Fullscreen HTML5Med Res (??MB)HTML4

U s in g Ch e a p A s ian El e c t r M o d u o nic l Par t 6e s AD9833-based Direct Digital Synthesiser By JIM ROWE This little signal generator module uses an Analog Devices AD9833 DDS chip and a 25MHz crystal oscillator. It can be programmed to generate sine, triangle or square waves up to 12.5MHz and it's all controlled via an SPI serial interface. 18  Silicon Chip Asia, especially China, some of them available at surprisingly low prices via internet markets like eBay and AliExpress. As a result, you can buy the tiny (18 x 13.5mm) AD9833-based DDS module shown in the photos, which includes a 25MHz crystal oscillator, for the princely sum of $7.88 each – including free delivery to Australia! That's really quite a bargain, which is why we're focusing our attention on it this month. To get an idea of how a DDS works, take a look at the panel titled “DDS in a Nutshell”, which can be found on pages 23 & 24 of this article. Inside the AD9833 The block diagram of Fig.1 shows what's inside that tiny MSOP-10 package. There's quite a lot, although some of the elements are mainly involved in giving the chip its flexibility in terms of output waveform and modulation capabilities. The main sections 2 CAP/2.5V 3 COMP 1 FULL-SCALE CONTROL ON-BOARD REFERENCE REGULATOR 2.5V FREQ0 REGISTER (28-BIT) MUX 1 28 PHASE ACCUMULATOR (28-BIT) Σ 12 SINE ROM 10-BIT DAC MUX 3 FREQ1 REGISTER (28-BIT) MSB S1 (12-BIT) PHASE0 REG PHASE1 REG MCLK 8 FSYNC 6 SDATA 7 SCLK DIVIDE BY 2 (12-BIT) VOUT MUX 4 10 200Ω CONTROL REGISTER (16-BIT) S2 SERIAL INPUT REGISTER (16-BIT) AGND 5 MUX 2 DGND irect Digital Synthesiser or DDS chips have been around for well over 20 years now but for much of that time they were fairly costly. Until recently, they didn't include an integral DAC (digital to analog converter), so you had to use their digital output to drive a separate DAC to generate the analog output signal. In the early 2000s, Analog Devices Incorporated (ADI) announced a new generation of complete DDS devices which did have an integral DAC, as well as offering high performance combined with a price tag significantly lower than what you formerly had to pay for a DDS+DAC combination. Although it's one of the low-cost, lower-performance devices in their range, the AD9833 provides a good example of just what can be achieved nowadays. When combined with a 25MHz crystal oscillator, it can be programmed to produce any output frequency from 0.1Hz to 12.5MHz in 0.1Hz increments, with a choice of three waveforms: sine, triangular or square. All this comes from a chip housed in a tiny MSOP-10 package, running from a supply voltage of 2.3-5.5V, dissipating only 12.65mW and currently with a price tag of $17.46 AUD plus GST in one-off quantities. That's significantly lower than earlier DDS chips. As we've seen in earlier articles in this series, there has also been a huge surge in the manufacture of many kinds of electronics modules in VDD D 4 9 Fig.1: block diagram of the AD9833 DDS IC. The critical blocks are yellow. The phase accumulator generates a series of addresses to look up in the ROM sine table and the resulting values are then fed to a 10-bit DAC which produces the output waveform. Other circuitry allows the output waveform to be changed to a triangle or square wave and also allows for frequency and phase shift keying. siliconchip.com.au involved in basic DDS operation are those shown with a pale yellow fill. Down at lower left in Fig.1 you can see the 16-bit shift register where data and instructions are loaded into the chip from almost any micro, via a standard SPI (Serial Peripheral Interface) bus. We'll discuss that in more detail later. Just above the serial input register is the control register, also 16-bit. This stores the control words, used to set up the configuration of the device, including the output waveform type, which of the two 28-bit frequency registers (FREQ0 or FREQ1) is used to set the DDS output frequency and also whether the phase is shifted by the content of 12-bit phase registers