Right, so Ozpropdev and I have learned a fair bit this morning, and its prob good news for you Rjo.
By default, my 123A7 board operates a little differently to Oz's, and looked to have the same problem Rjo's 2 boards showed.
It seems doing chip's operation of pushing 1 kohmed ground wire under ball A1 makes everything work, including the ws2812b leds, which are timing sensitive. Furthermore things continue to work for at least several seconds after removal of the PLL pull down, but after that few seconds you can see the timing go erratic, and the ws2812b's go flaky.
With such a high impedance, that time constant is probably dependent on surface wash residue etc, which would explain why it can go either way. I'm guessing some 123A7's naturally pull down, but most probably pull up, due to more vcc than gnd pins in the area.
Good to progress the understanding of what's going on. And good work putting the flaky pin in the corner rather than other 480 harder to get to pins!!
Edit: 3 neighbors are gnd, there goes that theory. Perhaps pin A1 is influenced by the plane immediately below top layer, or the nConfig track running past
I know this is kind of a late suggestion, but if A1 needs a pull-down and it's surrounded by ground balls. Could you just use a syringe to inject some paste loaded with carbon? later comments show that the 2k 1% spec value is actually more like 2k 200%.
Yeah, I suspect you could make that work, its just a bit scary from a non reversible point of view. At least you can pull a resistor lead/wire back out again
I have a sample of "conductive" 3d printer filament, presumably loaded with carbon black, which would probably do it, too
Yeah, I suspect you could make that work, its just a bit scary from a non reversible point of view. At least you can pull a resistor lead/wire back out again
Yes, relying on only conductive paste is a lottery at best, but a hybrid solution could work :
Put a small dab of conductive ink on the 0603 resistor end, before it is pushed in, and solder it under slight push-pressure.
Here the conductive ink is small, and augments the contact rather than defines the resistance.
It is very easy to put a 603 in there. ok you have to wiggle it in... but after the first couple of wiggles you have it in place. I didn't treat the inserted end in any way. And the connection is very firm... so I didn't try to fix it in place either. Conductive ink or paste between the protruding end and a scratched-off area of the solder resist mask is all that is required. Really very easy...
I couldn't talk to my P123 very reliably before and have had no problems since.
I just got my P123 board today. It's very tiny, compared to DE2-115. Not much bigger than nano really.
This is good because DE2-115 thing is enormous...
Don't see any sign of A1 mod on mine either.
But, I can't even see the A1 ball.
I think I'll just live with 50 MHz and get the A9 when that comes out.
BTW: Does the 50 MHz limit apply with the P1 code too?
I just got my P123 board today. It's very tiny, compared to DE2-115. Not much bigger than nano really.
This is good because DE2-115 thing is enormous...
Don't see any sign of A1 mod on mine either.
But, I can't even see the A1 ball.
I think I'll just live with 50 MHz and get the A9 when that comes out.
BTW: Does the 50 MHz limit apply with the P1 code too?
Yes. Anything higher than 50MHz will need PLLs working. Yesterday I talked to David Carrier who is going to implement the fix on the -A7 boards.
I suppose this is too late, but I kinda wish the P123 board had a uSD card jack...
I am glad you've added a SDRAM chip though. In order to do full screen graphics in VGA more or higher is going to take SDRAM, I think...
Also, you might want to include some standoffs so the board is off the table, just a though. I had some but others might not...
It would also be nice if the 40-Pin header provided 5V power. I was looking at a 7" LCD panel (driven by SSD1963) that needs 5V power for backlight. Don't see 5V output anywhere...
I'm working on a serial terminal that will be able to save and load files. Right now with no tricks and at 50Mhz, it is chugging along at 3MBaud:)
And that includes a few extra clocks for the rdlut/writelut
Comments
I know this is kind of a late suggestion, but if A1 needs a pull-down and it's surrounded by ground balls. Could you just use a syringe to inject some paste loaded with carbon? later comments show that the 2k 1% spec value is actually more like 2k 200%.
Yeah, I suspect you could make that work, its just a bit scary from a non reversible point of view. At least you can pull a resistor lead/wire back out again
I have a sample of "conductive" 3d printer filament, presumably loaded with carbon black, which would probably do it, too
Yes, relying on only conductive paste is a lottery at best, but a hybrid solution could work :
Put a small dab of conductive ink on the 0603 resistor end, before it is pushed in, and solder it under slight push-pressure.
Here the conductive ink is small, and augments the contact rather than defines the resistance.
I couldn't talk to my P123 very reliably before and have had no problems since.
This is good because DE2-115 thing is enormous...
Don't see any sign of A1 mod on mine either.
But, I can't even see the A1 ball.
I think I'll just live with 50 MHz and get the A9 when that comes out.
BTW: Does the 50 MHz limit apply with the P1 code too?
Yes. Anything higher than 50MHz will need PLLs working. Yesterday I talked to David Carrier who is going to implement the fix on the -A7 boards.
I am glad you've added a SDRAM chip though. In order to do full screen graphics in VGA more or higher is going to take SDRAM, I think...
Also, you might want to include some standoffs so the board is off the table, just a though. I had some but others might not...
It would also be nice if the 40-Pin header provided 5V power. I was looking at a 7" LCD panel (driven by SSD1963) that needs 5V power for backlight. Don't see 5V output anywhere...
I'm working on a serial terminal that will be able to save and load files. Right now with no tricks and at 50Mhz, it is chugging along at 3MBaud:)
And that includes a few extra clocks for the rdlut/writelut
Rich