cc430 tunning

- First I assume that your settings are ok since you manage to get the expected PER with a couple of EMs.

- Are you able to measure output power on your board?

- Could you post the schematic (PDF) and layout (gerber) of your design? Without knowing how you have made your board it is hard to know where we should point you.

Hi,

I only have an oscilloscope and some EM boards, don't know how to (if possible) measure the output power.

I already put the schematic and layout in another post, please check here, since no one responded I gave up waiting.

Any help would be very useful.

Thanks

Is there a significant frequency offset between the RX and TX units? Try with a wider RX filter BW. 

Try with the 1.2 kbps settings as they allow for a larger frequency offset than the 38.4 kbps settings. Also, try to increase the RX filter BW from 58 kHz to something higher (e.g 100 kHz).

Hi, I have more test results:

1.  If my board is on the RX side and EM board on TX side I have no errors (all packets are successfully received). I tested with 1.2kbps, 38kbps, 250kbps. The RSSI is ~24dBm, antennas are on the same desk.

2. If my board is on the TX side then I have errors on the EM RX side as I stated above. Lowering the TX rate to 1.2 bps results in a lower PER - from 64% to 50%. Increasing the RX filter BW does not produce any changes in the PER.

Dragos

And even more:

Now I realize that even if I use two EM boards I have the same results. Don't know why, I have to figure out because I tested some time ago with the same configuration with no errors.

I'll come back here with more results

I suspect a communication error with SmartRF Studio

I think the error comes from the SmartRF studio which is not able to handle a fast incoming package rate. I'll explain a bit:

I have two debuggers, one is  MSP-FET430UIF and the other is the debugger than comes with chronos watch package - eZ430U. The second one uses a spy by wire interface for debugging which is much slower than the normal JTAG interface. I can see the difference when I set a device in TX mode: while the device connected to FET is sending packets with the default delay of 150 ms (fast), the same device connected to the eZ debugger send packets much slower, with a delay between packets around one second.

I suspect the same case in RX mode. The device connected to the eZ debugger receive packets at a speed that overloads the spy by wire interface and SmartRf studio receive partial packages misleadingly reported as errors.

To prove it is simple, I have to put a longer delay between packets at TX side but due to a bug in smartrf I cannot change the default packet interval for packet sending - I can in the interface but it seems that the interval is ignored by the program.

Hi Dragos,

Regarding the delay between packets sent with SmartRF Studio you are right. There is a bug in the SW. The changes in UI is not taken into account. We need to fix this. If you send packets one by one with SmartRF Studio I hope you see that they are received OK.

Thanks for the feedback.

Regards,
Øyvind

I checked a bit more and realised that a fix wouldn't take too much time. It takes some time for the next release though but I prepared a patch I hope you can use. Attached you find a zip file with the required updates. replace the existing files with the one in the zip file. I did a quick test of it and it seems to work fine.

Regards,
Øyvind

Thanks

Hi, I tested these days different scenarios and the results are not satisfactory. The range is insufficient compared to the antenna #6 from the Antenna dev. kit. The RX performance is ~90% and TX is somewhere between 70-80% compared to #6.

In the docs it is stated that the performance levels of these two antennas are nearly the same. Do I need to fine tune the antenna?

I red a lot regarding antenna tuning but since I don't have a network analyzer my possibilities are severely limited, any advise would be very helpful.

Thanks,

Dragos

What is the potential volume for your product?

Sorry for the later post, edited: The product is a parking spot occupancy sensor, The physical volume/form factor is a cylinder with radius 7,5 cm (this can be changed) and height no more than 2, 4cm. The PCB will be potted in epoxy. 

Best regards,

Dragos

When putting your board in an epoxy, you will detune the antenna and move the resonance frequency. The size of your ground plane will also affect the resonance of your antenna. Have you looked at AN058  http://www.ti.com/lit/pdf/swra161? I understand you do not have a network analyzer, but you could try sweeping the frequency and look at the output power. Have you tried the kit antenna (whip antenna)?

The are some physical constraints which in this particular case cannot be avoided. But in the same time I'm not asking for the maximum range obtained in line of sight. What I need is 50m range which is achievable with antenna #6 with 0% PER in very harsh conditions (thick plastic cover, under a large car in a crowded street). The RSSI in these conditions was 85dBm at 9.6kbps.

I tested different frequencies by using the following setup:

At TX side I have my board and at the RX side the evaluation board. I tested some frequencies near 868 MHz with the following results:

Freq RSSI
------------

867.4 -86
867.5 -84
867.6 -82
867.7 -79
867.8 -76
867.9 -74
868.0 -44
868.1 -74
868.2 -76

The whip antenna is used on one side, on the other side (sensor side) I cannot use it since the physical dimensions are impossible to fit.

Can you do a larger freq sweep? Do you have a picture or something of your board? So we could see how your board looks like and also one of the casing? Have you tuned the antenna at all or kept it as is? Have you done conducted testing? Have you tried other antennas as a reference? 

