OLED needs to mature. it's yields are low atm. also i'm sure latency needs to come down. THERE is a metric shit ton characteristics of CRT that really help it out due to it's analogue nature. such things as low latency and amazing image reproduction.
cdoublejj, if you edit the file make sure to use a program that doesn't change anything except what you're explicitly editing or in other words, don't use Notepad. I believe Programmer's Notepad is recommended for editing the .dat file. I'd also get a program that does binary comparisons and make sure nothing else has changed before uploading it to the monitor.
I picked up a Dell P991 from craigslist a few days ago. It has the blue tint and flyback lines. Through some google searching I found the resistor hack and the guide on programming the EEPROM. I don't have the RS232 voltage adapter yet but I'll be getting it in a few days and will be setting up the software environment (gonna have to bust out the VMs for this one I think) this weekend.
Since I haven't played with WINDAS yet I don't know what exactly is changeable on the P991. Knowing that an analog voltage value can be changed is very exciting to me. The reason I picked this monitor up is because I fried a CRT TV while trying to measure signals (make sure to use isolation transformers and ground the probe to the chassis!). My undergrad senior project is to take an existing CRT and see how far it can be pushed in terms of refresh rate and maybe color depth (though the latter is hard to measure and I doubt I'll be getting microvolt accuracy out of anything). Nvidia graphics drivers let you set the resolution, refresh rate, and even timing parameters of the VGA output. However if you try to feed the P991 with an unsupported VGA format it won't take it. I'm wondering if there's a way to change what the P991 thinks is acceptable VGA formats through WINDAS. Does anyone have any insight on this? The reason I'm so curious about this is because my original method was to intercept and replace the R, G, B, Hscan, and Vscan signals (before they get to HV amps). This takes quite a bit of FPGA work and not to mention dangerous interaction with circuitry. I'm also weary of frying more CRTs since I now know how easy that is.
Also does anyone know where I can find the schematic or service manual for the P991? I couldn't find anything with a quick google search but I know they can be hidden away on the internet.
@willis936 said:
The reason I picked this monitor up is because I fried a CRT TV while trying to measure signals (make sure to use isolation transformers and ground the probe to the chassis!). My undergrad senior project is to take an existing CRT and see how far it can be pushed in terms of refresh rate and maybe color depth (though the latter is hard to measure and I doubt I'll be getting microvolt accuracy out of anything). Nvidia graphics drivers let you set the resolution, refresh rate, and even timing parameters of the VGA output. However if you try to feed the P991 with an unsupported VGA format it won't take it. I'm wondering if there's a way to change what the P991 thinks is acceptable VGA formats through WINDAS. Does anyone have any insight on this? The reason I'm so curious about this is because my original method was to intercept and replace the R, G, B, Hscan, and Vscan signals (before they get to HV amps). This takes quite a bit of FPGA work and not to mention dangerous interaction with circuitry. I'm also weary of frying more CRTs since I now know how easy that is.
Don't mess around with horizontal frequency; it will fry the flyback. Flyback transformers heat up more and die much faster by high horizontal frequencies. (You can even see the energy waste on a socket power meter.) My first Nokia 417TV burnt out after few years by overclocking (it was only rated for 800 lines modes, but on Linux I successfully used 1024 and more). So my new Nokia 417TV is solely used as videophile PAL TV and hasn't degraded much despite it runs roughly 14h per day. Flatscreens jerk way too much and make me seasick and nothing beats a CRT with fast scrolling classic 2D games.
