Mk14 vdu

[Mark1960]

#759 and #760 make more sense.

#759 shows a delay from NENIN to the spike from the 74ls27, which would be expected due to internal delay of the SCMP releasing the bus.

#760 shows one chip enable stays low, but the other switches from high to low at about the same point as NWDS switches from low to high, and timing between them appears to cause the spike.

Pullup on NWDS might make it rise a bit faster, but I think you already tried that.

Another possible fix is to slow down the chip enable, maybe a small capacitor on the chip enable. Something in the 47 to 150 pf range, which shouldn’t impact normal access time.

[SiriusHardware]

About that: The VDU is read-only, so it has no connection to NWDS, only to NRDS.

The spike indicates, or is supposed to indicate, when _CE1 and _CE2 and _NWDS are all low.

If _NWDS is low during NENIN, why is that? The SC/MP is (supposedly) off the bus when NENIN is high, and the VDU doesn't have any way to drive NWDS.

[Mark1960]

Originally Posted by SiriusHardware

About that: The VDU is read-only, so it has no connection to NWDS, only to NRDS.

The spike indicates, or is supposed to indicate, when _CE1 and _CE2 and _NWDS are all low.

If _NWDS is low during NENIN, why is that? The SC/MP is (supposedly) off the bus when NENIN is high, and the VDU doesn't have any way to drive NWDS.

I think what’s happening is that the SCMP is performing a write cycle and is kicked off the bus with NENIN, and during the kicking off the NWDS stays low a little longer, while the address lines change faster and address the other ram chip.

[SiriusHardware]

I finally recovered my power supply, so taking another look again. The first thing I tried was three schottky diodes pointing away from _NWDS and 0Fxx _CE1 and _CE2 with the diodes all joined at the 'away' end and taken to 0V via a 27K resistor.

The output from that is unusable, I'll leave it at that.

So then, poking around with just the MK14 and the bridge PCB and the OrtonView (with 'optimised' firmware FW523/FW532) chained together I observed, as Tim often has, that if I just leave that combination running, the third 7-segment digit in from the RHS eventually changes to a reverse 'C'. Unless we have two different problems going on, this is more than likely another symptom arising from the same problem which causes stray characters to be written to the RAM.

So maybe this is the reliable 'trigger event' we can use, the moment when one or all of segments a-d of the third display cell light up. Trigger from that, and examine what is happening at the same instant on A8-A11, NWDS and NENIN in turn. Repeat the observations on each line several times so we can see if each line is always doing the exact same thing at the moment when the extra display cell lights up.

Of course this is complicated by the fact that the display is multiplexed so we need to look for at least one of the segment drive outputs to the third display cell being high at the same time as the corresponding LED common cathode goes low. I invite suggestions as to the simplest way to detect that and derive some sort of output which can trigger one channel of the scope.

[Mark1960]

I think the way the MK14 display works is that the display is driven from data stored in RAM, so this is probably another example of RAM corruption.

I don’t think triggering from the display driver could help to isolate the problem in this case.

I’ll take a look through scios to confirm.

[Timbucus]

Well you know I said I ordered some clips... they may have come with a new toy...





I have been experimenting with what things are worth looking at and may contribute to pinning down the issues. OD on F00 and B00 are the same and also the same as NRDS it seems - looking at the Circuit diagram and some samples that is correct so no need to capture them separately.

Any suggestions welcome...

[SiriusHardware]

I think you're probably right, SCIOS must render the display from a RAM buffer, because once it starts to display the reverse 'C' it continues to display the reverse 'C', implying therefore that it is the display RAM buffer which has been written to.

Maybe then something like a 74LS154 with A8-A11 taken to its a, b, c, d inputs and enable G1 (active low) taken to _NRDS - this will generate an output pulse (low) on output 15 whenever there is a write to 0Fxx. If G2 (also active low) were taken to NENIN via an inverter, that would produce an output pulse only when there was a write to 0Fxx when NENIN was high.

Looking for writes to other address blocks would just be matter of monitoring one of the other 15 outputs from the IC.

[SiriusHardware]

I think the way the MK14 display works is that the display is driven from data stored in RAM

I think you're probably right, SCIOS must render the display from a RAM buffer, because once it starts to display the reverse 'C' it continues to display the reverse 'C', implying therefore that it is the display RAM buffer which has been written to.

Maybe then something like a 74154 with A8-A11 taken to its a, b, c, d inputs and enable G1 (active low) taken to _NRDS - this will generate an output pulse (low) on output 15 whenever there is a write to 0Fxx. If G2 (also active low) were taken to NENIN via an inverter, that would produce an output pulse only when there was a write to 0Fxx when NENIN was high.

Looking for writes to other address blocks would just be matter of monitoring one of the other 15 outputs from the IC.

[SiriusHardware]

Nice toy Tim, how sophisticated can the setup be? Can you set complex trigger circumstances, ie, trigger when NENIN high and _NWDS low, that kind of thing?