The board has now a SMA connector attached, I'll try to do some tests with #6 antenna and I will post the results here.

The enclosure for the prototype is made from polyurethane, the final shape and material could be a bit different.

I did't tuned the antenna, the matching components are as indicated in the swra416.pdf: ANT2=1pF and ANT3=12nH

I cannot do conducted testing since I don't have the tools to do it.

By "larger frequency sweep" you mean to zoom in the frequency around 868.00Mhz? Like 868.05, 06, etc?

Thanks,

Dragos

Yes, I did mean a finer resolution for the freq sweep, sorry for not beeing clear. Since from the RSSI values you gave the antenna did not look too bad, but I see that you have a completely different size on the gorund plane and with the casing, I would expect the antenna to need some different tuning components. Do you have a whip antenna from our kits you can try?  

Also, can you post your settings so that we can look at them? Do you now have a two way communication or one way? You have one of our kit as one node, correct? Do you see any difference between RX and TX performance that you saw earlier?

I'll respond quickly on what I already know:

1. The settings are standard in SmartRFStudio, basically I start it, select 38.4Kbps speed, optimized sensitivity then change the speed to 9.6Kbps, on TX side I put power C0 - 12dBm (I'm using wire wound inductors so I expect this to work). Then, I test continuous TX/RX transmission and packet transmission and I'm monitoring RSSI and PER levels. I've noticed that I don't see any improvement in RX if I make the RX filter BW larger.

2. Two way communication, I'm using my board as "sensor" and ALWAYS a dev board (em430f6137rf900) with whip antenna as a node. The function of the node is to retransmit packets from the sensor and to provide sync packets to the sensor (I'm using a sync mechanism which ensures max 1msec desync time)

Today I built another board, I did some tests, I'm puzzled a bit because when I'm testing the board in a concrete building with three walls in the middle, the board actually behaves better (2-3 dBm) than standard EM430... with whip OR #6.  I believe that, due to it's smaller size, this board get's de-tunned much easier that the others when I put it on ground, under a car and in a plastic encapsulation.

Should I continue with it or should I switch to #6 which seems that it can deliver what I need?

These are the results for the freq sweep near 868.00 MHz. Test setup:

TX: sensor with mini helical antenna, with plastic encapsulation on top, with standard SmartRF 38.4 kbps settings and changes:  9.6Kbps speed and 10dbm TX power, continuous TX

RX: em430f6137rf900 with whip antenna, standard SmartRF 38.4kbps settings and changes: 9.6kbps speed, continuous RX

Distance between devices: ~1m

867.8 -59
867.90 -33 -> -44 bounce
867.91 -29 -> -62 bounce
867.92 -27 -> -60 bounce
867.93 -27 -> -53 bounce
867.94 -26
867.95 -26
867.96 -26
867.97 -26
867.98 -27
867.99 -27
868.00 -27
868.01 -27
868.02 -27
868.03 -28
868.04 -29
868.05 -36
868.06 -27 bounce to -58
868.07 -60
decreasing

I will post now RX/TX tests from the sensor side to see if there are differences

Test #1

RX

sensor with mini helical antenna, with plastic encapsulation on top, floor level, with standard SmartRF 38.4 kbps settings and changes:  9.6Kbps speed, packet RX

TX

em430f6137rf900 with whip antenna, standard SmartRF 38.4kbps settings and changes: 9.6kbps speed, packet TX, 10dbm power.

Distance between devices: ~14m with three concrete walls in between.

Results for 100 packets:

Average RSSI: 93.9dbm

Received OK: 99, NOK 1, Error rate: 1%

Test #2

RX

em430f6137rf900 with whip antenna, standard SmartRF 38.4kbps settings and changes: 9.6kbps speed, packet RX.

TX

sensor with mini helical antenna, with plastic encapsulation on top, floor level, with standard SmartRF 38.4 kbps settings and changes:  9.6Kbps speed and 10dbm TX power, packet TX

Distance between devices: ~14m with three concrete walls in between.

Results

Average RSSI: 90.3 dBm

PER 1% (99 OK, 1 NOK with CRC error)

Test #3: attach a whip antenna to the sensor and do a TX/RX test in the same conditions as test #1 and #2 (but with improvised plastic encapsulation since the original one cannot fit)

Results

Sensor in TX mode: PER 0%, Average RSSI: 64dBm

Sensor in RX mode: PER 0%, average RSSI: 70 dBm

Test #4

Same conditions as test #1 but sensor without the plastic cap, on the floor level and lifted up.

Results:

On the floor level the RSSI is about -85dBm if I lift it up it goes to -69, -70 in the most favorable position. So, apparently my statement with "mini helical performs better than whip antenna in building" is highly questionable it depends on the distance and the position of the receiving side.

Test #5

Same conditions as test #1 but sensor without the plastic cap, on the floor level and lifted up.

Results

1. Most favorable position for the sensor:

RSSI: 75.5dBm at RX side

PER: 0%

2. Sensor on the floor:

RSSI: -92.5 dbm

PER: 3%