RTC on Pandaboard

Is the CJE RTC module for the Pandaboard supported yet in ROM builds please (and, if so, what module/version do I look for)? I currently use RISC OS 5.21 (03-Sep-2013).

Support has been built in for about a year. It is automatically detected but you do have to set the clock to not use a time server, I think it defaults to timeserver.
Support was broken for about one? month, It was fixed mid December. I think the 03.09.13 build should be o.k.

Support for external RTC modules was added on 2013-11-12 into ROOL CVS. Any ROM build after that date should have it.

Oops sorry I was thinking of the Pi.

I set the clock to “set manually” in the Time and Date configuration, i.e. not nettime,; but, when I powered the Pandaboard up the next day, it gave the time of when I shut down the day before. :-(
Panda has the RTC with battery and runs RISC OS version 5.21 dated 06-Jan-14.

@Will

Many thanks.

Support for external RTC modules was added on 2013-11-12 into ROOL CVS. Any ROM build after that date should have it.

Do you have a note of any milestones (such as this) in the Panda ROM builds? I would like to improve the user documentation by explaining when some of the facilities (such as this) were added.

For example is reset (by expansion connector switch) and reset (by software) supported yet?

@Chris

Here are some “milestone” of the last year for OMAP4:

2013-11-16 Audio module changed to use API 2.0 (for stereo reversal)

2013-11-12 added support for external RTC and power control hardware

2013-09-02 added SmartReflex driver

2013-09-01 added support for all 4 I2C interfaces

2013-04-13 added GPIO device

2013-04-07 Audio interface: removed samplerates which are not supported by hardware
           [88200 and 96000 are directly supported by hardware]

I have sent in some changes for solving the reset problem, but they are not yet added into CVS.

@Willi

2013-11-12 added support for external RTC and power control hardware

So can you tell me how to get the RTC going on my Pandaboard, because setting it to “Set manually” in the Configuration application doesn’t keep the time on rebooting the computer. My Pandaboard has the RTC hardware and battery and is running RISC OS V5.21 (28-Feb-14).

@David

Do you have a CMOS widget fitted?
If not try saving your CMOS settings (e.g. SaveCMOS ram:$.cmos) and write it to your boot drive.
Is the “Set manually” still set after a reboot?

Otherwise, do you have something in your boot sequence which is responsible for this behaviour?
Try !Reporter for analysing whats going on on your system.
Or alternatively start with a clean !Boot – after saving or renaming your current one – from the HardDisc4 image.

@Willi
Yes, the CMOS widget is fitted. I have just opened up the case to make sure. The widget is in the JTAG socket near the RTC gadget with battery. I will try your suggestions. I notice that Choices.Boot.Tasks.Timesetup has only some commented out lines after I configure to set time manually; but, when configured for nettime, there are several lines relating to nettime, like the time server. The permanently commented out lines that seem relevant are
Format:6
Locality:UTC+00:00_UK_’GMT’
ServerSelector:0

@Willi
I have used a clean !Boot from HardDisc4 of 4th March and added in only enough from the old !boot for the monitor settings. Set the Time and Date in the Configure application to “Set manually”. But on the next boot up, Alarm gives the time of when the Pandaboard was last shut down. In the Configure application, the Time and Date configuration is still set to “Set manually” after a reboot.
I think settings are being saved to CMOS as I have done a saveCMOS on two occasions, but with different settings (one with “Set manually” and the other with “Set from the network” and the two files are different, according to !SideDiff. For the latter, I set up Network in the Configure application.

Next, I got !Reporter set up to report on the boot up, but I couldn’t see anything wrong.

@David
Switching the time settings between “Set manually” and “Set from the network” does not change the content of the CMOS – at least on my Panda. !SideDiff has sometimes problems when comparing binary files. I have used !Diff (from ROOL’s DDE) to compare both versions of the CMOS content and they are identical. The only noticable change is in the file Choices.Boot.Tasks.TimeSetup, and I can confirm what you have already posted.
What does happen on your system when you remove the external RTC module? My Panda starts in this situation (using the internal RTC of TWL6030) with a default date of 2-Jan-1970.
There was a similar problem for the RPi Maybe you can use Jeffrey’s code for checking that the RTC module is working correctly. But for the RPi the problem was related to SDCMOS which is not part of the OMAP4 ROM from ROOL.

David,

The RTC should work even without running your !Boot. If your monitor can cope with it, then it might be useful to try disabling running !Boot to make sure there’s nothing there that’s causing the problem.

Also, can you read the temperature from the on-board temperature sensor on the RTC? Go to http://www.cjemicros.co.uk/micros/products/resources/PandaTempDisp.zip to download software to try it. If this doesn’t work, then it may be a faulty RTC board. I’m sure we’ve had numerous discussions by email, phone and at Shows regarding sending the RTC back to us for testing, but I don’t think you have done?

pp. Chris

@Willi and Chris
With TimeSetup set to Manual, I booted up Panda with shift key held down to avoid !Boot, but this made no difference: It still gave the time at last power-off. Thanks for PandaTempDisp and this appeared to work correctly as its output temperature did vary. Also, with the RTC board removed, PandaTempDisp gives error “No acknowledge from IIC device 1290”, which is to be expected, presumably.
Also, with the RTC board removed, I can confirm that the time after bootup is 2/1/1970.
I have run Jeffrey’s BASIC program without RTC board, but changing the dates and times in it to today’s date and time (22:30) and it output the following.
20140306 22235800
20140306 22300000
Press SPACE or …… to continue.

