Status 13 Dec

Testing the logic started out OK. Creating a file on Cimbaoth, it was read by the Raspberry Pi, which then signaled the power controller to cycle power which it did.

However, after running a while, it was clear that the power was also being cycled on occasion when it was not commanded.  This caused a need for more complex signaling between the Pi and the controller so it was harder to be misunderstood.

As of this point in time, the logic for recognizing a more complex signal, and for generating that signal, has been developed, but not yet added to the production controller.

---

ID	Estimate	Status	Description
917	2016-08-10	FIXED	Order NW Digital board
981	2016-09-11	WONTFIX	KC8ARJ as repeater
983	2016-09-11	FIXED	Order quality SD cards
986	2016-09-11	FIXED	Validate LTE
982	2016-09-12	FIXED	WiFi startup
994	2016-09-12	FIXED	Investigate hang time control
978	2016-09-13	FIXED	Select UDRC/Pi Case
984	2016-09-15	FIXED	SD card receive
995	2016-09-15	WONTFIX	Investigate hang time in source
979	2016-09-16	FIXED	Receive UDRC/Pi Case
987	2016-09-19	FIXED	Select LTE provider
980	2016-09-20	FIXED	Modify/Install UDRC/Pi Case
988	2016-09-20	FIXED	Order LTE Device
989	2016-10-04	FIXED	Receive LTE device
991	2016-10-10	FIXED	Log file strategy
1033	2016-10-15	FIXED	Power control strategy long
1032	2016-10-19	FIXED	Power control strategy short
1034	2016-10-22	FIXED	Design power control
1035	2016-10-23	FIXED	Breadboard power control
1037	2016-10-29	FIXED	Order power control parts
1040	2016-11-02	FIXED	Receive power control parts
1038	2016-11-03	FIXED	Order power control PCB
1036	2016-11-03	FIXED	Power control PCB
990	2016-11-06	FIXED	Configure/Test LTE Device
1039	2016-11-15	FIXED	Receive power control PCB
1041	2016-11-20	FIXED	Construct power control circuit
1044	2016-12-04	ASSIGNED	Package power control
985	2016-12-09	ASSIGNED	Schedule 2030 install
992	2016-12-14	ASSIGNED	Make backup SD
1042	2016-12-16	REOPENED	Develop short-term firmware
993	2016-12-24	ASSIGNED	Move to 2030
996	2017-01-07	ASSIGNED	Test 100 watt amp

---

---

ID	Estimate	Status	Description
1044	2016-12-04		Package power control

2016-10-27 09:41:50:  

Develop physical packaging for repeater power control circuit

2016-11-14 08:36:04:  

Holes cut, plugs/jacks mounted, now need to wait for PCB.

2016-11-22 08:12:49:  

Still need outlet

2016-11-26 11:58:05:  

Still need more time, hope to do today

2016-12-02 15:03:29:  

Still having trouble with packaging

2016-12-05 08:11:02:  

Got box to work (sort of), had been connected to a poor choice of
Raspberry Pi pin.  But a new issue emerges.  On boot, the RPi pin
is not an output, must be pulled up because it goes high in spite
of the PIC's weak pull-down. Once getit runs, it gets set low,
and a minute later the plug is pulled.

After about 20 seconds, the cycle starts all over.

---

ID	Estimate	Status	Description
985	2016-12-09		Schedule 2030 install

2016-09-12 10:30:56:  

Schedule a time to move the repeater to 2030

2016-11-26 11:58:56:  

Predecessor slipped so slip this one, too.

---

ID	Estimate	Status	Description
992	2016-12-14		Make backup SD

2016-09-11 09:49:16:  

Need to create a backup SD card immediately before taking
repeater to site

2016-09-12 10:32:15:  

Create a backup of the SD card.  Moved date to the day before
install since this wants to be the LAST thing we do.

2016-11-26 12:01:29:  

Must be 993 - 1 day

---

ID	Estimate	Status	Description
1042	2016-12-16		Develop short-term firmware

2016-10-27 09:41:49:  

Design and test firmware for the short term repeater power
control, both PIC and Pi

2016-11-14 10:23:54:  

Largely complete
commit 56e57f72d9b15bf1256c4f30ba4dd6a59e5bce3c

Still want to make a new branch with doxygen

2016-12-13 09:22:04:  

Original firmware worked, however, power cycles every 30 minutes
without the Pi software running, so something else was signalling
the controller.

Moving to a more involved signal, where three pulses of specific
length are required to signal the controller to cycle power.

