Symbiface II and 6128 Plus Expansion Bus

[Badstarr]

Hi guys!


Ok I have been messing round trying to get my new Symbiface 2 (2nd batch) to work with my 6128 Plus. I have successfully managed to get the card working, but with a caveat, the IDC cable is only around 5 inches long. If I use a cable any longer the RAM/ROM sends junk to the CPC. Using a small cable isnt too much of a hassle, it just means that I will have to rethink housing the card, though the length of the cable is just barely useable!


So why is this happening? I would have assumed that the expansion impedance on the Plus would be the same as the classic 6128/464? The problem occurs with all 3 of my Plus machines, I have cleaned the connections on all the machines so I doubt it's anything to do with dirt on the connections as the problem is so consistent.


I borrowed the centronics connector from my MegaFlash cable, and unfortunately I snapped the darned clip when in the vice when I was clamping it together so anyone know a good place to buy a new one? The ones I have found seem a little on the pricey side!

[TFM]

5 inches is pretty long. Try to keep ways short 

[Badstarr]

I actually had to go back to the original MegaFlash cable. The symbiface was still being a little fussy with the cable I made up, I tried shorter lengths and by the time I had it working 100% the cable was exactly the same length as the original MegaFlash one so I thought I may as well use the original cable!

If I use the original edge connector cable with my 6128 the Symbiface plays nice, the thing about that is that the cable is about 11" long so why is the 6128+ being fussy?

The thought had occurred to me that the addition of the cartridge circuitry on the Plus might be a factor?

[Munchausen]

I had similar problems with a standard 6128 (e.g. "Press Play and Any Key" when loading symbos). With another 6128 and my 464 it was fine. I thought there was something wrong with the 6128, but after Bryce has had a look at it, the board is fine, so it must have just been the cable.

Is there some way to get extra length on the cables to work? Higher quality, lower resistance cables are all I can think of. I have a lot of expansions, and though I don't currently connect them all at the same time, it is something I am planning to try shortly. It would be disappointing if it doesn't work!

[Badstarr]

I suppose the higher quality the cable the less resistance so longer runs can be used without as much loss in signal quality. However, I have a cable that I use to extend the MegaFlash cable, this is what I used initially to connect the SF with my Plus machines, I used the same cable with the edge connector cable on my classic CPCs essentially creating a cable over a foot and a half long and the Symbiface worked just fine.

I'm growing more certain that this is something to do with the Plus design. Now the weird thing is that when I put together my as yet unfinished GX4000 mod, I used basically the same cable, and the length of wire on the expansion bus is at least a 1 foot long, I haven't tried the symbiface with that machine yet, but the current expansions all work without issue.

So far its a mystery!

[Bryce]

The resistance of the cable isn't the problem. The problem is interference. The strands of wire work like an antenna, so if your monitor is spewing out all sorts of unwanted signals and the wire "receives" them, then this can cause bits to be interpreted wrongly and cause a crash. Using the cable in front of a CRT is worse than in front of a modern LCD. The length of the cable is important because the longer it is, the more chance it has of receiving unwanted signals, although sometimes lengthening the cable will solve the problem. If the cable happens to "tuned" (exactly the right length) for the unwanted frequencies, then it will receive them.
I wouldn't recommend using anything more than 25cm and I wouldn't recommend running the cable parallel to the monitor.
And always make sure the expansion port contacts are clean. Oxidated terminals will cause more resistance than a metre of cable can.

Bryce.

[Badstarr]

Hi Bryce! I had considered the possibility of RF interference, but thought I would rule it out as the problem isn't replicated on my classic 6128 or 464. I use my machines with a LCD TV or a LCD VGA monitor so I would think that there is no interference there.

As I trust your hypothesis more than my own, what do you think could be causing this to happen on the Plus machines? Obviously the design of the Plus machines is quite different ie the ASIC and the Cartridge circuitry but nothing particularly leaps out at me as being a significant enough deviation from the classic CPC to cause the problem? Noisy PSU perhaps?

