SDR Frequency Tuning Mechanics

This application note discusses the frequency tuning mechanics associated with Per Vices Software Defined Radio (SDR) products for both Tx and Rx functions. Due to the high instantaneous bandwidth of our products, along with tuning ranges that vary between DC to over 18 GHz, our hardware architecture supports various mechanisms to support frequency tuning. The specific mechanisms used for tuning into various frequencies for various device configurations will be discussed, along with expected results.

Note

The diagrams for Rx and Tx mechanics are not to scale, but are simply used to illustrate how Per Vices shifts signals (i.e. up-converts and down-converts signals) within the bandwidths of different components of the radio chain.

This application note has been broken down into separate sections for baseband, midband, and highband tuning for both Tx and Rx. The specific frequency range associated with a given tuning methodology is product dependent; we’ve aimed to provide specific ranges here. Though we’ve provided rough operational bounds, please refer to the device product page for authoritative information.

1.Overview

A signal may be shifted (i.e. downconverted/upconverted), either using a mixer, the convertor, or within the FPGA. As each element operates independently of each other, it is possible to shift the frequency at a number of different places, within both the analog and digital domains. This application note discusses the specific tuning mechanisms, along with the relevant frequency mechanics, for each mode of operation (baseband, midband and high-band) for both Tx and Rx.

We use the term Tx mechanics to describe the process by which a low frequency signal is shifted up from baseband, i.e. to an intermediate frequency, that is appropriate for conversion using a digital-to-analog (DAC) convertor, before being transmitted via power amplifier and antenna. In this chain, a specific signal path for transmission is selected based on the frequency of the signal (different signal paths are optimized for different frequency ranges). Within each radio chain, the frequency may be mixed by analog or digital mixers, effectively shifting the frequency up to what will be transmitted.

Conversely, we use the term Rx mechanics to describe the process by which a high frequency signal is shifted down from baseband, i.e.to an intermediate frequency, that is appropriate for sampling by a analog-to-digital convertor (ADC), as well as the subsequent frequency translations that may occur once the signal is digitized. Referring to Figure 1.1, we see a generalized overview of the frequency mechanics process.

Figure 1.1: Frequency Mechanics Process Overview

2. Baseband Operation

The baseband Tx and Rx chains connect the external SMA input or output to/from the FPGA and converter device through a series RFE amplifiers, filters, mixers, switches, and anti-aliasing or anti-imaging filters. The baseband chain is automatically enabled by configuring the SDR to frequencies below those supported by the analog up/down convertor, and within the range shown in Table 1a. The baseband frequency chains do not include any analog mixing stages.

Table 1: Baseband Frequency Bands
Device Min Frequency Max Frequency
Crimson DC 120 MHz
Cyan DC 500 MHz (may be extended to 800 MHz)

2.1. Baseband Transmit (Tx) Operation

During baseband Tx tuning, the external data port is connected to the FPGA where re-sampling and digital up conversion occur on the FPGA with a CORDIC mixer. There are no analog up- or down- conversion, or mixing stage facilities used when operating in baseband Tx mode.

Baseband Tx Tuning

During Tx operation, data originates from the host user application, passes through the FPGA-based interpolation routine, FPGA-based digital up conversion (using an numerically controlled oscillator (NCO)), before being converted into JESD204B serial data and sent to the DAC. Any user applied frequency shifting occurs after interpolation, prior to sending the JESD serial data to the DAC, as shown in Figure 2.1.1 This also shows the baseband radio front end (RFE) configuration, where the DAC converts the JESD204B data into an analog signal, and passes the signal through anti-imaging filters, before reaching the SMA port which can be connected to a power amplifier and antenna to transmit the signal.

Figure 2.1.1: Baseband Transmission (Tx) Signal Flow

Baseband Tx Mechanics

As the signals progress through the various components/stages of the SDR, the signal’s frequencies and bandwidths that we are dealing with change. Now that we have a good understanding of our circuitry, let’s look at what is happening to the frequency of the signals at each of the SDR stages.

Figure 2.1.2: Baseband Tx Frequency Mechanics

Figure 2.1.2. shows what needs to takes place in our radio to enable the transmission of an analog signal:

  1. Data Port Generated Samples: the left side of the image shows three waveforms (A, B and C) that we might be looking to transmit. Prior to the samples getting generated, the user defines a sample rate, which we will call user bandwidth (UBW). The sample rate serves to specify the user bandwidth; an interval [-UBW/2 , UBW/2] which is centred around 0 Hz. Since these waveforms will be offset by the NCO frequency at a later stage, the initial sine waves may (in some cases) have a negative frequency (like signal A in the diagram). Once generated, the samples will be sent to the SDR for further processing. However, it is important to note that sometimes not all of the samples in the user bandwidth will get transmitted (this will become more clear at later stages).
  2. Interpolation: after we have generated the user samples, the next step is to perform interpolation used to obtain a larger bandwidth. This new bandwidth specifies a larger interval (also centred around 0 Hz) that is defined by the sample rate of the device (325 MSPS for Crimson TNG, 1 GSPS for Cyan). The user bandwidth is always smaller than the conversion bandwidth. Interpolating the samples to a larger bandwidth is crucial for the next stage where the digital up conversion takes place.
  3. Digital Mixing(CORDIC): after interpolating the signal to the conversion bandwidth of the device, the FPGA can proceed to upconvert the samples. Both Crimson TNG and Cyan have CORDIC digital mixers that are capable of both up-conversion and down-conversion (DDC, DUC). Up-conversion is accomplished by mixing the Tx samples with a local oscillator found in the FPGA (set to what is referred to as the NCO frequency). This causes the frequency of all our signals to increase by the NCO frequency. Using the larger conversion bandwidth that we obtained from interpolation ensures that we can capture more of our mixing products. The signals are then serialized and framed via a JESD204B serial link.

    Note

    In some cases, mixing our generated signal with the NCO frequency results in a frequency that does not lie within the user bandwidth. Here the mixing product will still have an image that is rotated, but outside the analog transmit bandwidth: for baseband signals, we discard the negative frequency component (C), so this image isn’t transmitted at the SMA.

  4. Digital-to-Analog Converter (DAC): the DAC then converts the serial data to its analog form. The production of images is a fundamental and expected outcome of the Nyquist theorem and so we will still have images of our original signal at each Nyquist zone (i.e. at each multiple of our conversion bandwidth, we are likely to see images of the signal at their corresponding offsets).

  5. Anti-Imaging filters: these are used to suppress the images that would typically show up at higher Nyquist zones (multiples of the conversion bandwidth). It has a bandwidth that is ~80% of the DAC bandwidth.
  6. RF Gain: the final analog signal now has gain added to it.
  7. SMA Edge Launch: the gained analog signal can now be transmitted via power amplifier and antenna.

2.2. Baseband Receive (Rx) Operation

During baseband Rx tuning, the external SMA port is connected to the RFE through a series of amplifiers, filters and switches before reaching the ADC. In addition, re-sampling and digital down conversion occur on the FPGA using an NCO (CORDIC mixer). There are no analog down conversion facilities used when operating in baseband mode. The ADC converts the analog signal to digital, which is then sent to an FPGA which provides framing of the data before being sent over the data port to the host user application.

Baseband Rx Tuning

During Rx operation, a signal originates from the SMA, passes through amplifiers, then anti-aliasing filters, followed by digitization by the ADC into digital JESD204B serial data. This digital signal may then be shifted on either the ADC or FPGA, where is it parsed and passed to the host application over the Ethernet ports. Any user applied frequency shifting occurs after digitization, but prior to decimation, as shown in Figure 2.2.1.

Figure 2.2.1: Baseband Receiver (Rx) Signal Flow

For reception (SDR to user application), the receiver data flow is a mirror of the transmit data-flow we previously discussed. The analog signal incident to the SMA is amplified through a low noise amplifier (LNA), passes through an anti-aliasing filter, and is sampled at the ADC input channel. The digital signal is then passed to the FPGA, where it undergoes digital down conversion using an NCO (CORDIC), and then decimation, prior to being sent to the user application. During receive operation, frequency shifting occurs only prior to decimation.

Baseband Rx Mechanics

Figure 2.2.2: Baseband Rx Frequency Mechanics

Figure 2.2.2 shows what needs to takes place in our radio to receive an analog signal:

  1. RF Input/Gain: the first part of the image (see left) shows several pure sine waves (A-E) as they are picked up by the antenna and pass through initial gain amplifier.
  2. Lowpass Filter: a low pass filter cuts off any frequencies above a threshold (product dependent).
  3. Anti-aliasing filter: an analog anti-aliasing filter aims to restrict the incoming signal to only those that fall in the converter’s bandwidth (an AAF is ~80% of the converter bandwidth). This is important because the ADC’s capture bandwidth is limited by the fixed sampling rate; signals outside the capture bandwith would otherwise be aliased back into the digitized signal, which would be undesirable. The converter bandwidth specifies a large interval centred around 0 Hz that is defined by the sample rate of the device (325 MSPS for Crimson TNG, 1 GSPS for Cyan). We will shrink our reception interval in later stages so that we only capture the signals we are “tuned” to.
  4. Analog-to-digital converter (ADC): this converts the incoming signals into a digital form.
  5. Digital Mixing (CORDIC): at this point, the converter BW is large. To prepare for decimation, the samples are digitally down-converted. This decreases the frequency of all the signals from the CORDIC mixer on the FPGA. Both Crimson TNG and Cyan have CORDIC digital mixers that are capable of both up-conversion and down-conversion (DDC, DUC). Down-conversion is accomplished by mixing the Rx samples with a local oscillator found in the FPGA (set to what is referred to as the NCO frequency). Note that after this takes place, some of the frequencies might be negative.
  6. Decimation: prior to the samples being received, the user defines a sample rate (call it user bandwidth). The sample rate specifies this user bandwidth (UBW), an interval [-UBW/2 , UBW/2] which is centred around 0 Hz. Decimation ensures that all the incoming signals fall within the user bandwidth.
  7. Application: the sampled data can now be sent over Ethernet as frames and used for a specific application.

3. Midband Operation

The midband Tx and Rx chain connects the external SMA input or output to/from the FPGA and converter device through the RFE containing amplifiers, filters, mixers, switches, and anti-aliasing or anti-imaging filters. The midband chain is automatically enabled by configuring the SDR frequencies within the midband range, shown in Table 2 below.

Table 2: Midband Frequency Bands
Device Min Frequency Max Frequency
Crimson 120 MHz 6000 MHz
Cyan 500 MHz 6000 MHz

When compared to the baseband signal flow, there are two notable differences:

  1. The converter devices operate in dual channel mode to support complex (IQ) operation. Unlike the baseband case, this avoids discarding negative frequency (or the imaginary component) from samples.
  2. The RFE includes a complex, analog, mixing stage.

3.1. Midband Transmit (Tx) Operation

Midband Tx Tuning

During midband Tx operation, data originates from the host user application, passes through the FPGA-based interpolation routine, FPGA-based digital up conversion (using an NCO), before being converted into JESD serial data and sent to the DAC. Any user applied frequency shifting occurs after interpolation, prior to sending the JESD204B serial data to the DAC, as shown in Figure 3.1.1. This figure also shows the baseband radio front end (RFE) configuration, where the DAC converts the JESD204B data into an analog signal, and passes the signal through anti-imaging filters, IQ upconverter/complex mixer, before reaching the SMA port which can be connected to a power amplifier and antenna to transmit the signal.

Figure 3.1.1: Midband Transmission (Tx) Signal Flow

Midband Tx Mechanics

Figure 3.1.2 shows what needs to takes place in our radio to enable the transmission of an analog signal in midband. Note how negative frequencies are preserved after the interpolation/complex digital mixing.

Figure 3.1.2: Midband Tx Frequency Mechanics

  1. Data Port Generated Samples: the left side of the image shows three waveforms (A,B and C) that we might be looking to transmit. Prior to the samples getting generated, the user defines a sample rate, which we will call user bandwidth (UBW). The sample rate serves to specify the user bandwidth; an interval [-UBW/2 , UBW/2] which is centred around 0 Hz. Since these waveform will be offset by the NCO frequency at a later stage, the initial sine waves may (in some cases) have a negative frequency (like signal A in the diagram). Once generated, the samples will be sent to the SDR for further processing. However it is important to note that sometimes not all of the samples in the user bandwidth will get transmitted (this will become more clear at later stages).
  2. Interpolation: after we have generated the user samples, the next step is to perform interpolation to obtain a larger bandwidth. This new bandwidth specifies a larger interval (also centred around 0 Hz) that is defined by the sample rate of the device (325 MSPS for Crimson TNG, 1 GSPS for Cyan). The user bandwidth is always smaller than the conversion bandwidth. Interpolating the samples to a larger bandwidth is crucial for the next stage where the digital up-conversion takes place.
  3. Digital Mixing(CORDIC): after interpolating the signal to the conversion bandwidth of the device, the FPGA can proceed to upconvert the samples. Both Crimson TNG and Cyan have CORDIC digital mixers that are capable of both up-conversion and down-conversion (DDC, DUC). Up-conversion is accomplished by mixing the Tx samples with a local oscillator found in the FPGA (set to what is referred to as the NCO frequency). This causes all the frequency of all our signals to increase. Using the larger conversion bandwidth that we obtained from interpolation ensures that we can capture more of our mixing products. (The squiggley line on the frequency axis indicates that the signals (A,B and C) are actually shifted to a much higher bandwidth, and the diagram is for illustrative purposes only).

    Note

    In some cases, mixing our generated signal with the NCO frequency results in a frequency that does not lie within the user bandwidth. Here the mixing product will still have an image that is rotated to fit within our capture bandwidth.

