Powering iMX.23 through VDDIO pin?

[markl]

Hi all,

A question on power source...

Our board has a number of 3.3V peripherals on it, and combined they draw more than the i.MX can supply.  So, we had to add a separate 3.3V switching regulator.

Our board is externally powered from 5V, and we are currently using the VDD5V input on the i.MX to give it power.  However, this results in some power being dissipated constantly through i.MX linear regulators, no matter what mode the i.MX power supply is in.

And plus another downside of using the VDD5V input is that the i.MX doesn't have a lot of tolerance (5% @ >100mA) for voltage droop on it (table 2-3).

So, since we have a stable 3.3V supply available, is it possible to power the i.MX23 directly through the VDDIO pin?  This would allow more tolerance for 5V fluctuations since it would now be a function of the external 3.3V regulator, and it could possibly reduce the heat dissipation in the i.MX a little bit.

There is no battery in our application, so VDD4P2V doesn't matter, and we would just use the linear regulators to supply VDDA and VDDD and disable the DCDC converter (which we're already not using anyway).  Looking at the diagram of the power supply (figure 32-1) in the manual, it appears that power coming in through the VDDIO pin should be possible since the final stage of the linear regulators are all pass MOSFETs.

It's a little unconventional, I admit, but I can't think of any reason why it can't (or shouldn't) be done.

Thanks for any thoughts,

-mark

[danix]

Mark Litwack said:

Ok, so I tried this and I'll save everyone time replying.

In short, I couldn't get this to work.  It doesn't draw any current worth mentioning on VDDIO and obviously doesn't boot.

I'm guessing that something is looking for voltage coming into VDD4P2V or VDD5V before anything interesting starts happening.  Oh well.  board_parts_count++ ...

Mark, I've been designing a feature phone using i.MX233 and one thing I want to say is, it's one of the lowest power ARM SoCs out there! But you have to keep the following in mind to appreciate its low power capabilities

1. The linear regulators regulators should never be used if your board draws significant current, say > 1/2 the rating. as they have poor efficiency, thus will heat up the die. I use the linear regulators only when I'm charging the battery through USB-5V.

2. If your current demand is more than the DC-DC can supply, then using additional external DC-DC converter is the best option. let the integrated DC-DC power up the SoC and some light loads on board and the rest separately supplied from external DC-DC. I have GSM/GPRS, GPS,SDIO wifi and bluetooth chips on my tiny PCB and all are powered from an external DC-DC which can supply up to 3A @3V3, from a 3.6V, 1200mAh Li-ion battery which usually lasts up to 5days once fully charged!

I know this is so a late reply, but i posted anyway 'case it might help

--cheers!

   ---Dan

[mikeasa]

thanks mark.

for imx28; the power calculator from freescale; shows the power usage is about twice the imx233.(part of it is due to DDR2 chip).

with that; the chip's power supply efficiency is about 75%!!!(and they call this efficient too!).

It has the convenience to regulators on the cpu; but at highly low efficiency.

So now I am thinking of how to disable all the cpu's regulators and use all, from outside which increases board space and cost. but 75% eff. is just out of question and I must use DDR2 memory!

thanks again.

.

[markl]

Hi Mike,

No, we weren't able to get it work and punted on it since we needed to get the board spun and move on.  We pulled all 3.3V peripherals off the processor (including direct drive LEDs) so the only thing the processor is powering now is itself and memory through VDD5V.  It doesn't run very hot in this configuration.

Another option we considered was supplying power via VDD4P2V or BATT.  We would have had to add the components to enable the DCDC converter and incur more design time and component cost.  But you might want to look at that for your imx28x.  3.3V is the minimum for VDD4P2V/BATT, so I would look to increase that a little more for a comfortable margin, but of course that's yet another external regulator.

It's too bad they didn't spec the DCDC converter to operate with 5V to address non-portable applications.

We found the OnSemi NCP1529 to be a nice fixed/adj. high frequency regulator with a low parts count if you want to explore some external regulator options.

[mikeasa]

Hi Mark;

did you get to any results on this power supply case.

I am trying to come up with similar solutions to reduce the IC's power consumption; specially in Imx28x which seems to have a much bigger power consumption than imx233!

thanks

[markl]

Ok, so I tried this and I'll save everyone time replying.

In short, I couldn't get this to work.  It doesn't draw any current worth mentioning on VDDIO and obviously doesn't boot.

I'm guessing that something is looking for voltage coming into VDD4P2V or VDD5V before anything interesting starts happening.  Oh well.  board_parts_count++ ...

-mark