Nvidias Win7 drivers screwed my monitor

[Oakey]

Okay, here's the situation. I have a Gigabyte 7800GTX connected via DVI to a 24" Dell 2405FPW TFT.

Yesterday I installed Win7 and everything went smoothly. After install Win7 detected the 7800GTX and set the res to the monitors native 1920x1200. This remained for a few hours until eventually I made the mistake of downloading the latest Win7 Nvidia drivers and installing them.

After install my Dell was detected as a 'Generic NON PnP Monitor' and the maximum resolution I could select is 1024x768. I've tried rolling back the Nvidia drivers, installed the original MS ones, uninstalled the Nvidia software, tried new monitor drivers, different DVI cables, etc, etc. NOTHING resolves this issue.

On top of that, whenever I now boot I no longer see my BIOS / POST screen unless I plug a different monitor in. For example, if I plug my spare Samsung in (via DVI) the POST screen appears on there, Winlogon will then appear on the Dell. No matter what I do I can't get the POST screen to show on my Dell via DVI. On top of this, Win7 will happily detect the Samsung and set it to its native res.

So I Googled a bit and saw some other people with similar issues. I got a DVI > VGA Adapter and plugged that in my 7800 and into the Dell's VGA port. Windows correctly detects this as a 2405fpw and sets the native res as 1920x1200. If I unplug the VGA cable and plug the DVI one into it, Win7 detects it as 'Display Device on DVI' (no mention of Dell 2405fpw). I can also now select 1920x1200. Unfortunately, if I reboot it goes back to 1024x768 and won't go any higher which means I would have to constantly boot up in VGA, then switch the cables over. This, is not acceptable.

To add further insult to injury, the Dell's DVI port now refuses to work on anything other than my Win7 machine. I plugged it into a WinXP machine and it just goes into sleep. The Dells VGA port works fine on the XP maxchine, but absolutely nothing at all shows up over the DVI connection. The monitor just says it's in 'sleep mode' and to move the mouse or press a key (which does nothing). Plug it back into the Win7 machine and it works (albeit at 1024x768)

I am furious, everything was fine until I installed those drivers, now my monitor has gone completely tits. Oh and I did attempt a fresh install of Windows again but now when I try to do that the setup screen displays at something ridiculous like 640 in 8bit colours. Wonderful.

I can't even begin to understand how this can happen, it's like the Dell's DVI port has lost all PnP capability.

[shaithis]

Try holding the MENU and + buttons while powering up the screen...then select factory reset.

[Oakey]

Try holding the MENU and + buttons while powering up the screen...then select factory reset.

Tried that, no joy. Thanks though.

[Singh400]

Where did you download the drivers from Windows Update or nVidia.com?

[Syllopsium]

Could be corrupt EDID data (timing and resolution support data held internally in the monitor). Really the monitors should protect their data but sometimes they don't.

I wrote this a couple of years ago, and some of it is probably mildly out of date. In particular it's unlikely you'll have to use the beta powerstrip anymore - it'll be part of the full version. I plugged my VP730 into an ATI FireGL 8800 card and it stopped working(!).

'The DVI connection had failed the moment I plugged a new graphics card in - duff BIOS, voltage spike - who knows? In any case a search on the web revealed this was likely to be an EDID issue.

E-EDID, the Enhanced Extended Display Identification Data specification from VESA (PDF) formalises the reporting of optimal usage parameters for a display. These are sent/requested over the DDC pins in HD15(VGA)/DVI connections.

DVI requires EDID as part of the specification (VGA does not, unless it's manufactured after 1996) and if the EDID data is missing/invalid the connection will not work.

Unfortunately, in many cases - particularly on TFT monitors, there appears to be no method of write protecting the EDID. Thus, if a duff graphics driver, a broken BIOS or a power surge happens it's possible to corrupt the EDID.

Viewsonic produce a utility to query, edit and report the EDID data but it only works if the EDID is valid.

Fortunately Entech Taiwan produce a number of utilities, including Powerstrip. Monitor Asset Manager is free, and will enumerate all working EDID connections correctly.

Or, to be accurate, they will enumerate all *operational* connections. This means the broken EDID device has to be started and enabled using a different connection (HD15/VGA). Once powerstrip recognises both connections, the DVI->HD15 connector is hotswapped for a DVI->DVI connector and the data can be read..

