Recommend TEST PATTERN GENERATOR for CRT Calibration???

[Dochartaigh]

TLDR: Looking for recommendations for a CRT test signal pattern generator I can take on-the-go (battery powered), AND use it at home with my colorimeter for more in-depth color calibration. RGBS output for 15Khz SD and 31 signals is a must (and secondarily YPbPr/Component, and YC/CV for NTSC).

DISCLAIMER: I want the type of handheld device professionals used back in the day, to pair with my professional/broadcast BVM's. Do NOT want to have to buy, repurpose/program-CRTEmu/etc, and lug-around, a desktop or laptop computer for 15khz output. Nor a Raspberry Pi with RGBS hat (or similar) and all that would go with that... just want a smaller size box I simply plug into a monitor and I'm all set.

ALREADY OWN/TRIED: Extron VTG 300 - has limited patterns (need multiple IRE screens, and these do either 100, or newest models do 20/100) so this one is out. Sencore VP300 I just bought off eBay, despite saying it can do "Analog Composite Sync", doesn't seem to be able to do SD resolutions in RGBS (or I'm doing it wrong, or it's broke) -- will prob. return to eBay (open to ideas about how to make it do this though!).




LONGER VERSION:
I'm slowly getting into calibrating my CRT's (mostly BVM's and PVM's, but a consumer set or two down the line as well) with my colorimeter, but lack a proper signal generator. So far I've only done a white balance with bias/gain on my 2x BVM's via their very basic built-in 100/20 IRE patterns, measuring them with a ColorMunki colorimeter, using HCFR software. Need RGBS output (some don't have this for whatever reason), and ability to do 15khz and 31hz.

Can anybody recommend some models to get, or look out for? Around $150ish would be ideal. Portable is key, but if there's no options right now a desktop type is fine as long as it's not humongous. I know these are hard to come by now with HD/HDMI being prevalent for so long, but hoping something decent is on eBay now, or a common enough model that when I set alerts I can find one (hopefully portable) and not have to wait forever...



GENERAL QUESTIONS ON CALIBRATING A CRT:
Am I even on the right track with wanting to calibrate CRT's with a RGBS signal? If that Sencore can't even output RGBS (Extron can) maybe that's out of the norm for some reason?

Do people have to calibrate their monitors in RGBS, then again in YPbPr (then I guess also separately for S-Video, Composite, etc.) to compensate for the internal monitor circuits which convert each of these non-RGBS signal types to the pure RGBS that gets sent to the tube? I know there's tons of settings for each of these... just so new to this I'm still learning and that's one of my burning questions (and also why I want to initially calibrate for RGBS, which is the signal I use for 16 out of 17 consoles). Unfortunately the lone video series I can find on color calibrating a PVM (NONE on BVM's I could find)! he's using YPbPr (and 480p so he doesn't have to worry about SD output like I need), and seems to be tweaking all these internal values I don't think even apply for with RGBS...

I also figure that through the years I've seen SO many posts about YPbPr/Component being off in color, and SO many of the external YPbPr>RGBS converters (or RGBS>YPbPr) - even professional ones, seem to be complained about them being off-color ALL the time... and again, the monitor is converting to RGBS at the end anyway... so why not bypass the problematic and go straight RGBS.

[NoAffinity]

Something like this?

http://craftymech.com/arcade-test-pattern-generator/

Although that is ttl sp probably not right for a TV.

Or a dvdo vp30. They have a good variety of test patterns.

Sent from my SM-G955U using Tapatalk

[ShootTheCore]

Well, the MiSTer can be set for native RGB, Composite or Component analog output, and its console cores can run various versions of the 240p Test Suite. I've used the SNES 240p Test Suite pattern generators to calibrate displays before, but there's two drawbacks:

A) The test patterns have to be changed manually using a controller.
B) The SNES hardware has an incorrect gamma curve, and the MiSTer's FPGA SNES core is so accurate that it has the same inaccurate gamma curve as well. Using the MiSTer SNES 240p Test Suite to calibrate with your colorimeter will thus give you an inaccurate gamma curve.

Now that the MiSTer runs PlayStation 1, I expect the PS1 240p Test Suite would deliver more accurate gamma curve results than the SNES Suite. I'll try it out the next time I calibrate a set.

