Fullscreen HTML5Med Res (??MB)HTML4

Project by Andrew Woodfield, ZL2PD 2m VHF FM Test Signal Generator This Test Oscillator uses an AD9834 DDS chip in a somewhat unusual way, producing signals in the 2m band (144-148MHz) even though that is well above its 80MHz oscillator frequency. It can produce a CW signal at three different levels and optionally add frequency modulation. T he Analog Devices AD9834 direct digital synthesis (DDS) chip is a common feature of audio function generators and HF oscillators. This unusual variation delivers programmable FM digitally generated signals on channels across the 2m amateur radio band. It can even run from a single AA-size cell. The 2m amateur radio band covers 144-148MHz in Australia, New Zealand and North America, or 144146MHz across Europe. Designs using the inexpensive AD9834 have been published for over 20 years. Yet, looking closely, almost all designs mimic the circuits from AD application notes, with the differences primarily being the choice of microcontroller, display and PCB layout. Several years ago, I made a simple test board with an ATtiny45 microcontroller and a couple of pushbuttons. I tested the AD9834’s overclocking potential and several DDS output filter designs and quickly established that the AD9834-BRUZ chips could be overclocked at up to 85MHz, a useful margin over the 50MHz specification from the data sheet. The slightly more expensive -CRUZ devices are rated at 75MHz but are reported to handle 100MHz clocks. That was the limit of my interest in the AD9834. I had no immediate use for it and put it on my shelf. Features & Specifications ▬ 144 to 148MHz output in 500kHz steps ▬ Four additional user-programmable memory frequencies in the 2m band ▬ -45dBm, -75dBm and -105dBm output levels ▬ Carrier-only (no modulation) or FM ▬ ±3kHz FM deviation (1kHz tone) ▬ CTCSS and external modulation audio inputs ▬ Simple to align output filter; everything else is digital 44 Silicon Chip Australia's electronics magazine So I expected few surprises when a friend recently asked me to confirm the results he was measuring on his AD9834 signal generator design. Phase modulation As I worked on these tests, the claimed frequency modulation (FM) and phase modulation (PM) features headlined in the data sheet caught my eye. In my previous DDS designs with the AD9850 and AD9851 chips, their limited phase registers made them unsuitable for FM. The AD9834 variable-frequency oscillator (VFO) designs I’d seen only used the DDS chip as a variable RF oscillator or audio function generator. I’d never seen any mention of AD9834 phase modulation anywhere before. Looking more closely, I realised that the AD9834 has far more capable 12-bit phase registers than the AD9850/9851. That made me curious. Within an hour, I had phase modulation ‘working’ in a limited fashion. However, it wasn’t clear from the data sheet how to get a specific modulation level. An extensive search of Analog Devices application information showed they were also largely silent on the topic. As I continued to explore phase modulation, the calculations, and the other features of the chip, I realised it could achieve something quite different to the usual AD9834 design. This novel 2m CW/FM test oscillator is the result. DDS and alias frequencies The AD9834 is a DDS oscillator chip typical of many made by Analog Devices. The block diagram (Fig.1) shows the internal arrangement. It contains a pair of programmable Frequency Registers and a pair of Phase Registers to allow the selection of one of two combinations of frequency and phase. These registers control a powerful numerically controlled oscillator (NCO) using the chip’s 28-bit Phase Accumulator. The output of the NCO drives a lookup table and 10-bit DAC that converts the NCO output to a sinewave. The output frequency, fout, can be calculated as fout = Nreg × fclk ÷ 228, where fclk is the external clock frequency, and Nreg is the 28-bit digital value loaded into one of the two selectable frequency registers. The siliconchip.com.au associated Phase Registers allow the output signal phase to be shifted by a programmable phase angle. With an external crystal clock and a suitable low pass filter (LPF), the AD9834’s output is a reasonably clean, low-noise sinewave that can reach up to about 30MHz. It’s possible to generate frequencies with a resolution of 0.3Hz. The output frequency’s accuracy depends on the crystal oscillator’s precision and stability. Actually, the output produced by the DDS is significantly more complex than this. In the absence of the output LPF, the DDS process also generates an extended series of signals along with the main output signal. These can reach well beyond 300MHz, as shown in Fig.2 (also from the data sheet, again with a bit of colour added). The primary output signal (fout) is typically filtered by an LPF that heavily attenuates these other unwanted signals. The output may be programmed to any frequency up to half the external DDS clock frequency, eg, 40MHz if the DDS uses a clock frequency of 80MHz. Those unwanted products generated by the AD9834 are called ‘alias’ or ‘image’ signals. The first is on a frequency of fclk − fout. When the DDS clock is 80MHz, if fout is changed in steps from 1Hz to, say, 30MHz by changing the value programmed into