Pull-down on the Imp wake-up pin VS pull-up on sensor output

I’m trying to pair the Imp with the TI TMP102 temperature sensor. Imp is in deep sleep mode and wakes up when the temperature conditions are broken. I’ve described details & shared code here: http://goo.gl/maFMO

Everything works fine, except the power consumption. TMP102 has default state 1 switched to 0 if alert condition happens. When i reconfigure its logic to suite Imp wake up protocol, sensor starts to consume 2.6ma via 1k pull-up resistor. I replaced it with 40k and it works ok consuming 90μA. But i’m afraid that if i make it too big, it will conflict with the internal pull-down resistor on the Imp pin 1 in wake up configuration. Is it right? And if so, what should be the max pull-up resistor on the sensor to avoid conflict with the Imp?

Really, you need to use a buffer here, so you drive high (vs relying on a pull-up).

A simple 1g04 takes <=1uA.

The alternative would be polling temperature; when the wifi-off release comes out, you can do this pretty cheaply in power terms (turning wifi on is the big power hog)

Hugo, thanks for your help! I didn’t think about a buffer, it is much safer than my solution!

The alternative would be polling temperature; when the wifi-off release comes out, you can do this pretty cheaply in power terms (turning wifi on is the big power hog)
estimated release date ??

I noticed that the imp can wake up from the dead now (Power state: online=>offline), but in the dead state it still refuses to process, it’s dead, and so I lose data

Hoping it will be in the next few weeks. We’ll likely be looking for testing volunteers before than :slight_smile:

I don’t understand your comment about waking up from the dead?

When I get Power state: online=>offline I noticed that even after 15 minutes or more it comes back (Power state: offline=>online), something it did not do before (I had to reboot the imp).
The frequent online=>offline state started with the introduction of the firmware release 14. First I thought that the reason was local (bad wifi connection or so) but after changing my watchdog function to show the dBm signal, it was clear that this was not the issue.

I can live with those online=>offline states, if the Imp continues processing…

And yes, I love to be a volunteer in this case

But I think my comments are becoming a bit off-topic in this thread, sorry for that

@dolf if you do imp.enableblinkup(true) so that the LED is always flashing, I’d be very interested in knowing what the LED is doing during one of these outages…

it keeps blinking green, every second in the offline state

Hmm, that points to the imp not actually knowing it’s offline.

This happens when the link is idle and the router tears down the connection. The imp won’t work out it’s disconnected until it next tries to send something - either due to your code sending, eg the watchdog, or due to it sending a ping to the server which it does pseudorandomly.

If the server doesn’t receive the ping from the imp when the imp said it’d send it, the server will close the connection.

As soon as the imp tries to send something, and finds the server has closed the port (or the NAT entry has been removed), it’ll restart the connection process and reconnect.

What router do you use?

A THOMSON SpeedTouch 780, but I’m fairly sure that this didn’t happen so fast in the beginning

Hmm. How often do you have your watchdog firing in your code?

just once a minute :slight_smile:
…and in between it does it’s comunication also

Yeah, that should be plenty. It should never to out to lunch for longer than a minute in that case, which isn’t what you’re seeing, which points to some other type of issue.

I’ll PM you about testing a new release.