eBay 128K Spectrum RGB cables overheat the Spectrum!?
Hi everybody,
For a long time I've wondered why Spectrum 128K Toastracks were overheating when using my Retro Computer Shack RGB SCART cable.
Today I found out why.
Turns out that they got around the problem of there not being a +5V to use for the RGB switching signal by using the composite SYNC! The poor ULA is driving the RGB switching as well as the sync going to the TEA 2000. The net result is that the load is too much for the ULA, and it gets too hot, eventually overheating and causing graphics corruption and unreliability.
You can tell if the ULA is over-stressed by measuring the output voltage of pin 23 (CSYNC) on the ULA. It should be about 3.7V when using the composite output or the TV output. When it's driving the blanking signal, it's down to nearer 1.7V.
Therefore I would suggest that people do not use these cables. If you want a quick fix, just desolder pin 16 on the SCART plug. That'll give you composite, and if your TV allows you to select RGB/Composite, you can still switch to RGB. If you want RGB again, though, you'll need to wire it up to a 5V supply, using a resistor to lower it to a correct voltage like 3V.
It's definitely more convenient to do it the way those cables do, and it usually works... but at what cost? I think the cost is my Spectrum 128K's ULA, and that's too high....
For a long time I've wondered why Spectrum 128K Toastracks were overheating when using my Retro Computer Shack RGB SCART cable.
Today I found out why.
Turns out that they got around the problem of there not being a +5V to use for the RGB switching signal by using the composite SYNC! The poor ULA is driving the RGB switching as well as the sync going to the TEA 2000. The net result is that the load is too much for the ULA, and it gets too hot, eventually overheating and causing graphics corruption and unreliability.
You can tell if the ULA is over-stressed by measuring the output voltage of pin 23 (CSYNC) on the ULA. It should be about 3.7V when using the composite output or the TV output. When it's driving the blanking signal, it's down to nearer 1.7V.
Therefore I would suggest that people do not use these cables. If you want a quick fix, just desolder pin 16 on the SCART plug. That'll give you composite, and if your TV allows you to select RGB/Composite, you can still switch to RGB. If you want RGB again, though, you'll need to wire it up to a 5V supply, using a resistor to lower it to a correct voltage like 3V.
It's definitely more convenient to do it the way those cables do, and it usually works... but at what cost? I think the cost is my Spectrum 128K's ULA, and that's too high....
Post edited by Spirantho on
Comments
- IONIAN-GAMES.com -
Your blanking signal input does not satisfy the SCART specification for the Input Impedance to be at least 10.000 Ohm. See this:
http://www.fruitcake.plus.com/Sinclair/Spectrum128/SCARTCable/Spectrum128SCARTCable3.htm
And this:
http://www.fruitcake.plus.com/Sinclair/Spectrum128/SCARTCable/Spectrum128SCARTCable3.htm
You said Composite Sync and I thoght you were referring to CSYNC pin :D. What you call Composite Sync is Composite Video (CVBS) to me, hence the confusion.
Point is, though - pin 4 on the RGB output (the CSYNC) should definitely not be connected to pin 16 on the SCART block - even if it does happen to work.
All the cables I've made use the CSync for ... sync and I use a 1.5V AA battery for the blanking/switching pin.
Works on all TVs I've tried, by the way.
But this is against what Paul says in his pages... I'm confused. It happens I'm building a SCART adaptor for all my Speccies, following Paul's recommendations, and you stopped me dead in my tracks. :-o What should I do now?
From his own document:
That's from http://mts.speccy.cz/doc/128_rgb.pdf which is Paul Farrow's document!
I say again, though, no matter what anyone else says - do NOT use the CSYNC as an RGB switching signal. Compare the temperature of your ULA without this line connected with the temperature when it is connected, and you'll see what I mean. There's no way I'm going to put my ULAs through that temperature, that's for sure.
Great, now we are in sync :D As a matter of fact, my first reply was wrong because I misquoted the Blanking Pin Impedance as being >10K Ohm, while instead it's 75 Ohm (10K is actually the Function Switching Pin impedance)
I certainly agree with you. I just went away and didn't cross-check with the Spectrum schematics, otherwise I would have noticed that a 75 Ohm load was being driven by a TTL output which of course is a no-no.