PHASE0 or PHASE1. The main reason why the AD9833 has two frequency registers and two phase registers is to give it the capability of generating signals with frequency-shift keying (FSK) or phase-shift keying (PSK) modulation. Multiplexers MUX1 and MUX2 allow these options to be controlled using bits in the control register. So how are those 28-bit frequency/ phase increment registers FREQ0 and FREQ1 loaded with 28-bit data from the 16-bit serial input register? This is done by sending the data in two 14bit halves, in consecutive 16-bit words from the micro, with the lower half first and then the upper half. The AD9833 can be configured to accept the data this way simply by manipulating two bits in the control register. The same bits can also be used to configure it for setting either the lower or higher 14-bit “half word” alone, which can be useful for some applications (such as frequency sweeping). MUX3, MUX4 and switches S1 and S2 are all controlled by further bits in the control register. MUX3 simply allows the sine ROM to be bypassed, with the output from the phase accumulator fed directly to the DAC. This is how the AD9833 produces a triangle wave output, since the amplitude of a triangle wave is a linear function of its phase. For a square wave output, the DAC is disconnected from the chip's analog output (pin 10) using integrated switch S1, and instead makes use of the MSB (most significant bit) output of MUX3. This automatically gives a square wave output and MUX4 allows you to divide siliconchip.com.au VCC CON1 1 2 3 4 5 6 7 4.7µF 100nF 10nF 2 VDD VCC COMP DGND SDATA 6 SCLK 7 FSYNC 8 SDATA SCLK IC1 AD9833 MCLK 1 4 3 5 VCC OUT 25MHz XTAL OSC EN GND 1 2 100nF FSYNC AGND 10 VOUT VOUT 22pF CAP/2.5V AGND 9 DGND 4 3 100nF Fig.2: circuit of the AD9833-based DDS module used in this article. The AD9833 IC and 25MHz crystal oscillator plus a few passive components are mounted on a small PCB, with a SIL header to make control and output connections. its frequency by two, if needed. The integrated 200W resistor connected between the analog output pin and ground via switch S2 is used to convert the DAC's output current into a proportional voltage output. Since S2 is controlled in parallel with S1, this means that when S1 cuts the link between the DAC output and pin 10, S2 also removes the built-in 200W output shunt. This makes the chip's output voltage swing in square wave mode significantly higher than for the sine or triangular (DAC-derived) options. To be specific, the square wave output is around 5.2V peak to peak, while for sine or triangular waves the output drops to around 650mV peak-to-peak. The complete module Now refer to Fig.2, which shows the complete circuit for the 18 x 13.5mm module shown in the photo below. It simply comprises the AD9833 DDS chip (IC1) and its equally tiny (3 x 2.2mm) 25MHz crystal oscillator. It has six even smaller SMD capacitors, most of them used for filtering either the power supply rails or IC1's Vout pin. Seven-way SIL connector CON1 is used to make all of the signal and power connections to the module. Pins 1 & 2 are used to provide the module with 5V power, while pins 3-5 are used to convey the SPI commands and data to IC1 from the micro you're using to control it. And pins 6 & 7 are used to carry the analog output signal from IC1 out to wherever it's to be used. Limitations Before we talk about driving the module from a micro like an Arduino or a Micromite, we should discuss its limitations. Firstly, the aforementioned difference in output amplitude for square wave versus sine/triangle waves is a factor of about eight times, or 18dB. So if you want to use the module as the heart of a function generator, you will need to attenuate the square wave output by 18dB, to match the sine and triangle output levels. You will also need to pass the sine and triangular outputs through a low- The DDS module is shown at approximately twice actual size to provide greater clarity. From left to right, the pin connections are VCC, DGND, SDATA, SCLK, FSYNC, AGND and VOUT. April 