GPIO too weak

Hannes Petermaier

Prodigy 235 points

Other Parts Discussed in Thread: AM3352, TPS65217

Hi all,

i'm quite new within the "Sitara" business and have a problem on a custom Board to drive a GPIO-Pin.

We're using the Sitara AM3352 processor.

The Application only does setup GPIO2_4 Pin and afterwards it wants to toggle the pin.

Power Rail VDDSHV is connectet to 3.3V

I can see toggling as expectet on the scope but drive strength is much too low. Output only reaches about 2.4V and connecting a few megaohms to ground lets drop the output downto about 1.5V.

Same application on the beagleboard is working.

We've allready checked voltages and powersequencing without success.

Any idea whats going wrong here ?

over 12 years ago

0 Biser Gatchev-XID over 12 years ago

TI__Guru**** 393215 points

Hi Hannes,

the drive strength of this pin should be +/-6mA. If it isn't connected to any load there must be a problem on the board itself.

0 Hannes Petermaier over 12 years ago in reply to Biser Gatchev-XID

Prodigy 235 points

Hi,

I allready disconnected everything from the pin with same result, i agree with you that something must be wrong at my board but i havent any idea where to measure or look from where this behaviour is coming.

Any ideas which things on the hardware may be wrong?

0 Biser Gatchev-XID over 12 years ago in reply to Hannes Petermaier

TI__Guru**** 393215 points

I would search for shorts to adjacent balls first.

0 Hannes Petermaier over 12 years ago in reply to Biser Gatchev-XID

Prodigy 235 points

i allready looked for short circuits around the pins -> nothing. Furthermore a short circuit would not correlate with voltage depending on load current, it seems to me that I/O driver has a very high impedance - but why ?

I also tried a pin within another GPIO-Bank with same result.

regards,

hannes

0 Biser Gatchev-XID over 12 years ago in reply to Hannes Petermaier

TI__Guru**** 393215 points

How are your power supplies connected? Did you check that all power supplies are present and at proper voltage?

0 Hannes Petermaier over 12 years ago in reply to Biser Gatchev-XID

Prodigy 235 points

hi,

voltages are supplied from the PMIC (TPS65217) an connected as shown:

we allready verfied twice voltages and power-up sequence without any abnormality

0 Hannes Petermaier over 12 years ago in reply to Biser Gatchev-XID

Prodigy 235 points

Power is coming from the PMIC (TPS65217). We've checked voltages and sequencing twice and observed no abnormalitys.

Is there a possibilty to post some PDF of my schematic ?

0 Biser Gatchev-XID over 12 years ago in reply to Hannes Petermaier

TI__Guru**** 393215 points

I have send you a friend request with my e-mail, if you don't want to post it here.

0 Hannes Petermaier over 12 years ago in reply to Biser Gatchev-XID

Prodigy 235 points

Hi,

i've sent you an email with my schematic. maybe you can see if there is something wrong.

in between i made further experiments on my board an tried out booting from spi flash.

what a mess, also the spi pins have this strange behavior regarding voltage and impedance. so i did take another scope picture where i've watched the SPI Pins from the beginning of power up. I observed, that voltage is around 3.3V (as expected) for the first 22ms, afterward voltage drops down to 2.4V. But supply pins of AM3352 still have stable 3.3V.

Any explanation to this ?

best regards,

Hannes

0 Biser Gatchev-XID over 12 years ago in reply to Hannes Petermaier

TI__Guru**** 393215 points

Hi Hannes,

I've sent you an e-mail with some comments. I will copy them here:

I don't like much what I see around the PMIC. Please check the Beaglebone Black schematics:
http://circuitco.com/support/index.php?title=BeagleBoneBlack#Hardware_Files

Also there is a TPS65217 guide for AM335X:
http://www.ti.com/lit/ug/slvu551g/slvu551g.pdf

Some quick notes:
The ferrite beads L304...L306 will have negative impact and should be removed or replaced with 0Ohm resistors.
R303 and R412 should not be placed.
RTC_nPWRONRST should come from LDO_PGOOD.
5V should come in on AC input of PMIC, not on BAT input.

0 Hannes Petermaier over 12 years ago in reply to Biser Gatchev-XID

Prodigy 235 points

Hello Biser,

hope you had a nice weekend, many thanks for your advice.

We allready changed R303 und R412 before we powered up the first time.

At this time i tryed to replace L304..R306 with 0E resistors --> no change.
Next i tryed to connect RTC_nPWRONRST with LDO_PGOOD --> also no change.

As last action, i changed the stuff of 5V and AC input ... as you can imagine, that was a realy hard job --> also no change. Please have a look to the PMIC Datasheet on Page 36, they also connect AC/USB to ground an give +5V to BAT.

best regards,

Hannes

0 Biser Gatchev-XID over 12 years ago in reply to Hannes Petermaier

TI__Guru**** 393215 points

Hi Hannes,

Something's very wrong then. Please check the control signals between the PMIC and AM335X. I can't see from where the PMIC_nWAKEUP comes. Also do you have pullup resistors on the I2C lines?