The first line makes sense because I ran the program at that date and time (to the nearest minute) and the second line is due to my tweak of the program. After exiting from the program, the clock output on the iconbar (from Alarm) was 10:30:00 pm. 6/3/14 as expected. Similar results with RTC board fitted some minutes later.
I need to correct something I said in my post of 2 days back. When the date and time is set to “Set Manually” in the configure application, the lines in Choices.Boot.Tasks.Timesetup do not include ServerSelector:0 . This seems odd to me as that line is included (with several further lines) when “Set from the network” but commented out.

David,

Ok, from what you say, it would seem that RISC OS is able to talk to the RTC board and clock chip. The symptom of the time restarting from when you last switched off could possibly be down to a faulty crystal on the RTC board not ‘ticking’.

Once again I shall reiterate the offer that if you would like to send your RTC board back to us, we can test it for you and replace if necessary.

pp. Chris

If you’re having trouble, you might wish to try one of our RTC boards, which should work out of the box, and don’t require anything fancy extra to my knowledge. I can’t comment on CJE products, but I can say ours have been in the field for well over a year now with no problems. You will, however, need the pins in place on the Panda expansion header, as they aren’t present by default.

@Chris

I have just posted (first class) the RTC board including 3V cell battery to you for testing and replacement if necessary.
I know, when I have raised the mattter, you have offered to test the board; but the reason I have not taken you up on it until now is that there was always this problem with the OS not supporting the RTC and so it seemed that there may not be anything wrong with the RTC board.
By the way, the PandaTempDisplay gives a much lower temperature (16 degrees C lower, approximately) than CPUTmpMon, presumably because the latter measures the higher temperature of the CPU and the former measures what, the temperature of the air inside the case or the temperature of the RTC board?

At the risk of stating the obvious: the reading returned by the RTC board will be that of the sensor on the RTC board.

Since the RTC board hardly consumes any power, it’s reasonable to assume that the sensor will be at pretty close to the temperature of the RTC board.

The RTC board will be pretty much at the temperature of the air in its vicinity; very slightly higher because of heat conducted from the main board through the connectors between main and RTC boards.

And, since the thermal conductivity through the boards and connectors is very poor, that leaves the RTC board measuring pretty much the temperature of the air in its vicinity.

The air in the case is presumably not fanned, so it will be moving slowly, so you would see significant differences in air temperature within the case. But you probably don’t care.

(Long explanation in order to try to get you thinking through the physics involved.)

I have just posted (first class) the RTC board including 3V cell battery to you for testing and replacement if necessary.

Ok, thanks, will let you know how we get on.

I know, when I have raised the mattter, you have offered to test the board; but the reason I have not taken you up on it until now is that there was always this problem with the OS not supporting the RTC and so it seemed that there may not be anything wrong with the RTC board.

RTC support has been in ROM for over a year…

By the way, the PandaTempDisplay gives a much lower temperature (16 degrees C lower, approximately) than CPUTmpMon, presumably because the latter measures the higher temperature of the CPU and the former measures what, the temperature of the air inside the case or the temperature of the RTC board?

The temperature sensor on the RTC board basically measures the temperature of the sensor on the RTC board. As Dave Higton says, this will be effectively the localised temperature of the air surrounding the RTC board in the case. Obviously, this will be different to the temperature of the CPU core.

pp. Chris

@Dave and Chris

OK, so the temperature measured by PandaTempDisplay is as I guessed.

I was not the only one who thought that RTC support in the ROM has been broken for a long time: See Chris Hall’s post of 26th Feb that started this thread and Chris’s post in response that says that support was broken for about a month. I don’t remember any announcement being made unless buried in these forums and there is a lot here to go through if you are not sitting in front of a computer all day! If Chris Hall doesn’t know about it, I am sure I won’t! :-)

“See…Chris’s post in response that says that support was broken for about a month. " I followed that up with a correction to say I had been talking about the Pi RTC support, not the PandaBoard.

Also a clarification. RTC support has been in builds that Willi has done and supplied by us to purchasers of PandaROs but was only added to ROOL builds since 2013-11-12.

David,

Received your RTC board this morning thanks, and it was indeed a fault with the crystal. This has now been resolved. I’ve tested the board and time does now continue to tick by when the Pandaboard is turned off! I’ll post it straight back to you. Apologies for the inconvenience.

pp. Chris

Received RTC board yesterday, thanks. Yes, it keeps beating when the Pandaboard is off. Even overnight the RTC didn’t sleep and Alarm showed the correct time in the morning!
So, is it a new RTC board or did you manage to repair it?

So, is it a new RTC board or did you manage to repair it?

Remote diagnosis: dry joint on the connections to the crystal. Resoldered.

If Chris says yes I tick a small box.

So, is it a new RTC board or did you manage to repair it?

I was able to repair it.