2017  19 AD9833 SERIAL INPUT WORD FORMAT: D15 D14 D13 D12 D11 D10 D9 D8 D7 D6 D5 D4 D3 D2 D1 MSB D0 LSB TO WRITE TO THE CONTROL REGISTER: 0 0 B28 HLB FREQ PHASE SEL. SEL. 0 RESET SLEEP SLEEP OPBIT 1 12 EN 0 DIV2 0 MODE 0 MODE: 0 = SINEWAVE 1 = TRIANGULAR CONTROL REGISTER ADDRESS FREQUENCY REGISTER WRITE MODE: 1= WRITE AS TWO SUCCESSIVE 14-BIT WORDS 0= WRITE SINGLE 14-BIT WORD INTO MSB (HLB = 1) OR LSB (HLB = 0) HALVES OF FREQUENCY REGISTER DAC DATA MSB: 1 = NO DIVISION, 0 = DIVIDE BY 2 CONTROLS WHETHER DAC OUTPUT IS CONNECTED TO VOUT OR NOT: 0 = CONNECTED (SINE OR TRIANGLE), 1 = DISCONNECTED (SQUARE WAVE) HIGH OR LOW BITS SELECT: 1= MSB BITS, 0 = LSB BITS DAC POWER SAVING MODE: 0 = DAC POWERED UP, ACTIVE, 1 = DAC POWERED DOWN FREQUENCY REGISTER SELECT: 0 = FREQ0 REG, 1 = FREQ1 REG Scope 1 (above): a 1000Hz sinewave generated using an Arduino programmed with "AD9833_DDS_module_test.ino". MCLK ENABLE BIT: 0 = MCLK ENABLED (NORMAL OPERATION) 1 = MCLK DISABLED (DAC OUTPUT CONSTANT) PHASE REGISTER SELECT: 0 = PHASE0 REG, 1 = PHASE1 REG Fig.3 (left): the format of the 16-bit digital control data sent to the AD9833. The top two bits determine whether the remaining 14 bits are used to update the frequency, phase or control registers. The control register is used to change the output waveform type, switch between two different sets of frequencies and phases or go into a low-power sleep mode. RESET INTERNAL REGISTERS: 0 = NORMAL OPERATION, 1 = RESET TO WRITE TO A FREQUENCY REGISTER: 0/1 0/1 D13 D12 D11 D10 D9 D8 D7 D6 D5 D4 D3 D2 D1 D0 D5 D4 D3 D2 D1 D0 14 BITS OF DATA FREQ. REGISTER ADDRESS: 01 = FREQ0, 10 = FREQ1 TO WRITE TO A PHASE REGISTER: 1 1 0/1 X D11 D10 D9 D8 D7 D6 The underside of the DDS module with the 7-pin male header attached. Again, it's shown twice actual size, due to its small size (18 x 13.5mm). 12 BITS OF DATA PHASE REGISTER ADDRESS: 110 = PHASE REG0, 111 = PHASE REG1 Another limitation, as noted in the “DDS in a Nutshell” box, is that the maximum output frequency is half the sampling clock frequency; in this case, 12.5MHz. But because of the way a DDS works, it can only produce a clean square wave at this maximum frequency. If you want to get a reasonably smooth sine or triangular wave output, this will only be possible at frequencies below about 20% of the clock frequency, or in this case, a maximum of about 5MHz. Programming it When your program starts up, it will need to carry out a number of set-up tasks. These include: 1. Declare the micro's pins that are going to be used by the SPI interface and set them to their idle state (typically high). 2. Start the SPI interface, configured USB TYPE B MICRO ARDUINO UNO OR NANO, FREETRONICS ELEVEN OR LEOSTICK, DUINOTECH CLASSIC OR NANO, ETC IO1/TXD ICSP IO0/RXD IO3/PWM MISO 1 IO2/PWM IO4/PWM IO5/PWM IO7 IO6/PWM IO8 IO10/SS IO9/PWM IO12/MISO IO11/MOSI GND IO13/SCK AREF SCL SDA SS pass filter with a corner frequency of around 12-15MHz, to remove most of the DAC switching transients. After this processing, the outputs can all pass through a common buffer amplifier and output attenuator system. You don't need to worry about any of these niceties if you simply want to use the module as a programmable clock signal source. You can just program it to generate a square wave output and use it as is. VCC 2 +5V SCK 3 4 MOSI RST 5 6 GND DGND MOSI SCK ADC5/SCL ADC4/SDA ADC3 ADC2 ADC1 ADC0 VIN GND GND +5V +3.3V RESET DC VOLTS INPUT +5V SS Fig.4: hooking the AD9833-based DDS module up to an Arduino. It mainly involves making SPI connections via the 6-pin programming header. 20  Silicon Chip SDATA SCLK FSYNC AD9833 BASED DDS MODULE AGND OUT SINE, TRIANGULAR OR SQUARE WAVE OUT TO LP FILTER, BUFFER & ATTENUATOR siliconchip.com.au Scope 2 & 3: a 1000Hz triangular (left) and square (right) wave produced by running the "AD9833_DDS_module_test.ino" file on an Arduino or compatible device. Note the higher amplitude of the square wave output. for a clock rate of say 5MHz, the data to be sent MSB (most significant bit) first and using clock/data timing mode 2 (10 binary). If possible, it should also be set for the data to be exchanged in 16-bit words rather than bytes. 