Recommended maintance & MDV ULA Problem

[Martin_Head]

A couple of months ago. I had to dig out my old QL's, to look at a QL network problem. And of course none of them was working correctly any more.
All of them of them having at least one problem, And I have now got around to trying to fix them.

I am replacing all the network sockets, as they are all worn out. And I am thinking of replacing the electrolytic capacitor across the reset switch, as some of the QL's don't always start at power up.

Are there any other electroyltic capacitors, or other components I should think about replacing at the same time?

On one of the QL's, I have only one microdrive unit fitted, as a long time ago I had a problem that mdv1 made mdv2 run all the time. I moved mdv2 to into the mdv1 place and left the old mdv1 out. I'm now having a look at repairing the removed microdrive unit.

I suspect the ULA, as I am having some different measurements with a multimeter on the diode range, when comparing both drives.
I get different readings on pins 20 and 22 (COMMS OUT/IN) of the ULA.

I also have some odd readings on these pins on another pair of microdrives units which I thought were OK. So now I am getting a bit confused, as to what is right and wrong.

Is there a known problem with the ULA's?

I notice that the ULA's seem to be a bit thin on the ground now. So a repair may not be possible.

[martyn_hill]

Hi Martin!

I've never had a problem with RESET that required that Cap to be replaced - when I have hitn problems with reset, it was either down to the ZX8302 or something more fundamental with the main ZX8301, but replacing the Cap is a simple and cheap enough operation to eliminate it.

I am in the same process of replacing the Net sockets on some of my QLs which intermittently seem to interfere with communications between other stations when connected. To match the existing PCB footprint, I found the Cliff Electronics CL1382 available from Farnell which, although having a fourth leg, this can either be bent-out of the way or removed fully from the socket.

As for the MDV ULA - I have replaced a few now to rejuvenate dead or misbehaving units successfully, though it's not a job for the faint of heart.

I pick-up the MDV ULAs from time to time from eBay and SellMyRetro, but they are certainly getting more scarce.

That said, others here have fixed similar symptoms (constant running MDV, failure to format etc) by replacing either or both of the transistors (PNP and NPN) used in the Motor power and Erase ccts (TR4 & TR5, IIRC - but do check) on the main QL board.

Replacing the linear Voltage Regulator on the head-board (78m05) has also brought some unis back to life.

Can you share the unusual traces on COMMS-IN/OUT? Perhaps we can make sense of them together...

[Martin_Head]

Hi Martin!

I've never had a problem with RESET that required that Cap to be replaced - when I have hitn problems with reset, it was either down to the ZX8302 or something more fundamental with the main ZX8301, but replacing the Cap is a simple and cheap enough operation to eliminate it.

I've noticed that on one or two of the QL's, when I power up, The QL is inclined to not start/displays odd patterns etc. Pressing the reset switch fixes it. I thought the caps might be deteriorating with age.

I am in the same process of replacing the Net sockets on some of my QLs which intermittently seem to interfere with communications between other stations when connected. To match the existing PCB footprint, I found the Cliff Electronics CL1382 available from Farnell which, although having a fourth leg, this can either be bent-out of the way or removed fully from the socket.

I'm using the same one's from an Ebay seller. I'm sticking a small piece of insulation tape on the board under the cut off leg on one of the sockets as is very close to a track.

Can you share the unusual traces on COMMS-IN/OUT? Perhaps we can make sense of them together...

I need to do some more investigations. I'm not sure if the MDV1 unit I think is faulty, is making the MDV2 drive run. Or if the MDV2 unit is running by itself.

I've not quite figured out how the drive selection works. The manual talks about the MDSELCKN signal driving a shift register. But the same clock signal goes to all the microdrive units. How does the MDV unit know which one it is?

Looking at the MDSELCKN signal with a scope does not show much. All I could see was a very, and I mean very short pulse. How big should the pulse/pulses be? Maybe the ZX8302 is in trouble.

I need to make some measurement with a scope on a 'working' QL. And try swapping MDV units about, to test them individually.

I've also got other problems to try to sort out. One QL has a very noisy UHF modulator. Another QL built into a case has no UHF output, which I think is due to the built in switch mode power supply losing it's +12V line.

And I've also got a faulty Super Gold Card. Poking about with a scope, nothing seems to shout out 'I'm faulty'. I've got no circuit diagram, and I'm tempted to just change all the 74 chips, If you can still get them. Some time ago, I looked in RS and Farnell, and I think some were available, but I would have to buy in quantity.

