and per line and the number of lines per frame must rigidly match. If they don't, any behaviour (including the devices physical integrity) is undefinad and if you see anything it would be noise (I guess 99.999...% of these ill behaviours would be treated as "loss of sync" and a black screen with or w/o error message would be produced). You're not sampling an analog signal that comes from the HDMI/DP cable, you're like filling a two indexes array in RAM but instead of cells to store the values you have D/A converters and subpixels attached.

Imo it's a drivers problem. Inside the computer there're flexible means to first expand and then crop the image, intentionally or in error like in this case. The poor monitor is merely passive, it works hard but it doesn't have intelligence to do this, intentionally or in error. If it manages to establish connection at 3840 x 2160 x 60, it just receives, computes (*) and distributes 24,883,200 (3840x2160x3) digital values to the D/A converters, 60 times per second.

(*) I know basically nothing about the "auxiliary" channels to R, G and B, why they exist and what they do (I believe they're luminance/gamma/etc things), but a monitor must just be able to feed its D/A converters, whatever the calculations it must do with one or several inputs per subpixel.

I'm using now a monitor (the Samsung in my sign) for two computers, this modern Windows 11 miniPC and an old Windows 7 desktop. Both have slight problems with the video, this monitor or whatever. The Win11 does glitches, the image appears distorted for a split second (very very quickly, likely it lasts only one frame each time) a few times per hour, maybe 5 or so. Maybe it does it more at higher temps and/or in longer sessions. Idk which point of the transmission/displaying does this but the cable is a strong candidate. The Win 7 problem is more serious: about 1/10th of the bootups are done w/ black screen (after the first screens that display the RAM amont, disks and optical drives, a table of peripheral numerical parameters,... and the Win 7 splash screen with the lights flying and converging, the last alive screen when this happens). It once lost video (went to black screen) while doing a compressed disk image (a very heavy task), but the latter is very rare. I attribute this to the gfx card oldness. Regardless of this the rest of the computer behaves fine (for example the said disk/partitions image completed and verified fine in the usual time, I verified it again after rebooting just in case).

I did so me more research. It turns out that this is a known firmware problem with this monitor. Wonderful. I have 3 of these monitors. Acer never fixed it. Thankfully, my workaround of using the dummy EDID adapter works fine.

BTW, the FW bug is that Acer is marki9ng the EDID for 4K resolution on HDMI to be a "television" format which enables overscan.

So it uses a "compatible" mode that's only "partially compatible". Seems like the subpixels frequencies aren't rigidly established, the D/A converters can work at either the higher (w/o "overscan") or the lower (w/ overscan) subpixel frequencies, but it should be the computer or TV what establishes the "validity window" for each line to "achieve" (in error) the "expand/crop" effect: the computer sends "validity windows" for the "computer mode" but the monitor uses "TV mode" (so called "overscan") for D/A converting.

That flexibility seems to attend different physical or history requirements produced in both fields. With "physical" I mean Idk if it would be wise to just use the D/A converters with less rest between each line and the following. In error or not they have produced a "lazy monitor" lol: it works less time and at lower frequency. Your EDID thing "takes out the whip" with the "rests", unless I haven't understood right what the bug does.