X10 Community Forum

🔌General Home Automation => Automating Your House => User Modified Devices => Topic started by: dhouston on September 09, 2011, 04:40:20 PM

Title: RR Five-by-Five
Post by: dhouston on September 09, 2011, 04:40:20 PM
There's a DIY project I've wanted to do for some time but kept putting off. Now, given my age and health issues (70, serious spinal cord damage which I've lived with for about 50 years, Congestive Heart Failure for the past 9 years, lung cancer surgery 18 months ago, prostate surgery and a couple other minor surgeries in the past year plus a recent bout of atrial fibrillation), I think I need to get it done before I leave for med school. I'm wondering if there would be sufficient interest to justify the effort.

I call it the RR5x5 (i.e. five-by-five). It modifies the RR501, adding a daughterboard that replaces the PIC16C54 with two newer PICs and adds a few other chips. One PIC will handle PLC-IO and administrative tasks, communicating to the other chips via an I2C bus, a smaller PIC that would handle RF-IO & IR-IO, an 8KB FRAM, and a battery-backed RTC chip. It would transceive all housecodes (with user configuration). There would be an optically isolated serial port communicating with the main PIC, a galvanically isolated 50-ohm BNC antenna connector, an IR receiver and IR emitter controlled by the smaller PIC and possibly an RF transmitter, also controlled by the smaller PIC. The main PIC will have a bootloader so its firmware can be updated in the field. I've used the firmware (or variations on it) in the smaller PIC many times so don't think it will need updating. Thus, it will not have a bootloader but both PICs will also have ICSP pads so the firmware in both is updatable. WizNET has an external $25 Serial<->Ethernet adapter that could be used to put it on your LAN.

The RF transmitter will depend on whether I can find a small transmitter module with the same pinout & footprint for both 310MHz & 433.92MHz. I've bought 310MHz from WenShing in the past but the last time I inquired they wanted a financial statement for my company before they would even quote - that's nearly as bad as Google wanting my mobile phone number to create an email account. There is another Chinese company I've dealt with that can probably supply them but I've yet to request a quote.

This would not be a project for the inexperienced as desoldering the existing PIC can be daunting, even for those with experience. If I proceed with it, I'll provide step-by-step, illustrated instructions. You will have to be prepared to sacrifice the PIC16F54 if you have trouble desoldering it.

I do not know whether the European version of the transceiver uses a similar mainboard or, if it does, whether my daughterboard will fit in the housing. In the past, I've worked with a European dealer who might be interested in doing a European version, if needed.

I have X-10, RCA & IBM versions of the RR501 and all use the same board so it should work equally well with each. I'm about to order a new RR501 to see whether anything has changed.

The RR501 already has PLC-IO with 10Vpp output and a very good RF receiver so it provides a solid base to build on.

I haven't done any cost analysis yet but my seat-of-the-pants estimate is the daughterboard will cost $50-$75. It will be designed so it can be manufactured in China with only the need to program the two PICs upon receipt (the logistics of supplying preprogrammed chips scares me - in a former life long away and far ago, I managed a multimillion dollar international operation but we dealt with large, expensive machines that were difficult to misplace so I've little direct tracking experience with little things like 8-pin PICs in significant volume). I'm trying to design it so that the operation could be taken over by someone else once I get it to the point of release.

I'll provide free interface software for Windows, Linux and OSX to connect either via the serial port or an external Serial<->Ethernet adapter. I'm considering making all of the code open source but, since it's all written in various dialects of Basic, real developers might not understand it.

I need to get a feel for whether there might be sufficient interest to justify going ahead with it.

Oh, and I had planned an AirPad application to interface with it but I may shelve that since I'm not a real developer.  :'  

I'm sure I've forgotten to include a few points - I'll probably remember them tomorrow - it goes with being 70.
Title: Re: RR Five-by-Five
Post by: Brian H on September 09, 2011, 05:13:26 PM
The new RR501 is different.
Main power supply is now 40 Volts DC and not 18 volts in the older units.
I never measured the Logic Voltages but you may want to look at this thread until you get a later model.
The author's web photos do show the basic different PCB layout.

Title: Re: RR Five-by-Five
Post by: dhouston on September 09, 2011, 05:30:46 PM
Thanks Brian.

The physical layout is more important than anything else. As long as they still used a PIC, I'm not concerned about voltages - I'll just take my supply from the supply to their PIC. I am concerned with being able to rigidly mount the daughterboard and hope that can be done in the same manner on all models. If the newer versions uses an SMD PIC, that could change the equation significantly. I have an RR501 and TM751 (the latter just for my own edification - not as mod fodder) on order.

EDIT: I've looked at the pictures mentioned and don't see anything that significantly affects my design plans. That might change when I get the one I've ordered. They have relocated some components but none are in areas that impact on my design mechanically. However, a major concern is whether they reduced the PLC output. If it has been reduced from 10Vpp, the attractiveness of the RR501 is diminished - it's the 10Vpp of the original design that made me consider this in the first place.

A couple of points I forgot to include...

The RTC chip has a calibration register so its accuracy is excellent. In any event, it can always be reset via the serial port. The battery drain is super low - it can power the RTC for 10 years so, unless someone tosses theirs on the shelf for several years, it's not likely to need replacing. There is, however, a low battery flag. It also has a timestamp for when it switches to battery so when power is restored we can see how long it was out.