What is actually required is Soft MCCS. Powerstrip will retrieve the corrupt EDID data, but it won't be displayed unless it's valid. Soft MCCS won't either, but it *will* report the data in its log - see :

0047909E75.....DeviceID = [..snippy..]
0047909E76.....DeviceKey = [..snippy..]
00478FA451......Raw data received = 0xDB
00478FA452......00: 00FFFFFF FFFFFF00 5A631BE5 00000000
00478FA453......10: 230F0103 80221B78 2EC5C6A3 574A9C23
00478FA454......20: 124F54BF EF808180 8140714F 310A0101
00478FA455......30: 01010101 0101302A 00985100 2A403070
00478FA456......40: 1300520E 1100001E 000000FF 00505142
00478FA457......50: 30353335 30333537 300A0000 00FD0032
00478FA458......60: 4B1E520E 000A2020 20202020 000000FC
00478FA459......70: 7F7F7F7F 7F7F7F7F 7F7F7F7F 7F7F7F7F

It looks mostly ok. (The above hex may actually be a bit wrong, but is indicative of what you will see)

So what I did next was to manually type in and verify (make sure all the parameters are reasonable) all the bytes into the Viewsonic EDID editor, revealing the checksum to be corrupt (it should be 0x40 instead of 0xEB). The revised EDID report can then be saved to disk.

Next, taking the beta version of the registered Powerstrip (this costs money), select the DVI connection and get it to write the data saved from Viewsonic to the port. Voila, fixed!

[Oakey]

Syllopsium:

All of that goes way over my head. However, I've used softMCCS and when connected with DVI > VGA it detects the Dell 2405fpw over Analog.

If I plug the DVI cable in and scan again, it says it's "Not a Plug and Play Device" and all the sections under that say 'N/A'.

Does this sound like corrupt EDID?

Also, how do I turn off the Dell's service menu? I've unplugged the thing and left it for about 10 minutes but every time I power it on the Service Menu is still there.

Okay, my friend has the same monitor as me so tomorrow I'm going to get him to save the EDID info using Phoenix EDID Designer. I'd appreciate any help in writing the EDID data with Powerstrip as when I click 'update EDID' I get a few messages:

"AN EDID EEPROM was not detected on this monitor, do you wish to scan the bus for other EEPROMS"

Clicking yes gives:

"An EEPROM has been found at port 2 do you want to attempt to write this EEPROM?"

Now if I do the same with the VGA cable plugged in it gives me port 1 and 2, am I to assume port 1 is VGA and port 2 is DVI?

[Syllopsium]

It probably is a case of port 1 VGA, port 2 DVI but you may have to read the SoftMCCS log to be sure. It could well be your EDID data being corrupt, but I'm not guaranteeing it.

Unfortunately, whilst it helps having a friend with the same monitor, it's not valid to copy all his settings. Amongst other things the EDID information includes the serial number and the year/week of manufacture. The remainder of the information (timings, etc) should be the same so it is definitely useful.

I have no idea how to turn off Dell's service menu; I do not have a Dell TFT.

Whilst it sounds complex, once you've got your head round it it's not actually that difficult. It literally is a case of copying the data from the SoftMCCS log and checking for invalid entries. The EDID editor will guide you and tell you if things don't add up. It is a bit of a faff though.

Unfortunately my link to the Viewsonic EDID editor is no longer valid, but a google can still find it.

Of course the easy but slow way is to return it to Dell under warranty. Alternatively, Dell themselves might have an EDID repair tool which would make things an awful lot easier but you'd have to check their website and contact support.

PK

[Oakey]

Okay, here's my Old EDID:

EDID BYTES:
0x 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
------------------------------------------------
00 | 00 FF FF FF FF FF FF 00 10 AC 10 A0 53 43 33 31
10 | 22 0F 01 03 80 34 21 78 EE EE 50 A3 54 4C 9B 26
20 | 0F 50 54 A5 4B 00 81 80 A9 40 71 4F B3 00 01 01
30 | 01 01 01 01 01 01 28 3C 80 A0 70 B0 23 40 30 20
40 | 36 00 07 44 21 00 00 1A 00 00 00 FF 00 4D 36 37
50 | 35 34 35 38 47 31 33 43 53 20 00 00 00 FC 00 44
60 | 45 4C 4C 20 32 34 30 35 46 50 57 0A 00 00 00 FD
70 | 00 38 4C 1E 51 11 00 0A 20 20 20 20 20 20 00 19

