100x current consumption on 5V vs 3.3V

Tip / Sign in to post questions, reply, level up, and achieve exciting badges. Know more

cross mob
KaPa_2418181
Level 2
Level 2

Hi!

   

I'm learning about power saving using the PSoC 4. For this I am running the advertisement project from AN92584 on a BLE Pioneer Kit (CY8KIT-042-BLE-A). I am measuring current consumption on J15 on the main board with a Fluke 87 multimeter, board powered from USB.

   

What I am observing is that everything is as expected when I select 3.3V power using jumper J16 (default setting), in this case I measure a current consumption of only 1.5uA in deep sleep. Just as specified, perfect.

   

When I move the jumper to the 5V position though, my deep sleep consumption increases to over 106uA. I tried various things, like:
- changing the VDDA/D/R voltages to 5V in PSoC Creator 4.1 on the System page
- trying with cold/warm resets
- changing DEEPSLEEP_ONLY in code to 1 to rule out any software-induced effects

   

... but with no luck.

   

As far as I can tell, feeding the accompanying PSoC4 daughter board with 5V is perfectly valid and within specifications. I do expect a somewhat higher consumption on a higher voltage rail, but 100x higher current within voltage specifications seems very abnormal to me.

   

Any ideas? Is this normal? Is there some additional (software or jumper) setting I need to make? Or is my kit defective?

   

Thanks,

   

Karoly

0 Likes
1 Solution

Hi!

   

Sorry for not checking back but I also got some results from earlier. They match dsweet's, that is, if the BLE module is not powered from the base board, then the consumption is alright even with 5V (1.6uA). So the Pioneer base board is at fault here, at least partially, read on.

   

I tried to analyze what is wrong on the baseboard, and arrived at the following conclusion:

   

It has to do with the protection circuit of the F-RAM and possibly also with the PSoC4 chip. What happens is that first the module board is powered from 3.3V, and so C29 charges up. Now when you switch to 5V, the protection circuit will turn off VFRAM, but will not shunt it to the ground. So VFRAM stays floating, but is actually holding a potential due to the charged capacitors. This apparently causes a high leakage current on pin P5_0/P5_1, though I'm not yet sure where this current is flowing to. The leakage current decreases to "only" 90uA if you just write "1" to the pins (and zero leak if "0" is output), so I guess it is a float issue. The leakage also stops if you disconnect both P5_0 and P5_1.

   

So, conclusions:

   
        
  • You can only observe this issue if C29 is charged before switching to 5V. For example you won't be able to reproduce the issue if you are turning on the board to 5V after a long off-state. You need to power it from 3.3V, then from 5V.
  •     
  • Even if VFRAM is floating, this shouldn't cause any of the P5_x pins to leak, since they are all set to analog HiZ. I thought my module might be faulty, but since there are others experiencing the same issue, this might be a more generic module defect. Please Cypress, check this.
  •     
  • The best way to completely stop the leakage on the Pioneer board is to make sure VFRAM is discharged when powering from 5V. You can do this manually by shorting C29's poles, or automatically by outputing a logic "0" on P5_0 or P5_1.
  •    
   

The issue might also have to do with the working of the protection circuit, because if I measure the resistance on the unmounted R7 when running from 5V (which means - and I checked this by measuring the gate - Q6 is turned off), I measure a resistance of only 110KOhms. This is way too low. Again, I thought my board could be faulty, but it'd be best if others checked this too just to make sure this is not a board design issue.

   

I hope Cypress can clear things up even more, but I think I could offer some help. Anyway, to other Pioneer users, here is a warning:

   

Read and understand the above if you are trying to measure sleep current with J16 set to 5V. Otherwise your measurements could well be bogus.

View solution in original post

0 Likes
4 Replies