If I live long enough, I somewhen will construct an open source colour processor for improving colour curves, sharpness and possibly horizontal convergence and linearity of genuine analogue CRT TVs. Important is to buffer only 1 scanline to avoid the atrocious jerk of full-frame image dishancers (100Hz scandoublers etc.) known from 1990th TVs. I do not tolerate digital motion recognition or snow removal algorithms (those as well remove rain...) in a videophile CRT TV. If a TV can not handle lightgun games or a C64 lightpen, it does something wrong. The horizontal resolution should be at least full-HD with per RGB colour channel 16 or even 24 bit (i.e. much higher than 1990th trash), which nowadays in the age of HDTV is cheaply available. May be there will be additional modern framebuffer based algorithms (having 1 frame delay), but this should not be the main function and menus should display a warning message that such modes reduce image quality. The hardware cost should be somewhere around 50..100EUR (no moonpriced boutique stuff). May be I use something like a Raspberry-Pi PCB or even FPGA to minimize latency. Likely it will somehow use the ColorHCFR colorimeter software. The device shall be capable to store on external media (e.g. SD cards) settings for any count of different TV sets or projectors to be useful also for analogue TV set collectors. I don't know if even existing Linux settop boxes (Dreambox etc.) in PAL mode may have enough surplus CPU power to run such a colour processor in the background.
Again, do not play with the horizontal frequency! If you don't know what you are doing, better first read this FAQ instead of killing yourself:
Unless there are hardware based protection circuits built-in, higher frequency at the flyback can result in uncontrolled rise of CRT accelleration voltage (the infamous 25 to 30KV for those you likely have no meter) resulting in excessive xray emission. So if you (and all your neighbours) don't enjoy getting cancer and still want to procreate babies instead of mutated freaks, do not attempt this! Better find out which exact part in a Trinitron monitor makes the black turn pale (too bright) when cold and develope a temperature controls (e.g. heating resistor with sensor) for it.
I have always thought that the flyback transformer on complex CRT monitors (aas opposed to TVs and fixed frequency monitors) is driven by a separate circuit since it would be difficult to build a flyback transformer that could operate efficiently in the whole 30kHz - 120kHz frequency range. I neer checked that, so I may be wrong.
However, even if the transformer is driven by the horizontal deflection transistor, there is no aboslute need for it (the flyback transformer does not have to operate at the same frequency as the horizontal deflection as its outputs are used to power the heater and as high voltage DC).
So, he can make a separate driver for the flyback transformer that operates at the correct frequency and then play with the horizontal deflection which will then be limited by the frequency response of the cathode drivers (pixel clock) and the deflection coil.
I did find the service manual! For VGA timings that it accepts I'm seeing that the H-freq ranges from 31.468kHz (720x400 @ 70.084Hz) to 106.25kHz (1600x1200 @ 85Hz, that's a 24% higher pixel clock than 1080p60!). That FAQ that Windler posted is quite the read. I'll read through it before doing anything more than lowering G2 to lower the brightness.
If I know I shouldn't exceed 106kHz on the H-freq I should still have a theoretical max refresh rate of 236 Hz if I keep the resolution at 720x480. This is a rather unsexy pixel clock of 82MHz but it would give me something to try and talk about.
I'm not understanding what you're suggesting Pentium. Does the flyback not drive the horizontal scanning coils directly?
Also here's a quick thought: If I'm not mistaken the flyback has to work at such a high voltage because it has to go bring the beam back to the other side of the screen in the short Hblank period. I wonder how much further things could be pushed on CRT if you didn't need to scan back across after each line (and all of the way up/down at the end of each frame). Put another way the beam would snake across the screen. This would require flipping lines and frames in a weird way which would have been near impossible back in the day but in the digital era it wouldn't be hard. You'd also have to account for the fact that there would be a different "time-since-last-hit-with-the-beam" value for each pixel by changing the intensity (and knowing something about the phosphor). You might be able to get much higher pixel clocks out of CRTs that way.
The flyback transformer is used to provide high voltage DC for the anode (25-30kV), focusing and G2 (500V or so). It also usually is used to provide low voltage AC for the heaters.
The deflection coils are driven using low voltage. The flyback is driven from horizontal deflection purely as a cost saving measure - the TV/monitor aready has a high frequency oscillator (the H driver), so the transformer may be driven from it instead of building another circuit just for the flyback.
Some very old TVs used linear power supplies for the high voltage (around 5kV for B&W sets), essentially just another winding on the mains transformer. However, it was found out that this was very dangerous - such power supply could provide much more current than needed and could kill a human very easily, while the flyback transformer cannot provide so much current (it still can kill, but you have more chances of surviving). The fact that the flyback transformer is driven from the horizontal deflection also added a new safety feature - if the horizontal oscillator stops for whatever reason, the high voltage disappears and you do not burn a spot (or a line) in the screen.