0 Hannes Petermaier over 12 years ago in reply to Biser Gatchev-XID

Prodigy 235 points

Hello,

on I2C i have 4k7 pullup to 3.3V

PMIC_nWAKUP line initially was routed through an external reset controller. I've changed this (connecting EXT_WAKEUP from AM335x to nWAKEUP of the PMIC while having a 10k pullup to VDD_RTC) very early within our troubleshooting-process, but unfortunately without success.

any more ideas ?

best regards,

Hannes

0 Biser Gatchev-XID over 12 years ago in reply to Hannes Petermaier

TI__Guru**** 393215 points

I don't know Hannes. I have the feeling that the AM335X I/O cells are being supplied parasitically through the I/O lines, not from their proper power pins. Is this only one board you have tested?

0 Hannes Petermaier over 12 years ago in reply to Biser Gatchev-XID

Prodigy 235 points

hi biser,

this was also my first suspicion, therefore i verified voltage at the power rails --> ok.

I have many boards, i tryed another one with same result.

Do you know any registers which i can check to findout whats going wrong ?

best regards,

Hannes

0 Biser Gatchev-XID over 12 years ago in reply to Hannes Petermaier

TI__Guru**** 393215 points

Have you seen this: http://processors.wiki.ti.com/index.php/AM335x_board_bringup_tips?

0 Hannes Petermaier over 12 years ago in reply to Biser Gatchev-XID

Prodigy 235 points

hello biser,

yes we've allready did study this wiki page and checked the powersequencing twice. Also the "is the device alive" is passed, as you know we are allready running some code on the machine. Also DDR-Ram is working in between.

"Only" I/O Pins are actually not working :-(

best regards,

Hannes

0 peaves over 12 years ago in reply to Hannes Petermaier

TI__Guru 63145 points

Can you specify which other GPIO from another GPIO Bank you used when testing if another GPIO did the same thing?

I'm not sure if your definition of GPIO Bank is GPIO signal groups (GPIOx where x could be 0, 1, 2, or 3) or VDD supply rails (VDDSHVx where x could be 1, 2, 3, 4, 5, or 6).

I'm trying to determine if you tried another GPIO on a different VDD supply rail.

Have you tested more than one of your custom boards to see if this problem occurs on all of them?

Regards,
Paul

0 Hannes Petermaier over 12 years ago in reply to peaves

Prodigy 235 points

Hello Paul,

with GPIO-Bank i mean as you told 0,1,2,3 ... i have all rails connected to 3.3V.

In between i've testet all 4 Banks with same result, also SPI Interface (SPI1_nCS0 for example) behaves in that matter.

Also it seems that only positive switching isn't possible, i tryed to sink some current into the chips - that looks good.

Do you now any processor-register where i could look to detect the failure ?

many thanks.

Hannes

0 Hannes Petermaier over 12 years ago in reply to Hannes Petermaier

Prodigy 235 points

Hello again,

in between i took a measurement on the I/O pin to check the sink capability of the pin. This looks good, Pin is able to take up about 6mA.

Afterwards i tryed with switching the same resistor (500E) to ground and observed, that driver puts out a very short pulse up to 3.3V and drops downto 0.5V. see attached picture.

0 Hannes Petermaier over 12 years ago in reply to Hannes Petermaier

Prodigy 235 points

Hi folks,

We’ve fixed the problem - I/O pin is now switching as expected. :-)

We've checked all power rails with their sequencing and so on after powering up the board (without running any software), everything looked fine.

Finally we detected the problem within the 1.8V voltage rail from the processor which is supplied from the PMICs LDO3. Software (Starterware) from the Beaglebone board tunes this output pin from the PMIC up to 3.3V - means that our 1.8V rail was supplied with 3.3V.

In fact the processor was supplied with too much voltage on the 1.8V rail and entered any failure state while switching on a GPIO Output.

I don't know why this impacts the GPIO-drivers and all Powergood an ON-State Bits within the registers say that everything is fine.

Finally I want to say thanks’ to the TI employees for their help.

Best Retards,

Hannes

0 Biser Gatchev-XID over 12 years ago in reply to Hannes Petermaier

TI__Guru**** 393215 points

Grat news Hannes!

Glad you solved this issue, and thank you for taking time to update on the forum. I did check the Beaglebone schematics, but only the BB Black, if I had looked at the old BB White we could have found the problem earlier, sorry for that. As for GPIO imparct, the VDDS voltage (1.8V) powers the internal part of the I/O cells and by supplying 3.3V to VDDS the upper FET's of the buffer were probably almost shut-off, which would explain the voltage drop and weak drive strength you observed.

I can only hope you didn't "fry" too many boards. Keep in mind that the processors involved were overstressed and other issues may arise with them.

Processors

Processors forum

GPIO too weak