the AD9834’s Frequency Register, this first ‘alias’ signal (labelled fc − fout in Fig.2) is generated. It begins at 80MHz and reduces in frequency with each step, down to 50MHz (50MHz = 80MHz – 30MHz). These are shown in more detail in Fig.3. If the ‘wanted’ fout signal moves towards 40MHz (half the DDS clock frequency), this first alias output product becomes increasingly annoying. It also approaches 40MHz from above and must somehow be filtered out. That becomes more and more difficult as the desired fout signal rises above 30MHz and approaches 40MHz. This demands the use of a low pass filter with a very steep cutoff for most applications. Such filters usually start attenuating just above 30MHz, with the rejection increasing sharply to reach at least 60dB by 40MHz. Most designs use a complex 5th- or 7th-­ order output low-pass filter for this reason. siliconchip.com.au Fig.1: the AD9834 block diagram, reproduced from the data sheet (with some added colour). It is a typical DDS oscillator with a 28-bit phase accumulator that can generate accurate RF sinewaves up to 30MHz from an 80MHz external clock. Fig.2: without any added filters, the AD9834’s output signals extend well beyond 300MHz. Fig.3: the AD9834 output spectrum with an 80MHz DDS clock and the frequency register set to 14MHz. The arrows show the direction the unfiltered alias carriers travel as the frequency of the 14MHz fundamental increases. The green dashed line shows the typical high-order low-pass filter normally used to remove these other products from the output signal. Australia's electronics magazine October 2023  45 You can also see from Fig.3 that the DDS output level is not the same across the spectrum. The output level falls following a sin(x) ÷ x response. With an 80MHz clock, for example, the output at 30MHz is 2dB less than at 1MHz. The 50MHz ‘alias’ output generated when the main output is 30MHz is 6.5dB below the 1MHz level, and only 4.5dB below the 30MHz wanted output. So, without a good LPF, the AD9834 output at 30MHz would have severe distortion due to the desired signal mixing with the nearby unwanted 50MHz alias output product. 145MHz, 146MHz, 147MHz and 148MHz respectively, albeit at much lower levels. If a relatively narrow bandwidth LC bandpass filter is added to the AD9834 output and tuned to the 2m band, passing just these 2m-band signals is possible. An example of the response of such a filter for use with the AD9834 is shown in Fig.4. This output filter must also be designed to match the 200W output impedance of the AD9834 and allow for a 50W output load impedance, to match the expected loads in typical RF applications. Operating the AD9834 in the 2m band (144-148MHz) Modulation Fig.3 shows the other alias signals generated above the desired and first alias outputs. If the desired output is 14MHz and the first alias is 66MHz, the next alias output is 94MHz. More alias outputs are generated at 146MHz, 174MHz, 226MHz and beyond. The AD9834 output also contains a residual clock output at 80MHz. Since it is a square wave clock, it has a strong unwanted product at the third clock harmonic, 240MHz in this case. Fig.3 also shows the direction the aliases move as the main carrier increases in frequency. Alias outputs can appear in the 2m band, for example, if the AD9834 output is set in turn to 16MHz, 15MHz, 14MHz, 13MHz and 12MHz with an 80MHz DDS clock, the third-image alias output (‘super-Nyquist’ product) will be generated at 144MHz, With this filter selecting the 2m band signals from the AD9834 output, the next step was to see if it was possible to achieve frequency modulation with the chip. As noted earlier, the AD9834 data sheet highlights the possibility of phase (and thus frequency) modulation but gives no further detail. Analog Devices’ application notes did not provide any further details about how the AD9834 phase modulation registers might be used to achieve this. Also, despite a thorough search, I could not find any DDS design in which this feature was actually used. That led me to dig further into phase modulation. I analysed and tested the AD9834 Phase Registers to understand their impact on the DDS output signal. When FM was emerging on the 2m VHF amateur radio band from 1970 to 1980, phase modulation (PM) was generally considered the preferred Fig.4: a bandpass filter using discrete inductors and capacitors can give this response, which allows the 144148MHz alias output product to be selected while other alias, clock and fundamental signals are rejected. 