I am having an issue that the NRST pin stays at about 1.7v to 2v after power up. The DBG UART outputs ROMBOOT without the prompt (">"). It seems that the AT91SAM9261 is in reset.

The baord is based on the AT91SAM9261-EK and current only CPU+SDRAM on it. Core runs at 1.2v and all others at 3.3v. It worked for a while (with output "ROMBOOT and ">").

One weird thing is that I measued the capacitance between the NRST and GND using LCR. It shows 2uF, but there is no capacitor placed there. Don't understand why.

Getting frustrated on this issue. Was told lots of experienced people on Atmel parts here in this forum. I need your help. Thanks.

>>The DBG UART outputs ROMBOOT without the prompt (">"). It seems that the AT91SAM9261 is in reset.
>>The baord is based on the AT91SAM9261-EK and current only CPU+SDRAM on it. Core runs at 1.2v and all others at 3.3v. It worked for a while (with output "ROMBOOT and ">").

Is it in reset, or is it executing code pulled from NAND, DataFlash, MMC, or whatever (found a bootable signature and jumped to it), or trying too.

When my 9G20 starts, I get the ROMBoot, and then it runs my code pulled from NAND, no '>' prompt.

So either it had loaded some code, or is trying to probe a boot device. I'd guess I'd look at the pins for the boot devices, and the order ROMBoot is supposed to probe them. I don't suppose it's stuck in reset, because the ARM core is running long enough to spit out the "ROMBoot" string.

Have you tried using JTAG and breaking execution to see where it's at?

Check the 18.432 MHz, perhaps it's trying to bring up the USB.

Thanks for the reply. Really appreciate your help.

There is no NAND, Dataflash, or MMC etc., only CPU and SDRAM. So, I believe the ROMBOOT does not see anything else and tries to boot from SAM-BA.

JTAG cannot communicate with the target. I'll double-check the OSC to make sure the 18.432MHz does oscillate.

The thing I can not explain is that why the NRST pin gets stuck at 2v instead of going all the way up to 3.3v.

What kind of pull up do you have on the NRST line?
What voltage are you running the SDRAM at?
What else does the NRST connect to? JTAG, Ethernet, DataFlash?

The pullup on the NRST is 1K. The SDRAM is a 3.3v one. Right now, the NRST has nothing else connected.

Again, thanks for your help!

The NRST has an internal ~100K pull-up to VDDIOP.

You could check the voltage without the external pull-up, and check the supply side of the external pull-up.

You could check your external power-planes, and verify they are all connected and you don't have un-powered islands, or power back-feeding from other parts. Try probing an unpopulated PCB, or the solder-sample board.

If you have access to an xray inspection tool, you could check the BGA for voids/shorts.

The problems I've seen with people and NRST signals is where they actively drive them high, and internally the OC driver is trying to clamp them low, and they fight.

If it's clamped internally with just an external pull-up, it's going to be darn near zero.

Is the capacitance you measure on a raw device, or on the board. If on the board you are probably measuring the bulk/decoupling capacitors via the VDDIOP it's internally pulled up to.

Back to the original suggestions, validate that the boot against other devices is occurring (ie probe attempts). Jumper BMS so it tries to boot an external NOR, and validate if the EBI accesses are occurring.

Like I said, if the part is running stable enough to print "ROMBoot", and not constantly rebooting, I'm not sure NRST is even your real problem here, although it not being 3.3V is admittedly a tad odd. Just don't focus on it, go validate that other parts of the board are at the correct voltages, and the other pins are doing what might be expected, or not. By eliminating what is working, you might get closer to discovering what isn't and why.

I think you are right. There is some kind of short on NRST. I somehow mangaed to remove the extra solder near the NRST. NRST pin goes high as it should and the target board runs fine so far.

Now, I have another problem. I hook up the JTAG and power it up along with the target board. The target board runs fine and the JTAG (a USB one) is recognized by the PC ok as well. But, as soon as I run the OCDCommander, which is a simple debugger from Macraigor, the JTAG pulls the NRST low and it stays low. That stops the CPU from running again :-(

Can't offer you much help on the Macraigor, I'm mostly using JLINK/SAM-ICE Segger, or ULINK Keil pods these days. I've used MULTI-ICE, and wigglers in the distance past.

Double check the pull-up/pull-downs on the JTAG. The 100K pull-ups are a bit on the weak side, 33K or 47K would be my preference. Check your wiring expectations, the JLINKs push 5V out of pin 19, the JLINK software also has some amount of connectivity diagnostics, code downloading, running/single-step stuff. Try some other pods if you have access to them. Download current drivers/firmware.