I have been wondering if the ASIC is a little more fussy about voltages on its inputs than the DIP Gate Arrays, if it is then RF interference could explain it, in other words, the signal to noise ratio is enough to trip up the ASIC but not a Standard GA? Dammit I wish I had a scope!

I suppose I will have to spend some time investigating ...

[gerald]

Interference, voltage, timing issue .... You cannot guess what is going on without looking at the signals. Yes, a scope would be cool. A logic analyser too.

The fact that all your Plus have the problem and not a regular CPC would hint either the Plus generally, the cable or the SymbifaceII is the problem.

It would be nice to know, from owner of the SymbifaceII who use it with  a Plus, what is the cable length they use, and if they faced any problem with it. If they use cable longer than yours without problem, we might consider that the problem is not the Plus.

If you have a multimeter, you can check that the SymbifaceII is properly powered : a DC measure will show if the 5V is correct, an AC measure will give a rough idea of the voltage stability. Stability measurement should be done with some activity on the Symbiface (ie RAM test or Rom programming).

Now you may have corner case SymbifaceII, testing a second one would help. Wasn't the last batch delayed because of new component problems ?

[Badstarr]

Hi. Gerald! I have a Logic Analyser, I have never found a decent enough scope in my price range as of this time, I should really try to find one.

I'm fairly certain the simbiface will be getting enough juice as my Plus is running from a 5v 2.3 Amp PSU, and the Symbiface is connected to a 12v 2 Amp PSU and the 2.5" laptop hard disk is connected to its own 5v feed.

I'm about to bust out the test equipment and see if everything is as it should be, I will come back to report my findings...

[Badstarr]

Well I've tried everything I can think of, voltages seem nice and stable no matter how much I stress the Symbiface. I've tried using all the different IDC cable I can find, really nothing changes. With longer cables, when I try a RAM test I get random errors, ROMs seem to send corrupt data. With a cable around 4 inches, I can boot SymbOS but the hard disk access is hit and miss.

So this must just be RF like Bryce said and the shorter cable tunes it out, perhaps the Plus machines just don't cope as well as their older siblings with RF interference.

I dunno, maybe I should just be thankful it works albeit, with a darned short connection?

[Bryce]

No, I wouldn't limit it to RF problems. There are many other things it could be. For a start, the timing could be different. Depending on how the buses are being driven, there could be differences in the timing, voltage levels, rise times, many things can go wrong. Example: The time between the Read signal going low and Data being present on the data bus is critcial. If the ASIC happens to be doing this even a few nanoseconds too late, then it could cause problems. Another problem is "rise times". This is the time it takes to go from 0 to 1 or vice versa. If the change is too slow it can cause problems and the length of the connection can also make this problem worse. To know exactly what it is, you really need to take measurements on the actual hardware. (@Gerald: Before you start picking at the accuracy of my example, it's meant as a very simplified description of the type of issues hardware buses can have ).

Bryce.

[TFM]

Ok, this looks like an hardware issue. But nevertheless let me ask you which software did you use?[nb]I mean which software didn't work?[/nb]

[Badstarr]

I had hoped that this would be a simple matter of RF interference, my understanding of digital circuits and timing in rather patchy in some ways. I can understand data sheet timing graphs, but basically just well enough to know if ICs will play nice when I'm upgrading some RAM or choosing an EPROM/FLASH IC for my adventures in CPC land.

I suspect that all theories are in play here, I experimented with shortening the IDE cable attached to the Simbiface and now I can increase the expansion bus cable now to around 7" and there are no issues using the card.

I don't know exactly how the Symbiface hardware works so understanding timings has that extra level of complexity, I'm assuming RAM and ROM are juggled by the CPLD so if there is a timing issue it would most likely be there (?).

I'm of the option at this stage that there is a hardware issue that is compounded by some RF signals upsetting things. Or the voltage output that has been assigned to the CPLD is just enough in ideal circumstances but not quite enough to communicate 100% to the Plus for some reason.

I'm thinking out loud here...

