XCore Exchange

Hi Kster

Sorry to hear you having issues with you board design it effects all of us building boards.

I have been perusing your schematic (can you provide a PCB?) here are a couple of thoughts:

1) I see your power section is based around LTC3417 unlike the ref design. You drive the Run1 (1 volt enable) from the 3.3V supply output. Xmos warns that you should bring up and stabilise the 3.3v before ramping the 1v. I wonder how much time you config gives you and if its enough, definitely worth checking on a scope. The Run1 would probably trigger at less than 1 volt (0.85v typically) which may be to quick or subject to spikes even. Adding an RC to the Run1 pin could buy you more headroom on the 1volt start up. the L1 design checklist states:

"The VDDIO (and OTP_VDDIO if present) supply must be within specification (3.0V -
3.6V) before the VDD (core) supply is turned on. Specifically, the VDDIO supplies
should be within specification before VDD core reaches 0.4V."

2) You show a 4.7R pullup on the debug signal is that correct or a typo?

Just my $0.02
regards
Al

The LTC chip is copied exactly from the XC-1A (same as XC-5 and XC-1 design) design so I assume it is ok but I would be open to any mods.

The 4.7R is indeed a typo as I put a 4.7k there. Interestingly enough this relates to bigger problem that I put down a 4.7k resistor on the PLL_AVDD line instead of a 4.7R. I think this was my problem all along as it drops the voltage down to 350mV instead of 1V required by the PLL. I ordered some 0603 4.7R from digikey and hopefully everything will be ok after this.

Also the 4.7k debug resistor isn't on the reference design but is listed in the errata so it is probably good to have.

The reference design should be a gold standard so that should probably be updated if pullup resistors are needed (on debug, rst_n, trst_n).

kster59 wrote:I ordered some 0603 4.7R from digikey and hopefully everything will be ok after this.

In the meantime you could bypass this resistor with a wire strap, because AFAIK it's just to form an RC lowpass to get a cleaner voltage.

Do you mind to show us the PCB you made? I'm interested in a minimal board :)

kster59 wrote:Interestingly enough this relates to bigger problem that I put down a 4.7k resistor on the PLL_AVDD line instead of a 4.7R.

:D I made the same error on my first custom board.
I did however make the error in the schematic so I found it quite easily after checking the schematic a couple of time. After I replaced the 4k7 with two 10R in parallel (the lowest I had available) everything worked perfectly.

I did check your schematic but since you had the correct value in your schematic I did not comment on the problem I had. I regret that now, could have saved you some time... :(

I'm attaching my schematics if anyone is interested...
(I have not had time to update the schematic to the 4k7 error is still in there.)

I will add everything (schematics, PCB, source code) to my Sensor Node project as soon as I've finished the code and completed the build log.

Glad that you were able to follow my instructions for checking for signs of life. Sorry there isn't any on your board! The fact that you get something back suggests that the JTAG chain is at least complete.

PLL not working would be my best bet: short the 4R7 as skoe suggests.

Also the power up sequence may be causing problems as Folknology suggests.

skoe wrote:In the meantime you could bypass this resistor with a wire strap, because AFAIK it's just to form an RC lowpass to get a cleaner voltage.

That's right, it avoids digital noise getting to the PLL's supply. Shorting out the 4R7 pad should work fine to bring up the board.

The power on reset is controlled by the LTC3417 exactly like the XC-1A and XC-5.

Are these boards known to have problems?

Also, are pullups required on RST_n and TRST_n? I solder some on to test but I'm not sure about reliability.

The errata clearly states there should be a 4.7k resistor pullup on DEBUG but that is not there in the reference design so that should be fixed.

Woody wrote:Also the power up sequence may be causing problems as Folknology suggests.

Digikey is pretty fast. I pay $2 for shipping and it takes 2 days but I didn't want to wait so I shorted the 4.7R resistor just to test until the resistor gets here tomorrow.

Works great now!

I do not need pullup resistors on RST_n or TRST_n. Not sure about long term but I will report back if I observe any problems.

Please remember to post here if you find any problems so the others of us can learn from it.
Specially if it´s problems such as it work 95% of the time but malfunction 5%, and that was solved by a stronger pull-up etc.

kster59 wrote:Works great now!

Good news. Can you just clarify what the problem was. Was it using the 4K7 rather than 4R7?

A quick note on copying XMOS schematics - while the dev kits are functional boards they may not have the required power sequencing and power on reset requirements (as discussed in the Hardware Design Checklist) - so you might have issues.

Our reference design schematics are in the process of being updated to cover the requirements, so keep an eye out for that. We'll keep you posted!

Cheers,