46 Silicon Chip approach. There were claims of “better quality” modulation and “more natural sounding” voices. However, there was little to suggest any evidence supporting these claims. Practical issues, and the arrival of cheap varicap diodes, led to PM being quickly overwhelmed by FM. Varicaps were easy to use in oscillators and often reduced the component count, unfortunately sometimes at the expense of modulation linearity. PM quickly fell out of favour, and that may have led to the minimal information about phase modulation in the technical magazines, handbooks and reference textbooks of the period. One useful source from those days was William Orr’s classic “Radio Handbook”. The 1981 edition briefly described the method and provided a few examples; see my summary in Fig.5. First, the transmitter’s frequency deviation (ie, modulation) is directly proportional to the amplitude of the input audio signal level for both phase modulation and frequency modulation. With FM, the frequency deviation remains constant regardless of the input signal’s frequency. However, with PM, the deviation increases with increasing frequency. Since phase-modulated transmitters were initially more popular, the characteristic PM frequency response, later referred to as ‘pre-emphasis’, also required the reverse audio frequency characteristic to be implemented in the receiver, ie, ‘de-emphasis’. This Fig.5: the frequency deviation of a phase- or frequencymodulated transmitter depends on the modulation level, but phase-modulated transmitter deviation also depends on the input modulation frequency. (Adapted from Bill Orr’s “Radio Handbook”, 1981) Australia's electronics magazine siliconchip.com.au is usually achieved by a simple RC circuit located immediately after the FM receiver’s discriminator (detector) stage. FM transmitters required the addition of this pre-emphasis characteristic to work correctly with those FM receivers. A similar RC circuit was usually added just ahead of the FM modulator to mimic phase modulation. Since the effect of noise increases with audio frequency, adding pre-­ emphasis to FM (or just using phase modulation with its integral pre-­ emphasis characteristic) improves noise performance. AD9834 phase modulation With that background, let’s return to the AD9834. Phase modulation in the AD9834 is produced by making periodic changes to the value stored in the Phase Register (PHASE0/1 REG; see Fig.1). The Phase Register’s value results in a precise phase shift of the current DDS output signal. The DDS output frequency is determined by the value in the AD9834’s Frequency Register. The AD9834’s Phase Register value shifts the phase of the fout carrier by 2π ÷ 4096 multiplied by the value contained in the Phase Register. The Analog Devices data sheet doesn’t explicitly state this, but that’s what it does. As to the lack of any application of this information, any example, or further supporting detail, AD9834 users are left to fathom the usefulness of this relationship for themselves. As it turns out, by periodically storing a value proportional to the amplitude of an incoming audio signal in the AD9834’s 12-bit Phase Register, it is possible to produce the desired PM (and thus FM) signal. There should, by rights, be a fanfare of trumpets at this point in the story, but there’s another crucial detail. The phase shift that produces phase modulation in the AD9834 is the same at the fundamental output frequency as for all the other aliased carriers (see Fig.3). That is entirely different from the traditional phase modulators and transmitters described in reference books, those early FM broadcast transmitters, and the really old, sorry, ‘legacy’ 2m VHF ham transmitters. These traditional transmitters used a series of frequency multiplier stages to generate the required VHF carrier siliconchip.com.au AD9834 phase modulation details As phase deviation is proportional to both the frequency and amplitude of the modulating signal, the following equation can be used: Frequency deviation (in kHz) = phase shift (in radians) × modulation frequency (in kHz) For example, if the modulating signal’s frequency is 1kHz and we have a carrier phase shift of +½ radian, the resulting output signal’s frequency deviation is +500Hz (note that 2π radians = 360°). A standard signal generator setup for testing a 2m amateur radio VHF FM receiver (25kHz channel spacing) uses a 1kHz test tone and a carrier frequency deviation of ±3kHz. Therefore, we require a maximum phase shift on the AD9834 output carrier of 3kHz ÷ 1kHz = 3 radians. The 12-bit Phase Register in the AD9834 generates a phase shift (on the DDS output carrier) of π radians when PHASEREG = 2048. Therefore, to achieve 3 radians of phase deviation, the Phase Register must be loaded with a peak value of 3 ÷ π × 2048 = 1956. The 1kHz internal oscillator delivers a 3.7V peak-to-peak sinewave to the micro’s ADC0 analog input. This gives a peak ADC value in the 10-bit ADC register in the ATtiny45 of about 750. The ADC reference voltage is 5V, so 5V at the ADC input would result in a maximum reading of 1023. The software scales this 750 input value to give a peak Phase Register value of about 2250. That is a little higher than the calculated value of 1956 due to rounding errors in the simple integer calculation routine used. Tests with a professional-grade modulation meter confirmed this value produced ±3kHz deviation in the AD9834 output signal. If you use the external modulation input in this design, the maximum frequency deviation that can be achieved is about ±4.5kHz. This is due to the ADC measurement limit of 1023 (because of the 10-bit ADC in the ATtiny45) with a 5V peak-to-peak audio input. That input level must not be exceeded, or the ATtiny45 could be damaged. (before the arrival of phase-locked loops [PLLs]). A typical early 2m VHF ham transmitter might have a 12MHz crystal oscillator followed by a series of multiplier stages. They usually used a frequency tripler stage followed by two frequency doublers to multiply the oscillator frequency by 12, giving