AK8963C Compass Module Intermittent SPI Communication with PIC Microcontroller

Hi,I have a PCB with a couple of PICs.  PIC 1 has a number of peripherals connected by SPI, one of which is a Compass module (AK8963C from Iphone 5) This Module was seemingly working ok, but has started intermittently working lately.I'm sure I've tried everything to find the error, with help from 'experts' with no luck.Where could the problem be, please?Camerart

Terry,
Let's start with the basics. You may have covered this before but it doesn't hurt to go over it again.
Since it was (apparently) working before, I'd start looking for what changed:First check all the connections for bad or intermittent solder joints or wires.
Second, what were doing to/with this device shortly before it stopped working? In particular, were you modifying the software or exposing the device to something in the environment (hot/cold, wet, spark)?If you were modifying the software, have you tried reverting to previous version?If it was environmental, it's possible that the device may be damaged. Have you checked the data sheet for the limits?Speaking of damage. Does it get hot when you plug it in? That's usually a sign of damage (especially if it wasn't hot before).

Since it's working intermittently, my first guess would be an intermittent connection. Unless you were modifying the code in which case that may be more likely that you inadvertently changed a register that you shouldn't have.

If you have an oscilloscope, you should check the SPI signals. Look for what is different between working and not. Especially look for signals with "funny" levels.

Hi E,Thanks for your reply.I have 5x Compass peripherals and 3x different PCBs, and have checked all connections.I've checked with an Oscilloscope and Logic analyser.I've tried earlier programs.Nothing gets hot.Today I was thinking about 'funny' levels.  I am going to look up SPI pullup/down resistors.Cheers, T

Hi,I've just tried Digital analysis and oscilloscope SPI READings.The analyser always appears to recognise the READings, and report sensible numbers.The oscilloscope shows 'raggy' READings, when COMP is working or not working.The DATA is arriving at the PIC MISO PIN ok, on all occasions, but works or doesn't.C

Very puzzling...Since the analyzer always appears to recognize the readings and gives sensible numbers, it's likely that the compass is working.I'm starting to suspect a timing issue. One of the things about the SPI bus is that you have one clock to clock data into and out of your device. If there is enough delay in the connections to or from the device, the data coming back can get skewed enough that it's no longer properly aligned with the clock. This is more likely if you have long cable runs, buffers, or isolators in the path. One way to test for this problem is to significantly slow down the SPI bus and see if it starts working...You can also check the alignment between the clock and data using a two channel oscilloscope. Pay attention to setup and hold time requirements on the PIC.

हमारे ब्लॉग pradhanmantri-yogana.in/ पर आपका स्वागत है| यह ब्लॉग आपको Pradhan Mantri Yojana के बारे में जानकारी प्रदान करता है| इस ब्लॉग से आपको pradhan mantri Jan Dhan Yojana, Pradhan Mantri Kaushal Vikas Yojana और कई योजनाओ की जानकारी प्राप्त होगी|
https://pradhanmantri-yogana.in/

Hi E,I think you may be on to something with 'timing'The wires from PIC to COMP are appr 200mm.  I've connected the Oscilloscope appr 30mm from the PIC.Here's my first look:You can see that the COMPASS I.D. =0X48 The Analyser shows 0X48 being READ, and the Oscilloscope shows the MISO switching on the CLK switch.C.

So is the PIC reading on the rising edge or falling edge of the clock?

Hi S,Rising, ( I thought the MISO should have switched HIGH before the CLK switching HIGH) so now I think this could be correct.It's just been pointed out, that perhaps the MISO READing voltage isn't high enough.C.

Terry,Yes, the MISO does look to be less than 2V which is probably not enough to be a reliable logic high on the PIC.Is this compass running off the same supply as the PIC? If it's running at a lower voltage, you may need a level shifter. That would introduce some delay but give you a better level.

Hi E,This seems to be the problem, hopefully.The PIC needs 2.3V to switch and the MISO is 1.65V, so I'll investigate.Will a MOSFET work as a level shifter?  What do you think of this one?: https://uk.rs-online.com/web/p/mosfets/6395556/ ?C.

The problem with using a MOSFET as a level shifter is that you invert the signal. You could use two  MOSFETs though... You'll want to look for a MOSFET with a low gate threshold.I've used 74xx type buffers for level shifting but finding one with a 1.65V in to 2.3V out could be difficult....

Hi E,I'm surprised! are you sure?  I presume you are aware there are N and P type, and I thought with the correct one, connected to 3.3V and switched by the 1.65V, this would simply lift the voltage to 3.3V.It is possible that this wouldn't be needed, as there may be an error with my system causing the lower voltage, I've now got to re-check.C

With N-type (like the one you referenced), you normally connect the source to ground, the gate to the input signal and the drain to a pull-up resistor and the output. Since a high level on the gate turns the MOSFET on, it will cause a low level on the output.I'm not sure how to connect a P-type in this situation.There may be another way to wire a MOSFET so it doesn't invert but I can't think of it at the moment. Maybe a P-type as a pass transistor?

Hi E,From memory: If  drain is connected to 3.3v and the source to the output, and a resistor to GND and the gate to the input signal through a resistor, then when the input goes high, then so does the source and the output.Correct me if I'm wrong.C.

Terry,Yes, you are right that would work but I don't think you can get a higher voltage out that you put in. In order to turn the MOSFET on, the gate needs to be at a higher voltage than the source.Perhaps a depletion mode MOSFET would work. If I remember right they have a negative threshold where you have to pull the gate negative with respect to the source to turn it off.
Here is a buffer that should work.http://www.ti.com/lit/ds/symlink/sn74lv1t34.pdfAt 3.3V supply, the high level input is 1.39V and output is 2.49VPropagation delay is 9 ns max. Not the best but pretty good for a level shifter.

Hi E,Ok, thanks.If my method doesn't work, I'll order some of the level shifters.They look handy anyway.C

Hi E,To clarify:The intermittent error is on the SPI MISO PIN of ak8963 Compass module, and in some occasions,  the voltage will be too low to switch the MISO of the 18F4620 PIC.
I have seen the voltage start at 3.3V, then error to 1.65V.  I've 'played' with the low voltage settings, but can't change the error.
I added a level shifter as recommended, and watched the output go from 1.65V then waver up to 3.3V, and started working correctly.  During this the level shifter input has been 3.3V so far.  As if the level shifter was allowing the MISO voltage to rise to the correct level.
The correction isn't 100%, but nearly.T

T,Glad that you are making progress on this. Since the voltage out of the compass module changes over time and is higher with the level shifter inserted, I suspect that there is something in the PIC that is loading down that signal. Perhaps a pulldown that is enabled on that pin when it shouldn't be?
The results that you are getting also imply a really low drive strength for the compass output. If there is a way to increase the drive strength that might help as well.E

Hi  E,I've tried a pull up resistor, but it seemed to pull up the whole short square wave.  I'll check for internal pull ups within the PIC.I agree, it appears that the Level shifter is allowing the PIC voltage to settle high, then outing a high voltage.   I'll investigate other ways to do this.Thanks, E