The reason for such high voltage in a CRT is because electrons repel each other, so the beam gets wider over time. If you accelerate the electrons to something like 30% of the speed of light, they do not have time to move away from each other before hitting the screen.
AFAIK the horizontal frequency and high voltage are coupled 1. as a energy saving measure (else the deflection kickback voltage had to heat a shunt resistor) and 2. because the electric pulse and magnetic field of an asynchronous high voltage trafo would make the picture wobble terribly when not perfectly shielded, which makes it expensive. Websearch for vector monitors (well known from arcade videogames and the Vectrex game console). These run the HV trafo at an independent frequency and can move the beam in any pattern you like. (I still used 60Hz on my PC before a blown capacitor(?) made this mode fail on my SGI monitor. Even 50Hz was ok to me (despite visible flicker) but I otherwise can not bear motion blur and choppyness of modern flat TVs.) I wonder why nobody made a TV system with spiral beam pattern (like some radar monitors) instead of scanlines. May be that timing distortions looked too ugly when it caused rotations instead of wavy picture.
CRT heater voltage is (at least nowadays) not derived from the flyback but from the mains power supply because it needs relatively high current at low voltage. May be that early vacuum tube TVs without mains trafo (i.e. deriving all medium high voltages directly from rectified mains voltage connected to one chassis pole - very dangerous) used the flyback for heaters, but I don't think so. Vacuum tube TVs also often wired all heaters in series (like xmas faery lights) to use a higher voltage at lower current and thus cheaper trafo (or pure mains voltage through only a resistor). In the age of switching power supplies (all modern PC monitors) chassis don't use rectified mains voltage anymore.
Recently I repaired two portable TVs (B&W but transistor) and they both had heater powered from the flyback transformer. Then again, both TVs could run on DC power so the heater was powered from the flyback transformer as a cost saving measure (no need for a separate converter).
However, looking at a circuit diagram of my FW900 it does appear (I didn't analyze the circuit thoroughly) that the flyback transformer and the horizontal deflection coil are driven by separate circuits. It must be very difficult to build a flyback transformer that operates efficiently at both 30kHz and 120kHz (and all frequencies in between).
I'm getting ECS syntax errors in windas. I'm using a USB to serial converter that goes to an RS232 to TTL board. I'm fairly certain I've hooked the pinouts to the right place judging by the service manual and silk screen on the D board. When I look at the data file it's mostly 0s and skips with a few values here and there. I may have shorted some pins together at some point but it's difficult to be sure. Has anyone else run into this issue before? The model selected is the Dell P991 F99, I'm running windows 7-64, and windas is being run in compatibility mode for Windows 98 and as administrator. I've set the serial port to COM1 (when it was anything else it wouldn't work, the setting kept reverting to COM1).
Please let me know if anyone sees anything wrong with my setup. If not I'll just order an adapter that actually has female pin leads already on it...
Although I am not familiar with your adapter (I used a homemade transistor circuit on a real RS232 port), it is extremely important not to use the supply voltage from the monitor data connector. Although it may work in an ideal world, the monitor tends to switch it off in various situations (timeout, bad firmware flash etc.) and so can make things much more difficult. Thus only use power from the USB port or a battery, but NOT the data port of the monitor.
In my experience, USB to serial converters often fail, with two different ways to manifest:
1-poor design won't work at all with a particular motherboard
2-serial device recognizes the conversion process, and refuses to work
Serial ports on PCI or PCIe serial port or multipurpose cards just work, like motherboard serial ports, unlike USB to serial converters.
I have an old WinXP SP1 PC someone gave me that I use for nothing else than WinDAS, which I originally set up precisely because of possession of several washed out P991s.
@Gooberslot said:
cdoublejj, if you edit the file make sure to use a program that doesn't change anything except what you're explicitly editing or in other words, don't use Notepad. I believe Programmer's Notepad is recommended for editing the .dat file. I'd also get a program that does binary comparisons and make sure nothing else has changed before uploading it to the monitor.
that alone sounds like i would need to look up some more how to videos. I'm going to be uploading the firmware and a video of me loading it up in windas in hopes someone one out there smarter than i can make sense of it all.
