Electron unstable/fails to boot when Plus 1 is connected

[myelin]

After experimenting a lot last night with my flash cartridge, my Electron started getting really unstable. It would fail to boot (blank screen, or hang partway through printing the "Acorn Electron / BASIC" banner) or reset or hang randomly after startup. Unplugging my homemade PCB didn't fix it, but after leaving the machine off for a while, it seemed a bit better, although this morning the problem is still there. The ULA was quite hot to the touch the first time the instability occurred.

Removing the Plus 1 and starting the Electron up with nothing connected, it seems to run just fine (I can *LOAD a 16kB file from tape without a reboot). Plugging in the Plus 1 brings back the instability, whether my homemade PCB is connected or not. I tried removing the Plus 1's EPROM, but that didn't appear to change anything.

The power supply rails are within ~0.05V of +/- 5V. I haven't tried putting a scope on the reset line. There's some corrosion on the keyboard connector, but the Plus 1 and the expansion edge connector both look clean.

My gut feeling is that either I've fried something in the Plus 1, or the Electron's power supply can't handle the extra load (although I can't imagine the Plus 1 draws much current without the EPROM and with no cartridges plugged in).

I just had the idea of simulating extra load by putting a resistor between 5V and GND on the Electron; I'll give that a go when I get home tonight. I also have a logic analyzer (and a couple of PaulB's breakout cartridges), so I'll see about taking a look at the /RESET line to see if anything odd is happening there.

Any suggestions for other things to poke at?

[danielj]

Is it screwed together? My gut feeling is bad connection on the edge connector... Do try cleaning it with IPA.

d.

[myelin]

Is it screwed together? My gut feeling is bad connection on the edge connector... Do try cleaning it with IPA.

That could be it... this Electron was missing the plastic protector on the edge connector when it arrived, so it's probably been well used. Also the main screws were missing so I haven't had it properly fixed to the Plus 1.

[danielj]

The screws do help. My plus 3 is badly behaved minus the screws to hold it all nice and tight.

d.

[Coeus]

The screws do help. My plus 3 is badly behaved minus the screws to hold it all nice and tight.

I don't know the electron but I had a XZ81 which used an edge connector for the 16K RAM pack. That was a big source of unreliability as the SInclair design relied on the edge connector for mechanical support as well as electrical connection and that isn't what edge connectors are designed for. The IDC connectors on the BBC micro with proper pins were a big relief.

So, if the screws are what keep the connection between the Elecron and +1 stable then I'd fit them.

[richardtoohey]

I go for the connector as well - not had a huge amount of experience here but cleaning the connector and getting some screws did seem to help.

Think they are M4s (or something like that; I asked on the forums and some kindly soul told me.)

EDIT: this is the thread viewtopic.php?f=3&t=6546&p=63725 and it's M6

[1024MAK]

What happens with edge-connectors that are not firmly fixed is the contacts move as a result of vibration and movement of the computer or interface. As the contacts move, sometimes one or more may momentarily experience greater contact resistance (either open circuit or just a high resistance connecton). In the interface, if this is a control line or address line, it may change state (think of the effect of an address line floating high when it is suppose to be low...) causing the interface to function incorrectly. Maybe by putting data onto the databus when it should not, causing a bus clash. Or if the power supply or ground (0V) connections experience problems, all manor of strange things may happen...

The normal application of edge-connections is for a backplane in card rack systems. Here, in their intended application they are very reliable.

Mark

[myelin]

I cleaned the contacts with IPA, and screwed the Plus 1 to the Electron with two M6 x 30mm screws, but it didn't appear to help at all. I noticed some dirt or corrosion on four of the contacts, that wouldn't clean off, so that could be making things a bit flaky. I wonder if all the wiggling has loosened the springs in the connector in the Plus 1 to the point that they don't make good connections any more.

Didn't have time to do any more than that last night, but the plan of attack now is:

- Tin all the contacts on the Electron side with solder, to make them slightly thicker and provide a better contact surface.

- If that doesn't work, replace the connector on the Plus 1 side.

Also working on a PCB for a simple Electron expansion that will give two (maybe three) cartridge slots, and I have a board on the way from China that'll let me hook up an FPGA board to the expansion connector, so these will give me some better things to test with to see where the problem lies!

[danielj]

Before you tin gold plated contacts, can you check continuity from the ICs in the plus 1 through to the elk?

Schematic: viewtopic.php?t=10413#p125767

I did end up having to replace the edge connector on my plus 1...

d.

[daveejhitchins]

- Tin all the contacts on the Electron side with solder, to make them slightly thicker and provide a better contact surface.

NO! Please don't do that - you'll just make it far worse, in the long run! The contacts, in the Plus 1 edge connector, are so designed that you can't stress them! (OK - they may be damaged by other methods > DanielJ ) - there are physical 'stops' to prevent excessive bending. If you have some corrosion on the Electron edge connector, try a fiber pencil. If you don't have one > invest! They're invaluable for jobs like connectors - edge and ULA connectors will clean up nicely with one of these pens. Also get hold of some contact cleaner to spray the Plus 1 connector. Another method I use, for the electron edge connector, is one of the soft 'plastic' rubbers (erasers - for our overseas members).


