Q40 and Q60 video controller for flatscreen monitors

[Peter]

Hi,

is there someone who wants to supply Q40 and Q60 users with an alternative video controller chip for flatscreen monitors? Derek has so much work with the Q68 already, maybe better if somebody else steps in.

The new video controller creates a 1024x768 signal for the monitor. The Q40/Q60 resolutions are matched to pixels exactly, with the consequence that one third of the screen remains black. The vertical refresh rate was reduced to 50 Hz. This alternative video chip can only be a compromise, and there will be no guarantee to work perfectly everywhere. However, on all tested flatscreen monitors there were acceptable results by now.

For this change, one of the socketed PLCC chips on the Q40/Q60 motherboard must be replaced, marked in blue:

Partname of the chip is ispLSI1016-80LJ44 (or ispLSI1016-90LJ44). A Lattice programming adaptor is required.

Peter

[Derek_Stewart]

Hi Peter,

I have the Lattice programmer and can source the chips.

If no one else can do this, I will be able perform the programming update.

[Dave]

I have delved into video conversion and signal processing in some detail for the QL, I have a few questions about the original signal:

What are the frames per second and number of lines in the image?
What 'legal' signal is it most similar to?
Is it possible to alter the frame rate, and/or to supply a pixel clock signal? This could open up the option of DVI conversion. (I have never owned a QX0 and don't know if it uses a card or only custom logic.)

I'm just trying to understand the problem and why the LCD monitor won't sync the nonstandard signal. If it's just one aspect of the signal it may be easily fixable and retain full screen display. If there are multiple elements to the problem, then not. Having access to the pixel clock and being able to manipulate the frame rate to 30 FPS - or even better 60FPS - would be extremely useful.

[smsq4ever]

The new video controller creates a 1024x768 signal for the monitor. The Q40/Q60 resolutions are matched to pixels exactly, with the consequence that one third of the screen remains black. The vertical refresh rate was reduced to 50 Hz.

Alas, 50Hz is totally non-standard for VESA 1024x768 (which is normally 60, 70 or 75Hz) and most modern LCD monitors will fail syncing at so low a vertical refresh rate. My EIZO FlexScan EV2455 (specified for analogous signals from H=31 kHz to 81 kHz and V=55 Hz to 76 Hz) most probably won't (unless, on a 66MHz Q60 the vert rate could be made higher ?)...

What about a 800x600 @ 56, 60 or 72Hz resolution, instead ? Standard VESA mode, fits the Q60 video memory, won't suffer from 1/3 black screen syndrome, already supported in SMSQ/E for the Aurora (so easy to add for the Q60 as well)...

[Peter]

Alas, 50Hz is totally non-standard for VESA 1024x768 (which is normally 60, 70 or 75Hz)

Of course it is not a standard VESA mode. The timings are a compromise, worked out in long practical tests with about ten flatscreen monitors and TVs.

and most modern LCD monitors will fail syncing at so low a vertical refresh rate.

Not most, and I'm fed up with your agressive replies. Why do you need to come up in public with "will fail" before you actually tried a representative number? Or even one?

As I said, no guarantee to work everywhere, but there was no fail in reality yet. I'm tempted to stop distribution of the chips if this continues...

I need a friendly, constructive atmosphere. It took me a lot of time to bring at least some improvement for this difficult issue. And believe me, if there was an easy better solution with the existing chips, I would most likely know. I'm not willing to spend my precious spare time for free, if I have to face very negative claims, not backed by actual problems, already beforehand!

[smsq4ever]

Alas, 50Hz is totally non-standard for VESA 1024x768 (which is normally 60, 70 or 75Hz)

Of course it is not a standard VESA mode. The timings are a compromise, worked out in many practical tests with about ten flatscreen monitors and TVs.

Care to give the models, please ?

and most modern LCD monitors will fail syncing at so low a vertical refresh rate.

Not most, and I'm fed up with your agressive replies. Why do you need to come up in public with "will fail" before you actually tried a representative number? Or even one?

There is no aggressiveness in my replies, just plain factual logic with regards to the specifications of the monitors: among the three I got, only the oldest (built in 2003) is (was: it's dead) specified to sync down to 50Hz, while the two others (one built in 2010 and the other in 2017) are both specified for 55 Hz minimum.