And here's my friends EDID:

EDID BYTES:
0x 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
------------------------------------------------
00 | 00 FF FF FF FF FF FF 00 10 AC 10 A0 53 30 34 30
10 | 1F 0F 01 03 80 34 21 78 EE EE 50 A3 54 4C 9B 26
20 | 0F 50 54 A5 4B 00 81 80 A9 40 71 4F B3 00 01 01
30 | 01 01 01 01 01 01 28 3C 80 A0 70 B0 23 40 30 20
40 | 36 00 07 44 21 00 00 1A 00 00 00 FF 00 4D 36 37
50 | 35 34 35 37 52 30 34 30 53 20 00 00 00 FC 00 44
60 | 45 4C 4C 20 32 34 30 35 46 50 57 0A 00 00 00 FD
70 | 00 38 4C 1E 51 11 00 0A 20 20 20 20 20 20 00 38

I've highlighted the values that are different (the other values that differ being serial number, date of manufacture, etc).

Unfortunately, now when I go to Powerstrip and click update EDID it's only detecting the EDID EEPROM on port 1, I no longer get an option for port 2. Have I lost it completely now then?

[Syllopsium]

The values you highlight are part of the serial number.

Your EDID is invalid - the checksum (the very last byte) is corrupt and should be 09 instead of 19.

You're going to have to construct a valid EDID and write it using Powerstrip or similar.

If you're using the Viewsonic editor to create it, you may have problems in the last section entering certain values - in particular the Pixel Clock (hex value : 11, just after 1E 51). The monitor range limits are : Min vertical freq 56Hz. Max vertical Freq 76Hz Min Horiz Freq 30Hz Max Horiz Freq 81KHz Pixel Clock 170MHz if you want to edit them using the text box on the right, rather than the hex values in the grid.

I'd be surprised if the port is completely dead - probably a bit of fiddling around will enable it to be recognised again.

Do note that Dell may have a utility that fixes a corrupt EDID. It's worth asking..

PK

[Oakey]

The values you highlight are part of the serial number.

Your EDID is invalid - the checksum (the very last byte) is corrupt and should be 09 instead of 19.

You're going to have to construct a valid EDID and write it using Powerstrip or similar.

If you're using the Viewsonic editor to create it, you may have problems in the last section entering certain values - in particular the Pixel Clock (hex value : 11, just after 1E 51). The monitor range limits are : Min vertical freq 56Hz. Max vertical Freq 76Hz Min Horiz Freq 30Hz Max Horiz Freq 81KHz Pixel Clock 170MHz if you want to edit them using the text box on the right, rather than the hex values in the grid.

I'd be surprised if the port is completely dead - probably a bit of fiddling around will enable it to be recognised again.

Do note that Dell may have a utility that fixes a corrupt EDID. It's worth asking..

PK

I realised that was the serial as I used the Viewsonic App, it apeears the guide I was following was wrong.

I found the Dell DVI Recover utility and ran it, this is what was output to the Ports file:

Port 1:
00 FF FF FF FF FF FF 00 4C 2D ED 00 39 31 49 44 21 0E 01 03 80 26 1E 78 2A EE 95 A3 54 4C 99 26 0F 50 54 BF EF 80 81 80 71 4F 01 01 01 01 01 01 01 01 01 01 01 01 30 2A 00 98 51 00 2A 40 30 70 13 00 78 2D 11 00 00 1E 00 00 00 FD 00 38 4C 1E 51 0E 00 0A 20 20 20 20 20 20 00 00 00 FC 00 53 79 6E 63 4D 61 73 74 65 72 0A 20 20 00 00 00 FF 00 48 34 4A 58 38 30 32 38 30 30 0A 20 20 00 F9

Port 2:
E0 FF FF FF FF FF FF 54 54 AC D2 A0 D3 E3 B3 F9 EA CF D3 FF FF FF FF FF FF FF FF FF FF FF FF FF FF FF FF FD CB 80 81 80 A9 40 71 4F B3 00 01 01 01 01 01 01 01 01 28 3C 80 A0 70 B0 23 40 30 20 36 00 07 44 21 00 00 1A 00 00 00 FF 00 4D 36 37 35 34 35 38 47 31 33 43 53 20 00 00 00 FC 00 44 45 4C 4C 20 32 34 30 35 46 50 57 0A 00 00 00 FD 00 38 4C 1E 51 11 00 0A 20 20 20 20 20 20 00 19

The start of that Port 2 should be 00 and end in 00 (Like Port 1), not E0 and 54? Also, the one you say is incorrect, it keeps changing to 19?

