The HalfBee project begins!

[KLD]

Sean & Eric

It's now been more than three weeks since I hooked the humidity sensor up into my 1-wire network. And there has been not a single dropout as far as I can tell. To re-cap, I'm using owserver under Debian, and take single reading with owread (called from a Perl script). The data are then stored in a RRD managed by Munin. And a week ago I started collecting 5min interval data into my standard SQLite database as well. Hardware: sensors connected via CAT5 cable to a DS9490R, NSLU2 a.k.a. Slug running Debian 'lenny' prot for the ARM processor. 1-wire network with about 25 sensors, powered via the 5V output of the DS9490R.

Since I haven't got a separate hygrometer to compare the data to, I can't tell how accurate the readings are. As a control for the temperature part of the DHT sensor, I put a DS18B20 onto the same board. This control reads consistently 2°C lower than the DHT. Not bad for a combined sensor!

I haven't had time to follow up on my initial experiments with Digitemp yet.

Thank you for all your great work!

Klaus

[StBarnabas]

Klaus
thanks for your kind words. You should really thank Eric particularly for this implementation.  Not only has he written the 1-wire "stack" but he has also written the interface to the Humidity/Temperature sensor. Was going to do this but Eric got in there first! You need to think about other sensors you might wish to implement?
Sean
 

[KLD]

Sean,

re. other sensors:

Yes, especially since the mornings are now too dark to read the gas meter without a torch    So, I've followed with interest the work wrt counters. But I haven't even got round to taking my old bicycle computer (reed switch) to check whether there is a convenient magnet on the meter spindle.
Likewise for the 'leccy meter. At least there is a flashing led on the meter, and in one of those cardboard boxes in the corner of the office must be a photo resistor.

Nearly there!

Klaus
 

[KenB]

Eric,

Could you please confirm the functionality of the halfbee sensors so far.

As I understand it you can offer the following

Dual pulse counter  (DS2423 replacement)

Quad ADC

Quad Thermistor

Quad Input (switch)

Quad Output.


Am I wrong in my thinking that you can offer quad functions from the 8 pin PIC?



Ken

[ericw]

Ken,
We have a dual counter DS2423 clone and a quad input switch in your wishlist.
There are also a whole range of single input devices that take data (solar,pulses,power) and output them as a 'temperature' to look like a DS18B20

However to do anything using the ADC's, with the current 8 pin PIC12F683's we are limited to 3, because the fourth 'ADC pin' has to be used for the 1 wire bus interrupt. I believe that Sean has done the maths part of linearising the thermistor curve, if not then there is always the LM35 to use.

Any quad device in an 8 pin package would only leave 1 pin to change the ROM ID so only allowing 2 chips (without reprogramming) on a bus.
By using an equivalent higher pin count device (PIC16F688?) we could do 4 ADC related functions and also enable a larger range of addresses to be user selected.

In terms of output devices I have just done an output only DS2405, but of course it cannot have an open drain output, and probably is best to use a buffer FET.

I don't see a problem in doing a quad either addressed as 4 devices or a quad, addressed more like the DS2408.  Again it would probably be better to move to a high pin count device purely to gain the ability to gain the ability to select more romid's.

Eric

[KenB]

Eric,

Thanks for your prompt reply.  That clarifies the capabilities - plus the limitations of working with an 8 pin PIC, the main problem being setting the ROM ID.

A little, universal pcb which carries the PIC plus any of the sensor options would be a useful asset. A bit like the common sensor board that Nuelectronics do - but with the 8 pin PIC.

One way of getting around the ROM ID would be to use one pin to "set the ROM ID" mode, and with this pin pulled high, it reads the address from the 3 remaining pins which are temporarily switched to form the address.  The mode pin is then released and the device reverts to ADC function from the other pins.

I'm hoping to use halfbee devices to add functionality to the Navitrino project, without gobbling up precious I/O - namely pulse counters, additional temperature sensors, user switches and remote relays for HW and CH switching.



Ken

[ericw]

Ken
While generating the ROM id as a separate step would work, I suspect there would be fewer problems if the 14 pin equivalent of the 683 chip, PIC16F684 ( £1.50@1off, £1.20@10off at RS & Farnell  ) were used, and the romid set by jumpers. This chip (or the next generation) could be the standard for higher I/O count devices. (3/4Bee?)

I did design a small single sided PCB, which had the halfbee on plus a Veroboard type area of tracks. I hoped to get a block of 15 ( possibly 24 ) from Olimex.com for around 40 euros all in. However as there didn't too much interest in using halfbees I didn't pursue it - could be a bit of a chicken and egg situation however.

