Hardware Design and Development

[Penri]

Terry

The whole question of position 4 not working when all the at the screen are correct just doesn’t stack up but if that’s what happens then reality has to be faced. If the new layout works with shorter lead length then the physical layouts in the model building are going to be critical.
Good luck is the new layout.

[TerryJC]

Penri,

I agree, but I don't see any other explanation at the moment. I would suspect that my soldering / build to be faulty except that the each signal is identical when viewed on the scope at each position. The new layout will shorten the track lengths by a few cms, but it will also reduce the amount of track that is parallel to other, perhaps sensitive, tracks. If that is the problem, then lead lengths will not be such a problem if the wires are loosely laid.

I'll keep you posted.

[hamishmb]

Not seeing anything wrong software-wise.

[TerryJC]

I'm over halfway through the build of the new device. I just discovered that I appear to have run out (or lost) my stock of microUSB connectors, so I won't be finished for a day or two.

Whatever happens we will gain something from the exercise. If the fourth Position bursts into life, then there is something wrong with the build of the previous version. If it doesn't, then we simply can only drive three Screens.

[TerryJC]

I just discovered that I appear to have run out (or lost) my stock of microUSB connectors, so I won't be finished for a day or two.

I found them again But not before I'd ordered another batch from eBay

Whatever happens we will gain something from the exercise. If the fourth Position bursts into life, then there is something wrong with the build of the previous version. If it doesn't, then we simply can only drive three Screens.

The fourth Position still doesn't work and what's more, Position 3 doesn't work when all four Screens are connected, although it works when it is the only Screen connected.

I'm definitely leaning towards line length and / or loading now. My new layout (see attached) shortens the length of the connected tracks, but lengthens the wiring between the Pi and the tracks. This configuration seems to have made matters worse, but I can't quite see why since the signals still look OK on the scope. I'll do a bit more investigation tomorrow, but I may have to admit defeat. Maybe smaller caps across the terminating resistors might work.

Otherwise I think that we might get away with three Screens connected, but I think four is going to be a bit of a stretch.

[TerryJC]

This is beginning to look like a hiding to nothing. I've spent the morning trying different configurations, but haven't really got much further forward.

Here is the current situation. I've removed the two 220 pF capacitors across the terminating resistors, because they don't appear to make much difference. I then tried one, two, three and four Screens in various positions with the following results:

• No Screen works in Position 4, even on it's own.
• If I have three Screens connected at Positions 1, 2 and 3. only two work:
  • Position 1 always works.
  • With two Screens connected at Positions 1, 2 or 3, they both work.
  • With three Screens connected only two work.
  • If I swap the Screens at Positions 2 and 3, the non-working Screen changes position.
  • The non-working Screen works if connected with only one other Screen.

I've monitored all the signals again and they all seem identical at each position. If I monitor a given signal at the Veroboard connector on Channel 1 and also at the Screen, there is no sign of loss of amplitude of synchronisation. However, the overshoot increases a bit (approx 5% to approx 7%).

So this third prototype seems to be less functional than the second one, despite all the problems I had soldering that one until I got a new soldering iron tip. Neither prototype seems to support a fourth Screen.

Any other thoughts?

[TerryJC]

Another day, another set of results

This morning I booted up the system in the same configuration that I had left when I finished testing yesterday afternoon. Then, I could only get two out of three Screens to work in Positions 1, 2 and 3 with nothing connected to Position 4. This morning all three Screens initialised and worked perfectly. To be on the safe side, I rebooted and again, all three Screens worked.

Next, I shut down and connected a fourth Screen to Position 4. Only Screens 1 and 2 worked. I shut down again and rebooted; only 1 and 2 worked...

Clearly, these Screens are working on the edge of their capabilities. I wonder whether we could utilise a bus buffer such as the CD74HC368E? I don't actually see why it would work, but at least we would be certain that there is sufficient drive to the ganged connectors and they would be nearer the Screens.

Thoughts?

[TerryJC]

I wonder whether we could utilise a bus buffer such as the CD74HC368E?

That would have to be the CD74HC367E, since the CD74HC368E is inverting.

They are pretty cheap from China and pretty expensive from UK (factor of 5), unless we pay the handling + VAT that RS now applies.

[Penri]

Terry

I apologies for not replying earlier, I've had a few busy days and didn't make time to check the forum. Anyway if it wasn't you I would not believe a word of the goings on with the screens, they appear to defy all logic.

Anyway I think your suggestion of a buffer is sound but given your comments on the signals I don't see why it's needed.

On the sourcing issue I'm happy that WMT pick up the bill to "guarantee" that the part is trustworthy.


Penri

[TerryJC]

Order placed.