ADF7020 ISM BAND TRANSCEIVER module

Found out the ADF7020 radio module is quite popular in market, so just brought some back to get a hands on it.

The ADF7020 chip
This chip, once frustrated me, because it's such highly integrated and it make ones feels that most of the study in discrete RF circuit design is worthless, LNA, quadrature mixer, IF filter, IF amplifier, modulator, demodulator, ADC, power amp, PLL... all-in-one single chip. What a respectable piece of art. It looks like there are no space left for RF designer. But I have to overcome this frustration and test test this little modules.

The module

The module is supposed to be inserted between two UART device as a cable replacement device. It consists of ADF7020 ISM band transceiver, a MCU, which responds for initialization and translate the external UART signal to the data format required for ADF7020. A 11.0592MHz crystal , what an odd frequency, some matching circuit, a SMA connector, a power regulator, and 7 pin interface terminal.
The manufacturer provide with an PC tool to config the module, the tool controls the carrier frequency, RF output power, baud rate etc. I set the carrier to around 443MHz and trying to see if any RF comes out from the SMA connector.

The output

It just fire up RF output once power is turn on, clearly seeing the fundamental 441MHz output, with power level of 15dbm (adjusted figure due to -3db attenuator) and some harmonics, about be 30dbc lower than the fundamental. This is a good start, some real field test have to be done to see how this power level is translating to the actually usable distance.

HP8590D Spectrum analyzer death CRT

The CRT display of 8590D spectrum analyzer is dead. There are a BNC composite video output on the back. Looks like it can output to a TV or something.

Brought back a composite video to VGA converter, it turns out like this.

Mixer compare - NE602 vs BFR93 Discrete Glibert cell

The discrete transistor Gilbert cell mixer which is briefly tested a month before, is that mixer really having a decent performance? Let's compare it with a classic, popular NE602 mixer. Because both mixer do not have 50 ohm output impedance so this time a 4:1 transformer is added to make it more match with the 50 ohm spectrum analyzer.

The board for mixer comparison is consist of upper section discrete Gilbert cell and lower section of NE602 mixer. Left side is RF and LO input connector and the output connector is on the right side.

The 1.6 mm board is a mistake, 1.2 mm would be better.


Transistor Gilbert cell under test, with LO input 27MHz -20dbm and RF input 32MHz -20dbm,
beside the wanted 5MHz mixing product, many other frequency component is observed.


With same input signal. The NE602A mixer give way less spurious component, but the mixing product also lower in power. This may caused by the bias current of NE602 is smaller than the discrete mixer. It is observed that the LO isolation in NE602 is better.

This comparison attempt did not draw any conclusion, further analyst and more test is required to be done.

New NE602 mixer board arrived

NE602 mixer board, built to compare the performance of the previous discrete Gilbert cell mixer.

Datasheet stated that it comes with temperature compensated bias network. I should test this.

1W 27MHz RF power amplifier

The BLT50 transistor looks like it can handle a little bit more power, so I rise the bias current from 75mA to 300 mA, for 7.4V lithium battery as power supply, it dissipate 2.2W in idle. It's a little bit out of normal working range stated on datasheet. But, who care.

Worry for the high RF power would burn up the spectrum analyzer, 2 pieces of -3db attenuate is inserted between power amplifier output and spectrum analyzer.  

In initial design, the output is directly connected to the collector of transistor, in such configuration the max output power can comes to +27dbm (0.5W) with 150mA bias current. In later design the collector loading inductor is replaced by a 1:1.3 step up transformer and bias current rise to 300mA, which give a max power of +30dbm (1W). +24dbm is shown on spectrum analyzer because of the action of -6db attenuator. There are quite high level of higher harmonics because the transistor is driven beyond it's linear region. Some output filtering is needed.

The second -3db attenuator. Brought from a used equipment online store.

Input drive required to be +7dbm to give full output power of +30dbm.

A quick, simple, naive, 27MHz power amplifier experiment using BLT50 BJT

Brought a NXP BLT50 power transistor long time ago, today try to build a quick test circuit to see how it feels like. If this transistor works well, it would be put into the 27MHz remote control transmitter power amplifier stage. To maximize the voltage gain and reduce distortion,  the transistor is biased as class A at 75mA, and supply voltage 8.4V, it dissipate about 0.6W of heat, for a SOT223 package, it doesn't became too hot. At lower quiescent current, the maximum output power would be reduced.

The circuit is simple, no matching circuit, a small emitter resistor to provide a little bit thermal stability, the circuit is simulated before put into board, to reduce cut and try time.

I didn't want to spent a few days to wait for the PCB back from factory, so I use part of the board from 433MHz receiver project. The big inductor is the collector loading inductor, about 2uH inductance.

In 27MHz, 0dbm input, the amplifier produce an nice output spectrum. with -30dbc higher harmonics and 23dbm output power, which equal 200mW.

What is the range of this 200mW transmitter if it is put in the remote control transmitter?

May be I should go for a field test.

Long struggle of driving mini-circuit diode ring mixer

For a long time, driving the mini-circuit mixer LO port is a pain. Early attempts was using a transistor crystal oscillator to drive it, and later adding a 3:1 step-down transformer to drive it, and later adding series inductor to filter it. All attempts didn't give satisfied result. 

The problem is, when you probe the LO itself, the waveform looks clean, it's a sin wave, but once you connect the LO to the mixer, the waveform become bad, for bad, I means not looks like a sin wave anymore, it more looks like having many harmonics and the harmonics is just looks uncomfortably terrible. 

But why this issue stay so long? well, because the mixer seems work fine with those harmonics! It can works to produce the intended mixing product. Many version of remote control receiver was done.

I should try to fix this problem once again.

Waveform of mini-circuit ADE-1 mixer LO port, driven by 74LVC1GX04 crystal driver, at 27MHz. Overshoot is too heavy. Something needs to be done.

Spectrum view in IF port.

I also discovered that the LO harmonics leak through the IF port, the leaks like fireworks. All the way to 1.8GHz.

The breakthrough

 By using direct experimental method (blind guess and played around with it until it works), adding a big 470pF capacitor across the LO port to ground give a good result. The waveform looks more sinusoidal and less overshoot. Try adding shunt capacity is not the first time, but last time just not big enough! I guess the mixer LO port can be think of a small value inductor so adding a big cap to resonate out the inductance would do the job.

Shunting a 470pF capacitor on LO port smooth out the overshoot.

The LO harmonics leaks to IF port is significantly attenuated.

Last thing is to check the IF level, if adding the capacitor would make the IF level drop. And it doesn't drop. The conversion loss still keeps in -7dbm.