Jan Apprecciate your response. Here are a few followup questions

1) I am already using SPI, can I "remap' them to just output pins before going to sleep? Is that possible 2) can I also dynamically change LPC_SYSCON->SYSAHBCLKDIV to reduce or increase clock speed?

Thanks again.

Re: 1. Yeah, remap them before going to sleep. About 2... Not completely sure. I think there are other clocks derived from the main clock divider, so don't know what will happen when you change that around during execution...

Jan, Thanks. Just to be sure My code will like this?

<code>

Start

SPIDevice acc(P0_11, P0_16,P0_28,P0_10, P0_25)

Before Going to Sleep

DigitalOut dummy0(P0_11,0);

DigitalOut dummy1(P0_16,0);

After Coming back from Sleep

SPIDevice acc(P0_11, P0_16,P0_28,P0_10, P0_25)

</code>

I tried above, I think I cant do SPIDevice acc(P0_11, P0_16,P0_28,P0_10, P0_25) because I will be repeatedly recreating without releasing

Would you kindly show me how to reassignments please?

What do you mean with "I think I cant do SPIDevice acc(P0_11, P0_16,P0_28,P0_10, P0_25) because I will be repeatedly recreating without releasing" ?

Jon FYI: We have a SPIDevice class which is a subclass of SPI. IT takes 3 SPI pins and two more pins for 2 Chip Selects. Hence the SPIDevice (pin1-pin5)

Back to your question I create SPIDevice at the start of the program (rather at the definition level). I have to allocated these pins to DIGITALOUT before going to sleep and come back to SPIDevice when woken up.

I tried to do new SPIDEVICE(pin1..pin5) but it worked for two cycles and said no more SPI devices and the program stopped. So I am guessing I am not "releasing the old pins"

How can i do the reassignment?

Erik,

We had to briefly step off of this project and are coming back. I can give you a better estimate shortly.

Meanwhile

1) We are checking power on the entire device. It has a buck convertor using TPS54240 (to bring to 4v) and 4v to 3.3v LDO. All other components have been turned off (during this testing)

2) We are using your Wakeup function to wake up form deep sleep.

3) In the next few days we will test by turning BOD and WWDT off (which are running as per lpc824 doc)

Best Regards

Surendar

Without WWDT WakeUp won't work though ;). Brownout is not required and the larger one of the two. But if possible, try if you can check the power to the LPC824 itself.

Also what you can try it just using a very simple program that does not do more than call deepsleep, and set the pins that really need to be set properly for this.

I happened to have looked with someone else for a similar issue, where half of it was software which was doing something wrong, and the other half was a wrongly read datasheet of an LDO, and which consumed 30 times what the microcontroller consumed in deepsleep.

Erick,

Few more pointers. I hope you can help us

1)We are using MCP1801. do you anticipate a problem with that?

2) The current consumption wtih a simple program is about 1mA when on DeepSleep. I thought you said that itself was high. We will try to remove BOD and see that makes it any better

Have been trying to respond to this thread and the "Post a comment" started working just this AM and hence the delay

That LDO seems fine. Removing BOD should help, but not even close to 1mA. More like ten(s) of uAs.

One thing to make sure is no pull-up/pull-downs are using power. And also no pins high to other chips which have no supply voltage.

Erik, Heres some more info When I use my hardware wiih a simple digitalout (with no peripherals and practically no other code) and move to deepsleep (for 10 seconds) the current (as per our multimeter) is 3mA @12v and about 5mA@12v on regular operation (for 10 seconds again).

When I remove the digitalout ( and no other peripherals are used) the deepsleep current is 1mA@12v and on regular operation is about 2mA@12v

I have ensured that the actual digitalout pin is not connected to any circuit so I dont think "power draining" is a problem. Also putting DigitalOut with default high (1) or default low (0) does not seem to make a difference

Any help please?

Best regards Surendar

Do you have the standard LPC824 board and what do you measure on that one? 2mA is way too much, but it really shouldn't be affected by the state of DigitalOut. Are you power cycling the device after programming it? Problem is I don't have an LPC824 myself, so I really cannot check for myself.

Erik,

I will try on the main board.

What do you mean by power-cycling the the device after programming? We push to code to our board and restart it.

More infomation

1) We have added just the code to lpc824 and have also provided a 1OHM resistor to track the current at 3.3v

2) When no Digital pin outs are used its at < 1mA

3) When 24 Digital outs are defined (all turned to 0 before going to power-down) its about 11mA at 3.3v

4) If i use about 14 digitalout pins (all turned to 0 before going to power-down) its about 6mA at 3.3v

I will try with the dev board and let you know

Best Regards

Surendar

I mean after programming disconnect the power and reconnect it.

Erik

I feel kind of stupid after this multi-month run-around :) We think we found the problem.

We had pulled up a few pins for the following reasons

1) One of it was open drain (pin 10 or 11) and we using it actively in SPI MOSI. So we had pulled it upto 3.3v

2) We had pulled up the reset pin so as to not accidentally restart the chip

We were, by default pulling all of these down before going to sleep and I think that was causing the problem. We have since then kept these pins at high before going to sleep and the current is NOT 11mA (at 3.3) but more like 1mA. While its a significant improvement i thikn we can make it better and are continuing to research.

That leads to another question (Maybe I will also post the answer out of this in antoher thread in case people are benefited)

In case I need to switch pin function on the fly for the following situation, Is it ok to dynamically use the SWM to pinmux? This way will help me define these variables globally and avoid redefinitions.

1) SPI pin to digital pin (to move to zero before going to sleep), OR

2) move UART0 from one set of pins to another (so I can get 4 uarts instead of 3),

Best Regards

Surendar

PS: Truly appreciate your help.

Yeah that should be fine. Just make sure you don't create glitches by for example moving your RX pin to another pin that is low at that time, or during a UART transaction.

Erick, Johan and Jan, Please check

https://developer.mbed.org/questions/76195/LPC824-Power-Consumption-A-Check-List/