I know you said you didn't want a repurposed Raspberry Pi, but there is a fantastic pattern generator for the Raspberry Pi named PGenerator. Compared to SNES 240p Test Suite:
A) PGenerator has an accurate gamma curve.
B) PGenerator can be driven directly by the colorimeter software on your laptop and can change patterns automatically for you, saving a lot of time when measuring your grayscales.
C) Raspberry Pi is HDMI only normally so you have to work out an analog conversion solution separately.

FWIW, the PGenerator is what they recommend as the best inexpensive portable pattern generator at the AVS forums where they discuss this kind of stuff constantly:
https://www.avsforum.com/threads/the-ch ... d.3023002/

I'm an amateur CRT calibrator myself but I wrote a walkthrough of the tools I use and my calibration process on another forum here - it might be helpful:
https://www.arcade-projects.com/threads ... ost-274247

[orange808]

Barco/Folsom ImagePro HD is good video processor for getting test patterns onto CRTs. Not convenient or portable, though. Should handle all legacy standard signals and regions.

I have no idea what prices they trade at. Should be inexpensive, because the video processing is dated and bad. Test patterns are all it's good for, but you can use them with confidence. I'm not sure I trust a MiSTer and the 240p Test Suite to deliver proper test patterns for a professional grade calibration. Probably best to get used pro gear that was designed to handle the job.

[Dochartaigh]

Something like this?

http://craftymech.com/arcade-test-pattern-generator/

Although that is ttl sp probably not right for a TV.

Or a dvdo vp30. They have a good variety of test patterns.

Sent from my SM-G955U using Tapatalk

Thanks for the link. And yeah, I've seen that one before - only new(ish) production one I've seen actually, but sadly won't work for consumer sets with only Composite or S-Video inputs, right? No SMPTE bars (needed for BVM auto-calibration and how I always set black level first), no standard 100 and 10 IRE patterns... (it might have similar... but I searched for those exact keywords, and the patterns need to be EXACTLY he same... and no mention of those in the manual...). I guess these are just intended to be used for eye-balling things on an arcade monitor sans colorimeter - like basic things like geometry and convergence I guess?

VP30 isn't going to be portable unfortunately... and if I do succumb to a desk-only version if I don't find a portable, a VP30/50 (was, I turned off my alerts for them a year+ ago) even more rare to find than these signal generators it seems lol.

Well, the MiSTer can be set for native RGB, Composite or Component analog output, and its console cores can run various versions of the 240p Test Suite.

Searching this for hours upon hours, I can't tell you how many times a calibration post specifically says to NOT use the 240p test suite I was unhappy to find out... A Mister is also still a shit-show on output - do i use the I/O board for whatever signal? Or should I use "Direct Video" (which their github says not all cores can use either), converters to convert and/or merge sync for certain outputs... all things I'm trying to avoid as I still can't find out if a converter, sync combiner, or whatever changes the needs-to-be-1000%-exact pattern or not.

A Mister is also still $450+ with the parts shortages and inflated prices, right? And generally out of stock everywhere?

I know you said you didn't want a repurposed Raspberry Pi.......

.......C) Raspberry Pi is HDMI only normally so you have to work out an analog conversion solution separately.

This again has me asking "are those converters going to be 100% accurate"? Also need SD... so that means a scaler too? This is kinda a game to me... so if I'm going to be playing around with a BVM with a 2005 MSRP of $20K... I want to make sure the signal generator is 100% legit with a 100% proper EXACT signal...

I also know my multiple Mike Chi Raspberry Pi hats (which DO output 240p) do NOT work on BVM D9 or D14's (sync issue)... and want to say some other (either RGBS or yPbPr) hat also didn't pay nice with some others (pretty positive Pi2SCART doesn't like my Panasonic, but ok with a few others I hooked up that time).

Barco/Folsom ImagePro HD is good video processor for getting test patterns onto CRTs. Not convenient or portable, though. Should handle all legacy standard signals and regions.

Thanks, I'll keep it on my list... seem to be $250 though currently... and was hoping for a full fledged pattern generator, instead of picking up a big old honky scaler or something else that just happens to have some built-in patterns...

[ShootTheCore]

As I mentioned, I'm an amateur at calibration myself so I welcome this discussion as a learning opportunity.

Let me check my understanding. The colorimeter takes measurements from the display and feeds the measurements to software that both displays the measurements and compares them against a target baseline (ColorHCFR set to HDTV-Rec. 709 and D65 temperature in my case).