I had a quick look at the data sheets and for output devices it looks as though the way to go would be to use the protocol used by DS2413 & DS2408 this is easier to program at both the master and slave ends.

[KenB]

Eric,

As you say - "chicken and egg".  Have you considered some of the very low cost Chinese pcb services that are now being offered?

I will be talking with Sean this week, and we will be sure to be discussing the pro's and cons of moving to a larger package - the 3/4 bee or even full-bee.  But then when you do this, there is an equal argument to use "any" microcontroller, and this then starts to become a crossover product towards my Arduino clone slaves, strung together on a serial network.

There is no question that the average household requires a whole bunch of sensors, separated by the order of tens of metres, even if just to measure room temperatures on a room by room basis.  Very few houses are configured such that all of the relevant gas, elecricity and central heating systems are located in the same cupboard for easy monitoring, so a wired or wireless network is inevitable.  Equally so, you are likely to want the display/programmer in an easy accessible place - and not buried with the rest of the wiring in the back of the airing cupboard. Similarly, the PC used for monitoring all of this lot is going to be some distance away, in a spare bedroom or office space.  So for almost all applications, a wired network of distributed sensors, output devices and dataloggers is going to be required.

Any product cannot be all things to all men,  but frequently you have to put a stake in the ground and set up your stall.  Only by the efforts and perserverence of pioneers, do we end up with the useful products and the enabling technology for others to build upon.

Keep up the good work



Ken

[ericw]

Eric,
I will be talking with Sean this week, and we will be sure to be discussing the pro's and cons of moving to a larger package - the 3/4 bee or even full-bee.  But then when you do this, there is an equal argument to use "any" microcontroller, and this then starts to become a crossover product towards my Arduino clone slaves, strung together on a serial network.
Ken

Ken,
I think we are in danger of suffering from 'mission creep', the intention of doing the HalfBee was to enable non programmers to add additional types of sensors to off the shelf display software. As a requirement has arisen for more complex sensors, your quad thermistor and a possible thermal meter it makes sense to cover these by having a variation with more pins but keeping to the original philosophy.
None of the off the shelf software really is capable of driving outputs and running a control system (LogTemp has a limited capability, but having to use a PC is a bit OTT), so any output clones are a bit of a bonus.

I think that once you start to develop a control system then you are into the rather exclusive world of programming and I tend to agree that although it's possible with 1 wire its arguably not the best way to go. 
If you don't want to do a combined data collection and control system it would seem to be useful to consider how to pass information from a control system into the data collection via some sort of bridge.
 

[StBarnabas]

Eric
it it a shame you are so far away as it would be very useful if the three of us could meet up (Ken, StB) and yourself. I will be having hopefully interesting discussions with Ken over the next few days and not just about sensing networks.


By chance one thing I looked at today was a 3D accelerometer (about £4.00 or £10.00 on a board with three analogue outputs). This would be ideal for monitoring pumps, wind turbines etc.

Sean

[benofishal]

Hi,

I would like to thank StBarnabas for sorting me out with several PICs for the humidity sensor build, you have been very kind. I am just awaiting the delivery of the sensors themselves and will post how I get on here.

Thanks.

[red_kooga]

Hi Eric,
I am interested in the DTH11 PIC code if it is available, I bought 2 of these sensors hoping to use them with PIC's I have done some assembler and oshonsoft basic but just can't seem to get my head around the timing for reading from the DTH11 sensors.

Thanks richard

[ziporah]

Hello Eric, St Barnabas

I'm Jo Vanvoorden, located in Belgium. In my normal living, I'm a linux system administrator / team leader.
In my spare time, I've recently built a mendel prusa, I play all the time with electronics, but I'm not a programmer guru.

I've currently got a 1-wire network with about 20 ds1820's in house to monitor my temperatures all trough the house. (rooms, solar boiler, ...)
I would like to expand my network with electricity, gas and water counters. I would also love to connect my dht11 sensors to the network.
I was wondering if it would be possible to provide me with the hex / asm code files for the ds2423 replacements in order for me to extend my network.
I've got some C/ C++ knowledge, I've tried asm last year because I wanted to change some dmx asm code, to allow it to react more like an enttec usb pro, but my asm knowledge wasn't that fresh anymore, so that task failed.
I'm also in the progress of creating an energy monitor for my solar power system, the components have just arrived. I'll probably use some jeenodes to built that system.