[danb35]

Well, so much for the smoking gun I thought I found--CS10 doesn't go through the level shifter, just through a 10k resistor. Hmmm...

[werk_ag]

CS10, though, is showing much slower rise/fall, ...

This is normal as the CS10 signal doesn't pass through the logic level converter, attenuation is done only by a resistor.

There is something strange with the voltage levels. As the Arduino Nano is powered by only the USB 5V, and there is always a small voltage drop (0.7V) at the schottky diode, how could you get a Vpp of around 6.24V?

Despite the voltage readings (you should check that) the pics show that the logic level converter is doing its work.

I'm inclined to say that the problem is with the cables or even with the display.

What voltage do you get at the 3.3V pin? This is the reference voltage for the output of the logic level converter, and it will not be higher than that value.

[danb35]

The peak-to-peak voltage includes the over- and under-shoot, but the traces show the right at least approximate voltage (i.e., 2.5 divisions on ch1, 2 volts/div = 5 volts). I haven't yet dug through my scope's menus to see if there's a different measurement for peak-to-peak that would ignore the overshoot.

The 3.3v pin measures 3.29 volts.

I'll work on the wires a bit, and I have a shorter set on the way.

On the display, I notice that U1 is not populated. That's consistent with pictures posted with the eBay listing, but I see some other listings do show a device there. Is there any chance that's relevant?

[werk_ag]

On the display, I notice that U1 is not populated. That's consistent with pictures posted with the eBay listing, but I see some other listings do show a device there. Is there any chance that's relevant?

U1 is only installed on models with the touch panel. Its absence will not cause problems.

The peak-to-peak voltage includes the over- and under-shoot, but the traces show the right at least approximate voltage (i.e., 2.5 divisions on ch1, 2 volts/div = 5 volts). I haven't yet dug through my scope's menus to see if there's a different measurement for peak-to-peak that would ignore the overshoot.

I gave a more closer look at the images from your oscilloscope.
If you ignore the over- and under shoot, the signal voltage for Channel 1 is around 4.8 V (seems good), and for Channel 2 is around 3.3V (seems good).

[danb35]

A few new scope traces, with more appropriate measurements.
MOSI:
   

SCK:
   

DC:
   

Observations: Amplitude measurements seem entirely reasonable. Overshoot on the 5V line ranges from 14.9% to 18.6% (0.7 to 0.87 volts). Overshoot on the 3.3V output ranges from 22.4% to 31.3% (0.80 to 1.03 volts). Overshoot is measuring only the low-to-high transition, not the high-to-low transition. Thus, overshoot is higher both in relative (percent of amplitude) and absolute (number of volts) terms on the output of the level shifter.

I'm not at all saying this is the problem--I don't know enough to suggest that. It just looks a bit anomalous to me.

[strada916]

Can you probe the SCK and MOSI on the same time? With out clock reference those wave forms mean nothing. In fact make sure the clock is on all of them so there is a reference.

[werk_ag]

Can you probe the SCK and MOSI on the same time? With out clock reference those wave forms mean nothing. In fact make sure the clock is on all of them so there is a reference.

Why should I do that? Sincerely I don't see the point. Why complicate things?

Of all the system, the Wireless Units are the most simple units to build, and they really work. There are already dozen of units working 24/7

I can admit the possibility that your WD unit be faulty, and in private I have already proposed you a solution. Have you ever considered the possibility that you have a faulty display? Are you absolutely sure that your display is working properly?
Believe me, the probability of getting a not working 2.8" display is around 10 to 15 percent, if you check carefully, you will see that the majority of the sellers sells those displays (2.8" version) as "New other" condition, which means something...

[uncle_bob]

I'm with Werk and also think it's time to start substituting parts.

[danb35]

I imagine the point asking for SCK/MOSI together is to see that the clock and data lines are in sync and such. Easy enough to do, but I'm not sure how informative that would be.

I'm open to any possibility, really. I hadn't particularly considered the possibility that the TFT could be faulty, but I guess it's just as likely as anything else (and I had no idea the DOA rate was that high). While I've been trying to figure this out, the behavior has changed--now it's just solid white (when I apply power, the screen turns on solid white for a second or so, then goes black for a second or so, then goes back to solid white and stays that way). I've tried a new set of wires, but I'm still waiting for a shorter set.

I've been focusing largely on the WD board because (1) I built it, can replace most of the parts on it if necessary, and may well have made a mistake, and (2) I have access to a resource (i.e., you) that knows how it's supposed to work and can help in troubleshooting. It's also the part that I believed I had the best chance of being able to fix quickly, rather than waiting for something else to ship from China. But with all of that said, I think it's looking increasingly unlikely that the problem lies there. I'll see if I can find a replacement on eBay from a vendor that can ship quickly.

[danb35]

Well, the shorter (10 cm) set of wires arrived rather more quickly than I'd expected, so I tried hooking up the display to the board with them. No change--I'm now getting just a solid white screen. I think the wires themselves can be reasonably excluded--I've tried two different sets at the 30 cm length, and one at the 10 cm length, with no change in results. I've ordered another display; hopefully it will get here soon.

I've taken a couple of traces with MOSI and SCL showing--MOSI is in yellow, SCL in blue. Here they are: