Colour Cast

Hi David and others,
I have tried the Chimei monitor with the Phillips 170S MDF and it does work. I have also tried my Phillips monitor (not an 170S, but close to one, a reader suggested) and the Chimei MDF and the Phillips S170 MDF work with it as well. However all these combinations still show a bluish colour cast on F12, when returned. They also show colour 248 turning bright yellow on a consecutive use of F12. These changes are corrected by using the Display/Mode icon on the bottom left of the icon bar. The Monitor/MDF combinations above, all produce the same real mish-mash of colours when booted from a “Power boot up”, though these combinations only have the same colour cast on a “Reset boot up”. Theoretically they all run through the !Boot the same and yet the outcomes are quite different? Help!! The CMOS appears to be good as the CDROM icon is not lost on reboots, and the clock is keeping good time.

The other observation that intrigued me was when I chose a Chimei definition and it came back with a “double image” shaking all over the place. Then when I went to the Display icon I could choose 1024×768, 256 colours, but also choose the frequency, I could choose a frequency that did indeed stabilize everything and it looked great. No shake, and no colour cast at all. When I went back and looked at the Chimei MDF I had created, the “bad” frequency was ahead of the “good” frequency. When I edited the MDF so the good frequency preceded the “bad” frequency, I re-did the config and all was good. Reboots were predictable and usable, sadly all colour cast hassles persisted. Then I realised – the !boot+configure+screen app does not allow the frequency to be set? Just the Monitor, the resolution and the colour depth. Then I went back to Display app and for the current monitor (Chimei or Phillips) it offered the resolution, the colour depth and frequency. Both MDFs worked ok on both monitors from that app.

What do I conclude?
First: The order of the “modes” in the MDF is important as RISCOS Configure just takes the first Screen that matches the resolution, and does not allow a choice of any Frequency options. I have to edit the MDF to put the Resolution I want to work with its frequency first in the queue, or it stuffs up.
Second:
The Display app has four choices, the type of monitor, the resolution, colour depth, the frequency (sometimes more than one choice = good). This seems to be much better written than the what ever is used in the OS WimpMode command. It works.

It would appear first power boot up gets it really wrong, hopeless palette, nearly all black, really screwed. After a start up and Display correction, a Reset boot get it nearly ok, with just a colour cast. Then this recurring problem, returning from F12 causes the colour cast, two F12s in a row causes colour 248 to go to yellow. Display/Mode corrects it.

To the whole community:
Why is this so? And why am I the only one who has experienced it? I am desperate, like really desperate, I want to move on.

Silly question, but have you thoroughly checked all the applications running at startup, and/or being filer_boot-ed? I recall some graphics packages (Photodesk comes to mind) included gamme calibration facilities and applications which could seriously mess with the look of the display. Most people never touched those things, but if the computer isn’t “clean” it is possible that something is messing with things. I agree that this doesn’t fit all the data points, but given your problem isn’t a common/known one, I’m trying to think “left field”.

The other thing would be to try with/without VRAM, although you probably tested this when you were switching RAM in/out. I’m assuming you’ve tried different VGA cables as well as different monitors. Could also be worth trying the computer in a different room/location in case you have an odd interference issue?

I replaced the StrongARM with the original ARM CPU and the problem still occurred ( with 128MbRAM). I replaced the 128Mb of RAM (2×64Mb) with one stick of original Acorn 8Mb RAM and the same result. Pressing F12, and returning, caused the colour cast, Display/Mode fixed it.

I had also taken the precaution to re-install RISCOS 2.04 over again to get as clean a starting point as possible. The installation went smoothly and booted OK though power on booting results with the screen looking like this: http://www.laketyersbeach.net.au/RISCPC/RISCPC_01.jpg Using Display/Mode fixes it. this is not good.

I get the impression that F12 followed by Display/Mode fixing it might be a bit of a red herring and that it sounds like a hardware problem. The hardware will be in a different state by the time you do that compared to when the machine is first starting up.

So you’ve tried different ROMS, suggesting it’s not the ROM. You’ve tried a different CPU and RAM, so it’s not those. Like Andrew said, you need to try a different monitor and different cables. If it still happens then it seems to me it leaves the motherboard, VRAM, power supply, any of which could be faulty and would need swapping out to test. There’s also the hard drive of course. Also are you able to plug it in in a different building or at least a different part of the house?

Forgive me Andrew, I’ve just repeated half of the suggestions you made! It’s been one of those days… :(

I can’t quite pin down what is failing from the various anomalies which you describe, it seems there is some undesirable combination of the MDF timings with possible weakness of the VIDC circuitry.