[Syllopsium]

Yes. EDID data must start with 00 FF FF FF FF FF FF 00.

Your port 2 data from the DVI recover app looks fairly screwy. The data from SoftMCCS you posted earlier appears to be mostly reasonable aside from the checksum.

I can't say whether the error is frequently changing EDID data (unlikely), a hardware problem which doesn't transmit the correct EDID or an error in either the DVI recover app or SoftMCCS.

If it was me, I'd correct the SoftMCCS data and whack it back it via powerstrip. However, it's your risk and your monitor!

PK

[Oakey]

Yes. EDID data must start with 00 FF FF FF FF FF FF 00.

Your port 2 data from the DVI recover app looks fairly screwy. The data from SoftMCCS you posted earlier appears to be mostly reasonable aside from the checksum.

I can't say whether the error is frequently changing EDID data (unlikely), a hardware problem which doesn't transmit the correct EDID or an error in either the DVI recover app or SoftMCCS.

If it was me, I'd correct the SoftMCCS data and whack it back it via powerstrip. However, it's your risk and your monitor!

PK

So how do I change the checksum? Everytime I edit it it automatically goes back to 19, then says the Checksum is correct.

The other problem is, I used Powerstrip and it told me the EDID header was corrupt and asked if I wanted to fix it. I clicked yes and a minute or so later it told me it couldn't repair it because the EEPROM is either write protected or the cable is too long.

[Syllopsium]

So how do I change the checksum? Everytime I edit it it automatically goes back to 19, then says the Checksum is correct.

The other problem is, I used Powerstrip and it told me the EDID header was corrupt and asked if I wanted to fix it. I clicked yes and a minute or so later it told me it couldn't repair it because the EEPROM is either write protected or the cable is too long.

Oops - my mistake. The EDID data you posted for your monitor earlier is actually ok - I mistyped one value .

If that data is from your DVI connection rather then the VGA connection, it's not an EDID issue. You must retrieve it from the DVI connection to be sure it's correct though.

PK

[Oakey]

That was the EDID data from the Windows registry so that's possibly actually correct? The one I posted just before is probably the incorrect one read from the monitor with the Dell DVI tool?

As I say though, I don't think I can write anything new with Powerstrip

ETA: Powerstrip must be doing something as the name of the corrupt header has been changed to DELA010 which is would be correct

[Syllopsium]

Yes. The data you compared with your friend's data. Both are correct. The Dell data is clearly not as you can see even the manufacturer name is corrupt.

Obviously you'll need to be able to write it to the port.

When I was fixing my VP730, Powerstrip was unable to read the DVI port - only SoftMCCS was able to. Once the fixed EDID information was generated I was able to persuade Powerstrip to write it.

There is a long thread here in the Dell forums about similar issues :

http://en.community.dell.com/forums/...px?PageIndex=1

You may have to try certain combinations of the DVI_Recover tool and service mode (it appears the EDID data is not always writeable except in service mode for some monitors).

PK

[Oakey]

Okay, I'm in DOS using EDID_Writer:

When I try;

ddcw -m 1 -f 2405fpw.txt is says:

EDID read from device:

ETA: will update, seen more errors.

EDID read from file:

00 | 00 FF FF FF FF FF FF 00 10 AC 10 A0 53 43 33 31
10 | 22 0F 01 03 80 34 21 78 EE EE 50 A3 54 4C 9B 26
20 | 0F 50 54 A5 4B 00 81 80 A9 40 71 4F B3 00 01 01
30 | 01 01 01 01 01 01 28 3C 80 A0 70 B0 23 40 30 20
40 | 36 00 07 44 21 00 00 1A 00 00 00 FF 00 4D 36 37
50 | 35 34 35 38 47 31 33 43 53 20 00 00 00 FC 00 44
60 | 45 4C 4C 20 32 34 30 35 46 50 57 0A 00 00 00 FD
70 | 00 38 4C 1E 51 11 00 0A 20 20 20 20 20 20 00 19

Update Failure

So it would seem it's almost correct, it's now just the checksum that's corrupt? My obvious problem now is getting the correct EDID to write to the EEPROM. Is this something I should persist with or is it likely the EEPROM is buggered and I can't actually write to it?