3. Send initialisation commands to the AD9833 to set up its control register, the FREQ0 register and the PHASE0 register. These involve sending the following five 16-bit words (shown here in hexadecimal): • 0x2100 (resets all registers, sets control register for loading frequency registers via two 14-bit words) • 0x69F1 (lower word to set FREQ0 for 1000Hz) • 0x4000 (upper word to set FREQ0 for 1000Hz) • 0xC000 (writes 000 into PHASE0 register) • 0x2000 (write to control register to begin normal operation) With that, the DDS should produce a 1000Hz sinewave. To change to one of the other waveforms, you need to send the correct code to the control register. To change the output frequency, you need to send the appropriate pair of 14-bit words to one of the frequency registers. Note that these are sent lower word first, then upper word. To make programming the AD9833 a little easier, the basic coding for the control, frequency and phase registers is summarised in Fig.3. I also have written a couple of simple example programs to illustrate programming the AD9833 module; more about these shortly. First you'll need to know how the module can be connected to one of the popular micros. module up to almost any Arduino or Arduino clone. This takes advantage of the fact that most of the connections needed for interfacing to an SPI peripheral are made available on the 6-pin ICSP header fitted to most Arduino variants. The connections to the ICSP header are quite consistent over just about all Arduino variants, including the Uno, Leonardo and Nano, the Freetronics Eleven and LeoStick, and the Duinotech Classic or Nano. In fact, the only connection that's not available via the ICSP header is the one for SS/CS/FSYNC, which needs to be connected to the IO10/SS pin of an Arduino Uno, Freetronics Eleven or Duinotech Classic, as shown in Fig.4. With other variants, you should be able to find the corresponding pin without too much trouble. Even if you can't, the pin reference can be changed in your software sketch to match the pin you do elect to use. Driving it from an Arduino Fig.5, on the next page, shows how to drive the module from a Micromite. Fig.4 shows how to connect the siliconchip.com.au Arduino sample program One of my sample programs written for an Arduino is called “AD9833_ DDS_module_test.ino”. It simply initialises the AD9833, starts generating a 1000Hz sinewave (Scope 1) and then changes the waveform after five seconds, giving you a triangular wave (Scope 2), then a square wave (Scope 3) and finally a half-frequency square wave, before returning to a sinewave and repeating the sequence. If you look at the code, you can see how easy it is to control the AD9833 DDS module from an Arduino. Driving it from a Micromite By connecting the MOSI, SCK and SS/ FSYNC lines to Micromite pins 3, 25 and 22 as shown, MMBasic's built in SPI protocol commands will have no trouble in communicating with the module. One thing to note is that if you want to drive the AD9833 module from a Micromite in a BackPack that is already connected to an LCD touchscreen, there's a small complication arising from the fact that the LCD touchscreen also communicates with the Micromite via its SPI port. To prevent a conflict, your program needs to open the SPI port immediately before it sends commands or data to the module, and then close the port again immediately afterwards. Micromite sample program My other program is written for the Micromite, specifically, the Micromite LCD BackPack. It's called “Simple AD9833 FnGen.bas”. This one is a little more complicated and lets you control the AD9833's output frequency as well as the waveform, simply by using buttons and a virtual keypad on the Micromite's touch screen. It's quite easy to drive, and again should show you how the AD9833 can be controlled via a Micromite. Both this program and the Arduino program are available for download from the Silicon Chip website (www. siliconchip.com.au). Alternative