Comments
OLED needs to mature. it's yields are low atm. also i'm sure latency needs to come down. THERE is a metric shit ton characteristics of CRT that really help it out due to it's analogue nature. such things as low latency and amazing image reproduction.
i just can't find G2 in the registry tree in windas for my p992.
cdoublejj, if you edit the file make sure to use a program that doesn't change anything except what you're explicitly editing or in other words, don't use Notepad. I believe Programmer's Notepad is recommended for editing the .dat file. I'd also get a program that does binary comparisons and make sure nothing else has changed before uploading it to the monitor.
I picked up a Dell P991 from craigslist a few days ago. It has the blue tint and flyback lines. Through some google searching I found the resistor hack and the guide on programming the EEPROM. I don't have the RS232 voltage adapter yet but I'll be getting it in a few days and will be setting up the software environment (gonna have to bust out the VMs for this one I think) this weekend.
Since I haven't played with WINDAS yet I don't know what exactly is changeable on the P991. Knowing that an analog voltage value can be changed is very exciting to me. The reason I picked this monitor up is because I fried a CRT TV while trying to measure signals (make sure to use isolation transformers and ground the probe to the chassis!). My undergrad senior project is to take an existing CRT and see how far it can be pushed in terms of refresh rate and maybe color depth (though the latter is hard to measure and I doubt I'll be getting microvolt accuracy out of anything). Nvidia graphics drivers let you set the resolution, refresh rate, and even timing parameters of the VGA output. However if you try to feed the P991 with an unsupported VGA format it won't take it. I'm wondering if there's a way to change what the P991 thinks is acceptable VGA formats through WINDAS. Does anyone have any insight on this? The reason I'm so curious about this is because my original method was to intercept and replace the R, G, B, Hscan, and Vscan signals (before they get to HV amps). This takes quite a bit of FPGA work and not to mention dangerous interaction with circuitry. I'm also weary of frying more CRTs since I now know how easy that is.
Also does anyone know where I can find the schematic or service manual for the P991? I couldn't find anything with a quick google search but I know they can be hidden away on the internet.
Don't mess around with horizontal frequency; it will fry the flyback. Flyback transformers heat up more and die much faster by high horizontal frequencies. (You can even see the energy waste on a socket power meter.) My first Nokia 417TV burnt out after few years by overclocking (it was only rated for 800 lines modes, but on Linux I successfully used 1024 and more). So my new Nokia 417TV is solely used as videophile PAL TV and hasn't degraded much despite it runs roughly 14h per day. Flatscreens jerk way too much and make me seasick and nothing beats a CRT with fast scrolling classic 2D games.
If I live long enough, I somewhen will construct an open source colour processor for improving colour curves, sharpness and possibly horizontal convergence and linearity of genuine analogue CRT TVs. Important is to buffer only 1 scanline to avoid the atrocious jerk of full-frame image dishancers (100Hz scandoublers etc.) known from 1990th TVs. I do not tolerate digital motion recognition or snow removal algorithms (those as well remove rain...) in a videophile CRT TV. If a TV can not handle lightgun games or a C64 lightpen, it does something wrong. The horizontal resolution should be at least full-HD with per RGB colour channel 16 or even 24 bit (i.e. much higher than 1990th trash), which nowadays in the age of HDTV is cheaply available. May be there will be additional modern framebuffer based algorithms (having 1 frame delay), but this should not be the main function and menus should display a warning message that such modes reduce image quality. The hardware cost should be somewhere around 50..100EUR (no moonpriced boutique stuff). May be I use something like a Raspberry-Pi PCB or even FPGA to minimize latency. Likely it will somehow use the ColorHCFR colorimeter software. The device shall be capable to store on external media (e.g. SD cards) settings for any count of different TV sets or projectors to be useful also for analogue TV set collectors. I don't know if even existing Linux settop boxes (Dreambox etc.) in PAL mode may have enough surplus CPU power to run such a colour processor in the background.
