Hi,

this topic is about the PCB for a wireless master in the OpenHR20 
project.
A thread about the OpenHR20 firmware can be found at 
http://embdev.net/topic/118781

A german article about OpenHR20 can be found at 
http://www.mikrocontroller.net/articles/Heizungssteuerung_mit_Honeywell_HR20 
(information might be a bit outdated).

Some schematic of a PCB can be found in the SVN repository at 
http://openhr20.svn.sourceforge.net/viewvc/openhr20/rfmsrc/doc/master_board/

The 10-pin connector is the same as the one on HR20, so there is JTAG 
and a RX/TX. For connection to a router as WRT or FritzBox, a 
serial-to-USB converter might be necessary. But I'm not sure whether a 
FTDI driver would work on a FritzBox.


Feel free to contribute if you are interested in designing or if you 
already have a prototype and want to share your findings.

Regards
Marco

Ok, seems that the project Freetz [1], a custom firmware for FritzBoxes, 
supports the FTDI serial converters [2].

Would it be better to integrate a FTDI converter on the master board, or 
connect two boards using the 10-pin connector?

And is it ok to supply the master board from USB?



[1] http://trac.freetz.org/wiki/freetz (german)
[2] http://trac.freetz.org/wiki/libs/ftdi (german)

my first try (I obviously not completed routing yet ;)

ok, next version :)

If it is connected directly to a serial interface, the FTDI USB 
converter part can be cut.
All parts except the coil are mounted on the top, and even the coil 
could be mounted on top if it is isolated a bit.

I've no idea what kind of antenna should be used on the RFM12 modules, 
so feel free to contribute, also if you find some errors :)

br
Marco

Hi Stan,

Yes your board looks good. I have add ,at my master, a second led which 
show the transmission of the sync packet. This looks usefull for 
diagnostic purposes. Maybe it could usefull to connect some additional 
portpins to pad for expansions.

bye

Ronny

Why not use a mini USB ? like 
http://www.reichelt.de/?;ACTION=3;LA=444;GROUP=C126;GROUPID=3213;ARTICLE=52003;START=0;SORT=artnr;OFFSET=100;SID=28f4vsnawQARwAAEpdONg3c5eae9c8e6da654f04e58b63c91e9bb

Ronny

Hi Ronny,

great idea! Are these four enough? Is it ok to use the INT-inputs? They 
are located at a free corner ;)

br
Marco

Hi,

here are the changes for version 0.4:

jdobry wrote:
> - current SW need some communication buffers and use ATmega32 not
> ATmega16. Pinout is same, difference is volume of RAM (and Flash)
Atmega16 changed to ATmega32.

> - RTS# signal for RESET# can be problem if we want to use simple
> terminal for debugging. Can you add jumper into this signal?
done

> - Antenna can be lambda/4 wire, for 868 it is 86mm (L[mm] =
> 75000/frequency [MHz])
> - Antenna can be path on PCB, due to PCB effect it must be 10-20% shorter
> than wire. Wire is usually better.
ok, I keep the the via to solder a wire.

> FSK pin is not used and in this case it need pull-up resistor. My
> original design use pull-up inside MCU, it save soldering of one
> component. Just a trick, nothing more.
external pull-up resistor added and connection to ATmega removed.

> Please remove R2+C5 components. ATmega contain better startup reset
> circuits inside and it complicate external reset signal connection
> (capacity load).
well, if they are not populated there's just 1 mm of extra wire on the 
PCB.

Test point number four is replaced by an extra LED as suggested by 
Ronny.

Any more suggestions?

BR
Marco

here's some kind of preview :)

Of course L1 (at the bottom) is not a resistor, but it is the most 
similar package.


Another idea:
would it be better to rotate the RFM12 by 180° so that the antenna is 
located at the left edge of the PCB? Or is it ok right now because a 
vertical antenna can be used?
What do you think?

Thanks Jiri :)

I'll add a point for soldering a GND connection and tomorrow I'll place 
my order.