a final output at 144MHz. Consequently, the frequency deviation measured at 144MHz was twelve times that at 12MHz. That was very useful because these legacy phase and frequency modulators could not be phase-shifted (or frequency-­ shifted) very much. But, since the transmitter’s multiplier stages also multiplied the phase and frequency modulation deviation, the final output readily achieved the desired modulation deviation. It’s a different situation with a DDS. If AD9834 output (fout) is set to 14MHz, changing the Phase Register values appropriately in the AD9834 will produce a 1kHz tone with a 3kHz frequency deviation on that output. The 146MHz alias product will have an identical 3kHz frequency deviation due to the aliasing process in the DDS. Australia's electronics magazine There is no ‘multiplication effect’ like that in those traditional PM and FM transmitters. The alias outputs are all directly generated equally and simultaneously by the DDS NCO process. Furthermore, the AD9834’s phase modulator is a 100% digital process carried out in the NCO. The maximum frequency deviation is limited by the Phase Register length and the NCO process. The accompanying panel titled “AD9834 phase modulation details” describes how the Phase Register values generate the desired PM (and FM) deviation. Circuit details Fig.6 shows the circuit of this compact 2m FM Test Generator. The AD9834 is controlled by an Atmel/Microchip 8-pin ATtiny45 microcontroller using a three-wire SPI serial bus. The ATtiny45 has a hardware SPI interface, simplifying the software and increasing data transfer speed. This SPI interface only uses two (USCK and DO) of the usual three SPI lines because no data needs to be read from the AD9834. October 2023  47 The ATtiny45 is clocked at 16MHz using its internal RC oscillator and PLL. That releases all six I/O pins for this design (the other two are the 5V power supply). Practically all of the firmware operates using a series of software interrupt routines. Three pushbuttons (S1-S3) control the generator’s frequency and modulation. These all connect to pin 2 of the ATtiny45. Pressing any button triggers a software interrupt routine that measures the voltage at that pin to determine which button was pressed. The Frequency button (S2) allows the selection of one of eight fixed channels at 500kHz intervals from 144 to 148MHz. The Memory button (S3) selects one of four user-­programmable channels in the 144-148MHz range. Finally, the Modulation button (S1) turns the 1kHz modulation tone on and off. As described in the panel, this produces a frequency deviation of ±3kHz. A modified interrupt routine could support a rotary encoder for multi-channel frequency tuning, memory selection and other features. However, without a suitable display – there are just not enough pins – the design was intentionally kept ultra-simple and inexpensive. Generating a test tone The internally-generated 1kHz modulation test tone is produced by ‘bit-banging’ digital output pin 5 (PB0). Usually, one of the ATtiny’s internal timers would be used to do this. However, that wasn’t possible here because the timer-related pins were already handling the AD9834 SPI control bus. The bit-banging process produces an unusual low-harmonic PWM output. It is designed to null the 5th and 7th harmonics. As a result, it only requires a very modest 10kW/22nF RC filter to give a remarkably clean and accurate 1kHz sine wave. The sinewave measures around 3.7V peak-to-peak (with Vcc at 5V) when it arrives at the ATtiny45’s ADC0 input on pin 1. The 1kHz PWM tone is generated continuously at pin 5. However, it is only sampled (after the RC filter) at pin 1 with the ATtiny’s analog-to-­digital converter (ADC) when modulation is required. The sinewave’s amplitude is sampled 8000 times each second (8ksps). These samples are used to update the Phase Register in the AD9834. The relationship between the audio tone’s amplitude and frequency and the DDS-generated frequency deviation is explained in the panel above. The ±3kHz deviation for FM gives the usual 60%-of-peak-modulation Fig.6: two small chips and a handful of passive components are all that’s required in this high stability digitally modulated 2m FM Test Generator. Alignment just involves adjustment of the bandpass filter for maximum output. 48 Silicon Chip Australia's electronics magazine siliconchip.com.au level used for testing 25kHz VHF FM channels. By making this modulation process external to the ATtiny45, external modulation sources can also be used. If external modulation is selected with S4, the maximum input level of 5V peak-to-peak will produce at most ±4.5kHz deviation. For those using 12.5kHz FM channels, an external audio level of 3V peak-to-peak will give ±2.5kHz deviation with the Test Generator (peak deviation), and 2V peak-to-peak will give ±1.75kHz deviation (60% of peak deviation). Those figures all assume Vcc is close to 5V. The 8kHz sampling rate used in this design limits the modulating frequency to less than 4kHz (the Nyquist limit). However, the lack of any anti-aliasing filter in the software practically limits external modulation frequencies to less than 3kHz. Output signal generation An 80MHz external crystal oscillator clocks the AD9834. This frequency is close to optimal for this