CC2640R2L: Help with impedance matching

Hi Eyal,

Your approach seems correct.

To help you troubleshoot, could you verify a few things:

1. Could you clarify this part of the statement: "I've calibrated the VNA by connecting to the cable, putting 0 ohm resistor at L2, and then using C43 as Open\Short\Load. After which I placed Short again and added time delay to set the Smith Chart."

2. Could you show the baseline S11 of your antenna? This can give us a lot insight as to why the smith chart is not moving properly.

3. What values are you using for the tuning components? At 2.4 GHz, the effective values for capacitors are between 0.1 - 5pF and 1 - 10nH for inductors.

4. Can you connect a Thru between port 1 and port 2 of the VNA and show me the S11 and S21?

Best regards,

Bun

Hi Bun

1. After connecting an Open, Short and 50 Ohm load (using 2 parallel 100 ohm) at C43 and finishing the calibration, I connected a Short again at C43 and used time delay to move the marker on the Smith Chart to the correct position. You can see Edelay -7ps at the pictures below that signify the time delay I entered.

2. This is the measurement I get right after calibration and L2, L4, C46 have a 0 ohm resistor (the rest are not placed). The probe is at C5:

As you can see the marker is basically outside the unity circle but I chalked it up for some measurement error. I decided to add a 1.5pF capacitor at C43 just to see that the marker moves correctly and to get the marker away from the edge of the chart and I got this:

Which seems rather reasonable and close to theory. Now I added a 2.2nH inductor at L4 (and removed the 0 ohm previously there of course) which should have brought me close to the 20mS admittance circle, however what I got was this:

The marker barely moved, as if I didn't put anything there.

I've tried several filters and something like this always happens, where it seems like the inductors have no effect.

3. The values you mentioned are the ones I'm using.

4. I did not calibrate Thru during my initial tests. I've recalibrated today and this is the result:

Thanks a lot,

Eyal

Hi Eyal,

Thank you for the thorough images. This paints a clear picture for me. Your baseline performance seems a little too capacitive as a 2.2nH inductor did not move it. Combine that with the negative delay which should be at least 200-300ps depending on the length of your probe/pigtail. There might be a tiny open somewhere in your circuit. Try these few things:

1. Check for a discontinuity .

2. Try a 100nH on L4 to see if the smith chart moves. This would tell you that a higher inductance is needed.

3. Can you show me the S11 of the thru measurement?

Best regards,

Bun

Thanks for the suggestions Bun.

I've removed all components from the board and tried again. This time the inductors did move the marker, but still the marker movement differed wildly from theory. After many, many, MANY attempts to understand and study the behavior I gave up on this circuit for now.

Instead I turned to a different PCB to see if I could get better results. I've attached the schematic of the RF path on this PCB:

This time I calibrated the VNA with its own open\short\load calibration kit, and then attached the adapter and coax cable and added electrical delay. This is how it looks before adding delay (the coax cable is floating):

This is the through measurement you wanted to see:

I've added a 230ps delay and connected the cable to C78's pad by scraping the GND next to it. I connected a 2.4GHz antenna to J13, put a 0 ohm resistor at L11 and left everything else not assembled and I got this:

This is the SWR\S11 of the same configuration:

For verification, I tried to use a spectrum analyzer with a reflective bridge as well just to see if it matches. I calibrated the spectrum analyzer (only has open load calibration) the same way I calibrated the VNA:

When I tried to verify the VSWR read after placing the 0 ohm resistor I got this:

So there's quite a disagreement between the two, with the VNA claiming S11 to be  -14 dB at 2.44GHz and the spectrum analyzer says -10 dB.

Is it even practical to compare 2 devices like this to get verification or should I just stick to one device?

Moving on, I placed a 2.2pF capacitor at L11, which should have moved me downwards to the 20m admittance line, but instead I got this:

I don't quite know what to make of this.

I tried a 1.2nH inductor on L11 instead, which should have moved me upwards to the 20m admittance line, but instead I got this:

Again not really close to theory. You might think I was happy to see this because it shows as almost a perfect match, but considering the shenanigans I saw with this device I am highly doubtful that this is true. I checked this configuration with the spectrum analyzer and it showed me this:

Which is an even worse result than just having 0 ohm resistor.

I would appreciate any insight on what I might be doing wrong here

Hi Eyal,

You are on the right track in regards to troubleshooting and right now it is difficult to tell which equipment is correct. Since the test equipment is in question, the first thing to check is the calibration and from image #2 it seem could be better because I typically aim for S11 = -30dB or better and S21 = 0.02dB or better. Same goes for the spectrum analyzer. Afterwards please follow these steps as it is important:

1. Ensure the blue RF cable is not damaged as those types of thin cable are easily damaged and not as reliable as its thicker counterparts.

2. Grab a known good device/board with measured data (aka gold unit) to check the equipment. This is the quickest way to exonerate each equipment.

   a. If not, perhaps borrow a keysight/agilent VNA or PNA as I can vouch for these equipment.

3. Does the duct tape shift your performance? If so remove it

   a. Does wiggling any of the cables change your performance? If so, then there might be loose connection.

Best regards,

Bun

After more than a week of attempts, including a few hours of crawling under my desk into a fetal position and crying, I'm still baffled by this thing.

Realizing my main issue is probably adding the wrong electrical delay I took a different approach. Starting with zero ohm resistors I marked the impedance movement on a Smith Chart using different values of parallel capacitors, and then with different values of series inductors. Then, using a printed Smith Chart and a god damn pair of compasses I tried rotating all of these marks by phase until they all lined up correctly - meaning the 0 ohm resistor mark and capacitors marks are all on the same admittance curve and the 0 ohm resistor mark and inductors are on the same impedance curve. I then set the electrical delay on my VNA that corresponds to what I found.

Yet STILL I got inconsistent behaviour when adding components after that. And I don't mean small differences, I'm talking about the impedance going in the exact opposite direction of where it should when adding components.

I'm about to give up on this notion and just blindly add\remove components until the spectrum analyzer says SWR is acceptable.

Hi Eyal,

I understand this can be a frustrating process as I also been through it too. What you're experiencing is common for anyone going through RF journey. It appears parasitic/unwanted inductance/capacitance is dominating your performance. For example, if the ground pad of a shunt capacitor does not have immediate access to a ground via, then it can introduce a big enough parasitic impedance to ground which can throw off your matching.

Let's take a step back and methodically work through this issue one step at a time to rule out and correct each issues.

Step 0) submit your design review here: https://www.ti.com/tool/SIMPLELINK-2-4GHZ-DESIGN-REVIEWS. In addition to the schematic, please ensure to submit the layout in Gerber format with the layers properly labelled. While my team reviews this we can work to clear up the other issues in parallel.

Step 1) Measure a gold unit to ensure both equipment have matching results. The gold unit does not have to be your design/board just a board with a well known performance. It could also be one of our launchpads. This step is important as it will allow us to deem the test equipment and any attached cables reliable to use.

Step 2) Check for consistency. Once the equipment has been deemed reliable, check the baseline performance of 3 or more boards and install a component to check if all 3 boards behave the same way. If so, then it systematic and narrowed it down to the layout design.

Best regards,

Bun