An obvious fix is to remove all the different frequency variants of the desired mode in the MDF, and then RISC OS will have no choice but to use the working one, regardless of how it is selected.

Hi Andrew, I have tried two different VGA cables today, one with a pin missing (my usual cable) and a cable from the Mens Shed computer that has all pins present. Sadly no difference, boot up bad screen and colour cast on F12. I have no VRAM much earlier and got the same results. I did buy a 2Mb video RAM card to tackle this problem, and it appears to be fine (as shown by Display/Mode changes being spot on).

Hi Andrew and Andy, I also took the whole RISCPC into the Mens Shed today (got a few comments on archaic system’s, to which I replied this is a collector’s item!!!) to test out the power supply/interference problems. I was keen to try this suggestion as I have a 2kW solar cell system on my roof. Installed C:2010, so the inverter is a bit ancient in design and it shuts down if the local grid voltage exceeds 265V or thereabouts. This meant we could not even just use our own generated solar power, let alone export anything to the grid. It was a lose, lose!! So we got a 220V regulator put in to handle the high grid voltages. After that the ancient inverter only ever saw 220V, so never shut down, we got to use as much solar as we could, and maybe exported a bit as well. However 220V is the lowest voltage acceptable on the RISCPC power supply, so I took it to the shed for a full 240-250V supply. Sadly there was no change. Bad power on screen and blue cast on F12, and both fixed by Display/Mode/OK. I was thinking the RAM maybe unreliable with below par supply voltage, but I think that can be ruled out now. I think it also rules out some sort of interference at my place.

I feel if there is any hardware problem, then Display/Mode would not fix it. Plus once fixed, if F12 is not used other applications can be run and the screen stays just fine. It is just the return from F12 or Configure/Screen that does it. I have not run any apps that might play around with screen definitions etc. The RISCPC boots into a brand new vanilla !Boot in RISC OS 4.02, so very little has been changed. I have added in the RO5 monitor definitions to get the Philips S170 and I have hacked my own Chimei definition from first principles with a RPi connected to the Chimei monitor with a HDMI to VGA adapter. Both MDFs can do a great 1024×768 C256 on either monitor, ie they are largely compatible monitors and MDFs. In all four combinations they still show the problems with the power on boot screen and the return from F12 colour cast.

Hi David, I have [revious;y reduced the Chimei MDF file to one entry 1024×7687 C256 and doing so forced it to use the one definition, but the faults persist, bad boot up and colour cast on F12. No mode I have tried (and believe me I have tried a lot!) does not have this fault. I think I can remember choosing a 4grey scale set up very early on at one stage and it booted reliably from power on, and maybe F12 was OK too, but I can’t be sure that was true. I have tried 4greys Wimp since and it behaves badly as above as well. I must have been imagining it. I feels like I have tripped some setting some where that is stored and is haunting me.

Thanks again folks for applying your greater experience and listening to my reamblings, Rob

Men shed are you in aus?

and it shuts down if the local grid voltage exceeds 265V or thereabouts.

Good grief. The power supply is probably made for 230V ¹, +10%/-6% ². That’s about 256V max.

Australia’s power is supposed to be the same, so any decent inverter will shut down above that limit. Things that are running (computers to fridges to lightbulbs) may be damaged if the voltage is regularly too high.

It’s a known issue: https://www.solarquotes.com.au/blog/sa-power-networks-shonky-voltages-causing-headaches-adelaide-solar-owners/

¹ Until 1993 the UK was 240V and western Europe was 220V. Then they all standardised on 230V.

² https://www.legislation.gov.uk/uksi/2002/2665/regulation/27/made

I think India is still officially 240V, it certainly still was in the mid-1980s. In practice, in some of the villages, it can be anything from about 150V to about 300V – varying up and down as consumers turn things on and off. When it’s on at all – some villages or poor suburbs suffer from load shedding every day at peak times, sometimes for hours at a time.

Until 1993 the UK was 240V and western Europe was 220V. Then they all standardised on 230V.

No, the UK has always been, and continues to be 240V, and Europe is 220V, they just fudged the standard to 230V -6%, +10% (206V to 253V) which encompasses the nominal voltage ranges of both grids.

No, the UK has always been, and continues to be 240V, and Europe is 220V, they just fudged the standard to 230V

Have you measured your power? A look online suggests that some parts of the UK have had a 10V drop, other parts have not. I think it’s too do with the step down transformers and when they are replaced.

