OpenTRV: thermostatic radiator valves and remote boiler control

D

DamonHD

Senior Member
Lack of "nap" looks like a real showstopper for me using the 20X2 since I have to economise on power (my total budget is about 100uA) but my inner loop is a second, or 4 seconds at a real push. (And I want to be able to operate down to 2V or marginally below to operate from 2x AA rechargeables.)

I can possibly dream up a much more elaborate scheduling system to get round that, but that implies linked lists and similar fun which wouldn't be easy...

Rgds

Damon
 
S

srnet

Senior Member
Well, I guess have said it before, but I would have started off using an 28X2. If your code space is tight on the 18M2, then the 28X2 would have heaps, looks like there is plenty of space on that PCB for one. Later on you might manage to fit it all in one slot of a 28X2, so do a Mk2 PCB for a 20X2.

The extra I/O pins would help too, you mention the need for sleep and nap, well the RFM22 has a rather nice wake up timer that can use a cheap watch crystal, although the internal RC time is reasonably accurate too. Put the PICAXE to sleep with 'sleep 0' an get it to wakeup via an interrupt from the RFM22, from mS to months later. You should be able to get sleep current down to around 20uA, depending on the regulator used ...............
 
D

DamonHD

Senior Member
I believe from the PDFs that the 28X2 is only down to 2.1V supply, which would be marginal for 2xAA rechargeables: is that correct?

An additional problem we have is that the RFM22/RFM23 is now end of life, I believe...

Rgds

Damon
 
D

DamonHD

Senior Member
Stuart is also my source, but he seems to have reasonably hard evidence, albeit 2nd hand.

Rgds

Damon
 
S

srnet

Senior Member
I have never said that the RFM22 was 'end of line', so can we be careful to stick to what is known.

What you will find is that if you go to the Silicon Labs website and lookup the Si4432, you will see this;

"Si4430/31/32 devices remain in production, but are not recommended for new designs. For new designs, please use Si4460/1/3/4 RF Transceivers instead."

Simply put all they are saying is that they would prefer applications to move across the the Si446x, but they have given no indication that they intend to stop production of the Si4432 at all, and if it continues to sell they would have to be mad to cut off a lot of sales. The Si446x is not a direct replacement, so may not sell.

And just because someone like RS decides to stop selling one version of the RFM12, I would not be assuming that production of the RFM22 is about to stop !!!!.

Can't quite see how it can be justified to post on a blog as fact that some products are 'end of life' with apparently no conformation of the people making them !!!!!!
 
S

srnet

Senior Member
DamonHD said:
I believe from the PDFs that the 28X2 is only down to 2.1V supply, which would be marginal for 2xAA rechargeables: is that correct?
Click to expand...
I think so, there is a very low power version of the 18F25K22 (28X2) that goes down to 1.8v, but its not used for PICAXEs.

Given the issue I have mentioned before (power supply glitches affecting the RFM22) I would not be using the RFM22 in the first place. From my experience you seem to be doing most of the things that are most likely to cause it to reset\crash when transmit is turned on, a very undesirable feature in a 24\7 consumer device.
 
D

DamonHD

Senior Member
My TX problems seem to have evaporated, and my comms seems reliable for 3 separate devices controlling rads 24x7, even at full speed (32MHz) and on my early lash-ups with little or no decoupling, so I think I have some elbow room.

Our shiny nice new PCB does have lots of reasonably sensible decoupling on it, largely in line with your suggestions:

https://twitter.com/mikestir_uk/status/319917642435665921/photo/1

Rgds

Damon
 
S

srnet

Senior Member
DamonHD said:
Our shiny nice new PCB does have lots of reasonably sensible decoupling on it, largely in line with your suggestions:
Click to expand...
Really ?

Decoupling capacitors right next to the RFM22 power pins ?

Electrolytic decoupling cap ?

Regulator ?
 
S

srnet

Senior Member
I have always mounted the regulator and capacitors literally right next to the RFM power pins (within 3 or 4mm) and kept the track as thick and short as practical. With SMT devices and two sides of the board to play with, its not usually a problem.

However, I am now well confused, when you said you wanted operation from 2xAA NiMh at 1.8v, I assumed you were not using a regulator ?
 
D

DamonHD

Senior Member
All of this design except the DS18B20 should operate at 1.8V and we have the option to omit the regulator and wire the battery directly to the V+ rail. For now a 5V (USB) supply is being used, via the regulator, and in all my initial deployments there is a power socket close at hand. The aim is to allow it to be operated either off batteries or a cheap standard efficient USB phone charger, whichever suits the end user better.

The current aim is to replace the DB18B20 with a precision thermistor/resistor potential divider (and move it to an ADC pin).

I have a cheaper and smaller RTC lined up for the next revision which will also aim to reduce power consumption.

Rgds

Damon
 
S

srnet

Senior Member
OK, thanks for the clarification.

One of the things I have found that the RFM22 needs is a very stiff supply, by that I mean a supply that alters very little under load. This significantly reduces the power line sags caused by the sudden increase in current when the TX turns on. These sags (glitches) are the cause of the RFM22 resets I have been referring to.

A small Lipo battery is a good supply, these batteries have a very low internal resistance and will give you a very stiff supply. A (small) regulated supply is almost as good. And you still need to keep power tracks short and thick to reduce the sags seen by the RFM22 pins.

A pair of NiMh at the 'end of life' are, shall we say, not an optimum supply.

Having said that, you can detect the RFM22 reset problem, but to be sure you can fix it, you need to be able to control the SDN pin.
 
D

DamonHD

Senior Member
Thanks.

1) This particular design may or may not be the one that I hope gets deployed in significant numbers this coming winter. The new one may be a revised PICAXE design or (say) ATmega/Arduino or both. I hope I'll get to do extensive testing against batteries, including towards their end of charge.

2) The need to replace the RFM22/RFM23 in the design may make the point moot. I was considering stockpiling 100 of them but we may just directly build something around the EzRadioPro or C11xx chips ourselves which may not be that hard. But we'll continue to bear decoupling in mind.

Rgds

Damon
 
You must log in or register to reply here.
Top