I know from my understanding of FPGA programming that it is possible to choose the logic levels in and out of a FPGA so if this is also the case with CPLD perhaps the levels are not quite enough? It's probably the sound engineer in me, but I really do think this is a signal to noise ratio issue, too much noise, or too week a signal.

So far I have only used SymbOS with the Symbiface and run a memory test from BASIC. All my plus machines show the same behaviour, while my classic CPCs run great with the Symbiface even with stupidly long cables attached!

[gerald]

I experimented with shortening the IDE cable attached to the Simbiface and now I can increase the expansion bus cable now to around 7" and there are no issues using the card.

If we had a bit more information about the symbiface itself like schematic and pld content (source), it would be easier to guess the root cause of your problems.
But I do not think these are available.

Or the voltage output that has been assigned to the CPLD is just enough in ideal circumstances but not quite enough to communicate 100% to the Plus for some reason.

Regarding the PLD IO, the xilinx PLD used works with IO at 5V or 3.3V (XC9572 from available picture, PC84 and PC44). Can you confirm that they are not XC9672XL, which have 3.3V or 2.5V IO (however, they tolerate 5V as input).

I know from my understanding of FPGA programming that it is possible to choose the logic levels in and out of a FPGA so if this is also the case with CPLD perhaps the levels are not quite enough? It's probably the sound engineer in me, but I really do think this is a signal to noise ratio issue, too much noise, or too week a signal.

Your trouble can also be caused by a setup or hold issue on one single signal.

[TFM]

So far I have only used SymbOS with the Symbiface and run a memory test from BASIC. All my plus machines show the same behaviour, while my classic CPCs run great with the Symbiface even with stupidly long cables attached!

Ok, so it would be interesting to run some tests to check the SF2 integrity specifically on the CPC Plus. Such test programs can be written. Can you run FutureOS or do you encounter problems with it too?[nb]That can already give information about critical things. Well, I can only approach the problem from the software side.[/nb]

[Bryce]

The original Symbifaces were all supplied with XC9572, not the XL type. I'm not even sure the XL version was released back then? The XL version definitely wouldn't work.

Bryce.

[TFM]

Dr. Zed did some adaptions to cope with the situation. It was verified working on a regular CPC. I don't know about 6128 Plus.

[Badstarr]

@TFM, I havent installed FutureOS yet (stop wagging your finger at me    ) But I will install it next time I switch the machine on (probably tonight at some point). Have you a ready to go diagnostic that can test the Symbiface through FutureOS?

[TFM]

Not really, but if we can circle the problem I could create a test, but for the Firmware then.


My question about FutureOS was for the following:
- If it don't crashes the ROM content is fine, and jumps can made from ROM to ROM
- If discs work, then there is no problem with "too less power" when drive motors start working
- DIRectories are stored in RAM, if you can display them (w/o errors) the RAM works (at least a bit).
- If you can use the SF2-mouse, then the mouse is fine.
- If it displays time and date, the RTC is fine.


Well... had no better idea in quick. But if I do a testing routine, it would be good to know for which features to test. And that's probably the hard part about it.

[Bryce]

- If discs work, then there is no problem with "too less power" when drive motors start working

The SF2 has it's own power supply, as does the CPC. The drive functionality wouldn't give any indication of power supply problems.

Bryce.

[TFM]

That's right, but one never knows ;-)[nb]At lest not me, there may / could be some interference??[/nb]

[Bryce]

If there's a problem that's serious enough to pull both supplies below 5V then you'll know about it. It'll be the bit that the flames are coming out of!

Bryce.

[TFM]

Flames? They depend more on Ampere than on Volt. 

[Bryce]

Exactly, to get the output of both supplies to drop below 5V, the failure would need to be pulling Amps to the sum of what both supplies can deliver. Probably more than 6A! This would definitely be causing some "rapid oxidation" somewhere on the board.

Bryce.

[TFM]

Oh man! And I thought the AC would be the most draining. 


Well, so let's forget about power supply issues for the moment [nb]As long as his SF2 has an appropriate PSU.[/nb]