CC1310: RF HF Oscillator stops working

Hi Martin,

On the non-working devices, what is the VDDR voltage?

Did you by any chance probe the 24 MHz crystal on the non-working units before they failed? This can potentially damage the internal DCDC regulator.

Cheers,
Fredrik

Hi Fredrik,

The VDDR voltage on each of the failed boards is between 1.670V and 1.675V. This is comparable to the VDDR voltage of the LaunchXL board when the transmitter is not on. I did not probe the crystal until after each board had stopped working and I used a Tektronix FET oscilloscope probe with an input capacitance of <1pF.

Since yesterday, I have created a test firmware load that enables the HF XTAL oscillator on power up and loaded it into my failed units. The oscillator on each unit started up correctly so it is not the oscillator that is preventing the radio from working. Also, I can program the units and debug my code so I don't think that it is a problem with the JTAG.

While I was making the VDDR measurements, I connected one of my failed units to SmartRF Studio and tried to transmit a 915MHz CW at 14dBm. Right after I pressed the Start button on SmartRF Studio, I noticed a transmission at 915MHz for a fraction of a second and then after a couple of seconds SmartRF Studio gave me the status of 0xbad0 again for the response to the radio commands. I tried the CW again and I was able to transmit the carrier wave this time. However, after that, I could not get the unit to transmit again. I tried at 0dBm and I also power cycled both the unit and the JTAG and restarted SmartRF Studio but I keep getting the 0xbad0 status. I'm running my boards off of a bench top power supply with 1A current limit.

1. What does the 0xbad0 status from SmartRF Studio mean? Is that the actual response from the radio MCU or does that mean that SmartRF Studio did not get a response?

2. If it's from the radio, what would cause that status response? If it is not from the radio, is there any way to get more information from SmartRF Studio about the error?

3. What would cause the radio to stop transmitting immediately after starting the TX_TEST cmd? Why would it work some times and not others?

Thanks,

Marty

Hi Martin,

At least it does not sound like your devices are physically damaged. The intermittent nature of your problem is strange though. The 0xbad0 response comes from the RF core and is a bus fault message meaning that the system has tried to access something without success. What it usually means though is that there is a problem either with power supply or the clock.

My best guess here would be that there is some kind of weakness in the HW design. If you can send me your layout and schematic through a private message I would be happy to take a look.

Cheers,
Fredrik

Hi Martin,

I have taken a look at your design files and the PCB layout needs a respin. There are multiple design weaknesses that may be causing the issues you are seeing.

One of the major items is the layout of the DCDC components, so you could try to disable the DCDC and see if that improves anything. Since the DCDC regulator is a switching buck converter it is very important to place the output inductor and capacitor close to the switching output put. Equally important is to make sure the output capacitor has a short and low impedance ground path back to the CC1310.

Another problem in your design is decoupling. Decoupling capacitors must be placed as close to the pin they decouple. As with the DCDC, it is equally important to consider the ground return path. If this is on the opposite layer (on a 2-layer board) it is important that the ground path is not obstructed by another signal trace.

Also, the layout in the reference design must be followed closely for the RF components to ensure proper RF performance.

Even though these wiki articles are written for CC26xx they still applies to CC1310:
processors.wiki.ti.com/.../CC26xx_HW_training_Layout_Considerations.pdf
processors.wiki.ti.com/.../CC26xx_HW_Checklist

To summarize:
- DCDC components must be placed as in the reference design (close to DCDC-SW)
- Decoupling caps must be placed as in the reference design with focus on ground paths
- VDDR (pin 28) is missing one 100 nF decoupling cap
- DCOUPL cap is missing (this is required for proper operation, but not including it will most likely not permanently damage the device)
- RF components must be placed according to the reference design
- Add more ground vias, close to critical ground nodes specifically and all over the board generally

Regards,
Fredrik

Thanks, Fredrik. I believe that this will resolve our issue. Is there a 2-layer reference design for the CC1310 4x4 IC? I only see 4-layer reference designs on the website. If possible, we would like to use a 2-layer design since this is a consumer product.

Marty

Hi Marty,

Sorry for the delayed response here. We only have a 2-layer reference design for the 7x7 package, and our general advice is to use 4-layer PCB. This is because we have seen that the third harmonic levels can be a bit (too) high when using 2-layer boards due to increased inductance in the ground vias. If you really want to use 2-layer PCB it should not be thicker than 0.8 mm and you should be generous with the ground vias. The 7x7 reference design can also be used as a reference for the 4x4 package in this case.

Regards,
Fredrik