Fullscreen HTML5Med Res (??MB)HTML4

Using Cheap Asian Electronic Modules By Jim Rowe Geekcreit’s 35MHz4.4GHz Signal Generator This self-contained module is based on the Analog Devices ADF4351 wideband digital synthesiser chip. It has an onboard microcontroller unit (MCU), OLED display and pushbuttons to set the desired frequency and adjust the output level. All it needs is a 5V DC power supply. If the ADF4351 sounds familiar, that’s because it was also used in the digitally-controlled oscillator we reviewed (May 2018; siliconchip.com. au/Article/11073). But whereas the earlier unit needed to be controlled via a separate microcontroller such as an Arduino or a Micromite, this one is a self-contained instrument, delivered ready to use. It is larger than the earlier one, measuring 88 x 67mm compared to 48 x 36.5mm. But the price isn’t all that much higher, currently setting you back $48 plus $7 for shipping to Australia. It can be purchased from Banggood (siliconchip.com.au/link/ab83). As shown in the photos, it comes with two cables: a USB Type-A to mini-B cable and a 240mm-long DC cable with a plug on one end to match the module’s DC input socket. It also comes fitted with four 5mm-long Nylon mounting spacers and matching screws. But no case is supplied, so you’ll either need to use it as a ‘bare’ module, or come up with your own arrangement. On the PCB, there’s an STM32F103 MCU (visible at lower left), a small siliconchip.com.au OLED (organic light-emitting diode) display with a 128 x 64 pixel 25mm (1-inch) diagonal screen, and a total of seven pushbutton switches. The five at lower right control the module, while the one in the centre resets the MCU. The one near the upper left with a square body and blue actuator is the ON/OFF switch. The ADF4351 synthesiser chip and its surrounding components are all in the upper right-hand corner of the PCB. The two nearby edge-mounted SMA sockets are the RF outputs, while the vertical SMA socket near the centre of the PCB is an input for an optional external master clock, an alternative to the onboard 100MHz crystal oscillator. The ADF4351 chip at the heart of the module is a digital ‘phase-locked loop’ or PLL device, and a pretty fancy one at that. But there isn’t space here to give you a full explanation of PLLs and how the ADF4351 itself works. So if you want to know more about these aspects, refer to the May 2018 article A close-up of the 1-inch OLED screen when using the “Point” command from the main screen. Australia’s electronics magazine December 2021  43 (siliconchip.com.au/Article/11073), which has a comprehensive explanation. The data sheet for the ADF4351 can be found at siliconchip.com.au/ link/aajc UG-435, which you can download from their website (siliconchip.com. au/link/ab82). A brief rundown Lack of instructions How it works The ADF4351 is a wideband digital synthesiser IC with a ‘fractional-N’ PLL, allowing it to be programmed to produce any desired output frequency between 35MHz and 4.4GHz. It is locked to a ‘master clock’ crystal oscillator of typically 25MHz or 100MHz. It can be programmed to change the output frequency in steps as small as 10kHz, and can also provide an output sweeping over a range of frequencies in steps of the same minimum size. The whole chip is controlled/programmed via a simple three-wire serial peripheral interface (SPI), in this case, via the onboard STM32F103 MCU. The Geekcreit 35-4400MHz signal generator module comes with very little user information, so you have to work a lot out for yourself. All you get is a brief summary of its main specs and features, and you can download a circuit diagram that is not easy to decipher. So before I began testing the module, I spent a couple of hours redrawing the circuit so that we can all see how it works, shown in Fig.1. Like the earlier module, this one is fairly closely based on Analog Devices’ evaluation board for the ADF4351. That is described in their User Guide In Fig.1, the ADF4351 (IC2) is on the right, with its onboard 100MHz master clock oscillator to its left. These form the actual VHF-UHF RF synthesiser ‘heart’ of the module. The two complementary RF outputs emerge from pins 12 and 13 of IC2, and are fed via 1nF capacitors to the two SMA output sockets at far right. The 3.3V DC supply to pins 12 and 13 flows via inductors L2 and L3. Only the RF output from pin 12 of IC2 (RFout+) has an onboard 51W terminating resistor. The other components on the righthand side of Fig.1 are to provide IC2 44 Silicon Chip Australia’s electronics magazine siliconchip.com.au Fig.1: the circuit diagram for the Geekcreit signal generator module. with power, set its operating mode, or feed it control signals. For example, the components between