Time to remove the Blanking connection and see if both the Philips CM8833 and my TV support RGB/composite switching in other ways. More on this later this evening.
The composite signal may be at maximum 5V. There is a 68 Ohm resistor between the output pin of the ULA and the RGB socket of the 128k Spectrum. In worst case, if this pin of the RGB socket is shorted to ground the maximum power dissipation inside the ULA caused by this shortage would be if the internal resistance of the ULA at this pin would be also 68 Ohm. The power dissipation of the ULA caused by this event would be 5*5/( 68+68 )/2= 92 mW. An additional power of 92 mW will not heat the ULA that much I am sure - and this is the real worst case - the load at pin 16 of SCART is about 75 Ohm to ground, which would cause about 43 mW in maximum (you can calculate it the same way). You could reduce this worst case power by changing the 68 Ohm resistor by a 120 Ohm resistor which should be low enough to supply SCART pin 16 with more than 1 V. With a 120 Ohm resistor (56 Ohm could be inserted inside the SCART plug in series to pin 16) the maximum additional power dissipation for the ULA is about 32 mW.
All of these thoughts imply that the TV set is ok and there is no positive current coming out of pin 16 of the SCART socket. This pin has to act as a load of about 75 Ohm.
Greets Ingo.
Assume ULA pin voltage is 3.7V
Assume input resistance of TV is 75Ω
So 3.7V / (68Ω + 75Ω) = 25.87mA
5V - 3.7V = 1.3V
1.3V X 25.87mA = 33.6mW (ULA additional power dissipation)
As above, but assume ULA pin voltage is 1.7V
So 1.7V / (68Ω + 75Ω) = 11.89mA
5V - 1.7V = 3.3V
3.3V X 11.89mA = 39.2mW
Mark
Repair Guides. Spanish Hardware site.
WoS - can't download? Info here...
former Meulie Spectrum Archive but no longer available :-(
Spectranet: the TNFS directory thread
! Standby alert !
“There are four lights!”
Step up to red alert. Sir, are you absolutely sure? It does mean changing the bulb!
Looking forward to summer in Somerset later in the year :)
the 68 Ohm inside the 128k is placed there from factory. So I think that the load of (68 + 75) Ohm is calculated to be ok for the ULA.
Greets Ingo.
First of all, the document referred to was put together without my knowledge by someone simply collating information from my website. It appears to have been created in 2001 and presumably has not been updated over the past 12 years and so doesn't contains all the latest updates made to my SCART web pages. The creation of the document was not authorised by me and so I cannot vouch for its accuracy or comment on its content. I do not promote information from my website appearing on other sites for the very reason of avoiding multiple versions getting out of sync.
I am not involved with these (or other) SCART leads being sold and so have no knowledge on how they are wired.
The concern of overheating of the ULA and damaging is a very valid concern, and I welcome any constructive feedback on my SCART design (which is why I present all design calculations so they may be scrutinised and queried). My SCART web pages discuss a range of options available for creating the necessary control voltages for the SCART socket. The approach of using CSYNC is the simplest but does, as has been pointed out, have issues. This is why more elaborate approaches are presented afterwards. Each tries to overcome the fact that the monitor socket was never designed to drive the SCART input of a TV, and aims to do so in a manner that is not too cumbersome to use. However, it sounds like I need to enhance my web site to describe the implications of each approach in more detail.
As has been pointed out, the output is actually driving a load of 68 + 75 = 143 ohms. However, this is still very low and affects the output voltage from the ULA (as shown in the charts on my website).
The approach I have used in my own SCART cable for the 128 (and in those I used to build for other people many years ago) is shown in the final diagram at the bottom of this page:
http://www.fruitcake.plus.com/Sinclair/Spectrum128/SCARTCable/Spectrum128SCARTCable4.htm
Here is the diagram and as can be seen it uses a voltage regulator to obtain the signal to feed into the BLANKING pin:
The approach results in a cable with a power socket protruding from the SCART plug into which an external 12V power supply may be connected. If the power supply is not connected then the control signals are not presented to the SCART socket and so it is up to the TV whether it allows manually selection of the input. If a 12V supply is connected then it will drive the FUNCTION SWITCHING input directly and is converted down to a voltage level suitable for the BLANKING input. Therefore, the strain placed on the ULA by using CSYNC is avoided, and the resultant cable fully complies with the SCART specification and so should work with the RGB SCART socket of all TVs.