[Timbucus]

Originally Posted by SiriusHardware

Nice toy Tim, how sophisticated can the setup be? Can you set complex trigger circumstances, ie, trigger when NENIN high and _NWDS low, that kind of thing?

Not that I have found with the supplied software as yet - but, once I get it working with PulseView yes you can I think.

[Slothie]

Originally Posted by Timbucus

Not that I have found with the supplied software as yet - but, once I get it working with PulseView yes you can I think.

Is that the logic viewer in the Sigrok project? I used that a while back and it was pretty good.

[Mark1960]

SCIOS is using F00 to F08 to store the segment settings for the display, which is then output to the display during keyboard scanning. F00 is the rightmost digit 1 and F08 is the leftmost digit 9.

Yes that is nine digits. If you have a nine digit display on your MK14 and connect the left most digit common to the 7445 then you can use all nine digits. I think the later version SCIOS does not initialise F08 so my MK14 displays rubbish in that digit on power up. Looking at the code for the first version SCIOS it does seem to initialise the leftmost digit to zero, all segments off, so it seems that at some point the intro kit and keyboard kit from nat semi might of been expected to use a nine digit display.

[Timbucus]

Originally Posted by Slothie

Is that the logic viewer in the Sigrok project? I used that a while back and it was pretty good.

Yes it is part of Sigrok but, it does not recognize my device at the moment... for £24 it is a bargain - the clips are the largest cost at £16 so I am sure I can get it working...

[Timbucus]

Originally Posted by Mark1960

SCIOS is using F00 to F08 to store the segment settings for the display, which is then output to the display during keyboard scanning. F00 is the rightmost digit 1 and F08 is the leftmost digit 9.

Yes that is nine digits. If you have a nine digit display on your MK14 and connect the left most digit common to the 7445 then you can use all nine digits. I think the later version SCIOS does not initialise F08 so my MK14 displays rubbish in that digit on power up. Looking at the code for the first version SCIOS it does seem to initialise the leftmost digit to zero, all segments off, so it seems that at some point the intro kit and keyboard kit from nat semi might of been expected to use a nine digit display.

That would tie in with random write to a portion of F00 in this case F03 gets the reverse C

[SiriusHardware]

F02, probably.

0F00, 0F01, 0F02... third digit in from the right. How many inputs does that gizmo have, just the eight? Are they available with more channels?

[Mark1960]

Originally Posted by Timbucus

Any suggestions welcome...

Remember those logic analysers only show high or low, so depending on the switching threshold they can show different logic level than that detected by the inputs of your circuit under test. Also a chance they could miss some of those runt glitches.

I thought you meant one of those DIL test clips that can clip on the top of a chip and make it easier to connect a scope. I have a 24 pin and 16 pin in my toolbox from many many years ago, still gets occasional use.

[SiriusHardware]

Incidentally the Pickit 2 (with very late versions of the standalone support software) can be used as a three channel logic capture device but I think two of the inputs have fairly strong pullups, internal to the Pickit. I found it useful in the past for capturing the output stream from remote controls, that kind of thing.

Also, again with another feature of very late versions of the support software for the Pickit 2, two of the inputs of the unit can be used to 'watch' RS232 data in both directions by connecting them to the RX and TX lines at TTL level.

[SiriusHardware]

Originally Posted by Mark1960

I thought you meant one of those DIL test clips that can clip on the top of a chip and make it easier to connect a scope.

That was what I had in mind. You can still get them but they are pretty expensive and I couldn't find one in 18-pin DIP (I think there was a 20-pin one).

[Timbucus]

Originally Posted by Mark1960

Remember those logic analysers only show high or low, so depending on the switching threshold they can show different logic level than that detected by the inputs of your circuit under test. Also a chance they could miss some of those runt glitches.

I thought you meant one of those DIL test clips that can clip on the top of a chip and make it easier to connect a scope. I have a 24 pin and 16 pin in my toolbox from many many years ago, still gets occasional use.

Indeed that lack of subtle was the reason I did not buy one in the first place when I went for the scope. I did indeed go looking for the DIL clips but, even Farnell has them on back order and they are just not available but, I found the little clip set in the photo - I have been using them with the scope as well and they are great at clinging to the individual pins.

Of course the Amazon 'frequently bought together' got me and led me the the £8 AZ-Delivery device... and I thought what the hell...

It may help anyway - with it fully hooked up though I do not get the corrupt character on CHARSET in F27 under the M of TIMBUCUS - the extra capacitance and effect of the measurement tool means no effect to see...

Reassuringly if I switch in the 10K the two corrupt chars come back and a third one appears a bit further down the B block - I will hazard a guess that is a moved version of the F27 corruption and the lengthening of the NWDS pullup is still long enough (despite the extra connection) to give rise the the phantom write into both chips.

[Timbucus]

Also got to Level 7 on MINEFIELD so it takes away the crash on that as well.

I do still see the Reverse C in F02 as Sirius correctly pointed out - not long after resetting the machine