---

ID	Estimate	Status	Description
993	2016-12-24		Move to 2030

2016-09-12 10:54:35:  

Install the repeater at the permanent site at 2030 building.

2016-11-26 11:59:18:  

Predecessor slipped so slip this one, too.

---

ID	Estimate	Status	Description
996	2017-01-07		Test 100 watt amp

2016-09-12 10:49:37:  

Repeater will initially be set for 20 watts. It is unknown at
this time whether the addition of an external amplifier will
result in excess desense. After the repeater is at the site, an
amp will be tested for coverage.

2016-11-26 11:59:58:  

Predecessor slipped so slip this one, too.

---

Power Control Revisited

The previous power control version used a PIC24FJ64GM002, a 28-pin part with 64K of flash.  For the simple problem at hand, that PIC was gross overkill, but it happened to be what I had on hand.

The PIC24F04KA200 is much better suited to the task.  It still is something of an overkill.  Although 4K of flash is pretty tiny for a PIC24, the application ended up taking less than a tenth of that.  Still, it's 14-pin package and low price make it a better choice.

After some experimentation, during which I got concerned about the base current, I decided to put each of the relays on their own output.  The PIC24F04KA200 can source 18 mA per output pin, and the final base current was measured at slightly over 4 mA so there is plenty of headroom.

Basically, the board is supplied with 5 volts which is needed for the relays.  The relays are operated by transistors so the PIC doesn't need to see the full current of the relays which it would not be capable of handling.  A regulator supplies 3 volts for the PIC, chosen because the PI outputs are 3 volts.

On a signal from the Pi, the PIC waits long enough for the Pi to shut down, interrupts the power for a short time, then re-applies the power which causes the Pi to reboot (the Pi is powered by the repeater).

The final circuit looks like this:

Power Controller Schematic

A printed circuit board was laid out, the layout looks like:

PCB Layout As Designed

The 6-pin header on the left is for programming the PIC, not shown on the schematic.  Also not shown on the schematic are two ways to apply power; the two-pin header on the left and the microUSB on the bottom.  Depending on how the packaging works out, one or the other will be used.

The "realistic" view of the PCB is:

"Photo-Realistic" View of Printed Circuit Board

The actual board is not expected to arrive until November 15.  I selected the high-priced spread (Accutrace) because they are about two weeks faster than my normal supplier (MakerStudio).  They are also extremely diligent.  I covered some of the microUSB pins with solder mask because that makes it easier to solder by hand.  They immediately emailed me with a PDF showing the issue and asking if that is what I really wanted.  That isn't normal but I'm surprised they picked up on that.

Power Control

With the better understanding of the Karma WiFi came the realization that some of the control we would like won't be available since we can't connect into the repeater over Internet, only out.

So, we would need some "pull" mechanism for control.  Thinking through that made us also recognize that, because the Raspberry Pi is powered by the repeater, removing power from the repeater also removes power from the Pi, meaning that we have no control at all.  This is messy especially because it takes time to arrange a physical visit to the site.

It isn't all dark.  We have various commands we can perform on the input.  And although we can't shut the repeater down, we can make it stop repeating which is almost as good.

However, we have seen the occasional case where we would like to cycle the power on the repeater.  That will take some external logic since when the power is off there is nothing available to turn in back on. Even after we get "real" Internet, we still have the issue of loosing control when we cycle power.

We have concluded that a small controller would do the job.  When the Pi sends a signal to the controller, it would wait long enough for the Pi to shutdown cleanly, then remove power for a while, perhaps 15 seconds or so, then re-apply power.

Block Diagram

Another approach would be to have a second Pi or similar box that could handle the sequencing but have separate power from the repeater.  The first approach looked simpler.

The power controller would be fairly simple; a couple of relays for the power, a transistor to drive the relay, and a PIC to receive the signal from the Pi and operate the relay.  The PIC I selected is overkill for the task, but it is still inexpensive and was a 3 volt PIC I had several of in the parts drawer. (The Pi is a 3 volt device, and although 5 volts is needed for the relay, it is simpler to drive a transistor with a 3 volt PIC than it is to level shift the input.)

Schematic

The circuitry has been breadboarded and tested. It remains now to lay out the printed circuit board, order the parts and PCB, and assemble the production version.

I have added the appropriate tasks to the bug list.  I find it helpful when doing things like this to be very granular, so there are a lot of extra bubbles on the map, but most are short or involve no effort (e.g. wait for PCBs).

The detailed list, including the completed items, will come in a later post.

Not shown here, nor in the to-do list, is the possibility of constructing a parallel controller for the W8QN repeater, run off the same Pi. There are available output pins.