Again, do not play with the horizontal frequency! If you don't know what you are doing, better first read this FAQ instead of killing yourself:
http://www.repairfaq.org/sam/monfaq.htm
Unless there are hardware based protection circuits built-in, higher frequency at the flyback can result in uncontrolled rise of CRT accelleration voltage (the infamous 25 to 30KV for those you likely have no meter) resulting in excessive xray emission. So if you (and all your neighbours) don't enjoy getting cancer and still want to procreate babies instead of mutated freaks, do not attempt this! Better find out which exact part in a Trinitron monitor makes the black turn pale (too bright) when cold and develope a temperature controls (e.g. heating resistor with sensor) for it.
I have always thought that the flyback transformer on complex CRT monitors (aas opposed to TVs and fixed frequency monitors) is driven by a separate circuit since it would be difficult to build a flyback transformer that could operate efficiently in the whole 30kHz - 120kHz frequency range. I neer checked that, so I may be wrong.
However, even if the transformer is driven by the horizontal deflection transistor, there is no aboslute need for it (the flyback transformer does not have to operate at the same frequency as the horizontal deflection as its outputs are used to power the heater and as high voltage DC).
So, he can make a separate driver for the flyback transformer that operates at the correct frequency and then play with the horizontal deflection which will then be limited by the frequency response of the cathode drivers (pixel clock) and the deflection coil.
I did find the service manual! For VGA timings that it accepts I'm seeing that the H-freq ranges from 31.468kHz (720x400 @ 70.084Hz) to 106.25kHz (1600x1200 @ 85Hz, that's a 24% higher pixel clock than 1080p60!). That FAQ that Windler posted is quite the read. I'll read through it before doing anything more than lowering G2 to lower the brightness.
If I know I shouldn't exceed 106kHz on the H-freq I should still have a theoretical max refresh rate of 236 Hz if I keep the resolution at 720x480. This is a rather unsexy pixel clock of 82MHz but it would give me something to try and talk about.
I'm not understanding what you're suggesting Pentium. Does the flyback not drive the horizontal scanning coils directly?
Also here's a quick thought: If I'm not mistaken the flyback has to work at such a high voltage because it has to go bring the beam back to the other side of the screen in the short Hblank period. I wonder how much further things could be pushed on CRT if you didn't need to scan back across after each line (and all of the way up/down at the end of each frame). Put another way the beam would snake across the screen. This would require flipping lines and frames in a weird way which would have been near impossible back in the day but in the digital era it wouldn't be hard. You'd also have to account for the fact that there would be a different "time-since-last-hit-with-the-beam" value for each pixel by changing the intensity (and knowing something about the phosphor). You might be able to get much higher pixel clocks out of CRTs that way.
The flyback transformer is used to provide high voltage DC for the anode (25-30kV), focusing and G2 (500V or so). It also usually is used to provide low voltage AC for the heaters.
The deflection coils are driven using low voltage. The flyback is driven from horizontal deflection purely as a cost saving measure - the TV/monitor aready has a high frequency oscillator (the H driver), so the transformer may be driven from it instead of building another circuit just for the flyback.
Some very old TVs used linear power supplies for the high voltage (around 5kV for B&W sets), essentially just another winding on the mains transformer. However, it was found out that this was very dangerous - such power supply could provide much more current than needed and could kill a human very easily, while the flyback transformer cannot provide so much current (it still can kill, but you have more chances of surviving). The fact that the flyback transformer is driven from the horizontal deflection also added a new safety feature - if the horizontal oscillator stops for whatever reason, the high voltage disappears and you do not burn a spot (or a line) in the screen.
The reason for such high voltage in a CRT is because electrons repel each other, so the beam gets wider over time. If you accelerate the electrons to something like 30% of the speed of light, they do not have time to move away from each other before hitting the screen.