module On page 68 of this issue, we have published a Micromite-based DDS Function Generator project by Geoff Graham which also uses the AD9833. Geoff has used a slightly different module which includes the ability to vary April 2017  21 eight bits of the data are the new potentiometer position while the upper eight bits contain two command bits, two channel selection bits and four “don't care” bits, which it ignores. The command bits allow you to select whether you are setting the pot wiper position or commanding the IC go to into a power-down mode. We suggest you check its datasheet for details, however, you really only need to send one of two different commands to this device when using this module: 0x11xy – set wiper position to 8-bit value xy 0x2100 – shut down potentiometer, saving power and disabling the output signal For example, the command 0x11FF will set the output level to maximum, command 0x1180 will set the output level to 50% and 0x1101 will set it to the minimum non-zero level. The attenuated output signal is available at pin 6 and this goes to the non-inverting input of an AD8051 high-frequency op amp which is configured with a gain of six times, providing an output swing of around 3V peak-to-peak for sine and triangle waves. This signal is fed to both a 2-pin header connector and SMA socket via another 0W resistor. Other variations in this module are that it has 100W protection resistors for GND +5V +3.3V VCC 26 25 24 22 SCK DGND MOSI SCK SS SS 21 MICROMITE SDATA SCLK FSYNC 18 AGND 17 OUT AD9833 BASED DDS MODULE 16 14 (MISO) SINE, TRIANGULAR OR SQUARE WAVE OUT TO LP FILTER, BUFFER & ATTENUATOR 10 9 5 4 3 MOSI RESET the output level and also has a lowimpedance buffered output. The module Geoff has used is shown below, to the left of its circuit diagram, Fig.6. It differs from the simpler module shown in Fig.2 in that it has the output signal from the AD9833 fed to a separate SIL connector. From there the signal is routed to an MCP41010 10kW digital potentiometer IC via a 0W resistor, which acts as a digitally controlled attenuator. The output of this attenuator is fed to an AD8051 rail-to-rail op amp and together these constitute a PGA, or Fig.5: connecting the module to a Micromite is similarly easy; all you need to do is wire up its two power pins, the three SPI pins and the signal output connections. Programmable gain amplifier (not pin grid array). The digital pot also communicates via SPI and in fact its clock and data pins are wired up in parallel with the AD9833's so it is on the same SPI bus. The only difference in communication is rather than pulling the FSY pin low, as you do to communicate with the AD9833, you pull the CS pin low to communicate with the MCP41010. Like the AD9833, the MCP41010 is controlled by writing 16-bit data words to it. For the MCP41010, the bottom VCC 4.7µF 1 2 3 4 5 6 100nF DGND 10nF 2 VDD VCC COMP 4 4 x 100Ω FSY 8 CLK 7 DAT 6 FSYNC SCLK IC1 AD9833 10Ω 1 4.7µF 100nF MCLK 3 5 25MHz XTAL OSC OUT GND 1 2 SDATA CS 3 CAP/2.5V AGND 100nF 9 VOUT 10 0Ω DGND 4 5 6 22pF 7 Fig.6: the circuit of the alternative DDS module shown above, which is also widely available. This one incorporates an MCP41010 digital potentiometer to allow the output amplitude to be controlled as well as an AD8051 high-speed op amp buffer to keep the output impedance low and provide some gain to allow a higher maximum output signal level. The module is shown at aproximately 1.5 times its actual size of 32 x 32mm. 22  Silicon Chip EN VCC 3 VOUT AGND 1 2 100nF 4 VDD PA0 MCP41010 CS SI PW0 SCK PB0 GND 5 AD8051 2 1 0Ω 8 1 3 2 4 PGA AGND 1 2 5k 1k siliconchip.com.au the four control pin inputs, the supply for the 25MHz crystal oscillator has a 10W isolating resistor that also forms a low-pass filter in combination with the added 4.7µF ceramic capacitor and there is a 2-pin header which makes the output of the AD9833 available, before it enters the attenuator. There are more details on how to use this module in this month's article on the Micromite-based DDS Function Generator, but