pins 7 and 20 at upper right form the ADF4351’s lowpass loop feedback filter (to optimise its performance), while the capacitors at pins 19, 23 and 24 bypass key reference points in its internal circuitry. The digital control signals from IC1 that direct IC2’s operation are fed to pins 1, 2, 3 and 4 at centre left, labelled CLK, DATA, LE and CE. The only other signal that passes back from IC2 is the LD (lock detect) signal from pin 25, which is high when IC2 is locked to the requested frequency. As well as being fed back to the MCU, this signal is also used to illuminate LED2, the blue lock indicator. The power supply section is at upper left in Fig.1. This accepts either 5V DC from mini USB socket CON2, or 5-15V DC from concentric DC socket CON1. This flows via on-off switch S7 to power indicator LED1 and the rest of the circuit. The incoming supply powers REG1 and REG2, both of which are LT1763 LDO (low drop-out) 3.3V linear regulators. REG1 provides 3.3V to the control siliconchip.com.au circuitry, while REG2 generates a separate 3.3V supply for synthesiser IC2. The incoming supply to REG2 is via T1-T2, a balanced decoupling transformer wound on a small ferrite balun core. As mentioned earlier, the control circuitry is based around IC1, an STM32F103C8T6 microcontroller, and the 128 x 64-pixel OLED display module to its right. The four digital signals to control synthesiser IC2 connect to pins 25-28 of IC1, while the LD signal from IC2 is fed back to pin 29. Pushbuttons S1-S5 at lower left select the operating mode of the synthesiser, its operating frequency, output level and so on. The MCU provides a series of menus and indications on the OLED display to make this reasonably straightforward. The OLED display is driven via an SPI serial control link from pins 14-17 of IC1. The instruction and master clock for IC1 is generated by an internal oscillator using 8MHz crystal X2, connected to pins 5 and 6. Pushbutton S6 manually resets IC1 if necessary. The D- and D+ data lines from mini USB socket CON2 are Australia’s electronics magazine connected to pins 32 and 33 of IC1, so its firmware can be updated from a PC if needed. It’s also possible to communicate with IC1 via a second serial link connected to pins 34 and 37, brought out to the pins of CON5. This is not a physical connector, but provision on the module’s PCB for fitting a four-pin SIL header. Trying it out When I received the unit and tried powering it up, there were a couple of problems. The first of these was that the DC supply cable provided with it turned out to have an open circuit in its red (positive) lead. So I had to discard it and substitute a known-good cable. Then when I powered it up, I found that the module was on regardless of whether power switch S7 was pressed or not. The cause turned out to be a solder bridge under the PCB joining its two active pins permanently. Luckily, I fixed that easily with a soldering iron. I also tried powering the unit from a 5V USB plug pack, using the USB Type A-to-mini Type B cable provided, which worked fine. December 2021  45 Fig.2: plot of the output level vs frequency when terminated by 50W. When the module is first powered up, the OLED screen shows its function menu, or more accurately, the top of it – listing the first three functions: 1. Point: used to set the module’s frequency to a particular figure, like 4375.05MHz 2. Sweep: used to set the start and stop frequencies for sweeping over a range 3. Step Fre: not clearly explained, but seems to be used to set the frequency steps used during sweeping Then if you continue pressing the down (DWN) button, S5, you find the remaining two options: 4. Step Time: not clearly explained, but it appears to be for setting the time between steps when sweeping 5. dB Set: see below When you press the OK button (S4) to select this last option, you get a screen giving a choice of four RF Power settings: +5dB, +2dB, -1dB or -4dB. These appear to be provided to allow ‘fine adjustment’ of the module’s RF output level. When I tried checking these output level options with the module’s frequency set to 1GHz using my Agilent V3500A power meter, I obtained the following results: • With the +5dB setting, the meter registered +3.60dBm • With the +2dB setting, it registered -0.24dBm • With the -1dB setting, it registered -1.99dBm • With the -4dB setting, it registered -4.52dBm These were all measured with the meter connected to the RFout+ On starting the module, the OLED display lists the five available functions. 