Hello People,

I will modify my code to work with your LED port and send a patch in the 
evening.

Ronny

Marco G:

You can compare rendered image with real here: 
http://openhr20.svn.sourceforge.net/viewvc/openhr20/rfmsrc/doc/master_board/master_without_ftdi.jpg?revision=274
Only one difference except missing FTDI is used inductor. I was use 
10uH.

Could you send me source files for update documentation?
Today I plan to replace old prototype board with your PCB and in this 
case source code will not fit into schematics in SVN. Thanks, Jiri

OPPS!

I found problem. This design can be used without FTDI with 3.3V power.
But with FTDI and USB power RFM12B will be overpowered.
Recomended power range for RFM12B is from 2.2V to 3.8V
5V from USB will not damage it (maximum is 6V) but it is outside range.

Jiri

Yes you're right, I intended to use a RFM12 (without 'B') on master 
because of the 5 V USB supply.

AFAIK there's no difference between RFM12 and RFM12B, but I'm not sure.

here is differences between RFM12 and RFM12B:
http://news.jeelabs.org/2009/05/06/rfm12-vs-rfm12b-revisited/

Its easy to fix: remove L1 and use the 3.3V Ftdi Internal 
Voltageregulator it drive up to 50mA - is that enough for the rfm12???

@Ralf H.:

50mA is enought

- RFM12B have 26mA maximum
- ATMEGA32 (5V/10MhZ) need 15mA maximum
- and remaining 19mA for LEDs

Problem is that you must crop VCC layer on back side, because FTDI must 
be powered from 5V, remainnig part from 3V3

PS: I was also use RFM12B and I miss information that pull up on FSK is 
not needed. It is not problem on this PCB, but I was can save soldering 
of one wire and resistor inside valves.

Jiri Dobry wrote:
> Could you send me source files for update documentation?
> Today I plan to replace old prototype board with your PCB and in this
> case source code will not fit into schematics in SVN. Thanks, Jiri

I'll send them to you this evening :)

Ralf H. wrote:
> Its easy to fix: remove L1 and use the 3.3V Ftdi Internal
> Voltageregulator

It's not that easy:
VCCIO must be connected to 3V3OUT so that the UART pins are driven with 
3.3 volt. Otherwise TX would still have 5 volt high level which would 
cause problems with ATmega.

> VCCIO must be connected to 3V3OUT so that the UART pins are driven with
> 3.3 volt. Otherwise TX would still have 5 volt high level which would
> cause problems with ATmega.

I thing that it is easy, see to the pictures.

- 1 wire to crop.
- 2 new wires
- do not use inductor

Connection on bottom side for modification.

sorry, I haven't got any left.

But I would fix the bug mentioned above and order some new ones. The bad 
thing is that they need at least 2 weeks to arrive...

Hello,

i have one board more then i need. but not the simular design. its a 
little bit smaller. and some components are in 0402. I get these boards 
next week.

Ralf

I've added my schematics to the repository, including a fixed version 
0.6

FTDI TX/TX LED have wrong connection. This signal is inverted and open 
collector.

That might be the reason why that LED never showed anything on my board. 
Thanks for the finding!

Has anyone tested this circuit with the FTDI and powered from USB? I am 
facing the problem that there seems to be no communication from the FTDI 
to the ATMEGA or from the FTDI to the PC.
I can not receive or send anything from the PC.

I am running it for nearly 3 years now, powered from USB. With the 
corrections shown above.

I have used the board and schematic 0.6 from the SVN without any 
changes. Which settings are you using for the UART connection? I tried 
38400 baud with 8n1 and do not receive a single Byte in my terminal 
program.

Which Atmel do you exactly use on your hardware? The schematic says only 
"Mega32-A" which is not really a correct part number. Maybe I have 
ordered the wrong controller? But I was able to flash an debug the 
software on the controller. Sadly I have no experience with Atmel 
controllers and the AVR Studio..

Version 0.6 should be fine. 38400 baud and 8N1 is also fine.