application because it eases the RF filtering task slightly, and suitable crystal oscillators are readily available at low cost. As described earlier, a low-loss, highly selective bandpass filter is required to extract the wanted super-Nyquist 2m-band RF signal. A pair of high-Q air-wound inductors are used. These are essential to produce the desired result. The filter is designed to give about 40dB of attenuation at the nearest alias bands close to 100MHz and 200MHz while introducing no more than 4dB passband loss. Happily, these inductors are quick and easy to make at minimal cost. Despite the wide variation in output levels generated by the DDS process, the resulting 2m band output levels (post-filter) are about -40dBm ±2dBm. The directly generated carrier, the residual clock and the other aliases are attenuated by 25dB or more, and spurious products are at least 30dB below the output level. A pair of switched attenuators provide three output levels suitable for receiver tests. The highest output level, -45dBm ±2dBm, places an FM receiver well into limiting without overloading, producing a 1kHz demodulated audio tone with a very good signal-to-noise ratio. siliconchip.com.au Enabling one attenuator (either) gives a signal level of about -75dBm. This is close to the typical ‘corner’ of FM receiver performance where limiting begins to improve the receiver’s signal-to-noise usefully. Adding the second attenuator gives a test signal of about -105dBm. This is close to that used in typical 12dB SINAD receiver sensitivity and squelch gating tests. The absolute accuracy of these levels depends in part on the output filter alignment and the construction method. The prototype was housed in a 3D-printed enclosure, which provides limited shielding. That limits the absolute accuracy of the signal level and the absolute accuracy of some measurements, so if you’re after precision, you will need a metal case for shielding. Power supply My initial plan was to derive the 5V supply for the ATtiny45 and AD9834 using a 7805 linear regulator. While the AD9834 data sheet states it is a “low current DDS device” (20mA at 5V), it needs an external clock generator, in this case at 80MHz. These typically consume 30-70mA, although a few will operate with as little as 10mA. Hence, a 78L05 might not be sufficient when operating from a 9-12V DC supply. An alternative is to use a switchmode regulator module like the one shown in Photo 1. This can be mounted in place of the 7805 linear regulator on the PCB. The input voltage can be from 6V to 15V DC, and the 5V output can deliver up to 500mA. It improves efficiency and remains cool during operation. The prototype is powered by a single 1.5V AA alkaline cell. This is only suitable for intermittent use due to the limited capacity of the cell but allows a compact 3D-printed PLA enclosure to be used. A boost regulator module steps the cell voltage up to 5V (shown in Photo 2). Finally, powering the Generator from a single Li-ion or LiPo cell is also possible. They have a nominal fully charged output voltage of 4.2V and an operating end-point voltage of 3.5V. I tested the Generator with supply voltages from 3-5V. The output level remained constant within 0.2dB across that voltage range! If using a Li-ion or LiPo cell, there is also the option to integrate a small Australia's electronics magazine USB charger PCB (see Photo 3). Some versions include an automatic battery disconnect feature to ensure the battery does not operate below 3.5V, which could damage it. Construction The Test Generator is built on a small 50 × 70mm PCB, coded 06107231, that hosts a mix of SMD and through-hole parts. The board has a near-­ continuous top-side ground plane with all the SMD parts mounted on the underside. This arrangement keeps the unit compact while allowing for easier testing and modification during development. It also produced improved RF performance over other approaches. The layout is shown in Fig.7. Start by fitting the 23 SMD components on the underside. It is easiest to start with the AD9834. Position it over its pads, tack one lead, then double-check that its pin 1 orientation is correct and all pins are correctly centred on their pads before soldering the rest. Adding a little bit of flux paste to the pads and on the leads will make soldering it much easier. If you accidentally bridge any of its pins (which is easy to do), add a bit of Photo 1: a DD4012SA small switching buck regulator (top and bottom shown) can deliver 5V DC to run the chips efficiently. Photo 2: this tiny 5V boost regulator module is used to step up the cell voltage. Photo 3: this TP4056 module charges a Li-ion or LiPo cell and automatically disconnects the load if the terminal voltage falls too low. October 2023  49 bit, scrape the enamel from the wire ends to allow for the soldered PCB connections. Flip the board over and continue construction by adding the ATtiny45’s socket (watching its orientation), the two trimmer capacitors, the two inductors and the two electrolytic capacitors on the top side of the PCB. Now mount the crystal oscillator module. The PCB allows for either full or half-sized oscillator modules to be used. Next, install the three pushbuttons, the two slide switches, the toggle switch, and finally, the regulator (a 7805 or one of the other options). 