2018-04-17

My Experience with pocketVNA

If you are working with some IoT hardware projects then you probably need a VNA for antenna matching. In the past, VNA is so expensive that most small companies can't afford it. But we have better choice now due to some PC-based VNAs are available in recent years.

So far, I am working on four IoT projects. One uses 915MHz while the other three use 2.45GHz. In order to fine-tune the PCB antennas, I spent a lot of time to survey a VNA suitable for me. Here is the requirements:
  1. For antenna matching, I don't need a very accurate VNA. The bandwidth of antenna is usually defined at -10dB. So, noise floor less than -30dB is pretty good enough.
  2. The scan speed is not important. I can wait for tens of seconds for a scan to complete because I won't use it frequently.
  3. The price must be as low as possible since I have to buy it with my own money.
  4. VNA with only one port is acceptable. But it will be better to have two ports.
  5. It will be better if the VNA is powered directly by USB port.
The first one I found is MegiQ VNA-0460, it costs €3,990. I try to borrow one from distributor for evaluation. Well, the result is very strange. After calibration, I measured the 2.45GHz antenna on the CalKit. But the minimal Return Loss occurred at 1.95GHz. Very strange! I asked the distributor and MegiQ for help... but no one really wants to help me... probably because I am a very small customer. After emails back and forth for more than three months. I decided to give it up.

MegiQ only provide application for Windows. If you are a MAC user like me, you probably want a macOS native application. The UI design is useful, but not good enough. The detection of VNA device when plug/unplug it in/from USB port is not working well. Sometimes, I have to turn off everything and re-launch them for it to be recognized.

After MegiQ, I try to survey other vendors. I contacted Anritsu. They have office in Taiwan. I thought, well, the service should be much better, but I was wrong. Anritsu's sales called me to see what I need... then... no response any more.

There is another vender called Copper Mountain. But I don't contact them. So, I can't comment on it.

PocketVNA


Finally, I found a company named pocketVNA, located in Germany. They only sell one product, the pocketVNA. The price is very low, only $490. They also provide a very simple and useful calibration kit with only $22. There is only one question: is it usable? The good news is they provide macOS  application. So, I downloaded it to play around. mmm... it is well designed. So, I decided to give it a try.

After contacting pocketVNA by email, I got response the next day. They only sell it through website. So, I payed the money by PayPal and got the VNA device after one week.


PocketVNA gives you a chance to adjust the averaging parameters while scanning. The over-sampling process is helpful to reduce noise. The major disadvantage of pocketVNA is the scan speed. It is pretty slow when compared with MegiQ VNA-0460. So, I try a lot of parameter combinations to find out a useful configuration with acceptable scanning time. Here is my suggestion for Calibration,
  • From: 4MHz
  • To: 4GHz
  • Steps: 1000
  • Averaging: 1
  • Adc AVG: 10
With this configuration, it will spend 16 seconds for each calibration. We have to do OSL calibration for Open, Short and Load conditions. So the total time is 48 seconds plus the time to change calibration connectors. It is a little more than 1 minute to do an OSL calibration.

There will be one calibrated point for every 4MHz. So, we have the following calibrated frequencies: 4M, 8M, 12M ... 3.992G, 3.996G, 4G. Notice that interpolation will be used for frequencies in between.

Then, use the following parameters for Live Measurement,
  • From: 4MHz
  • To: 4GHz
  • Steps: 1000
  • Average: 10
The time for a complete scan is about 44 seconds. It is acceptable for me. If you want a shorter time, the "Average" count can be reduced.

Calibration Issue


PocketVNA has a very useful function, you can assign the port extension length manually. The calibration kit let you calibrate to the end of the port connector. But the cable connects from the port to the target board will cause a phase delay. So, the port extension length can be used to compensate the phase delay.

The real situation is, radio signal will also be attenuated inside the extension cable, it is usually 0 to -3 dB and not uniform. The best way is to calibrate the cable all together. The problem is we usually won't put a SMA connector on the target board. So, the target end of the cable is not SMA connector. The solution for this problem is to use a smaller connector on target board, a U.FL (IPEX) connector can be used. Then we can calibrate VNA to the U.FL connector at the end of the extension cable.

OK, now the problem becomes where to get the U.FL calibration kit? Most venders provide calibration kit with SMA connectors. MegiQ have calibration kit with U.FL connector. It is actually a PCB board with some U.FL connectors. So, I decide to make one for myself. Here is my own version:


In my calibration kit, I can either use U.FL connector or solder the cable on the "pads" directly. This is very useful for target boards without U.FL connector. For this case, a extension cable is usually directly soldered on the feed point of the antenna on PCB.

Finally, we have to understand that the internal RF components inside VNA is very sensitive to temperature. So, the VNA must warn up for 30 minutes before using. Btw, after 10 minutes measurement, it is better to re-do the calibration.