Assuming that the colorimeter and target comparison software are both accurate, if they both say that my display is calibrated to the target then isn't the display calibrated regardless of what test pattern generator I'm using?

I understand that analog video is subject to variation across signal generation devices due to voltage differences (ie a Raspberry Pi might have more color saturation than a MiSTer if its analog RGB circuitry allows more current through) but you'll encounter that same problem with professional dedicated signal generator devices as well - you're not using the signal generator to play games or watch movies.

My mindset is that the display should ideally be calibrated against the video generation device you're going to use on it the most. If you're going to be primarily using a MiSTer to play games on your display, use that same MiSTer as your signal generator when calibrating. Ditto for a Raspberry Pi.

[Dochartaigh]

I understand that analog video is subject to variation across signal generation devices due to voltage differences (ie a Raspberry Pi might have more color saturation than a MiSTer if its analog RGB circuitry allows more current through) but you'll encounter that same problem with professional dedicated signal generator devices as well.

My mindset is that the display should ideally be calibrated against the video generation device you're going to use on it the most. If you're going to be primarily using a MiSTer to play games on your display, use that same MiSTer as your signal generator when calibrating. Ditto for a Raspberry Pi.

It's ALWAYS best to start with a rock-solid baseline. What if that initial calibration signal going to the display is off? Then your adjustments are off and skewed towards the off-ness of that particular console... then if you play another console and that one is off in another (maybe opposite) direction... you're then stacking tolerances as they say which never turns out good... at that point what is even the point of calibrating to a well-known, "correct", baseline to begin with? (if what you begin with isn't initially exactly to-spec?). I can't control my consoles being funky... but I can at least get the monitor's baseline to be correct I figure.

Like you said it yourself that the SNES has a bad gamma curve... and in my eyes the SNES is pretty much the gold standard so that stinks (I literally use its grid specifically to calibrate EVERY monitor I've ever owned so at least that's served me well)... I've also noticed when I tried to calibrate off any console's 240p test suite that certain R, G, or B levels will fluctuate a LOT more than I get when running from my Extron or Sencore (both professional testers), or the internal signals from my BVM's.

Like using my Extron tester I get this below result for 100 IRE, and it doesn't fluctuate more than maybe 0.1 to 0.2% for any RGB value (not moving much at all for a lot of the time too) - pretty darn rock solid.



...but when I run it from any of my consoles with 240p suite (tried NES, Genesis, and SNES), I can get up to +/-2%+ variation every couple seconds on one or more colors (think SNES red was the biggest constant swing back and forth... might be hardware issue, who knows). So just in my cursory experiments in comparing 3x consoles, to 4x pro signal generators (Extron, Sencore, 2x internal signal from a BVM-20F1U and D20F1U) that at the very least has shown that I get if not more accurate, at the very least way more STABLE testing conditions... which is again why I would like to stick with pro gear (you know... besides also being decked out wall-to-wall in professional gear and wanting to stick to the same motif lol).

[ShootTheCore]

Well, if you want a rock-solid baseline and want to stack the fewest tolerances then I really don’t think you can do better than an R-Pi 4 running PGenerator. The AVS forum guys throughly tested and confirmed it as bit-accurate against the $1000 Accupel DGA-6000 generator, supports every output resolution up to and including 4K, and can easily be made portable with a USB power bank. You will have to deal with converting the digital output to analog for a CRT but at least that should be the only place for potential variance in your pattern generation chain.

https://www.avsforum.com/threads/the-ch ... d.3023002/

[Dochartaigh]

Well, if you want a rock-solid baseline and want to stack the fewest tolerances then I really don’t think you can do better than an R-Pi 4 running PGenerator. The AVS forum guys throughly tested and confirmed it as bit-accurate against the $1000 Accupel DGA-6000 generator, supports every output resolution up to and including 4K

Can they do 15khz though? That's the majority of my monitors. I checked that topic, the linked one to the developer's topic, and searched google for "pgenerator" "15khz".... and didn't see a single hit, so I'm thinking not unfortunately.

Also really hoping for same functionality as the portable testers I already have signal-wise. That means RGBS, YPbPr, RGBHV, S-Video, and Composite (and one which could ALSO do HDMi would be awesome! - many ones from the SD to HD switch era seem to have DVI-I which can do both analog and digital signals which is cool).