As for my sockets? Since the mass rewiring that took place with the rollout of the smart meters, I’ve been getting a reliable ~230V (rather than ~215V dropping to ~185V when I put the kettle on!).
Just measured it, like right now, and it’s a steady 232V.

So harmonisation is a little more reality than just that which is written on paper.

Yeah, Lakes Entrance Mens Shed. East Gippsland, Australia, We have a Google page. The domestic solar power revolution in Australia is being choked by antiquated line transmission technology. We cover nearly all our daily power use at the Men’s Shed with a 7kW system. No battery needed as we are only daytime use.
We don’t get enough second hand computer gear for me to comb through. It is sad. Thrift shops can’t sell untested electrical gear, so most old stuff just straight to the tip :-(

No, the UK has always been, and continues to be 240V, and Europe is 220V, they just fudged the standard to 230V -6%, +10% (206V to 253V) which encompasses the nominal voltage ranges of both grids.

No, since the 230V nominal standard came in, newly installed transformers on both sides of the channel have tended to be 230Vac. If you have 240Vac at your sockets in the UK, then the substation feeding your house is very likely to have a pre-harmonisation transformer installed within it.

Edit That said, I should probably pedant myself before someone else does, to note that the voltage that you measure at any point in time will be very heavily dependant on the the quality of the infrastructure between your house and the substation, not to mention how well balanced the loads on the three phases of the transformer that is supplying your house are. But the nominal output of new transformers feeding three equally balanced phases will be 230Vac on any one phase to neutral.

Just measured our voltage here in Greenock. 231V, dropping to 230V when I put on the kettle and all four rings on the cooker (just for a couple of moments). I don’t think the change was coincidental, since it went back to 231V as soon as I turned them all off again. It could be much less than a volt change, or almost two volts, since my meter only has three significant figures.

(I did the measurement on a socket that’s on a different circuit from the cooker. The kettle was plugged into the socket on the cooker switch. Some time I might try it with the meter in the cooker switch socket.)

Our local substation’s transformer was replaced not all that long ago, after a lightning strike.

Just to refocus the discussion (I love power solar supply discussions, as I pay for the other stuff) back onto the RISCPC + RISCOS dilemma, how can I debug a “bad palette” start from power on and a colour shift (ie palette “corruption”) on an F12, or Configure/Screen or Configure/anything to do with desktop?
I really want to move on, not be obsessed with this problem, and get into setting up proper networking (Internet, local FTP, NFS etc) and then maybe RISCOS5 (I have bought the ROMs etc 3 years ago,curently a UFO (ie un-finished objective).
It probably seems like I have the impatience of a five year old, and it is hard for people to remotely understand and suggest strategies, but I am willing to give them all a try, so what do people think?
Should I: Put up with it (it does not appear to effect enything else) and move on, or struggle for an answer?
Cheers, Rob

Personally, I’d put up with it for the moment and move on – a solution might turn up at some point in the future, possibly with a D’oh moment…

Agreed Clive, I could find more rewarding challenges. I cant see anybody helping me though if I don’t keep the question “alive”, it will simply drop out of view. At the moment the alternative I run the RISC of boring the tits off everyone in the quest for a solution, again an unproductive outcome. If someone has an new angle I could try, I would desperately love to hear from them, however in the interim, without new ideas to follow up from the Community group I am going to let it rest myself, but still really keen to hear of fresh ideas (pls :-) ). I think I will try setting up my final goal of RO5.x(a sort of follow up to all the above 4.02 hassles).

Thanks so much for those who have found time to think about the problem and try to help me.

Cheers Rob.

PS People used to tell I was really good with computers, but really I don’t know a lot or nor am I particularly bright, but boy am I persistent and stubborn, never give up.

@Rob
As Andrew said I really think it’s a problem of a misplaced file on startup.
Try command : *WimpMode X800 Y600 C256 F60 or or the one you want depending on your monitor
In my file:
SCSI::SSD.$.!Boot.Choices.Boot.PreDesk.Configure.Monitor
this command is used.
My file content:
LoadModeFile Resources:$.Resources.ScreenMode.Monitors.EDID0
WimpMode X2560 Y1440 C16M F60
You can test this file by running it, and check if the problem persists.

@Rob. Looking through your pictures, I’ve seen issues like that, mostly connected to cabling, equipment and environmental issues. Are there any ex-Acorn dealers or computer repair shops you could take it to? It is starting to sound like one of those uneconomical repairs, as we used to call it; perhaps not? Good luck, fingers crossed for you.

