Chief Blur Buster
Owner of BlurBusters
- Joined
- Aug 18, 2017
- Messages
- 430
Referring to top-edge compressed linearity issue or top-edge folding issue
This may not be applicable to this type of issue, but generically seen on almost any CRT when:
- When you try to put LCD timings to a CRT.
- When you enter too-small sync/porch numbers in Manual mode in a Custom Resolution Utility
- Or when CRT vertical-reset capacitors get slightly too weak to reset electron gun fast enough
Sometimes the simple workaround for linearity at edge is to use a larger blanking interval. The blanking interval's too short of an interval for the CRT gun to move from bottom back to top edge.
If the problem is a bit of top-edge linearity problem, try small increases to Vertical Back Porch in a Custom Resolution Utility. If your linearity problem spreads over 50 pixels, then perhaps increase Vertical Back Porch by about 50.
You can also experiment by transferring Front Porch to Back Porch so you have more time interval on the other side of a sync signal. Usually on older CRTs you need more time after the Vertical Sync, to allow time for the CRT electron gun to reset back to the top edge, and if it's not reving up back to a constant speed when starting the top-to-bottom sweep, it can create a top-edge linearity issue (compressed linearity at top, or a fold-over effect)
The numbers in a Custom Resolution Utility can easily be human-visualized as overscan area beyond edges of screen, in this signal structure layout:
Pixels are delivered left-to-right, top-to-bottom, like a calendar or a book. The back porches is the overscan area as the "CRT electron gun acceleration area" after the reset (triggered by Sync signal). It's in the overscan area above top edge (vertical) and in the overscan area to left edge (horizontal). So any compressed linearity there, is indicative of not enough beam acceleration.
The fix, naturally, is to embiggen the Back Porch (overscan area) -- to give more hidden overscan area as "acceleration room" for the CRT gun out of the overscan area, so that it is a constant speed by the time it emerges below top edge (vertical) or from left edge (horizontal). Constant speed = linearity issue disappears.
So fudging around the numbers can add/remove extra time for CRT electron gun to move itself and re-accelerate (horizontally or veritcally) so it's a constant speed before showing picture data -- preventing linearity issue caused by not enough time for CRT gun to accelerate before displaying picture data
Cables deliver these pixels over this sequence over the last 100 years, from 1930s analog TVs to 2020s DisplayPort, left-to-right, top-to-bottom. Yesterday, sync/porches were analog commands/timing to control the CRT electron beam. Today for digital flat panels, they're just defacto equivalent to digital comma-separators. But the signal is the same layout for 100 years -- even a 240 Hz FreeSync and G-SYNC signal is transmitted in this signal layout and sequence too! Complete with porches and sync pixels, even 100 years later.
Be noted, increasing Vertical Back Porch increases your horizontal scan rate without decreases to refresh rate. So you may need to back off by a few Hz in refresh rate, to get enough Horizontal Scanrate specroom to hide an edge-only linearity issue.
If you're already maxed-out in refresh rate spec-wise, then for a 5% top-edge-only linearity problem, may require a 5%-reduction in refresh rate if you use the "bigger blanking interval" trick to hide a CRT top-edge linearity problem.
To ease the math calculation of "transfer-bandwidth-to--bigger-VBI", you can use ToastyX CRU and lock the Horizontal Scan Rate (Horizontal Refresh Rate) while increasing Vertical Back Porch a few pixels at a time (try about the pixel height region of your linearity problem to begin with, then adjust slightly bigger/smaller until your "too-little-time-for-electron-gun" linearity problem is now safely hidden in the VBI). ToastyX will automatically reduce refresh rate proportionally to the Vertical Total increasing.
If your linearity problem at top edge is slowly getting worse and worse, it's sometimes a hallmark of a CRT electron gun reset becoming weaker and weaker (becoming too slow to reset during the time interval in a VBI) -- this will require repairwork, hopefully just replacing caps (easy) that are now unable to give a strong enough gun-position-reset kick fast enough.
But it also happens when you use "Automatic" instead of "Manual" in a custom resolution utility, and those blanking intervals are often too small for CRTs. Then it's possibly simply a VBI too small for CRT spec, even would have had the problem when the CRT was new.
On the other hand, a minor degradation may have happened since the display was new -- then making VBI bigger with a larger Back Porch (more time to accelerate the scanout to a constant velocity = fixes the linearity). This is fine for those small 5% top-edge-only linearity problems, but keep a hawk eye for gradually-degrading linearity issue.