...You're simply not getting all those common signal types out of a Raspberry Pi anymore unless some new product I don't know about came out (since RetroTink Ultimate was discontinued years ago). Then we're back to the same question I still don't know the answer to: IF those different hats even exist/are available still, would those different hat's and/or HDMI-to-whatever converters (and or maybe chaining a scaler + a converter...) change the signal and make it no longer accurate???? I just don't know.

Also, once we're at the point where I need a Pi, maybe a scaler?, maybe multiple converters?, different power supplies for all this stuff, a controller/keyboard or whatever... you're kinda at the same point of a much larger desktop type signal generator, so say goodbye to the portability factor.

**edit, also wanted to thank you for all your suggestions - really don't mean to be shutting down all them! just know how I want to use one of these and what will or won't work for me!

[ShootTheCore]

PGenerator just follows whatever resolution you set the main R-Pi up with, and the R-Pi does support 15 kHz output. Take a look at the video_mode setting in config.says documented here:
https://www.raspberrypi.com/documentati ... g_txt.html

There’s also a pretty good YouTube tutorial on setting PGenerator up for a CRT and calibrating with it here:
https://youtu.be/D6z0wS5oRoE

As for accurate analog signals out of the R-Pi, I’ll have to do some research on that myself. PGenerator gets a lot of support so I imagine that there’s a good solution available. I know that the R-Pi can drive RGB analog signals directly from its GPIO if it’s configured properly, and there are analog connectors (ie SCART, JAMMA) that connect to the GPIO, so it should be able to output an accurate native analog signal without having to pipe the HDMI output through a D/A convertor.

[Dochartaigh]

As for accurate analog signals out of the R-Pi, I’ll have to do some research on that myself. PGenerator gets a lot of support so I imagine that there’s a good solution available. I know that the R-Pi can drive RGB analog signals directly from its GPIO if it’s configured properly, and there are analog connectors (ie SCART, JAMMA) that connect to the GPIO, so it should be able to output an accurate native analog signal without having to pipe the HDMI output through a D/A convertor.

I'll bet 99% of PGenerator's are using HDMI (just like everything else nowaday). Digital signals like HDMI are generally speaking nearly always perfect as it's just moving along 1's and 0's... This is most likely why they play super nice with modern systems. ...Analog is a totally different animal with how it is created, transmits, is received, and decoded though...

...Especially on a Pi. I can tell you right now the Pi has always been a shit-show, period (I've built over 250 Raspberry Pi systems back in the day, know them very well). Pi3's can output composite stock, and RGB off the GPIO with limited extra circuitry.... problem is even RGBHV/VGA can be a crap shoot to get the best signal from them, VGA 666 boards claiming they did it all at the highest bandwidth, then others proved them wrong, then for a time only the RetroTink had the best quality (want to say 24-bit for some reason?), then somebody came out with a VGA 666 clone that was actually just as good... For other outputs like S-Video and YPbPr, I don't even know if there's anything out there anymore since RetroTink Ultimate was discontinued a long time ago... MANY of those still gave non-standard signals which BVM's didn't like (which I've already mentioned several times before in this topic), with NO fix.

RP4 literally took what? 2 years just to get its first RGBS solution (underlying kernel issues or something over my head like that). I haven't followed them much lately, but I know people were having a hell of a time getting RP3 hats to even run on a RP4, and the only RGBS solution I know is the RGB-PI, which their OS is STILL not out of their beta firmware (I have one, but have barely touched...). Again, I'm not up to date but I don't think there's even a YPbPr solution that doesn't involve an external signal converter box as far as I know... haven't heard of a multi one with Composite and S-Video too, like my existing signal generators can do.

[dr_myslihiiri]

Extron VTG 300 - has limited patterns (need multiple IRE screens, and these do either 100, or newest models do 20/100) so this one is out.

IDK whether this helps you any because the device doesn't otherwise meet your specs, but the next model up (the VTG-400D/DVI) allows the level of some test patterns to be adjusted. From the manual (p. 22):

From the default cycle menu, the video level of specific test patterns can be adjusted in 1% increments (1 IRE) from 0 through 100% (0 to 100 IRE). Currently, there are six test patterns that support this adjustable level feature.

• Window (variable level)*
• Flat field*
• Flat field with targets*
• Checkerboard*
• Contrast transfer function
• Hum bar detect

* These test patterns have globally adjustable video levels, i.e., a change to one results in an identical change to all of the others

Unfortunately the lone video series I can find on color calibrating a PVM (NONE on BVM's I could find)! he's using YPbPr (and 480p so he doesn't have to worry about SD output like I need), and seems to be tweaking all these internal values I don't think even apply for with RGBS...