Using an external power supply is the approach I recommend for SCART cables intended for any model of 128K Spectrum (or QL).
Note that the issue of low loading also affects the colour signals from the monitor socket, although no where as severely as using CSYNC for the BLANKING input. Ideally all signals from the monitor socket should be buffered through an external circuit and then all issues of loading are avoided, but this would require an external power source be used. The loading of the colour signal outputs can be reduced by increasing the value of the resistors placed inline with them in the cable, but this will have the negative effect of darkening the picture. However, this may be compensated for by simply increasing the brightness controls of the TV.
Indeed this is right. I quickly stated that driving a 75 ohm directly is not suitable for a TTL output, and I should have published a correction after determining what you correctly state, but I didn't. Apologies! As a matter of fact, my own experiments with a 128K classic and a Philips CM8833 show a current drain of about 50mA for all the RGB signals combined, meaning
an average of 10 mA per pin which looks well within limits to me. Image is rock solid and colours appear well balanced.
It seems that more research is worth doing. When doing my own experiments, I connected a modern 4:3 TV/VGA LCD monitor, and found that it defaulted to composite input rather than RGB, apparently ignoring the CSYNC signal or, more probabily, putting an eccessive load which resulted in insufficient amplitude of the control signal. More investigation is pending as soon as I have some more spare time, but it is reasonable to think that SCART standards are generally not so well complied with...
It is highly possible that an active approach (video amplifiers/decouplers) with a separate power supply might be the only universally suitable solution.
P.S. not related to the discussion: RAM tester ROM has been just successfully used servicing a +2A, with an appropriate change in the ROMCS pin arrangement. :D
My other TV didn't used RGB if the 220 ohm series resistor was included. Finally I decided to use the USB port of the modern LCD monitor/TV to get the correct voltage for blanking.
In this topic you can find the temperature of the Toastrack ULA (with CSYNC connected to SCART blanking...):
http://www.worldofspectrum.org/forums/showthread.php?t=45721&highlight=temperatures
Now I repeated the measurement with externally powered SCART blanking. The ULA temperature was now 63.4 degrees celsius, with the CSYNC connected to blanking it was 62.2 degrees celsius. (Maybe the external temperature was not exacly the same, that's why the temperatures are a little bit different.) According to this no significantly higher temperatures...
BUT: As I have switched to external power for blanking, picture became more stable. Before the modifications I have seen some random black spots on the white background and if 128K Spectrum features were extensively used (e.g. I like to use Paul Farrow's excellent ZX81 emulator) the screen blanked for moments randomly. Sometimes the left side of the picture flickered a little. Now everything looks calm and nice! :)
It looks like that there is not overheating, but the situation changed by using external blanking voltage. Maybe the too low blanking voltage caused the picture instability? I don't know...
Of course I've since rewired the cable back to as it was before, because the 128 I bought it for I've also rewired, so the CSYNC pin is now just a 2V line for the RGB switching, but for most people with unmodified, this should give the same picture quality with less load on the ULA - a good thing.
Still unlikely to be as good as my 128 with an actual 2V line (leeched from the RS232 port's 12V line via a 330 Ohm resistor), but better than before, and shouldn't damage your 128 now. :)
Interestingly, I recently fixed up my Atari ST (and I don't mean STfm or anything - I mean ST) as the ST cable had the same problem. The picture would keep jumping to the left momentarily as the TV wouldn't be able to make up its mind whether it was in RGB or composite mode (not surprisingly when you consider that every time the display chip sent a sync signal, the TV would think the display was changing from RGB to composite!). Same make of cable, incidentally.
Most retro computers were designed when TV's either did not have SCART sockets, or where most TVs could be manually switched via their remote control. So little if any proper provision was made for DC control signals suitable for driving the SCART switching control circuit inputs.