144MHz to 148MHz. Enter your choice of frequencies in the blue cells. It’s best not to touch anything else! After entering the four memory channel frequencies into the spreadsheet provided (left side of Screen 1). Scroll down to the bottom of the worksheet and click on the green Write EEP File button (Screen 2). The spreadsheet then generates and saves the 2mTestGenFreq.EEP file in the same directory as the spreadsheet file. If you don’t have Excel, you can open the file in a free package like LibreOffice, and everything will work except for the final file-saving step; pressing the green button will do nothing. Instead, after updating the frequencies, check the text just to the right of that button. You will see three lines that start with colons. Click your mouse on the left side of the first line that starts with a colon (just to the right of the colon), then drag it down to the third line and release the button. Press CTRL+C (or the equivalent command to copy to the clipboard), then create a new text file, open it and press CTRL+V (to paste those lines into it). Save that file and then rename it from a .txt extension to .eep. That gives the same result as Excel does when pressing the button. Either way, rename the resulting EEP file so you know what it’s for. Otherwise, the next time you use the spreadsheet, it will overwrite your previous file. Generating the EEPROM file Programming the ATtiny45 I have created a spreadsheet to allow the easy entry and programming of the four user-selected 2m frequencies. These may be on any frequency from The HEX file for the Test Generator is available for download from the Silicon Chip website, along with the BASCOM source code. You can Fig.7: the Test Generator is built on a compact 50 × 70mm double-sided PCB with SMDs on the underside and the through-hole components on the top. flux paste to the bridge and apply some clean solder wick with your soldering iron. Once it gets hot enough and the flux starts to smoke, the excess solder will be pulled into the wick, leaving clean solder joints without a bridge. Repeat as necessary until all the solder joints look good under magnification. The remaining SMDs can then be fitted. All can be soldered in place by hand with a fine-tipped soldering iron. Fit the edge-mount SMA coaxial connector after that. Make the two inductors using 0.4mm diameter (26SWG) enamelled copper wire wound on a 5mm diameter drill bit shaft. Close-wind 10 turns for each, then stretch each coil slightly until each measures 13mm long. Allow 10mm of extra wire at each end of each coil for the connections. While keeping each coil on the drill Fuse settings for the ATtiny45 1. Memory Frequency Tables – ENTER YOUR FREQUENCIES HERE 2m TG Frequency Data Ch 01 144,285,000 Ch 02 145,775,000 Ch 03 146,900,000 Ch 04 147,250,000 HEX EFCAB8 D90E68 C7E3E0 C28CB0 Enter your four memory frequencies in the BLUE cells Calculated HEX value B8 68 E0 B0 EEPROM Data CA EF 0E D9 E3 C7 8C C2 00 00 00 00 INSERT MEMORY FREQUENCIES HERE Byte Value Lock byte 0xFF Extended 0xFF byte Four bytes of data per frequency to be stored in the EEPROM Screen 1: a shot of the spreadsheet which is used to generate the data required to program the user-selected 2m frequencies. High byte 0x57 Low byte 0xE1 Click on this button AFTER you have entered all four frequencies into the cells in Section 2 above Write EEP File Screen 2: clicking on the green button automatically generates the ATtiny45 EEP file, which contains your four desired frequencies. 50 Silicon Chip Notes Australia's electronics magazine RSTDISBL = 0 (set), EESAVE = 0 (set) CKSEL = 0001 for 16MHz internal RC oscillator, CKDIV8 = 1 (disabled) siliconchip.com.au load the EEP file with the four user-­ defined frequencies into the chip at the same time. Program your ATtiny45 with the HEX and EEP files using a suitable programmer (USBasp etc). After that, program the configuration fuses. Table 1 (“Fuse settings for the ATtiny45”) shows the required fuse settings. My article on the Shirt Pocket DDS Oscillator in the September 2020 issue included a small programming adaptor that can be used to program an ATtiny chip out of the circuit in conjunction with a suitable serial programmer (see p47; siliconchip.au/Article/14563). The PCB is still available from the Silicon Chip Online Shop; see siliconchip. au/Shop/8/5642 You could also use our May & June 2012 PIC/AVR Programming Adaptor (siliconchip.au/Series/24) or build an adaptor on a breadboard or small piece of protoboard. Reprogramming the ATtiny45 If you want to change the memory channel frequencies, you can’t just put the chip back into a regular programming adaptor since the RESET pin is disabled. The ATtiny45 must first be erased using a special HV programmer. I have designed a simple chip eraser and fuse restorer (“CEFR”) to do this. You can read about how to build it on my website at www.zl2pd.com/CEFR_ Fuse_Reset_Tool.html – it requires no special parts and can be powered from a USB socket or external 5V USB power supply. Another well-known DIY fuse resetting tool is the Fuse Doctor (see https://github.com/SukkoPera/ avr-fusebit-doctor). Final assembly Depending on the enclosure and Parts List – VHF / FM Test Signal Generator 1 double-sided PCB coded 06107231, 50 × 70mm 