besides needing to program the digital pot with the attenuation value, its control is pretty much identical to the description above. Final comments In the June 2016 issue of Silicon Chip, on pages 86-87, we published a “Touch-screen Function Generator” design by NSW reader Dan Amos. Mr Amos' design used an AD9833 module driven by a Micromite with an LCD BackPack but he also added a digital potentiometer, an output buffer amplifier and even an incremental encoder for adjusting either the output frequency or its amplitude. He also provided the MMBasic source code for his program, and a user manual as a PDF file – both of which can be downloaded from the Silicon Chip website. So that project and its software is also an excellent example to get you started on using the AD9833 DDS module. One last comment before closing. As well as being able to generate fixed frequency, FSK and PSK modulated signals, the AD9833 can also be programmed to generate swept-frequency signals. In fact, the Micromite DDS Function Generator in this issue does just that, so refer to that article on page 68 for more details on frequency sweeping. DDS in a Nutshell This simplified explanation should give you some insight into how a DDS works. A DDS is based around one or more look-up tables stored in read-only memory (ROM). These contain a set of high-resolution digital samples of a single wave cycle. Let's consider the case where the table contains a sinewave. The values from the ROM table are fed to a DAC (digital-to-analog converter), so that for each entry in the table, the DAC will produce an analog DC voltage corresponding to the value of the sample stored in that address. As a result, if a counter is used to cycle through the table entries continuously, the DAC output is a continuous sinewave. Let's say that the table contains 1000 entries which represent a single sinewave cycle and the counter which indexes the table is incre- mented at a rate of 1MHz. This means that the output will be a sinewave at 1MHz ÷ 1000 = 1kHz. By changing the rate at which the counter increases, we can change the output frequency. Since the DDS chip operates from a fixed external clock, in order to vary the rate at which the DDS runs through its ROM table, a fancier counter configuration known as a “phase accumulator” is used. This is shown in Fig.7 and it consists of a binary adder feeding an accumulator register. The important point to note is that the phase accumulator register has 28 bits of precision while the sample table, with 4096 entries, only requires a 12-bit number to index its entries. Hence there are an additional 16 bits of fractional phase data in the register and these effectively indicate PHASE ACCUMULATOR 28-BIT FREQUENCY REGISTER 28 BINARY ADDER + 28 28-BIT 28 ACCUMULATOR REGISTER 12 28 28 4096 ENTRY WAVEFORM SAMPLE TABLE (ROM) 10 10-BIT DAC ANALOG OUTPUT Fig.7: the basic layout of a simple DDS. The Phase 28 Accumulator adds the contents of the frequency register to the accumulator register on each FREQUENCY MASTER CLOCK PROGRAMMING INPUT clock cycle. The top 12 bits of the accumulator register is then used to look up an entry in the waveform table ROM, producing a 10-bit digital amplitude value which is subsequently fed to the digital-to-analog converter (DAC) to generate the analog output signal. siliconchip.com.au output phase values in-between those represented by the values in the table. The binary adder has two 28-bit inputs, one of which is the current phase value from the accumulator register. The other input comes from the frequency register at far left, also 28 bits wide. This is the register which we use to set the DDS output frequency.At each clock cycle, the value in the frequency register is added to the value in the accumulator register and this result is stored back in the accumulator register. As a result, as long as the frequency register value doesn't change, the accumulator register increases by the same amount on each clock cycle. With a 28-bit phase accumulator register, a value of zero indicates a