Hope this helps understand left-edge and top-edge compressed-linearity and how to quickly fix them;
YMMV -- but sometimes a simple fix, sometimes not.
This may not be applicable to this type of issue, but generically seen on almost any CRT when:
- When you try to put LCD timings to a CRT.
- When you enter too-small sync/porch numbers in Manual mode in a Custom Resolution Utility
- Or when CRT vertical-reset capacitors get slightly too weak to reset electron gun fast enough
Sometimes the simple workaround for linearity at edge is to use a larger blanking interval. The blanking interval's too short of an interval for the CRT gun to move from bottom back to top edge.
If the problem is a bit of top-edge linearity problem, try small increases to Vertical Back Porch in a Custom Resolution Utility. If your linearity problem spreads over 50 pixels, then perhaps increase Vertical Back Porch by about 50.
You can also experiment by transferring Front Porch to Back Porch so you have more time interval on the other side of a sync signal. Usually on older CRTs you need more time after the Vertical Sync, to allow time for the CRT electron gun to reset back to the top edge, and if it's not reving up back to a constant speed when starting the top-to-bottom sweep, it can create a top-edge linearity issue (compressed linearity at top, or a fold-over effect)
The numbers in a Custom Resolution Utility can easily be human-visualized as overscan area beyond edges of screen, in this signal structure layout:
Pixels are delivered left-to-right, top-to-bottom, like a calendar or a book. The back porches is the overscan area as the "CRT electron gun acceleration area" after the reset (triggered by Sync signal). It's in the overscan area above top edge (vertical) and in the overscan area to left edge (horizontal). So any compressed linearity there, is indicative of not enough beam acceleration.
The fix, naturally, is to embiggen the Back Porch (overscan area) -- to give more hidden overscan area as "acceleration room" for the CRT gun out of the overscan area, so that it is a constant speed by the time it emerges below top edge (vertical) or from left edge (horizontal). Constant speed = linearity issue disappears.
So fudging around the numbers can add/remove extra time for CRT electron gun to move itself and re-accelerate (horizontally or veritcally) so it's a constant speed before showing picture data -- preventing linearity issue caused by not enough time for CRT gun to accelerate before displaying picture data
Cables deliver these pixels over this sequence over the last 100 years, from 1930s analog TVs to 2020s DisplayPort, left-to-right, top-to-bottom. Yesterday, sync/porches were analog commands/timing to control the CRT electron beam. Today for digital flat panels, they're just defacto equivalent to digital comma-separators. But the signal is the same layout for 100 years -- even a 240 Hz FreeSync and G-SYNC signal is transmitted in this signal layout and sequence too! Complete with porches and sync pixels, even 100 years later.
Be noted, increasing Vertical Back Porch increases your horizontal scan rate without decreases to refresh rate. So you may need to back off by a few Hz in refresh rate, to get enough Horizontal Scanrate specroom to hide an edge-only linearity issue.
If you're already maxed-out in refresh rate spec-wise, then for a 5% top-edge-only linearity problem, may require a 5%-reduction in refresh rate if you use the "bigger blanking interval" trick to hide a CRT top-edge linearity problem.
To ease the math calculation of "transfer-bandwidth-to--bigger-VBI", you can use ToastyX CRU and lock the Horizontal Scan Rate (Horizontal Refresh Rate) while increasing Vertical Back Porch a few pixels at a time (try about the pixel height region of your linearity problem to begin with, then adjust slightly bigger/smaller until your "too-little-time-for-electron-gun" linearity problem is now safely hidden in the VBI). ToastyX will automatically reduce refresh rate proportionally to the Vertical Total increasing.
If your linearity problem at top edge is slowly getting worse and worse, it's sometimes a hallmark of a CRT electron gun reset becoming weaker and weaker (becoming too slow to reset during the time interval in a VBI) -- this will require repairwork, hopefully just replacing caps (easy) that are now unable to give a strong enough gun-position-reset kick fast enough.
But it also happens when you use "Automatic" instead of "Manual" in a custom resolution utility, and those blanking intervals are often too small for CRTs. Then it's possibly simply a VBI too small for CRT spec, even would have had the problem when the CRT was new.
On the other hand, a minor degradation may have happened since the display was new -- then making VBI bigger with a larger Back Porch (more time to accelerate the scanout to a constant velocity = fixes the linearity). This is fine for those small 5% top-edge-only linearity problems, but keep a hawk eye for gradually-degrading linearity issue.
Hope this helps understand left-edge and top-edge compressed-linearity and how to quickly fix them;
YMMV -- but sometimes a simple fix, sometimes not.
Last edited: