For contact info, please see the Sci.Electronics.Repair FAQ Email Links Page.
Copyright © 1994-2004
All Rights Reserved
Reproduction of this document in whole or in part is permitted if both of the following conditions are satisfied:
1.This notice is included in its entirety at the beginning.
2.There is no charge except to cover the costs of copying.
The purpose of the FF Mon FAQ is to provide unbiased information to aid in making an informed decision when considering the acquisition of a fixed frequency monitor for use on a PC.
Contributions of a non-commercial nature to enhance this document are welcome.
Don't arbitrarily attach the monitor cable to a socket just because they physically mate or have the same number of pins!
The designers of fixed frequency monitors originally designed for specific workstation applications may very likely have taken even less care in dealing with unexpected scan rates as their environment is closed and such treatment isn't expected. Spectacular and expensive failure is quite possible.
It can be very risky to experiment with scan rate settings either because you have no idea of what is valid or to 'explore the envelope'. Either of these could result in expensive repairs. The only hint you may get just before the smoke comes out are unusually poor geometry or instability but it would be foolish to depend on these possible warnings.
We will not be responsible for any damage to your monitor or ego that may result from such experiments!
Questions like this come up all the time on comp.sys.ibm.pc.hardware.video.
These are either fixed frequency monitors or are incompatible with the common VGA/SVGA 'standards' in some other ways.
This document provides unbiased information on the problems and possible solutions to dealing with these monitors on PCs (where the term 'PC' mostly refers to DOS/Windows based x86 and Pentium boxes but may include Macs and Linux based systems as well).
Note that some/many/most newer workstation monitors may indeed NOT be fixed Therefore, before spending a lot of time, effort, and money to adapt a fixed frequency monitor that isn't, attempt to determine its specifications from the manufacturer - all you may need is a cable! That workstation monitor may be identical internally to a one sold for PCs but with a different model number and video caonnector. Why is this now the case? Simple: Because it is cheaper for a monitor manufacturer to produce a single standard model rather than a one for the large PC market and another for the much smaller workstation market - even if it is somewhat simpler with a lower parts cost.
Monitors like the IBM 9517 are not fixed frequency but are XGA compatible. This was an IBM abortion and not compatible with VGA/SVGA even for booting your PC.
These types of monitors are generally manufactured by the best names in the industry such as Hitachi, Mitsubishi, Philips, Sony, etc. - and are thus often of very high quality. The specifications of these monitors may exceed those of any but the very top-of-the-line monitors used on PCs. The origianl cost of these monitors was probably much higher than an equivalently sized PC monitor as well. They become available as high performance workstations (whose technology advances nearly as quickly as that of PCs) are decommissioned or upgraded. The cost to you now is usually very low since they just take up space and you know how bean counters at big companies like to have all their beans lined up in a nice neat row :-)
Some fixed freqeuncy monitors may be from Apple Macintosh computers as well.
It would be nice if all you needed was a cable to use one of these beauties on a PC. Unfortunately, there is often much more involved in making these freebie monitors conveniently usable on a PC under DOS, Windows, or Win95.
Some of these comments - a la scan rates - may not apply to PCs running oprating systems like Linux where scan rate switching isn't required (at least after boot).
And, just because a monitor has BNC connectors (or does not have a VGA/SVGA connector or cable) does not necessarily mean that is is a fixed frequency monitor and therefore a problem. Many top quality monitors only have BNC connectors and might be fully compatible with most video cards running PC/DOS/Winddows. The only way to be sure is to obtain the detailed video, sync, and scan rate specifications.
Are you totally confused yet? :-) The purpose of the FF Monitor FAQ is to try to sort through all the confusing compatibility issues and alternative solutions to enable you to use that fabulous monitor.
The following are the principle difficulties in using a fixed frequency monitor on a PC:
Note: for the remainder of this document, I use the term 'Windows' to refer to MS Windows 3.1, WFWG 3.11, and Win95/98 interchangeably.
If all you run is Windows - never any DOS games or other applications that require you to suspend to a full screen DOS shell or run in native DOS mode - then you can always use a second VGA monitor for booting and then just switch over to the high resolution monitor once Windows comes up. Or, just assume your system **will** come up and forgo a display until the Windows desktop appears.
Unfortunately, there are a lot of DOS applications still used so this not a solution for everyone.
However, if you mostly use your PC for Autocad or Lotus, then this is a perfectly reasonable option if you have a suitable video card. (However, read on).
For monitors with BNC connectors, it may be possible to determine capabilities by counting them:
Monitors with a 13W3 connector will generally accept composite sync though the other options may be possibilities as well.
Some video cards (like the ATI GUP, GPT, and others) can be programmed in their SETUP or INSTALL program (or possibly from a command line option) to generate composite sync on the H or V sync wire. These will then work (at least with respect to sync) with a monitor requiring either separate or composite syncs. A few high-end cards can generate sync-on-green as well.
Sync polarity (whether the pulses are negative or positive going) may be an issue depending on the design of the monitor. However, most suitable video cards can be programmed for either polarity.
Therefore, depending on your video card, the sync issue may be a non-issue.
Otherwise, an adapter will be needed. Unfortunately, this is not just a cable as circuitry is required to combine the signals. If you are electronically handy, it is a simple matter to construct a suitable circuit but if you are not, this may be a show-stopper unless you can locate a commercial product. Note that the term 'electronically handy' means a bit more than knowing how to read the resistor color code. The circuits are very simple. However, for the adapter to work well at the very high video bandwidths of the typical (1280x1024) display, you must use the proper 75 ohm coax and connectors, and assemble the circuitry itself in a shielded metal box, if possible. Otherwise, there could be degredation of the displayed video - ghosting, ringing, and less than optimal image sharpness. If you will be making a cable from scratch, there will be some precision (very tiny pins) crimping or soldering needed to construct the VGA and/or 13W3 (typically) connectors as well.
(From: Royce Liao (liaor@uci.edu).)
ALL modern video adapters (PCI/AGP) incorporate programmable dot clock generators. So *any* decent video card should be able to output the necessary scan-rate video signal. Unfortunately, unless you have access to a special refresh-rate utility, and you *know* the precise frequency, this info still won't help your situation.
Scitech Display Doctor has a Windows based display utility which works with majority of contemporary video cards. The control center lets you tweak video refresh rates down to 0.1Hz, and adjust H/V sync polarity.
Some models appear to be quite competitive in terms of graphics performance (Windows accelerated, etc.) so these may represent an attractive alternative even for high performance applications like Autocad.
After specifying the monitor type and/or scan rate parameters for your monitor, the behavior of the card should be essentially transparent to your software. That is, programs think they are talking to a VGA/SVGA card but the output of the card drives your fixed frequency workstation monitor properly at all times - including booting, DOS games, Windows, etc.
Cables and adapters in addition to a PCI card for using a fixed frequency monitors on a PC.
(From: Malik (psxpic@malik.eng.net).)
Against all odds i thought what the hell and tried it anyway... And hey, presto! it works perfectly, no problems.
I had to feed a vertical sync into the monitor which would not have been possible because the feed for this was not present on the plug. However it was possible to remove one of the unwanted wires in the RGB lead and reconnect it to the unused Vsync input.
It appears Sony just fit the appropriate lead/connectors to the monitor depending on its purpose then badge it for Sun, etc.
The model number is GDM-17E20 its absolutely a superb monitor... If you can get one I suggest you do.
I know there is a 17E10 and 17E11 that are older.... the situation maybe the same with these.
The FAQ has received news.answers approval, so it should be archived at rtfm.mit.edu and all mirrors, as well as in news.answers and comp.answers. There used to be a Web site with the FAQ but it appears to have vanished.
(From: Tony Chau (tonychau@netcore.ca).)
If you insist, the adjustment would be called something like horizontal oscillator, horizontal frequency, or horizontal hold. If you do tweak, mark everything beforehand just in case you need to get back to the original settings.
WARNING: Make sure you understand the issues involved in working inside a monitor! See the document: Safety Guidelines for High Voltage and/or Line Powered Equipment. Something that looks innocent can really ruin your whole day!
There is also some risk to the monitor - changing it too far may result in damage either immediately (the horizontal output transistor or power supply may blow) or increase component stress reducing reliability and shortening its life. There is no way to know without looking at the design.
The IBM (Sony?) display 6091-19 is a 5-BNC type, and is, indeed, a fixed frequency type. Therefore, you'll need certain special modes available in a display adapter to accommodate it, beside the correct cable, of course.
I'm pretty certain that IBM 6091 was actually made by Sony, and, in their better displays, they still use the 5-BNC (separate signal) video cable, i.e., on their current SE series, etc. I'm certain you could buy a Sony (or other) replacement cable that would do the job, tying your display to a PC-type 15-pin VGA/SVGA "D" video connector, once you know you have a video card that will support it.
(From: Justin Thiessen (thiessen@physics.arizona.edu).)
I *know* the officially sanctioned video modes for my version (the 6091-19 (_not_ the 6091-19i)) limit the refresh rate to 67 Hz, but the one I have happily syncs to quite a few non-standard modes:
# # Following modelines are for linux and work under XFree86 3.3.6 # # H - 69.84 kHz; V - 61.37 Hz; Pixel Clock 138 Mhz # Generated by Colas XFree Modeline Generator Modeline "1464x1098" 138 1464 1524 1844 1976 1098 1100 1112 1138 -hsync -vsync # H - 71.15 kHz; V - 79.58 Hz; Pixel Clock 108 Mhz # Borrowed from somewhere Modeline "1152x864" 108.00 1152 1156 1300 1518 864 865 868 894 -hsync -vsync # H - 72.76 kHz; V - 59.84 Hz; Pixel Clock 78 Mhz # Thrashed out on my own - small black space at sides of screen Modeline "768x576" 78.00 768 792 888 1072 576 579 585 608 -hsync -vsync DoubleScan # H - 66.67 kHz; V - 66.14 Hz; Pixel Clock 64 Mhz # Thrashed out on my own - almost fills screen Modeline "640x480" 64.00 640 728 856 960 480 482 484 504 -hsync -vsync DoubleScanNote the use of the DoubleScan parameter on the last couple of modes, which just (as I understand) draws each scan line twice, allowing the use of (much) lower apparent modes than the 6091-19 is technically capable of displaying. Also, if I'm clear on what's happening here, then the "real" pixel clocks for the last two modes are twice the number listed in the modeline. (Each line is drawn twice, therefore there are twice as many pixels actually being plotted as the modeline indicates, yielding double the clock.) Please correct me if I'm wrong on this. This indicates, however that the monitor is capable of handling signals with bandwidth up to 156 Mhz, quite a bit higher than its official rating.
And, truthfully, as much as I relish the real estate of the higher resolution modes, the actual dpi of the monitor combines with the relatively low refresh rates to make them a bit harder on the eyes than I'd prefer. The 1152x864 mode is (by my guess) probably the one best-suited to the actual physical limits of the monitor. If I manage to get 2nd monitor hooked up and running under X I'll probably drop both of them down to this relatively eye-friendly mode.
I still don't get a boot-up screen, nor do I ever expect to without some hardware hacking. The low-end modes are just fun for trying out games. Since my video card (ATI Xpert98) has an apparent text-mode dot clock limit of 60 Mhz (according to the SVGATextMode documentation), I haven't been able to get a text screen yet, although exploring the DoubleScan parameter in that context may eventually yield results.
Now I take no responsibility for just how strangely I have all the pots inside the monitor tweaked. I never use anything that doesn't function in "Mode 3", and if I do happen to flip the switch to Mode 1/Mode 2, any signal that actually resolves is, at best, displaced and suffers severe fish-eye effects, but in mode 3, my monitor looks pretty good for being 10 years old. I also have no idea how much I'm shortening the life of the monitor by running it at these scan rates, nor do I have any clue as to just how far over actual spec I'm pushing it. (Who knows what the internal Sony/IBM engineering specs were for this beast?) I'd give a lot for a manual, but such seem to be scarce for most of the old workstation monitors available.
As a side note, a great deal of info is available on the 6091-19 on a few web sites:
Well, I can't stop tinkering.
I still have a IBM 6091-19, though now it's the 2nd head of a dual-head setup, and in an effort to match up the capabilities of my two monitors, (gdm20e20 rebadged by sun + the 6091-19) I've scraped out a few more modelines to fit the VESA-standard resolutions.
Here are some more:
# V - 64 Hz ModeLine "ibm1400x1050" 133.38 1400 1480 1800 1912 1050 1052 1064 1090 -hsync -vsync # V - 71.21 Hz ModeLine "ibm1280x960" 122 1280 1324 1716 1720 960 963 974 996 -hsync -vsync # V - 79.58 Hz ModeLine "ibm1152x864" 108.00 1152 1172 1316 1518 864 865 868 894 -hsync -vsync # V - 86.26 Hz ModeLine "ibm1024x768" 103 1024 1124 1364 1500 768 770 782 796 -hsync -vsync # V - 59.90 Hz ModeLine "ibm800x600" 81 800 804 908 1080 600 604 622 626 -hsync -vsync DoubleScan
Including the modeline above for 640x480, these cover the VESA standard resolutions (though not at the suggested refresh rates) from 640x480 to 1152x864. Also included are a nice 1280x960 mode (with a desirable 4:3 pixel ratio) and a large but very readable 1400x1050 mode.
Note that all these modes fill the screen, with the exception of 1024x768, which has some (quite small) black space on the left and right, and 800x600, which has a slight bit more black space on the left and right, but is still quite usable.
I am now using a Matrox Millenium II to drive the monitor, if it is of any concern.
(From: Helmut P. Einfalt (hp.einfalt@t-online.de).)
The IBM 6091-xx series can be driven by Cornerstone ImageAccel cards, at least by various ImageAccel 1 cards, including the MCA version MC1608C/11. And, above all, there are drivers for DOS, WfW, Win9x, NT3.51, *and* NT 4.0. A detailed account on how to install the MC1608C/11 and a 6091-19 (the procedure being the same for all other 6091s) can be found at
Cirrus BIOS Modifications to use the IBM 6091-19i Monitor
(From: u1061771156@csi.com)
I've gone through the more interesting task of adjusting the BIOS ROM on my video card so the boot screen and DOS work nicely as well. (It's a Cirrus 5429 based card). I still have the Vesa modes to fix, next weekend maybe. ;-)
The IBM 6091-19i has 4 modes of operation (1-4). The first three are based around 1024 lines of vertical resolution and are good for "desktop" use (I'm using the 1024x1024 mode under X as you can't get much more that 1024 Pels across with a 5429 since it's limited to 85MHz clock. As I currently only have 1M ram that fits neatly with 8bpp too.)
Mode 4 (the mode that was introduced new with the 6091-19i and was not supported in the old 6091-19) is more interesting for VGA use. It runs 63 kHz horizontal rate, and 120 Hz vertical rate, with a vertical total of 528. This is a similar geometry to the VGA modes, except about twice the rates. In practice I find the monitor will sync quite happily with vertical frequencies of 120 to 140 Hz and a bit beyond. (Incidentally this mode was originally intended for 3-D viewing with LCD shutter glasses synced to the vertical, and left/right images displayed on alternate frames by software.)
So the first step was to double the Pel clock (and hence all timings). After some serious disassembly, I found the first three PLL multiplier/divisor values for the Pel clocks are in a little table. The table contains the six values: 4A 2B 5B 2F 42 1F. For the particular BIOS version I have (CL-GD5429 Bios Version 1.00a) this is at BIOS address 0812, on an older BIOS, I found the same table with the same values at a slightly different address. If you do a search in any Cirrus 542X rom it's probably there somewhere. Paraphrasing a little (see the .pdf on the Cirrus Website) the Pel clock is generated by taking a 14.318 MHz crystal, and multiplying it by the first value and dividing by the second (ignoring the bottom bit) e.g. the first is 14.318*74/42 = 25.22 MHz and the second is 14.318*91/46=28.32 MHz. The bottom bit of the denominator byte is a "divide by 2" bit, if cleared the frequency is doubled. That's so convenient!! Hence step one is to clear these bottom bits, making the table into 4A 2A 5B 2E 42 1E and magically the monitor syncs up in all VGA modes!
(There are a few peculiarities to note when actually editing the EPROM. Firstly, like all Bios extension ROMs, there is a checksum at the end so the sum of all bytes in the rom is zero. Hence you have to add 3 to this to make the checksum correct. Secondly, the EPROM is a 27C512-150 which is 64Kx8 when the Bios is only 32K. There are two identical copies in it, only the later one is used as the A15 pin is wired high on the adapter card. Thirdly, and most annoyingly, the PC address line A0 is connected to eprom pin A14, and all the others shift down one. So Bios location 0812 (hex) is in the eprom at address 8409 when it's on the programmer. Why on earth did they do that?!)
If all you want is to be able to see your boot messages, this could be enough. However the image is a few columns too wide for the screen so you can't see both sides at the same time (the front panel control lets you move it side to side a bit.) So I've increased the frequencies a little more to make the Pels closer together. The table I'm currently running with contains 51 2A 62 2E 47 1E To maintain the correct scan rates and centering one has to increase the vertical total (CRTC register 0) as well as these frequency tweaks. This meant finding the VGA CRTC mode tables, a search for 2D 27 28 90 2B A0 finds the first one (mode 0 40x25 text). Each mode table (there are 1D of them) follows in steps of 40 hex bytes. Some experimentation showed that there are 3 different timings to change:
Modes starting 2D xx xx xx xx A0 I changed to 31 xx xx xx xx B1 Modes starting 2D xx xx xx xx 80 I changed to 32 xx xx xx xx B1 Modes starting 5F I changed to 66With these changes, all VGA modes produce a nicely centered image that is close to filling the screen (about 1cm black top and bottom).
There are a few mode tables that didn't start 2D or 5F which I have left alone for now, and affect Vesa modes, and there are other vesa tables elsewhere in the eprom that ought to be changed one day.
Note: I've been working on a Cirrus 5429 chip. Doing the same with a 5428 gave somewhat erratic results (shimmering characters) sometimes. I think this is due to the Sequencer(0x1F) register which sets the memory clock rate being initialized slower on the 5428 (1E versus 22 hex) so it's failing to get enough bytes out of ram in time to display. 22 is out of spec range for the 5428 but putting it up to that as an experiment seemed to fix it. ;-) I didn't find where that register is loaded in the EPROM though, but that fix is fine for a little DOS program to fiddle the modes. As I intend using the 5429 I haven't pursued it any further.
The only obvious side effect I'm left with is that the cursor flashes too fast. I can live with that! ;-) Maybe there's a register somewhere to fix that too?
Of course any noxious game that meddle with the registers direct may well screw up (but note that the Pel divisor registers are not part of regular VGA so may survive.)
I note that sync polarity doesn't affect the monitor at all. I guess its circuits accept either polarity and invert if necessary. Also, sync rates out of range don't appear to damage the monitor, it just loses sync quietly and the picture goes haywire. ;-)
I have the original "User Guide" and "Maintenance Guide" books which makes life a lot easier.
P.S. I see that the IBM 9517 can do 1024x768 if you can get to 85Hz vertical interlaced, maybe starting with a 1240x768 interlaced mode and fiddling the frequency up in a similar way that I did would work for the original question? Of course interlaced modes are yucky unless you're viewing images...
This monitor is not VGA/SVGA but XGA2, as used in IBM PCs with the short-lived MCA bus.
There are said to be ISA (PCI?) cards that drive them, but a friend (ironically, ex-IBM, but not the 'toy computer' end) who bought a stack of the monitors cheap never found any.
It is near-enough to impossible to convert 'properly' and can't be made to run at the 31.5 start-up scan-rate without the HOT (Horizontal Output Transistor) dropping dead.
Note that he also sells both the cable as well as just the required GAL (CMOS programmable logic device) already programmed so providing this information is truly in the spirit of these FAQs and the "open source movement". Thanks. However, purchasing the completed unit may indeed be the best use of your resources.
(From: Berg Hawkins.)
I sell a sync-on-green cable for single frequency monitors. It is $75 and will run any of a half dozen monitors from most standard video cards on the PC and Macintosh G3/G4. The cable contains a chip, a CMOS GAL programmed with the appropriate equations. It fits in a plastic housing that goes over/around a 15 pin VGA connector which plugs into the video card. A source of +5 VDC from a Molex connector tied to one of the power supply outputs on the PC drive it.
I have been selling it for about 7 years and have done hundreds of monitors. The equations are free, so is the schematic. I will even mail the programmed chips to anyone who does not have a GAL programmer for a small fee. The best is to buy the whole cable since it takes me about 45 minutes to make one and involves some close-in soldering work.
I have done Sony GDM1950s, Hitachi 4119s and CM2086s, IBM 6091s and other monitors.
Important note: This setup WILL NOT allow DOS full screen or most video games or DVDs to be played, all other Windows applications will work with most any video card.
The equations are included in the .PDS file, they go right into the AMD PALASM compiler and have been tested. The output of the compiler is in the .JED file, this is what is fed to the PAL programmer. It has also been tested. Note that ONLY CMOS GALs such as 16V8s are to be used since the circuit needs to SOURCE a few milliamps into the GREEN color output. This is done by means of an approximately 330-360 Ohm resistor (1/4 Watt) connected between pin 12 of the GAL and the GREEN video signal. One weakness of this circuit is that power from a 4 pin Molex connector on the PC's power supply output needs to be applied to pin 20 of the GAL. Also note that power to the chip MUST be applied BEFORE a signal is input, otherwise the chip can latchup and be damaged. This is accomplished by connecting the power Molex connector while the PC is powered off, then plugging in the VGA connector at any time.
Any speed 16V8 can be used, down to -25 parts, speed is not critical.
A crude diagram showing the chip, the VGA connector and the BNCs is included in the file diagram.txt.
It is best if all unused GAL inputs are connected to ground. I can program and mail the 16V8 chip and the 330 Ohm resistor for $5. I sell the complete cable for $75 (recommended since it takes me about 45 minutes to make one - I have been making these for 7 years and have some experience in this. It involves some close-in soldering work.
I also sell 20" single-frequency monitors, refurbished and tested for $200 to $250 with 6 month warranty.
The chip can be used without the resistor if composite sync is needed (a small minority of monitors will not sync-on-green, in this case connect pin 12 directly to the horizontal sync input of the monitor. Some monitors need POSITIVE composite sync, in this case pin 13 can be output to the composite sync of the monitor. In either of the 2 above cases, 4 BNCs will be needed instead of 3.
I have also successfully used this circuit to connect to monitors under Linux. Macintosh PCs also work, including PowerMacs, G3s and G4s. Over 8 types of 17" and 20" single-frequency monitors have been successfully driven by this circuit.
Send me email if you have problems syncing up a monitor. My phone number is 1-310-839-6430 in Los Angeles.
VGA Connector Coaxes
from PC video card to Monitor
Pin
1 -------------------------------------------------------------> RED
2 -------------------------------------------------+-----------> GREEN
3 -------------------------------------------------|-----------> BLUE
4 NC |
5 ---------+ |
6 ---------|---------------------------------------|-----------> RED RETURN
7 ---------|---------------------------------------|-----------> GREEN RETURN
8 ---------|---------------------------------------|-----------> BLUE RETURN
9 NC | |
10 NC | / 330 Ohm to
11 NC | \ 360 Ohm
12 NC | / 1/4 Watt
| To +5V from PC >----+ |
| 4 pin Molex conn. | |
| +----U----+ | | Top of chip
13 ---------------------------------|1 20|-+ | has notch
14 ---------------------------------|2 19|- NC | indicated by U
To Pin10 (+sync) OR Pin20 (-sync) -|3 1 18|- NC |
To Pin10 (+sync) OR Pin20 (-sync) -|4 6 17|- NC |
15 NC | GND -|5 V 16|- NC |
| GND -|6 8 15|- NC |
| GND -|7 14|- NC |
| GND -|8 13|- | Use pin 13 for +Csync
| GND -|9 12|-----+ Use pin 12 for -Csync
+-----------------------|10 11|- GND
+---------+
;PALASM Design Description
TITLE HVB2G2.PDS
PATTERN A
REVISION 3.0
AUTHOR Berg Hawkins, 1-310-839-6430, Los Angeles.
COMPANY
DATE 8/21/96
CHIP VIDEO_SYNC PAL16V8
; POS/NEG SEPARATE INPUT(selected with POL).
; SYNC ON GREEN OUTPUT, connect to the GREEN video output
; (pin 2 on VGA connector) through a 330-360 Ohm 1/4 watt resistor.
; Pin 12 can also drive separate/composite sync input directly, i.e.
; Hitachi 4119
;
; Histor
; HVB2G0.pds: Initial design
; HVB2G1.pds: removed extra 3 outputs, added LED output to pull down LED
; HVB2G2.pds: Added separate sync polarity control for H/V sync this takes
care of NEW Cirrus Logic CLMODE (i.e., H- V+)
PIN 1 HSYNC COMBINATORIAL ; Horizontal sync input from VGA pin 13
PIN 2 VSYNC COMBINATORIAL ; Vertical sync input from VGA pin 14
PIN 3 POLH COMBINATORIAL ; Polarity control for horizontal sync
PIN 4 POLV COMBINATORIAL ; Polarity control for vertical sync
PIN 10 GND ; Connect to pin 5 of the VGA connector
PIN 12 SYNC_ON_GREEN1 COMBINATORIAL ; composite sync or sync-on-green out,
; OUTPUT can also drive a composite sync input on the
; monitor directly
PIN 13 LED COMBINATORIAL ; drives LED, active low when OK, can also
; be used to sync monitors that need POSITIVE composite
; sync input
PIN 20 VCC ; Connect +5 VDC from the PC's power output (red cable) INPUT
;------------------------- Boolean Equation Segment ------------------
EQUATIONS
SYNC_ON_GREEN1 = /HSYNC*/POLH*/VSYNC*/POLV + /HSYNC*/POLH* VSYNC* POLV +
HSYNC* POLH*/VSYNC*/POLV + HSYNC* POLH* VSYNC* POLV
LED = /SYNC_ON_GREEN1
;----------------------------------- Simulation Segment ------------
SIMULATION
TRACE_ON SYNC_ON_GREEN1 HSYNC VSYNC LED POLV POLH
SETF /HSYNC /VSYNC POLH POLV
CHECK /SYNC_ON_GREEN1 LED
SETF HSYNC /VSYNC POLH
CHECK /SYNC_ON_GREEN1 LED
SETF /HSYNC VSYNC=20
CHECK /SYNC_ON_GREEN1 LED
SETF HSYNC VSYNC POLH POLV
CHECK SYNC_ON_GREEN1 /LED
SETF /HSYNC /VSYNC /POLH /POLV
CHECK SYNC_ON_GREEN1 /LED
SETF HSYNC /VSYNC
CHECK /SYNC_ON_GREEN1 LED
SETF /HSYNC VSYNC
CHECK /SYNC_ON_GREEN1 LED
SETF HSYNC VSYNC
CHECK /SYNC_ON_GREEN1 LED
TRACE_OFF
PALASM4 PAL ASSEMBLER - MARKET RELEASE 1.5a (8-20-92) (C) - COPYRIGHT ADVANCED MICRO DEVICES INC., 1992 TITLE :HVB2G2.PDS AUTHOR : Berg Hawkins, 1-310-839-6430 PATTERN :A COMPANY: REVISION:3.0 DATE :8/21/96 PAL16V8 VIDEO_SYNC* QP20* QF2194* G0*F0* L0000 00000000000000000000000000000000* L0032 00000000000000000000000000000000* L0064 00000000000000000000000000000000* L0096 00000000000000000000000000000000* L0128 00000000000000000000000000000000* L0160 00000000000000000000000000000000* L0192 00000000000000000000000000000000* L0224 00000000000000000000000000000000* L0256 00000000000000000000000000000000* L0288 00000000000000000000000000000000* L0320 00000000000000000000000000000000* L0352 00000000000000000000000000000000* L0384 00000000000000000000000000000000* L0416 00000000000000000000000000000000* L0448 00000000000000000000000000000000* L0480 00000000000000000000000000000000* L0512 00000000000000000000000000000000* L0544 00000000000000000000000000000000* L0576 00000000000000000000000000000000* L0608 00000000000000000000000000000000* L0640 00000000000000000000000000000000* L0672 00000000000000000000000000000000* L0704 00000000000000000000000000000000* L0736 00000000000000000000000000000000* L0768 00000000000000000000000000000000* L0800 00000000000000000000000000000000* L0832 00000000000000000000000000000000* L0864 00000000000000000000000000000000* L0896 00000000000000000000000000000000* L0928 00000000000000000000000000000000* L0960 00000000000000000000000000000000* L0992 00000000000000000000000000000000* L1024 00000000000000000000000000000000* L1056 00000000000000000000000000000000* L1088 00000000000000000000000000000000* L1120 00000000000000000000000000000000* L1152 00000000000000000000000000000000* L1184 00000000000000000000000000000000* L1216 00000000000000000000000000000000* L1248 00000000000000000000000000000000* L1280 00000000000000000000000000000000* L1312 00000000000000000000000000000000* L1344 00000000000000000000000000000000* L1376 00000000000000000000000000000000* L1408 00000000000000000000000000000000* L1440 00000000000000000000000000000000* L1472 00000000000000000000000000000000* L1504 00000000000000000000000000000000* L1536 11111111111111111111111111101111* L1568 00000000000000000000000000000000* L1600 00000000000000000000000000000000* L1632 00000000000000000000000000000000* L1664 00000000000000000000000000000000* L1696 00000000000000000000000000000000* L1728 00000000000000000000000000000000* L1760 00000000000000000000000000000000* L1792 01010111011111111111111111111111* L1824 01101011011111111111111111111111* L1856 10010111101111111111111111111111* L1888 10101011101111111111111111111111* L1920 00000000000000000000000000000000* L1952 00000000000000000000000000000000* L1984 00000000000000000000000000000000* L2016 00000000000000000000000000000000* L2048 11111111000000000000000000000000* L2080 00000000000000000000000000000000* L2112 00000000111111001111111111111111* L2144 11111111111111111111111111111111* L2176 111111111111111110* C1C97* 0D34(Checksum may be incorrect - a character was included that doesn't display properly.)
(From: Steven Leinwand (steve_leinwand@hp.com).)
It is true that the most expensive solution (a special video card) is usually the best solution. I bought one a while back, and in the interest of 'truth in advertising' let me describe some of the drawbacks you will experience. In all fairness, I will mention that I haven't tried all the video boards out there, and am basing my comments entirely on my experience. Your mileage may vary.
Fixed frequency monitors like those commonly supplied with Sun workstations cannot change video modes like multiscan monitors. In order to get them to work in DOS modes, video card vendors like Mirage and Photon modify the VGA bios on their cards to 'emulate' *some* DOS modes at a fixed scan frequency. This works well in *some* modes, works poorly in others, and doesn't work at all in some.
These BIOS mods interfered with motherboard timing on two VLB motherboards. I tested it with four video cards, the problem was the VGA BIOS). Video wouldn't sync after warm-booting. I had to shut the machine off, and wait about 10 minutes. In all fairness, I have not heard of similar problems on ISA and/or PCI cards, so that problem may be on that vendor's VLB boards only.
It usually works best in Windows, where the display is always in the same graphics mode, and mode switching isn't an issue. It works for *some* DOS programs, depending on what video mode they try to put the display in. It works poorest on games, which seem to insist on using weird and/or undocumented VGA video modes. At least half my games either wouldn't work, or their image was so small, as to negate any benefit of having a large monitor.
** Modes less than the monitor resolution will usually be displayed at 1/2 the screen size of the monitor **
This was a surprise to me, and was never mentioned in any of the ads for video boards. I've been told this is a fact of physics, and cannot be overcome in fixed frequency monitors. (Editor's note: it can be overcome using a scan converter but this is much more complex than a BIOS change! See the "Notes on Video Conversion" document for further information on scan converters.)
I finally sold both monitor and video card, bit the bullet, and bought a 17" SVGA monitor and card. If I didn't have the VGA BIOS timing problems with the VLB version, I might have been able to re-use the video card, but was tired of the hassle. Since I got the monitor for free (obsolete product destined for the dumpster), it wasn't that bad. My advice is ask detailed questions before you buy, get a money-back guarantee, and test with all your applications. If you can't live with the results, exercise your guarantee.
Also check out support capabilities. One of those vendors I mentioned was very responsive and helpful. The other one started out awful, but made some progress in responsiveness over time.
I have a SONY GDM-1961 (a.k.a. VRT 19-HA) fixed frequency (sync-on-green) monitor. This monitor displays 1280x1024, but I have recently been able to tweak it to display 1024x768, 800x600, and 640x480!
So now I can play Quake full screen under NT and Linux :-)
The clue is to reduce the visible resolution and add the missing pixels to the front and back porch. The image of course doesn't fill the entire screen, but it's a *lot* better than having none at all.
I have made a page with my experiences with making a fixed-sync monitor work on linux, NT and win95 to help others with access to these extraordinarily cheap and large workstation monitors:
I had to learn what signals my IBM 6091 monitor wanted versus what signals my video card provided by experimenting with a scope and a pulse generator for a day or so. In my case, the monitor wanted inverted polarity on the horizontal sync line. I found that I could provide this by triggering a lab pulse generator from the video card's horizontal sync output, and using the resulting (inverted) pulse to sync the monitor. Fortunately, the timing jitter in the pulse generator was low enough that no horizontal jitter was noticeable on the display.
I used the monitor this (rather nerdy) way for a while, and then I noticed a small ad in Nuts and Volts magazine for a moderate cost ($200) translating video card made by a company called Ming. I ordered one, and it turned out to be a modified (new video BIOS chip, a few wire jumpers) Jaton 58P card, which uses the Tseng Labs ET6000 chip.
This has proven to be a good solution to my fixed frequency monitor problem. I ran comparative benchmarks using the Landmark PCPRO test program, and the speed of this card - when writing to video RAM - is similar to other high performance (Matrox Millenium 2, Diamond Stealth 3D 2000), cards I have tested recently. Ming makes cards for several different fixed frequency monitors, so they could be a viable choice for many of your readers. You can check their web page at http://www.riverside.quik.com/ming for more details. (From: Fabio Quenel (info@tesitalia.it).) I've managed to get this monitor to work in Windows98 at 1152x864, 76 Hz, in true color (24 bit) with a 4 MB Matrox Mistique card (170 Mhz RAMDAC). I had tried with the 1280X1024 resolution but the refresh was 67 Hz and it really was straining for my eyes (also the definition lacked a bit). In DOS mode the monitor doesn't work and I don't know if thare is a way (if you know please tell me).
(From: Sam).
Since the 6091's lowest horizontal scan rate is 63 kHz, you would need over a 120 Hz vertical scan rate for the 640x400 boot screen. Even if you could set up your video card for 120 Hz or more refresh, the 6091 almost certainly won't work at too much beyond its 67 Hz vertical scan rate specification.Here is what my mga.mon file of the Matrox seems like:
[User-Defined.Customized IBM 6091 Mode 3 new] 1280X1024 = NI, *User-Defined_Customized_IBM_6091_Mode_3_new_,(1280X1024) 1152X864 = NI, *User-Defined_Customized_IBM_6091_Mode_3_new_,(1152X864) 1024X768 = NI, *User-Defined_Customized_IBM_6091_Mode_3_new_,(1024X768) [*User-Defined_Customized_IBM_6091_Mode_3_new_,(1280X1024)] PIXEL_CLK = 119089 H_DISP = 1280 H_FPORCH = 16 H_SYNC = 160 H_BPORCH = 240 H_SYNC_POL = 0 V_DISP = 1024 V_FPORCH = 3 V_SYNC = 3 V_BPORCH = 18 V_SYNC_POL = 0 INTERLACE_ENABLE = 0 [*User-Defined_Customized_IBM_6091_Mode_3_new_,(1152X864)] PIXEL_CLK = 112064 H_DISP = 1152 H_FPORCH = 72 H_SYNC = 96 H_BPORCH = 288 H_SYNC_POL = 0 V_DISP = 864 V_FPORCH = 2 V_SYNC = 15 V_BPORCH = 36 V_SYNC_POL = 0 INTERLACE_ENABLE = 0 [*User-Defined_Customized_IBM_6091_Mode_3_new_,(1024X768)] PIXEL_CLK = 94252 H_DISP = 1024 H_FPORCH = 32 H_SYNC = 192 H_BPORCH = 128 H_SYNC_POL = 0 V_DISP = 768 V_FPORCH = 11 V_SYNC = 15 V_BPORCH = 33 V_SYNC_POL = 0 INTERLACE_ENABLE = 0
Well, I did this to two of them about a year ago. Of course I wrote down what I did, so I'll try to reconstruct for you what I did.
There are two modifications, one is the H-FREQ control, the other is the H SHIFT VGA control.
To boot, you need it to display the standard 31.5 kHz VGA text mode.
Turning the H FREQ pot (on the rear edge of the PCB) will eventually give you a display, but it is off-screen to the right. You then adjust the H SHIFT VGA pot (on the right of the PCB) and find that this pot has insufficient range to center the display. The H FREQ pot is 4.7K from the factory. It is in series with an 18K SMD resistor on the solder side of the PCB. I found that I had to turn the 4.7K pot to max resistance to move the display to the left, so I removed the 18K resistor and the 4.7K pot, and replaced them with a 22K resistor and a 10K pot.
I was then able to center the display for text mode. The only problem is that the horizontal linearity in text mode is now off. The letters near the right of the screen are narrower than elsewhere. The problem is not serious (a casual observer does not see it) so I can live with that. Moreover, I never use text mode anyway (except during booting and for BIOS setup, etc.) The 320*200 mode (games) also works correctly after this modification.
Now, we want it to sync to highter refresh rates, specifically, 1024*768. That is the mode I adjusted it for. All the 1024*768 parameters are separately adjustable from the ones for the other modes, except the H FREQ. The pots that govern this mode are all labeled "4,5", meaning, I presume, that they govern the operation of what the monitor manufacturer calls mode 4 and mode 5. I found these to coincide with H frequencies of over approximately 50 kHz.
I found that by tweaking the H freq, I could get the text mode to be all right, *or* the 1024 mode, but not both at the same time. (BTW, I run 1024 with a pixel clock of 77 MHz, H freq. of 57.63, and V freq. of 68.13 Hz. I don't know what the specs for this monitor are, so I thought it was prudent to stay on the low side.)
So I needed two instead of one H FREQ pots. I also found that there is a relay click if you throw something over 50 kHz at this monitor. Hence, I tapped the relay coil to give me the 0/12 volt signal to tell me what mode the monitor is displaying. This signal I take to a little PCB that holds my two H FREQ pots, an inverter, and two NPN transistors that switch in one or the other pot instead of the original one.
The schematic is shown in ASCII below. Or, see the Gif version provided by: Puiu Chiselita (puiu@microinfo.logicnet.ro).
TO H-FREQ
o
|
+12 VDC +------+-----+
o | |
| / /
| \ 680 680 \
| / /
100K 1N4148 |/ E | |
RELAY o---+--/\/\--|<|--|<|--| PNP +--+ +--+
| 1N4148 |\ C | | | |
| | / | | /
| / 2.5K \<-+ +->\ 2.5K
| 100K \ / /
| / | |
| | |/ C C \| 1N4148 100K
| +----| NPN NPN |---+--|<|--/\/\--+
| | |\ E E /| | |
| / | | / |
| 100K \ | | \ 100K |
| / | | / |
| | | | | |
| +------+------+-----+-----+ |
| _|_ Gnd |
| - |
+------------------------------------------------------------+
Transistors can be any general purpose type:
Now I can tweak the H FREQ pot for both modes separately. To sum up, my 9517 now displays VGA text modes, 320*200, and 1024*768 in approx 70 Hz. When it is warm, I can also get it to display 640*480. I could probably do a fix for that too, but haven't bothered since I don't need that mode. Moreover, I don'thave schematics, so I'm a bit hesitant to continue modifying it, when it already works adequately for my use right now. The H linearity in the 320 and 1024 modes, by the way, exhibits no problems.
This solved monitor number one, and oddly enough, when I did monitor number two. I was able to get it working after installing only mod number one. A little help from the VGA card (Diamond Speedstar if I remember correctly) did the rest. So maybe you are in luck and need only one mod. A flexible VGA card definitely helps. (Or good setup-software for the VGA card.)
When you are turning all the little presets, be careful not to short the wipers to the chassis... use a plastic tool! The wipers are not insulated from the screwdriver slots. You can easily destroy some of the SMD transistors by doing that. That's what I did, and I had some VERY lucky guesswork in unsoldering them and replacing them by normal transistors (that is what I tend to do when I find a dead SMD transistor...)
These are beautiful monitors and well worth modifying. Fortunately, not a lot of people are able to do this, and this means that they are available rather cheaply. :)
(From: Ken Jones (k.jones@coastlight.com).)
Disclaimer: I will not be responsible for any damage to your monitor or ego that might be caused by the use of this circuit.
I don't know if this works with all VGA Cards. I've tested this with my Matrox Millenium and Mystique VGA Card, and it worked well. But there is a tricky part!!! For all those, who are trying this circuit with their Matrox Card, be sure to accomplish the following steps:
Now, your Monitor could work. If not, (as it happened to me...) do the following:
(From: Ross (rross@hotkey.net.au).)
I've had one of these working on a PC. From memory, the refresh rate is 1280x1024 at 60 Hz (therefore pretty crappy). You can either use a fixed frequency video card (expensive) or as I did, a Matrox Millenium which allows custom refresh rates in 1 Hz increments. Simply set up the computer on another monitor, and plug in the GDM-1950. The cable should not be a problem, easy to get here in Australia, so no problems in the states.
(From: Bill (bill_h@azstarnet.com).)
I have one - seems to me it was around 72 kHz/75 Hz refresh, but I'd have to check to be sure. I use a Quantum128 (costs around $150) but comes with a $30 adapter (Griffin Technologies) that makes the sync_on_green (this would be for 3 BNC type monitors) AND Composite sync (for the 4 BNC type monitors). You don't need this adapter if you're planning to use only the GDM-1950. If you buy the board, I'd be sure to get this adapter anyway. It comes in handy for checking the 'other' types of fixed frequency monitors.
The cable you need is pretty simple - just 'normal' VGA to the 5 BNC's.
This same cable is/was used on some sort of MAC's, and I found one in the close-out pile of a retailer dropping MAC's (for ten bucks).
(From: Joe McCarthy (mccarthy@si87.com).)
The GDM-1950 was sold by Sony to many different OEMs. The standard GDM-1950 is a 64 kHz, 60 Hz monitor designed to run 1280x1024 resolution. Radius bought a lot of these monitors and tweaked them to the MAC frequency of 63 kHz, 75 Hz, at 1024x768.
The most important thing is to get the sync pin(s) connected and the horizontal scan rate as close to the required value.
There is much more info and links at:
(From Paul Langemeijer).If you have a GDM-1960, you can make an even easier modification to the monitor. A thing I might add to the problem of the sync-on-green, is the 'modifying' of your monitor. Well actually it isn't that hard. If you remove some of the panels (the one near the connectors with the monitor turned off!), I found holes for two more BNC connectors. You can put them in, attach a bridge and put in two resistors (the holes for these are obvious, the numbers are above them). And you've got a 5-BNC monitor (and it works). You should use 75 Ohm resistors and the monitor syncs perfectly at 1024*768 at 85 Hz (Standard VESA videomode).
(From: Pascal Specht (specht@roguewave.fr).)
I'm now running the GDM-1960 from my DECstation 5000 on my PC.
I found an even easier way (if you can live without the three VIDEO OUT BNC's). Simply reuse two of them for H and V-sync - No resistors, no holes to bore. This applies to my Digital GDM-1960 D3. Simply follow up the cables going from these 'video out' BNC's and take them off at the other end. Put these ends now into the obvious holes for the two predesigned places intended to take in the missing BNC connectors. Add the bridge to make the HSync connected. That's it. It perfectly works on my nVidia RIVA 128 with 1024x768 at 85 Hz, but also 1280x1024 at 60 Hz.
I just wanted to share some info about a Sun monitor I recently acquired. It's one of the new 20" multisync monitors you get together with Ultras with part number 365-1335 or GDM-20E20 made by Sony. Searching the net told me it should be possible to connect it to my PC but gave no info on how. So, armed with a screwdriver, an soldering iron and a lot of curiosity a started to investigate my new monitor, and after some iterations I got the following "new" pinout of the 13W3 connector.
Analog: 13W3 connector:
+-----------------
| +-------------
| | +---------
| | | +-----
| | | | +- horizontal sync (new multisync only)
| | | | |
| | | | | grey
red | | | | | green blue
| 1o 2o 3o 4o 5o | |
(O) (O) (O)
6o 7o 8o 9o 10o
| | | | |
| | | | +--- sync common (gnd)
| | | +-------
| | +-----------
| +--------------- vertical sync (new multisync only)
+-------------------
The monitor works with a PC driver from Sony named "SONY Multiscan 20SE" and
is able to do at least 1280x1024 @ 75Hz, that's at least what my video card is
capable of.
Looking inside it I also found markings for one of the PC-monitor buses used for autoconfig but I have not bothered to get that to work. Any info on that would be appreciated but it does work nicely without it. Converting Sun Workstation Monitors for use on MACs (From: Phil Lee (plee@orbis.net).)
Here is a cheap, less than $300 total, solution for converting legacy Sun 20" workstation monitors for Macintosh use. Over the summer, there were large batches of Sun OEM Sony monitors, model GDM-20D10 being auctioned by ubid.com for about $225. They were being bundled with a Mirage Diablo 3D AGP, mirage-mmc.com, board and the package was advertised as compatible with PCs only.
The Sony 20D10 is a "limited multi-sync" monitor which is still satisfactory for the standard Mac two page display resolution 1152 x 870 @ 75 Hz (horiz=68.7KHz) for desktop publishing. Two cable related add-ons was required to convert it for Mac use.
Not bad for repurposing legacy equipment.
There were other Sun (Sony OEM GDM-1962b) monitors also being auctioned. Other WS mfgrs such as HP, DEC, and IBM have similar monitors. They may work as well. I chose the 20D10 as they seemed to be newer.
Also see the section: Sun 365-1335 or Sony GDM-20E20 monitor on PC and the following for more technical info:
Here is a cheap, less than $300 total, solution for converting legacy Sun 20" workstation monitors for Macintosh use. Over the summer, there were large batches of Sun OEM Sony monitors, model GDM-20D10 being auctioned by ubid.com for about $225. They were being bundled with a Mirage Diablo 3D AGP, mirage-mmc.com, board and the package was advertised as compatible with PCs only.
The Sony 20D10 is a "limited multi-sync" monitor which is still satisfactory for the standard Mac two page display resolution 1152 x 870 @ 75 Hz (horiz=68.7KHz) for desktop publishing. Two cable related add-ons was required to convert it for Mac use.
Not bad for repurposing legacy equipment.
There were other Sun (Sony OEM GDM-1962b) monitors also being auctioned. Other WS mfgrs such as HP, DEC, and IBM have similar monitors. They may work as well. I chose the 20D10 as they seemed to be newer.
Also see the section: Sun 365-1335 or Sony GDM-20E20 monitor on PC and the following for more technical info:
The SGI (Mitsubishi) HL7965KW-SG monitor is widely advertised as a
Fixed-Sync, Sync on Green monitor in the workstation world. Well, most
people know that this is actually a multisync monitor but not many people
may know that it actually has
I don't know if they appear on the video input board, but if you remove the
board where the 13W3 connectors are on you can actually see, beautifully
silk-screened on the main PCB, near the connectors that once went to the
input board, pins labeled:
R G B and HSYNC, VSYNC!
So this is actually NOT a sync-on-green monitor as everyone says. It's a
separate sync monitor! (Yes I have this and it works beautifully in Windows
up to 1600x1200x60hz. At 1280x1024 it does 75Hz.
However, the pincushioning will need adjustment (I don't know how) but if
you are willing to live with a little distortion at the sides it's perfectly
OK. You'll also need to solder in some 75 ohm resistors where the RGB goes
in (conveniently, there are SMT pads there!) otherwise you'll get extremely
severe ghosting. You might also want to replace a new 13W3 plug to cover up
the rather unsightly holes left from removing the input board, although
these can get *very* expensive and *almost impossible* to get.
And this is a monster of a monitor, weighing more than a 21" Sony CPD-G500!
There's actually a warning label at the back saying "Caution: 23kg" with a
picture of those "10 ton weights" that you see so much in cartoons :)
All that said, the video amps look like miniature tanks, they're fully
encased in shiny aluminum blocks and there are a ton of ferrite blocks and
shielding screens inside.
Although mine was manufactured in 1993, it still gives a better picture than
a slightly-less-than-new Philips monitor.
-- end V1.68a --