A lot of the retro computer communities have problems with their respective computer and a SCART TV not working together. If it is not the number one problem, it is not far behind.
Not helped by the ignorance of the communities in how TV and video systems actually work. Let alone how the SCART signal switching works.
[/gets off soapbox]
Mark
Repair Guides. Spanish Hardware site.
WoS - can't download? Info here...
former Meulie Spectrum Archive but no longer available :-(
Spectranet: the TNFS directory thread
! Standby alert !
“There are four lights!”
Step up to red alert. Sir, are you absolutely sure? It does mean changing the bulb!
Looking forward to summer in Somerset later in the year :)
The toastrack and Grey +2 both label the output as "RGB". (The grey +2 does have links on the PCB to allow it to be wired up as "peritel" option. I can only assume this was done for the french language version since that can't do RF out)
Only the Black +2 and +3 label it "RGB/PERITEL" and provide a 12v pin with a series resistor. If you connect this to the blanking pin it forms a potential divider with the TV seeing 0.8v. It's unknown whether the intention was to use this 12v pin for the blanking signal or not, but if so the resistor is just a little too big, so the voltage is below spec.
If you changed R44 for say a 680 ohm resistor this should put the blanking pin voltage over 1v as desired.
The other oddity is that both 12v pins on the socket are via the resistor, which means that you can't use one for the status voltage on pin 8 (9.5-12v) and the other for blanking, because connecting the blanking pin up drags the voltage down to 0.8 volts. To be useful it would need one pin before the resistor and the other after (or have no resistor at all and put 680 ohms in series with the blanking pin inside the cable).
I've also contacted the owners of the following documents to request they update them (or remove them) as they now contain out of date information copied from my website:
http://mts.speccy.cz/doc/128_rgb.pdf
http://www.benophetinternet.nl/hobby/vanmezelf/ZX%20Spectrum%20128K%20video%20fixes%20and%20video%20cables.pdf
http://k1.dyndns.org/Vintage/Sinclair/86/Peripherials/ZX%20Spectrum%20128%20Scart%20Lead.pdf
You do not recommend using the 12V output on the RS232 socket (with added resistor), because of the high load (50 mA).
A sidenote: 'inline resistor' in my opinion is an extra resistor in series to the output signal.
But in fact we are speaking about resistor dividers, so I would prefer using that terminology.
I think there is an easy solution to that load which is overlooked.
Here it is:
V3 = 2.5 = 12 * (R3 / (180 + R2 + R3) )
R3 = ( (180 + R2) * (2.5 / 12) ) / (1 - (2.5 / 12) )
Now let's use 560 ohm for R2.
R3 = ( (180 + 560) * (2.5 / 12) ) / (1 - (2.5 / 12) ) = 195 ohm.
The current now is: I = 12 / (180 + 560 + 195) = 12 / 935 = 1,2 mA.
Which is very reasonable.
So I think this would be a better solution than the external supply, but only when a cable with the correct connector for the RS232 socket is available.
What do you think Paul?
Ingo.
Agreed, but many do not want to alter their ZX Spectrum.
It looks like every cable should come with schematics to avoid surprise.
If now CSYNC is used as sync (SCART pin20) that means no more composite video (which maybe good when TV scans signals on SCART, now it will find RGB only).
One more question - Paul says, SCART for 128K is not suitable for 128K+2 gray, since +2 has already BRIGHT mixed with RGB. However most cables including RetroComputerShack are marketed 128K/128K+2(gray).
Do they mix BRIGHT twice ( and overload BRIGHT signal from ULA :confused: ) ?
That's true! Thanks for the hint!
http://www.retroisle.com/sinclair/zxspectrum/Technical/Hardware/ZXSpectrum%2B2Schematics.png
vs.
http://www.retroisle.com/sinclair/zxspectrum/Technical/Hardware/Investronica%20analog%20model%202-1.png
I have to change my documents to that knowledge too...
I don't think a toastrack cable with added parts to mix the bright levels, will do any good on a +2.
So I would expect there would be two different cables offered.