phase in radians of zero while its maximum value of 228 - 1 (268,435,455) represents a phase of just under 2π. When the value of the accumulator register exceeds 228 - 1, it rolls back around to zero, hence maintaining 0 < phase < 2π. Each time it “rolls over”, that represents one complete cycle from the output. With a frequency register value of 1, it will take 228 clock cycles for this to happen. With a master clock of 25MHz, that means the output frequency will be 25MHz ÷ 228 = 0.09313Hz or just under 0.1Hz. With a frequency register value of 2, it will take 228 ÷ 2 clock cycles to roll over, giving an output frequency of 25MHz ÷ 227 = 0.186Hz and so on. continued next page April 2017  23 So the output frequency resolution with this configuration is just under 0.1Hz. But how are such low frequencies possible with only a 4096-entry table? Well, only the top 12 bits of the 28-bit accumulator register are used to index the ROM table. This means with the minimum frequency value of one, it will only roll over to the next entry in the table once every 2(28-12) = 65,536 clock cycles. Hence, each value from the table is sent to the DAC 65,536 times before progressing to the next one, giving a very low frequency. At higher frequencies, in this case above 25MHz ÷ 4096 (6.103kHz), values in the table will be skipped when necessary in order to increase the output frequency. In other words, the counter which indexes the table may increase at a rate of 1, 2, 3, 4 times per clock, or somewhere inbetween, by skipping the occasional table entry. For example, to produce an output of 12.207kHz, every second entry from the ROM table is sent to the DAC (12.207kHz = 25MHz ÷ [4096 ÷ 2]). Based on the above, we can calculate the output frequency as: Fo = Dph × Mclk ÷ Rac For the DDS shown in the diagram, with a 28-bit phase accumulator having a resolution (Rac) of 228 (= 268,435,456) and with a master clock frequency (Mclk) of 25MHz, this simplifies down to: Fo = Dph × 0.09313 By the way, while we're showing the table as containing 4096 entries, note that due to symmetry, it's only actually necessary to store the values representing one quarter of a sinewave. The second quadrant of a sinewave is a mirror-image of the first, so this can be achieved by running through a quarter sine table backwards, while the third and fourth quadrant are simply an inverted version of the first and second, and these can be obtained by negating the values from the first two quadrants. You'll also notice that the samples stored in the ROM are shown as having a resolution of 10 bits, to suit the 10-bit DAC, which can produce an analog output with 1024 different voltage levels. One more thing to bear in mind. Because a DDS achieves higher output frequencies by skipping samples in the waveform ROM, at higher output frequencies, the sampling resolution effectively drops. This continues until you reach the “Nyquist frequency” of half the master clock frequency (ie, 12.5MHz) above which the output from the DAC actually starts to drop in frequency. So the theoretical maximum frequency for a DDS is half that of the master clock. But in practice, because of the above, if you want a reasonably smooth sinewave output that doesn't need too much low-pass filtering, it's a good idea to limit the maximum output frequency to about 20% of the master clock frequency; say 5MHz for a 25MHz SC master clock. Distributors of quality test and measurement equipment. Signal Hound – USB-based spectrum analysers and tracking generators to 12GHz. Virtins Technologies DSO – Up to 80MHz dual input plus digital trace and signal generator Nuand BladeRF – 60kHz– 3.8GHz SDR Tx and Rx Bitscope Logic Probes – 100MHz bandwidth mixed signal scope and waveform generator Manufacturers of the Flamingo 25kg fixed-wing UAV. Payload integration services available. Australian UAV Technologies Pty Ltd ABN: 65 165 321 862 T/A Silvertone Electronics 1/21 Nagle Street, Wagga Wagga NSW 2650 Ph 02 6931 8252 contact<at>silvertone.com.au www.silvertone.com.au 24  Silicon Chip siliconchip.com.au