Hi Andrew and Jean-Michel,

I will try the file as you suggested Jean-Michel, it is midnight here at the moment (on my phone 😐) so I will try it about 8hrs time.☺️👍

I suspect it is some misplaced, or sub optimal file or command or configuration, very subtle and hard to track. That is why I am tempted to just forge a head and try RO5 as my next step. I feel my RISCPC skills are back up to scratch enough to try it. I seems heavy handed, but maybe with other options exhausted, it is time. Otherwise it just sits its padded postage bag.

Your question Andrew about Acorn dealers in Australia brought tears to my eyes and a smile on my lips. It reminded of the days Acorn confidence, and what happened to the dealers in the far flung colonies as Acorn lost the PC wars. Dealer support in Australia died about 1992. I struggled on in school but when they issued us with Windows PC Laptops for all teachers I knew the game was up. The Acorn machines were sidelined and pensioned off, ie chucked out. So no, there is nobody in Australia that I know of that would have skills and know-how better than you guys.

I have tried quite a few hardware options, leads monitors efc, but no permanent modifications apart from the CMOS battery to NiMH. I am reluctant to do much physical mods as I don’t have any hardware support to fix any mistakes. One bad one could finish my quest completely.

Thank you for your encouraging words, I will try the file and then RO5, and was said above “Fingers Crossed”.

Cheers, Rob

Hi Andrew and Jean-Michel,

I will try the file as you suggested Jean-Michel, it is midnight here at the moment (on my phone 😐) so I will try it about 8hrs time.☺️👍

I suspect it is some misplaced, or sub optimal file or command or configuration, very subtle and hard to track. That is why I am tempted to just forge a head and try RO5 as my next step. I feel my RISCPC skills are back up to scratch enough to try it. It seems heavy handed, but maybe with other options exhausted, it is time. Otherwise it just sits its padded postage bag.

Your question Andrew about Acorn dealers in Australia brought tears to my eyes and a smile on my lips. It reminded me of the days Acorn confidence, and what happened to the dealers in the far flung colonies as Acorn lost the PC public relations wars. Dealer support in Australia died about 1992. I struggled on in school but when they issued us with Windows PC Laptops for all teachers I knew the game was up. The Acorn machines were sidelined and pensioned off, ie chucked out. So no, there is nobody in Australia that I know of that would have skills and know-how better than you guys.

I have tried quite a few hardware options, leads monitors efc, but no permanent modifications apart from the CMOS battery to NiMH. I am reluctant to do much physical mods as I don’t have any hardware support to fix any mistakes. One bad one could finish my quest completely.

Thank you for your encouraging words, I will try the file, and then RO5, and was said above “Fingers Crossed”.

Cheers, Rob

Until 1993 the UK was 240V and western Europe was 220V. Then they all standardised on 230V.

When I had solar panels installed here, the installer commented that the mains voltage was around 250 volts, and recommended fitting a “Volt Doctor” to bring it down a bit.

I was thinking “constant voltage transformer”, what I got was an auto-transformer, reducing whatever the input was by about 10%, so now the house runs on about 225 volts.

The difference – I haven’t had a toaster burn out since (and several did before) and the kettle takes noticeably longer to boil.

We are about 100 metres from the substation. The estate and substation were built around 1955.

Hi folks,
If you read this as a previous post before this edit, I apologise. I am pretty sure I was barking up the wrong tree. Putting the F65 on the end of the VRAM files’s wimpmode statement was just crashing the statement, so the wimpmode was not acted on at all so of course it did not get a chance to alter the wimp set up hence no change in colour cast and why that file could not correct it. Whatever theF setting was in that line it was just crashing the wimpmode function. A false dawn…..

I thought I had found a trigger that might help me debug it. The reboot with any thing other than “wimpmode X1024 Y768 C256” was a crash, resulting in blind typing etc

I tried putting a different Frequency in the MDF for the Chimei but the RPC must calculate the frequency from first principles from the x_res, y_res, pixel_rate, h_timings, v_timings etc and not the statement of timings in the #line. Because, even though I changed that line, the frequency was still stated as 60Hz. I just two of them.

However what does the BandLimit file do? Can it be tweaked for a monitor or is a given for the RPC hardware setup?

The RAM size says 130Mb (2*64mb+2Mb) at start up so I am sure the 2,Mb video RAM is seen, but the System report always says 1024k screen, where has the other 1024k gone?

Rob