I am using this one: ATMega32-16 AU
http://www.reichelt.de/Atmel-ATMega-AVRs/ATMEGA-32-16-TQ/3//index.html?ACTION=3&GROUPID=2959&ARTICLE=50553&SHOW=1&START=0&OFFSET=500&;

Does the FTDI enumerate correctly? Have you used the FTDI programmer 
utility to change the EEPROM seetings to 500 mA?

Yes the FTDI is correctly recognized by the PC and I am able to connect 
to the serial line with putty. I have changed the power setting in the 
devices EEPROM now, but this made no change. I still do not receive 
anything from the ATMEL controller.
I checked the voltage on the Atmega and there are correct 3.3V on the 
device. Also I am able to debug the controller and there does not seem 
to be any problem with the hardware.

I am using this controller:  ATMEGA32A-AU

I don't know.. I have found a fuses section in the code and checked the 
controller fuses against these settings. Now I got these fuses:

FUSES =
{
    .low = 0xA0,
    .high = 0x91,
};

The RFM12 module is also soldered and has the correct voltage supply. I 
do not hava a oscilloscope so I can not check the clock. But would it be 
possible to debug without proper external clock?

Hi,
my problem with the master is still not solved. It seems that the UART 
generated by the ATMega ist not with the correct speed.I have attached a 
picdump from my logic analyser. As you can see the trransmit from the 
controller to the FTDI is really slow. On the RXD line sou can see a 
transmission from the  FTDI to the ATMega with 38400 baud.

Has anyone a clue why the controller sends so slow? I have checked the 
clock from RFM module which is at the correct 10MHz.

Alright my problem is solved!
In the init routine for the UART I had to switch the setting of the 
baudrate and the other settings. First the UART is set up properly and 
then the baudrate is written into the registers. I don't know why but it 
works now!

Well now comes the real work. How can I figure out if the master is 
working properly? I have found several commands in the code like 'H' or 
'Y' but there are few more. Is there any documentation about the 
commands?

Is it correct that I have to set the time with the 'H' command first 
before the master tries to send anything?

You need to run daemon.php in \repo\rfmsrc\frontend\tools to interact 
with the master HW.

I used WAMP package for testing on my Windows machine.
Now is runs on my Fritzbox 7270v3.

How have you used this on windows? PHP is normally not able to access 
the serial interface on windows.

You're right. I have used SerProxy and then connected to it like:

$fp=fsockopen("127.0.0.1",5334);

http://www.lspace.nildram.co.uk/freeware.html

Okay, seems to work now - somehow..

Do I understand it correct, that the daemon.php is the interface to the 
serial console. It takes command from the database and puts them on the 
console? The frontend uses the same database to creat commands which are 
then processed by the daemon?

When I use a terminal program to view the serial output of the master 
then there are sometimes lines which begin with "ERR". Are these 
received packets from unkown nodes?

All your observations are correct :)

I get about 5 times more ERR packets than regular packets (with 5 
valves), so I think that is ok.

I have some master boards left. If anyone is interested, please let me 
know.

Hi,

I have changed the connection at the FTDI Communication LED and resized 
the whole Board. Now it's not possible to cut it into two pieces, if you 
would use it only with 3v3 uart. But I think thats not a problem - the 
whole Board is now smaller then the "uart only" part in version 0.6. 
Also I deleted the testpoints (TP1, TP2, TP3). I think the testing phase 
is over and if anybody need an debug output, he can solder it direct to 
the atmel.

In the attachment is the new layout - i think anybody could commit it to 
the svn (or git [or both]).

Greets,
Michael

Tomek Z. wrote:
> I have some master boards left. If anyone is interested, please let me
> know.

Is there anybody else, who got a board left? Plz drop me a line.

Greetz
Martin

Hello,

I had ordered some master boards with my new layout... I think, you can 
have one of them. But they are shipped from china an haven't arrived 
till now.

If you want one, write me a message and i will notice you, if i have the 
boards here.

Greets,
Michael