As I said, no guarantee to work everywhere, but there was no fail in reality yet.

If you could quote the models, we could check against their specs and deduce the chances that the 50Hz 1024x768 mode could be displayed on similarly specified monitors. There's always the possibility that the maker kept a small frequency margin (but 5Hz, i.e. 10% seems like a very big margin for a vertical refresh rate).

I'm tempted to stop distribution of the chips if this continues...

Amazing that some people cannot stand the slightest remark...

I need a friendly, constructive atmosphere.

I always have been constructive, but I'm not the kind of guy to flatter people and try to hide the slightest doubt when I see a potential issue. Being constructive is also about giving your point of view, even when (and in fact especially when) it does not match others' point of view.

I will also point out that I proposed an alternative solution (800x600). I *am* being constructive. I have been in the past too: our first conversation about the Q60 video mode issue dates back from 2016 (March, the 11th), in the ql-users list. I still have the archives if you want them, and in the lengthy (and constructive) conversation we had back then, I already mentioned the 800x600 solution, and even gave you the pointers for the things you considered as an obstacle to this video mode (the changes in SMSQ/E and Linux), citing the files and their few constants that would need a change in SMSQ/E and Linux sources.

It took me a lot of time to bring at least some improvement for this difficult issue. And believe me, if there was an easy better solution with the existing chips, I would most likely know. I'm not willing to spend my precious spare time for free, if I have to face negative claims, not backed by actual problems, already beforehand!

I'm just "voicing" (more like writing) my worries before it would be too late (the chips programmed and shipped) with the risk that the end users would find the result inadequate for their monitors.

I am not trying to denigrate your work or anything, much to the contrary: I'm trying to help you find the best solution for you and the end Q40/60 users (me included, of course).

So, please, stop reacting like if I was attacking you (I'm not !) or trying to denigrate your work of love for your baby. Just take my observations for what they are. Maybe I'm paranoid and everything will work fine for everyone (in fact, I would love that to happen), but I tend to believe what I see and for now, all what I see is that the modern LCD monitors are not specified for 50Hz vertical refresh rates.

[Dave]

I proposed an alternative solution (800x600).

800x600 is not feasible, based on the dot clock limitation Peter said he was already up against.

Code: Select all

512x256x72  =​  9,437,184 dot clock
​800x600x50  =​ 28,800,000     "
​800x600x25  =​ 14,400,000     "
​800x600x20 =   9,600,000     "

As the dot clock is the practical limitation and he was already up against it, 800x600 would be limited to around 20 FPS. Maybe the hardware is capable of much higher dot clock rates within the FPGA but Peter hasn't indicated either way.

I'm not that familiar with QX0 video capabilities other than what Peter's said, so I am just going off the details given in this discussion.

[Peter]

T.G. your reply just shows that you learned nothing. You had stated in public that most modern LCD monitors will fail with the new chip. Without testing a single device. If that is not very negative, then what is?

You had the chance to limit the damage, but chose to keep up your destructive statement. I'm not going to distribute the chip against that kind of public opposition. And I will think twice before spending my precious spare time again.

[smsq4ever]

I proposed an alternative solution (800x600).

800x600 is not feasible, based on the dot clock limitation Peter said he was already up against.

Code: Select all

512x256x72  =​  9,437,184 dot clock
​800x600x50  =​ 28,800,000     "
​800x600x25  =​ 14,400,000     "
​800x600x20 =   9,600,000     "

Excepted that the Q60 resolution is 1024x512; not 512x256. 1024x512 is a larger resolution than 800x600 (in number of pixels): it's a 37MHz dot clock.

[smsq4ever]

T.G. your reply just shows that you learned nothing. You had stated in public that most modern LCD monitors will fail with the new chip. Without testing a single device. If that is not very negative, then what is?

Prove me wrong: give me the model numbers of the monitors you successfully tested the video mode against...

You had the chance to limit the damage, but chose to keep up your destructive statement. I'm not going to distribute the chip against that kind of public opposition. And I will think twice before spending my precious spare time again.

So, you give up just because of me ?... Or are you implicitly admitting that 50 Hz can't pass modern LCD monitors requirements ?