Dave H

[1024MAK]

Dave means one of these fibre pencils.
As Dave says, great for cleaning all manner of electrical contacts. They should be available from any good electronic supplier.

Mark

[myelin]

Thanks for the advice! Fiber pencil and contact cleaner ordered

[danielj]

I had a bent connector pin Fortunately I also had a load of those right angle connectors!

d.

[myelin]

Going through a lot of my old hardware right now to get it ready for exhibition at VCF West with 8bitkick and Lion, and the key thing to get going is my Plus 1... time to take another look at it

I tried the fibre pencil approach when Dave suggested it (over a year ago now!) but it's still flaky: I plugged it in to an Electron for the first time in a while and it booted fine, but after plugging in the MGC, all I get is a fixed but glitchy looking red/yellow pattern on screen. Test #1 is going to be to record what's going on with an FX2 board connected to a cartridge socket, followed by opening the thing up and checking continuity on all the pins, to see if the connector needs replacing.

[myelin]

Figured out part of what's going on... scoping up the reset pin, it's glitching low in sync with the rising edge of PHI0:

I realized that I can get the same red/yellow pattern on screen by holding down BREAK on startup, so this confirms that the problem is reset related.

I first suspected ground bounce, but switching to analog, it looks like something's pulling the reset line down to ~1.2V:

[daveejhitchins]

If you haven't found the Plus 1 schematic, it's attached. The reset line goes to 4 x ICs! Could be a short or a faulty IC i.e. one of the input's doing the pulling-down . . .

0119,100-C Electron Plus 1 PCB Circuit Diagram.jpg.zip

(2.73 MiB) Downloaded 20 times

Dave H

[myelin]

If you haven't found the Plus 1 schematic, it's attached. The reset line goes to 4 x ICs! Could be a short or a faulty IC i.e. one of the input's doing the pulling-down . . .

Yep... one for the bank select latch, and three for the parallel port (main buffer, status input, and strobe output).

I'm guessing a faulty IC, because a short to anything actually driven should pull it all the way down to ground, I would imagine. Pity I don't have any 14-pin sockets to play with! Guess it's desoldering time...

[hoglet]

I'm guessing a faulty IC, because a short to anything actually driven should pull it all the way down to ground, I would imagine. Pity I don't have any 14-pin sockets to play with! Guess it's desoldering time...

Usually with this sort of issue, there is a low resistance path to ground inside the failed chip.

Start by measuring the resistance with the Plus one disconnected. It's probably a few hundred ohms. (Try the meter lead both ways around).

If you do have a low resistance path, pick one of the four chips at random (the hottest if there is any difference), and snip it's RST pin close to the board, then bend it out to break the connection.

Re-measure (both the isolated and remaining chips) and repeat until failed chip identified.

It should be possible to bend the non-failed pins back, and bridge with solder.

Dave

[myelin]

Usually with this sort of issue, there is a low resistance path to ground inside the failed chip.

Start by measuring the resistance with the Plus one disconnected. It's probably a few hundred ohms. (Try the meter lead both ways around).

Good thinking! My meter measured a total open circuit in one direction, and about 7Mohm in the other, so I guess that's not the case.

Powering up the Plus 1 from a lab power supply at 5V (using alligator clips on the electrolytic capacitor's leads), the unit draws 0.15A, and the reset line floats at 4.25V.

I've checked for shorts between pins on the expansion connector, and there don't seem to be any, whether it's plugged in to the Electron or disconnected.

With the Plus 1 disconnected and unpowered, there doesn't seem to be a low resistance path to reset from any of the other pins of those four chips, and the resistance to ground is > 18Mohm for all pins on the expansion connector.

So... all the sanity checks look pretty good so far! I guess the next step is to repeat the resistance check with it connected to the Electron, and then probably to scope up all the chip enables for anything connected to the data bus, to see if something's causing a clash there.

[myelin]

Gave up and just started snipping pins... finally found that IC6 was the culprit. Snipping its pin 1 made the machine boot again, although the Plus 1 ROM and MGC were invisible. Scoping IC6 pin 1 showed that it was sitting around 0.1V, which would have kept the bank select register cleared. Tying it to 5V with a wire and some alligator clips brought it to life

Thanks for the diagnosis technique, hoglet!

[hoglet]

Glad you got it sorted.

Dave

[1024MAK]

I have found with some failures of 74LSxxx logic chip, that input pins can become partly electrically connected to output pins. Of course, it's not easy to tell what else is going on inside the package. So if you do have a situation where you think a chip input pin is causing a problem, as well as testing the resistance between the suspect pin an both Vcc and GND, also test to all the other pins.

One further point, most modern digital multimeters use a low test voltage on the resistance ranges so that (silicon) semiconductor junctions don't turn on and distort the results. But when testing on a suspect or failed chip, you may well want to turn on any structure that has due to a failure become a PN junction. So also test on the diode test range.

Alternatively, use a +5V or +3.3V test supply and feed this to the device under test via a 4.7K or 10K resistor with a multimeter in series set to the 2mA / 2000uA (or equivalent) range.

Mark