1 3D-printed case & front panel label 1 set of AA cell contacts 1 AA alkaline cell 1 80MHz crystal oscillator, full or half-size DIP type (X01) [AliExpress siliconchip.au/link/abmb] 3 PCB-mount momentary tactile pushbutton switches (S1-S3) [Altronics S1126A, Jaycar SP0609] 4 solder tag miniature DPDT slide switches (S4-S7) [Altronics S2010, Jaycar SS0852] 1 8-pin DIL IC socket (for IC1) 1 SMA edge connector (CON1) 3 2-pin headers, 2.54mm pitch (optional) (CON2-CON4) 1 3-pin right-angle header, 2.54mm pitch (if REG1 has none) 2 No.4 × 5mm self-tapping screws 2 M3 x 8mm panhead machine screws 2 M3 x 8mm countersunk head machine screws 2 10mm-long M3-tapped Nylon spacers 1 400mm length of 0.4mm diameter/26SWG enamelled copper wire (for L1 & L2) [Altronics W0404, Jaycar WW4014] various lengths of light-duty hookup wire Semiconductors 1 ATtiny45-20PU 8-bit microcontroller programmed with 0610723A.HEX, DIP-8 (IC1) 1 AD9834-BRUZ or -CRUZ DDS signal generator IC, TSSOP-20 (IC2) [AliExpress siliconchip.au/link/abmc] 1 3-pin 5V output boost module (REG1) (for AA cell operation) [Silicon Chip SC6780, AliExpress siliconchip.au/link/abmd] OR 1 3-pin 5V output buck module (REG1) (for 6.5-40V DC operation) [Silicon Chip SC6781, AliExpress siliconchip.au/link/abme] OR 1 7805 5V 1A linear regulator, TO-220 (for 8-16V DC operation) Capacitors (all SMD M2012/0805 50V ceramic unless noted) 1 10μF 50V/63V radial electrolytic [Altronics R5065, Jaycar RE6075] 1 1μF 50V/63V radial electrolytic [Altronics R5018, Jaycar RE6032] 5 100nF X7R 1 22nF X7R 1 10nF X7R 1 120pF NP0/C0G 2 47pF NP0/C0G 2 6-20pF PCB-mount trimmer capacitors (VC1, VC2) [Altronics R4005] Resistors (all 1% SMD M2012/0805 size) 2 10kW 1 6.8kW 1 3.9kW 1 1.8kW 2 820W 4 51W 1 220W Photo 4: my simple Chip Eraser and Fuse Restorer (CEFR) resets ATtiny25/45/85 fuses back to the factory default settings so that you can change the memory channel frequencies. Parts availability We don’t have a kit for this project but we can supply the PCB, programmed microcontroller and buck or boost module. The remaining parts can be found at your usual suppliers or from the sources listed above. October 2023  51 Photo 5: the PCB fits in a 3D-printed enclosure and is powered by a single 1.5V AA cell. The boost regulator module is to the left of the power switch, while the output filter is above and alongside the regulator. The AD9834 and other SMD parts are on the underside of the PCB, but their locations are marked on the component side. power supply option you select, add the power switch and power wiring to suit. In the prototype, the AA cell is located in the lower part of a 3D-printed case designed for the board. STL files for the enclosure are available for download along with the software, see siliconchip.com.au/ Shop/6/266 The ‘battery shrapnel’ (those metal tabs at each end of the battery compartment) slide into the slots designed for them, and the power switch likewise slots into place in the case. The PCB can then be mounted. Two 5mm-long, 3mm diameter self-tapping screws pass from the underside into two 6mm-long Nylon spacers to hold the PCB in place. Four more screws hold the front panel in place from the front. Testing and operation Insert the programmed ATtiny45 into the socket, ensuring its pin 1 end is at the notched end of the socket; if in doubt, check Fig.7. Switch both attenuator switches on the Test Generator to the left-most positions (minimum attenuation). Place a 2m VHF FM handheld within about 500mm of the assembled PCB and set it to receive on 146.000MHz. Unmute the receiver so 52 Silicon Chip you can hear channel noise and adjust the volume to a suitable level. Connect and turn on the power to the Test Generator. The handheld’s receiver should immediately go quiet, and the handheld’s signal strength meter (if there is one) should indicate a very high level (S9 or better, typically). If this does not occur, turn off the power and carefully check your construction. Assuming your Test Generator passed this test, briefly press and release the Modulation button. You should hear a clean 1kHz tone in the FM handheld’s speaker audio. While monitoring the audio and observing the handheld’s signal level meter, adjust the two trimmer capacitors on the Test Generator to achieve the maximum signal level. You might need to move the handheld several metres away (or more) so you begin to hear a little noise on the received signal. A slight improvement in tuning can sometimes be achieved by very slightly compressing or stretching one or both of the coils, but this is seldom necessary. The generator starts with the output set to 146.0MHz without modulation. Each press of the Frequency button increments the frequency by 500kHz. If the current frequency is 148.0MHz, the next press will change the output to 144.0MHz, and the 500kHz frequency steps resume again through the 2m band. Briefly press the Frequency button once to change the Test Generator to 146.5MHz. Retune the handheld’s receiver frequency to 146.5MHz; the modulated signal should be audible on this channel. Press the Modulation button briefly to verify that the modulation can be turned on and off as desired. Ensure What about phase modulation? You might have noticed I wrote about phase modulation, but the design only supports CW or