RF Power, Switches, Reflections and Impedance Matching

[trevorth]

I have a 2.4GHz board with a class 1 bluetooth PA IC (Skyworks) and RF switches to bypass the PA for receive.

I have the prototype but am having the following problems:

1). When in transmit via the PA path, the output is -30dBm.
It should be +25dBm at the output of the PA, thus I expect +20-22dBM at the connector/antenna after the filter/switch.

2). When in transmit not via the TX path (i.e. bypass the PA), the output is -60dBm as though theres nothing there.
I can set the transceiver to 0dBm but just dont seem anything.

When the transciever is outputting 0dBm and the amplifier is on (+20dBm), data (verified by a CRC check) can be received by another module in close proxmity. So it is outputting some valid RF.

The following peaks are observed on the spectrum analyser:

2.33GHz -65dBm
2.40GHz -51dBm (Carrier)
2.44GHz -38dBm
2.48GHz -54dBm
2.54GHz -52dBm
2.59GHz -50dBm

Questions

I have verified the logic to the switches but cannot verify the switch is closed/open as both paths show the same resistance, <3ohm. Ive tried a voltage across the seitch with the same uncertainty. Is there something special about these RF switches which means they only provide isolation at high frequency?

All the traces are very short (<10mm). Can impedance matching along the short paths and at the output really cause so much power loss? And also cause a reflection to have a greater power level as seen on the spectrum anlayzer?

If I can verify the RF switches are operating correctly (which I think they are) the other issue that I can think of is the 12pF capacitors but I did check they would provide little impedance at that frequency.

Regarding a Vector Network Analyser and impedance matching, are there any guides on how to use one to determine antenna matching and where to take measurements from i.e. do I need to make another prototype board with connectors along the path? Is this an aspect thats normally simulated?

Thanks in advance for any guidance.

 

[PA4TIM]

https://www.pa4tim.nl/?p=1594 A tutorial about VNA use.
3 Ohm is rather high in ratio to 50 Ohm and this is DC resistance I presume. Then AC resistance will be much higher due to skinloss.
You talk about a spectrum analyser and reflections, so you have one with Tracking generator ? What kinda coupler or bridge you use ? And did you veryfy the return loss with an open/short/load test ?
More professional as mu tutorial are three books:
- impedance measurement handbook from Agilent
- Fundemantals of vector network analisys from Hiebel (designer for R&S)
- Microwave component measurements from J.P.Dunsmore (designs aboout 30 years VNA's for Agilent)

 

[E Kafeman]

Assume that you not tries to measure TX while it is jumping around in frequency. To be able to impedance match correctly must you be able to control if it is in TX/RX-mode and set it up non frequency jumping at typical 2400, 2440 and 2480 MHz.

If it not is frequency jumping at your measurements, seems TX to be more or less dead. CC2510 is a bit ESD sensitive at its RF output. If i remember correctly should DC level at RF out at CC2510 be around 1.1 Volt at both pins but if a RF-stage is blown does DC voltage typically goes down to 0.5 Volt unloaded. Compare with a working one.

If you have an VNA that have a GPIB PC interface, have I showed one way to impedance match an BT antenna.
The software that is used in the video to compute matching values is able to set any measured impedance as impedance reference so it can relative easy do impedance matching between different RF stages, without any need of a 50 Ohm reference along the path.
I usually get pre-production working PCB boards for impedance matching. Solder measurement cables at matching points, cut PCB traces that already is matched so that you only is measuring in one Tx/RX direction at the time... It is for me a relative complicated job requiring good VNA handling and RF knowledge around transmission lines an how stray capacitance behaves at these frequencies, avoid unbalanced RF current in measurements cables and similar things.

 

[FvM]

I have verified the logic to the switches but cannot verify the switch is closed/open as both paths show the same resistance, <3ohm. Ive tried a voltage across the seitch with the same uncertainty. Is there something special about these RF switches which means they only provide isolation at high frequency?

GaAs switches can work at low frequency or even DC, but with lower voltage level. The usual GaAs switch bias scheme however requires series capacitors and don't work for DC. Switch operation can be nevertheless verified with DC.

My impression so far:
- You don't understand GaAs switch operation principle (transistor level circuit, bias requirements). That's not surprizing because the data sheet is rather uncommunicative in this regard.
- You possibly connected the switches incorrectly

If it's not the switch connection, the problem is most likely caused by an inappropriate matching network. There might be a PA problem as well.

Although a VNA is preferred for circuit fine tuning (and also suitable for fault finding), you shouldn't need it to get a TX output above the ridiculous - 30 dBm.