The relay is sacrificed to provide room for the antenna (BNC) and serial (1/8" stereo jack) connectors. The outlet is always on.

For those who didn't catch the connotation of 5x5...
The isolated antenna connector allows for a better antenna and, when needed, a wideband preamp.
I've tested the eggbeater w/preamp to +300' (100m) although not with the RR501 RF receiver section. I've also tested with two preamps cascaded, further increasing the range, again not with the RR501 receiver section but I'm sure the eggbeater with cascaded preamps will discover if there are other X-10 RF users in the neighborhood.  :'

Microchip will program PICs with uploaded files before shipping. I'll probably try this with the Chinese supplier. We'll still have the ICSP pads to reprogram the PICs should things go awry.

My cost estimate is deliberately on the high side (I think) and I can refine it as things progress. It does, however need to have enough profit to attract someone to take it over once I'm at med school.

There may be other points I forgot.  

EDIT: For those unfamiliar with the term FRAM, it is a type of EEPROM that can withstand billions of erase/write cycles so we can use it not only for macros & timers but to track levels of all 256 addresses, allowing smoother transitions between dim levels for older switches and modules as well as saving the desired on level and restoring the system after outages. It costs more than regular EEPROM and, last I looked, the next size up is 32KB and it gets quite costly while 8KB is in the $3-4 range. Unfortunately, it is only available in SMD - otherwise I would provide a socket and let users decide how much they want (and want to spend).
Title: Re: RR Five-by-Five
Post by: dhouston on September 10, 2011, 09:43:52 AM
Europe already has an all housecode transceiver and their regulations allow for much, much higher RF signal strength so there may be less need for a modification like this. Plus, from the outward appearance of the TM13UAH, especially the antenna placement, I doubt that a board designed for the RR501 would work in it.
Title: Re: RR Five-by-Five
Post by: dhouston on September 11, 2011, 04:56:22 PM
Ramtron has a chip that combines 8KB FRAM and a battery-backed RTC. It saves about $1 over separate chips but that will probably just cover a 5V-3.3V LDO as the new chip is only available for low voltage. The ST M41T81S that I intended to use had a 1Hz squarewave output that I would have used as a time strobe for the main PIC. Without it I'll just use a counter in the PIC and query the RTC whenever the counter rolls over. And I think I'll just add some CMOS EEPROM - it's cheap so we will have 16KB for macros/timers plus 8KB FRAM for other things - tracking levels for all 256 addresses, randomizing events 30 minutes to give that lived in look while the house is unoccupied, generating delays in seconds, etc. The RTC has a calibration register so its accuracy should be just as good as the ST chip. If you use an ethernet/serial adapter, we can even sync the time to the internet a few times per day. I'll have to handle saving the power down time in code, whereas the ST RTC does it automatically, but the Ramtron chip has some features that should help with that (at least from my initial read of the datasheet).

Has anyone run out of memory in a CM15? This will have 3x the total memory of the CM15 which is likely to be overkill but the 16KB of CMOS EEPROM only adds about $0.50 in cost.
Title: Re: RR Five-by-Five
Post by: dhouston on September 14, 2011, 04:32:47 PM
The new RR501 is different.

I received the new transceivers today - date codes are Feb & Apr of this year but I don't recall which was which.

I don't see any issues with designing one PCB to work with both versions of the RR501 but the apparent lack of interest makes it look like I'll be designing something for my own use only.
Title: Re: RR Five-by-Five
Post by: janbo on September 14, 2011, 05:28:32 PM
I am having a hard time understanding the advantage of this over the CM15A. Please enlighten me.

Title: Re: RR Five-by-Five
Post by: dhouston on September 14, 2011, 07:08:32 PM
I am having a hard time understanding the advantage of this over the CM15A. Please enlighten me.
RS232 makes it much easier to connect to embedded devices than the USB connection needed for the CM15A. There was a thread on the forum in the past few days where someone making an embedded device that translates between protocols was bemoaning the lack of a CM15A with RS232.

10Vpp PLC output vs 6Vpp PLC output of the CM15A.

Full access to the RF input. RSSI readout - having a measure of signal strength can help with antenna decisions for transmitters. There are many HA related devices that use frequencies in the 310-312MHz range. These can be monitored and used to trigger actions.

REAL real-time access to the PLC activity. All X-10 devices report only valid X-10 activity - seeing everything, including collisions, can be a big aid in troubleshooting. RSSI readout.

3x the memory with 8KB that can be treated like RAM.

Battery-backed RTC that keeps accurate time during power outages and reports the time of the outage when power is restored.

I may think of a few other features.
Title: Re: RR Five-by-Five
Post by: janbo on September 14, 2011, 07:14:48 PM

Thanks for the reply. I now have a little better understanding.

Title: Re: RR Five-by-Five
Post by: dhouston on September 17, 2011, 01:01:16 PM
I have revised the RR5x5TM specs slightly. I've decided to use a single ZBasic chip, which has low-level X-10 functions built-in, instead of two PICs. As I'm also using ZBasic chips in my roZettaTM/ZarduinoTM projects it will be easier to maintain both for me now and for others after I leave for med school. Here are some links...