FM. So what’s going on? Phase and frequency modulation are two sides of the same ‘angle modulation’ coin. If a modulator is a true phase modulator, the input audio signal is integrated before modulation to produce FM. Likewise, differentiating the input signal to a genuine frequency modulator results in phase modulation. Hams (amateur radio operators) all talk of “FM”, regardless of how it’s generated in their radios. When using analog angle modulation (“FM”), it was also always easier to measure frequency deviation when setting up transmitters and everything associated with them. Nobody ever measured phase shift. So everyone talks about FM and frequency deviation when talking about analog angle modulation. More importantly for the 2m Test Generator is that, for a single tone, FM and PM are indistinguishable. Curiously then, but logically in context, we only talk about (and measure) specific instantaneous phase shifts when it comes to data angle modulation. Hence PSK, QPSK and 8PSK, where we see a digital application of relatively large phase step modulation. You might recall that I mentioned the possibility of AM. Not many folk use AM on 2m, but interestingly, combining AM and PM makes it possible to generate 16-QAM with a modest amount of extra effort. But that is an idea for another day. Australia's electronics magazine siliconchip.com.au modulation is on again before proceeding to the next test. If the Memory button is pressed, the generator will deliver one of four programmed frequencies starting with the first memory channel. Pressing the Frequency button will return the oscillator output to the currently selected 500kHz frequency increment. Change the handheld receiver frequency to the frequency you programmed as Memory Channel 1 by pressing the Memory button briefly. The modulated signal should now appear on this channel. You can continue to press the Frequency or Memory buttons to select and test the other Generator frequencies. Important notes Signals generated on some frequencies can produce spurious in-band and out-of-band products. This is to be expected from such a simple DDS generator. Analog Devices warn about this in the data sheet, too. These additional signals on the output can be generated by the DDS clock, its harmonics and the mixing of one or more aliases. They are typically at least 25dB (aliases) or 30dB (spurious) below the desired output. That’s similar to a few legacy commercial RF signal generators. Given the absence of anti-aliasing audio filtering, avoid the temptation to feed microphone audio into the external audio input or to add a low-power RF amplifier to the Test Generator. It is not suitable for use as a 2m FM transmitter. While usable for basic testing, it will not meet any regulatory test standards for FM transmitters. External modulation You can use an external audio source, such as an audio oscillator, to modulate the Test Generator. This can be applied to the CON2 input on the PCB. Ensure the level is between 0V and the supply voltage (ie, 5V maximum with a 5V DC supply). That means the signal must have a ~2.5V DC bias. Signals beyond that limit can cause damage. Monitor the output on a nearby 2m handheld transceiver or receiver to confirm that the modulating tone can be heard. If you want to add a socket to feed in an external modulation signal, we suggest you couple the signal from that socket to CON2 via a 100nF capacitor and connect 100kW resistors from the CON2 signal pin to 5V and GND points on the PCB to get the correct biasing. CTCSS operation A suitable CTCSS encoder can also be connected to the CON4 CTCSS input on the PCB. This may be used, for example, to test an FM receiver’s CTCSS decoder. To do this, set the handheld to a suitable channel and set the handheld CTCSS decoder to an appropriate Fig.8: the front panel graphics can be printed and laminated, then glued to the front of the case. CTCSS frequency, eg, 123.0Hz. Ensure the CTCSS encoder’s maximum output level is no more than about 0.5V peak-to-peak. Select internal modulation as the source on the Test Generator, and push S1 briefly to turn on the FM modulation. When the external CTCSS encoder is operating, the receiver’s CTCSS decoder should detect the CTCSS tone, unmute the receiver, and the Test Generator’s 1kHz tone should be heard. Turning the external CTCSS encoder off or reducing its level below about 0.1V peak-to-peak should cause the receiver’s CTCSS decoder to mute the receiver audio. However, CTCSS decoders can be very sensitive and may continue to detect a valid tone with CTCSS tone input levels of even 5mV peak-to-peak! Final comments Photo 6: the underside of the PCB with all the SMDs fitted. The bridged pins (9 & 10) on the AD9834 chip are both joined to GND, so I left it like that. siliconchip.com.au Australia's electronics magazine The difficulty I encountered in finding basic information and application examples on DDS phase modulation surprised me, given that the data sheets heavily promote this feature. However, achieving precise phase (and frequency) modulation levels with the AD9834 ultimately turned out to be relatively simple. I hope you find the details and design to be of interest. This compact FM Test Signal Generator is fun to use and a great conversation piece. SC October 2023  53