46 Silicon Chip Australia’s electronics magazine connector, using a very short (<20mm) SMA-SMA coupler. So the reference (0dB) level appears to be around -1dBm, and while these settings are not particularly accurate, they give you the ability to adjust the unit’s output level somewhat. The next step was to measure its RF output over the whole frequency range, again using the V3500A power meter. I did these measurements again using the SMA-SMA coupler, connected between the power meter and first the module’s RFout+ connector, and then the RFout- connector. In each case, the output not being measured was terminated in 50W, to hopefully prevent any standing wave disturbances. As you can see from Fig.2, the level from the RFout+ connector is about 4dBm lower than that from the RFoutconnector, probably due to the loading from the onboard 51W terminating resistor across RFout+. But apart from that, both plots are relatively flat, rising slowly by about 2-3dB between 40MHz and 1GHz, and then wobbling a bit to return very close to the 1GHz level at 4.4GHz. So overall, both outputs were within the range of -4dBm to +4dBm over the entire frequency range. Next, I checked the module’s RF output signal purity at several different frequencies, using my Signal Hound USB-SA44B spectrum analyser with the latest version of Signal Hound’s ‘Spike’ software. The results were reasonably acceptable, bearing in mind that the module’s outputs are essentially square waves with significant harmonic content, along with the inevitable spurs you tend to get from any PLL-type synthesiser. To illustrate this, Fig.3 shows the module’s output at 2.5GHz, with the analyser set for a 60MHz span (ie, 30MHz either side of 2.5GHz). The main output is a reasonably clean peak reaching about +1.5dBm in the centre, with two small spurs at about -55dBm around 25MHz either side. So far, it looks reasonably clean. But now look at Fig.4, which shows what the analyser displays when set to span over the total frequency range from 50MHz to 4.4GHz, with the module still set to 2.5GHz. Several additional spurs are visible, spaced at about 620MHz apart on either side of the main output, with amplitudes varying between about siliconchip.com.au -6dBm and -18dBm. So the output is not nearly so clean as Fig.3 suggests. The full-scan plots don’t look so bad with the module set to higher frequencies, though. For example, Fig.5 shows the result when the frequency is set to 3.0GHz, while Fig.6 shows a similarly clean plot when it is set to 4.0GHz. On the other hand, Fig.7 shows the result with a full scan showing what happens when the module is set to produce a 100MHz signal. There’s now a virtual ‘forest’ of spurs, varying in amplitude from -10dBm down to about -49dBm in alternating steps. Not a pretty picture! Because of the lack of information regarding how to get the module sweeping or stepping from one frequency to another, I gave up trying to test those functions. Fig.3: a graph of the module’s output at 2.5GHz with a 60MHz span provides a reasonably clean plot with just two small spurs at the edges. Fig.4: the span is now set over the range 50MHz to 4.4GHz with the same 2.5GHz output. Note the additional spurs approximately 600MHz apart. Summary So although the Geekcreit 35MHz4.4GHz signal generator module is a low-cost, self-contained unit that can generate output signals of around 0dBm (1mW) over that wide frequency range, it does have a few drawbacks and limitations. One of these is the lack of much information on operating the module, especially with regard to getting it to perform sweeping. Another is the large number of ‘spur’ components in the outputs, especially when it’s generating a frequency below about 1GHz. That is because its outputs are essentially square waves, rather than the sinewaves that are needed for many signal generator applications. Filtering these to produce a smoother signal is virtually impossible due to the wide range of possible output frequencies; however, external filters could be used if you need cleaner signals at specific frequencies. And finally, because of its lack of any shielding (especially for the RF generation circuitry around the ADF4351), it would be tough to achieve accurate control over its output level. But overall, the module would still be useful, for example, if you want to generate digital clock signals over a very wide range of frequencies. Just bear in mind that to use it as the basis of a practical VHF/UHF signal generator, you’d have to add shielding, output filtering and a wide-range output attenuator system. SC siliconchip.com.au Fig.5: setting the output frequency higher to 3GHz also provides a clean plot. Fig.6: the output frequency now set to 4GHz. Fig.7: setting the module to an output frequency of 100MHz produces a large number of spurs at the harmonic frequencies (ie, multiples of 100MHz) with varying amplitude. Australia’s electronics magazine December 2021  47