[martyn_hill]

Hi Martin!

No harm in replacing the Cap in the reset cct, I'm sure.

If the problem goes away after a further reset, it could be the 15MHz oscillator driving the ZX8302 ageing and not stabilising 'quickly enough.' But that's just speculation.

As for the method used to select each MDV unit, I think I can explain that:

As each MDV ULA includes part of a shift register, when MDSELN is high at the point of a positive going MDVCLK hitting that unit, the ULA sets it's 'outbound' MDSELN high that feeds both its own Power/Erase cct as well as the next unit (physically) in the chain - you'll notice that only MDVCLK is common to all units - the input to each MDSELN is only tied to the preceeding ULA's output MDSELN.

By arranging in SW to set the MDSELN high briefly, sending a MDVCLK 'pulse' then setting MDSELN low again, the first unit is selected.

However, if you continue to send pulses down MDVCLK, that MDSELN signal effectively propagates down the chain, one unit per each clock pulse.

EDIT: Whilst sending the second and subsequent required number of Clk pulses to reach the desired unit, you leave the MDSELN line low.

You may notice that, when selecting a 'higher' MDV unit, the LED of the 'lower' ones glows oh-so-briefly - in fact to reach say MDV3, both MDVs 1 & 2 both get activated very briefly. Not long enough to start to spin their own motors.

The Minerva source - or rather Lau's superb commentary - is a great document to study to understand this mechanism.

EDIT #2: It was Lau's thorough analysis that enabled me, quite 'easily' to port the first parts of the MDV driver to my 'Frankenstein's Monster' Q68-in-a-QL-Case-with-MDV project. Sufficient so far to read MDVs with the great help of a slightly modified version of Danielle's MDump v2 code... But now I'm just showing-off...

I don't recall the precise value, but I believe the 'standard' mandates at least a 25ms pulse on MDVCLK to be recognised by the respective ULA. Lau documents that in the Minny source. (in the 'md' sub folder.)

Good luck!

[Martin_Head]

Thanks for the description. I've got to sit down and do testing on a working unit, to see what I should expect.

I've been looking at these COMMS IN/OUT signals, as I found during cold testing with a meter, different readings on these pins on the two microdrive units in the faulty QL. However I also get different readings on the two microdrive units in another supposed good QL.

So I don't know if I have two faulty microdrive units, or if the meter is trigging the internal driver transistors in the ULA. Maybe they are FET's?

[martyn_hill]

Hi again Martin

Not sure how a multimeter would.ipact your readings, but I regularly use my simple/cheapo USB digital analyser with PulseView software to monitor those signals.

Monitoring the Erase and WP lines requires a bit of circuitry to prepare the signals for capture, but the 'sort of digital' lines like COMMS-IN/OUT (aka MDSELN in some literature), MDVCLK and the 2 RAW data lines is easy enough.

I can share some sample traces (or images thereof) if that'd help? Just PM if so.

[martyn_hill]

Hi Martin

For what its worth, attached here is an image from an old capture-trace I took of the MDV signal lines at the very start of a 'DIR mdv3_', as seen at the MDV-Extension port. MDV3 was a ZX unit, attached externally to the QL, as you'd imagine.

A few points:
a) I was wrong about the edge used in the MDV Clk signal - we see in the trace that the MDV ULA detects the MDSELN signal (aka COMMS-IN) on the falling Clk edge, not rising as I wrote previously.
b) You can see in the trace that the MDSELN (inbound) to MDV3 - which is actually the outbound MDSELN from MDV2 - is activated after the second Clk pulse - as it passes out of the 2nd MDV ULA. Only then following the 3rd falling Clk is it activated on MDV3's MDSELN outbound/COMMS-OUT - where it remains in-force as there are no further Clk pulses. It is MDSelDn (out) - as titled in the image - that drives the Motor/Erase (and WP) to MDV3 via it's own pair of PNP/NPN transistors.
c) The Clk pulse does indeed appear to be in the 25 us range (oops! I did say 25 ms before - meant 25 us!)

Good luck!

[Martin_Head]

Hi again Martin

Not sure how a multimeter would.ipact your readings, but I regularly use my simple/cheapo USB digital analyser with PulseView software to monitor those signals.

There's something at the back of my memory about how FET's can be turned on during cold testing with a meter. But I might be wrong. It's been a while since repairing electronics was my day job.
I will have to point my good old Avo 8 at it, rather than a digital meter.