AFAIK the horizontal frequency and high voltage are coupled 1. as a energy saving measure (else the deflection kickback voltage had to heat a shunt resistor) and 2. because the electric pulse and magnetic field of an asynchronous high voltage trafo would make the picture wobble terribly when not perfectly shielded, which makes it expensive. Websearch for vector monitors (well known from arcade videogames and the Vectrex game console). These run the HV trafo at an independent frequency and can move the beam in any pattern you like. (I still used 60Hz on my PC before a blown capacitor(?) made this mode fail on my SGI monitor. Even 50Hz was ok to me (despite visible flicker) but I otherwise can not bear motion blur and choppyness of modern flat TVs.) I wonder why nobody made a TV system with spiral beam pattern (like some radar monitors) instead of scanlines. May be that timing distortions looked too ugly when it caused rotations instead of wavy picture.
CRT heater voltage is (at least nowadays) not derived from the flyback but from the mains power supply because it needs relatively high current at low voltage. May be that early vacuum tube TVs without mains trafo (i.e. deriving all medium high voltages directly from rectified mains voltage connected to one chassis pole - very dangerous) used the flyback for heaters, but I don't think so. Vacuum tube TVs also often wired all heaters in series (like xmas faery lights) to use a higher voltage at lower current and thus cheaper trafo (or pure mains voltage through only a resistor). In the age of switching power supplies (all modern PC monitors) chassis don't use rectified mains voltage anymore.
Recently I repaired two portable TVs (B&W but transistor) and they both had heater powered from the flyback transformer. Then again, both TVs could run on DC power so the heater was powered from the flyback transformer as a cost saving measure (no need for a separate converter).
However, looking at a circuit diagram of my FW900 it does appear (I didn't analyze the circuit thoroughly) that the flyback transformer and the horizontal deflection coil are driven by separate circuits. It must be very difficult to build a flyback transformer that operates efficiently at both 30kHz and 120kHz (and all frequencies in between).
I'm getting ECS syntax errors in windas. I'm using a USB to serial converter that goes to an RS232 to TTL board. I'm fairly certain I've hooked the pinouts to the right place judging by the service manual and silk screen on the D board. When I look at the data file it's mostly 0s and skips with a few values here and there. I may have shorted some pins together at some point but it's difficult to be sure. Has anyone else run into this issue before? The model selected is the Dell P991 F99, I'm running windows 7-64, and windas is being run in compatibility mode for Windows 98 and as administrator. I've set the serial port to COM1 (when it was anything else it wouldn't work, the setting kept reverting to COM1).
Please let me know if anyone sees anything wrong with my setup. If not I'll just order an adapter that actually has female pin leads already on it...
USB cable:
http://www.amazon.com/gp/product/B0007T27H8/ref=oh_aui_detailpage_o00_s00?ie=UTF8&psc=1
voltage adapter:
http://www.amazon.com/gp/product/B005D5T292/ref=oh_aui_detailpage_o01_s00?ie=UTF8&psc=1
P991 pin header:
http://i121.photobucket.com/albums/o236/willis936/IMAG0233_zpsyd6aldsk.jpg~original
USB/voltage adapter:
http://i121.photobucket.com/albums/o236/willis936/IMAG0235_zpslg9gvbtb.jpg~original
voltage adapter pin out:
http://i121.photobucket.com/albums/o236/willis936/IMAG0236_zpsthnqcfdr.jpg~original
Although I am not familiar with your adapter (I used a homemade transistor circuit on a real RS232 port), it is extremely important not to use the supply voltage from the monitor data connector. Although it may work in an ideal world, the monitor tends to switch it off in various situations (timeout, bad firmware flash etc.) and so can make things much more difficult. Thus only use power from the USB port or a battery, but NOT the data port of the monitor.
In my experience, USB to serial converters often fail, with two different ways to manifest:
1-poor design won't work at all with a particular motherboard
2-serial device recognizes the conversion process, and refuses to work
Serial ports on PCI or PCIe serial port or multipurpose cards just work, like motherboard serial ports, unlike USB to serial converters.
I have an old WinXP SP1 PC someone gave me that I use for nothing else than WinDAS, which I originally set up precisely because of possession of several washed out P991s.
that alone sounds like i would need to look up some more how to videos. I'm going to be uploading the firmware and a video of me loading it up in windas in hopes someone one out there smarter than i can make sense of it all.
HEY!!! How come when using resistors why don't we just use a big ass POT so it ca be adjusted later on?