Is this the series you're referring to? It happens to be the only in-depth tutorial on calibrating CRT displays I've bumped into in my own research as well (in addition to CurtPalme.com's old guide). If anyone can point to other in-depth resources on the subject worth checking out, please do.

[Dochartaigh]

IDK whether this helps you any because the device doesn't otherwise meet your specs, but the next model up (the VTG-400D/DVI) allows the level of some test patterns to be adjusted.

Thank you, but well aware. Unfortunately those are unobtanium. Haven't been available for a long time. One up for $1300 right now though (and yes, I know that's one of 'those' eBay sellers asking crazy mark-up)... with the last being sold months ago for around $700... which stinks as pre-covid could have picked one up multiple times for around $100-150ish maybe. Those aren't battery powered either, but I would have made an exception since they're still pretty small, and still an all-in-one device like I'm looking for.

Is this the series you're referring to? It happens to be the only in-depth tutorial on calibrating CRT displays I've bumped into in my own research as well (in addition to CurtPalme.com's old guide). If anyone can point to other in-depth resources on the subject worth checking out, please do.

Yup, that's the PVM one I was talking about.

Wish somebody made one on the BVM's... I know you really need an oscilloscope and advanced electrical knowledge and such to really calibrate them properly (PVM's too for that matter), but that's out of a LOT of peoples expertise level... where even a novice can watch a video and use a colorimeter basically and mess with the service/maintenance menu's just perfectly fine at least...

[orange808]

Thanks, I'll keep it on my list... seem to be $250 though currently... and was hoping for a full fledged pattern generator, instead of picking up a big old honky scaler or something else that just happens to have some built-in patterns...

Makes sense. Let me know if you find something more portable at a reasonable price. I would also like something more convenient.

I prefer that particular kind of machine (the ImagePro II is also good) because it was marketed directly to video professionals for events. So, the test patterns were a major part of the design. Any serious audio/video pro would have a stack, so the rack size machine made sense.

Beyond that, Barco manufactures high end projectors and professional monitors, so their "in house" video processor was designed to test and configure their gear. The test patterns are so rubust, because typical projector configuration is significantly more involved than a CRT or panel display. (You just turn on a dedicated display and the image appears; with a beamer, I would have to write a book to explain all the considerations.)

Of course, all kinds of CRT's and legacy signals were still in use (if just barely) when Barco made those video processors, so you know the machines were specifically crafted for that use case, actually tested using Barco's own pro display gear, and used in real life at live events. It's not an afterthought or a bonus.

I have confidence the patterns are right on all the outputs and in all the color spaces, so you'll get your calibration right. Obviously, if the test pattern is wrong, everything else is wrong.

[ShootTheCore]

I did some asking around both on Twitter and in the RetroTech USA CRT Repair Discord where a lot of calibration folks hang out (including StickFreaks who made the YouTube calibration tutorial series linked to earlier).

The two options recommended to me were:

A) A Sony DVD player and a calibration disc, such as either the Spears & Munsil or the Danmons FreeCalRec601. The Sony players are specifically mentioned because they have an accurate DAC. The Sony PS3 is mentioned frequently as a great DVD player option since it has both HDMI and analog output capabilities and the accurate DAC.
https://github.com/danmons/FreeCalRec601

B) Raspberry Pi running PGenerator and the HD Fury 3 DAC. As mentioned before, PGenerator is already thoroughly tested to be bit-accurate at the digital output level, and the HD Fury 3 is a very accurate DAC that won't scale or manipulate the digital output. There's a DAC comparison table here:
https://docs.google.com/spreadsheets/d/ ... edit#gid=0

It turns out that using direct analog out of the Raspberry Pi through the GPIO port as I considered before won't work with PGenerator because the PGenerator is hard-coded to look at the HDMI information.

[dr_myslihiiri]

Unfortunately those are unobtanium. Haven't been available for a long time. One up for $1300 right now though (and yes, I know that's one of 'those' eBay sellers asking crazy mark-up)... with the last being sold months ago for around $700... which stinks as pre-covid could have picked one up multiple times for around $100-150ish maybe.

FWIW, one was recently listed on eBay Kleinanzeigen (a German online marketplace) for €200 (~$215), so I suppose bargains can still be had if you know where to look (eBay proper not being one of those places, obviously).