  4. Digital-to-analog Converter (DAC): the DAC then converts the signals to their analog form. The production of images is a fundamental and expected outcome of the Nyquist theorem and so we will still have images of our original signal at each Nyquist zone (i.e. at each multiple of our conversion bandwidth, we are likely to see images of the signal at their corresponding offsets).

  5. Anti-Imaging filters: these are used to suppress the images that would typically show up at higher nyquist zones (multiples of the conversion bandwidth). It has a bandwidth that is ~80% of the DAC bandwidth.
  6. Analog (IQ) Upconverter/Mixer: this mixer shifts the signals to a higher frequency (upconverts the signal with a frequency synthesizer that generates an LO) at a much greater bandwidth than the converter, and is complex.
  7. RF Gain: the final analog signal has gain added to it. (The squiggley line on the frequency axis indicates that the signals (A,B and C) are actually shifted to a much higher bandwidth, and the diagram is for illustrative purposes only).
  8. SMA Edge Launch: the gained analog signal can now be transmitted via power amplifier and antenna.

3.2. Midband Recieve (Rx) Operation

During midband Rx tuning, the external SMA port is connected to the RFE through a series of amplifiers, filters and switches before reaching the ADC. In addition, re-sampling and digital down conversion occur on the FPGA using an CORDIC mixer. There is further analog down conversion facilities used when operating in midband mode. The ADC converts the analog signal to digital, which is then sent to an FPGA which provides framing of the data before being sent over the data port to the host user application.

Midband Rx Tuning

During midband Rx operation, a signal originates from the SMA, goes through a low-pass filter, before passing through amplifiers, then anti-aliasing filters, followed by digitization by the ADC into digital IQ pairs and transmitted using JESD204B serial data protocol. This digital signal may then be shifted on either the ADC or FPGA, where is it parsed and passed to the host application over the Ethernet ports. Any user applied frequency shifting occurs after digitization, but prior to decimation, as shown in Figure 3.2.1.

Figure 3.2.1: Midband Receiver (Rx) Signal Flow

Midband Rx Mechanics

Figure 3.2.2: Midband Rx Frequency Mechanics

Figure 3.2.2 shows what needs to takes place in our radio to receive an analog signal.

  1. RF Input/Gain: the first part of the image (see left) shows several pure sine waves (A-E) as they are picked up by the antenna and passing through an initial gain amplifier.
  2. Low-pass Filter: the LPF then cuts off and attenuates all signals above 6GHz.
  3. Analog RF Mixer: the signal is then further downconverted using a frequency synthesizer (generates an LO). (The squiggley line on the frequency axis indicates that the signals (A,B and C) are actually shifted to a much lower bandwidth, and the diagram is for illustrative purposes only).
  4. Anti-aliasing filter: an analog anti-aliasing filter aims to restrict the incoming signals to only those that fall in the converter’s domain (AAF bandwidth is ~80% that of the ADCs). This is important because the ADC has a finite operating range. The converter bandwidth specifies a large interval centred around 0 Hz that is defined by the sample rate of the device (325 MSPS for Crimson TNG, 1 GSPS for Cyan). We will shrink our reception interval in later stages so that we only capture the signals we are “tuned” to.
  5. Analog-to-Digital Converter (ADC): this converts the incoming signals into a digital form.
  6. Digital Mixing (CORDIC): at this point, the converter BW is large. To prepare for decimation, the samples are digitally down-converted. This decreases the frequency of all the signals by the NCO frequency set on the FPGA. Both Crimson TNG and Cyan have CORDIC digital mixers that are capable of both up-conversion and down-conversion (DDC, DUC). Down-conversion is accomplished by mixing the Rx samples with a local oscillator found in the FPGA (set to what is referred to as the NCO frequency). Note that after this takes place, some of the frequencies might be negative. (The squiggley line on the frequency axis indicates that the signals (A,B and C) are actually shifted to a much lower bandwidth, and the diagram is for illustrative purposes only).
  7. Decimation: prior to the samples being received, the user defines a sample rate (call it user bandwidth). The sample rate specifies this user bandwidth (UBW), an interval [-UBW/2 , UBW/2] which is centred around 0 Hz. Decimation ensures that all the incoming signals fall within the user bandwidth.
  8. Application: the sampled data can now be used for a specific application.

4. Highband Operation

To transmit and receive high frequency waves, we need to make use of an additional local oscillator and mixer on the LOgen board. This allows for even higher frequency operation, since our architectures implement a superheterodyne architecture, as discussed on this wikipedia page. This effectively means that the analog frequency conversion is performed in two stages, using a complex base-band stage that is mixed, using an IQ convertor, to a known, real valued IF stage, which is subsequently converted to a final RF stage. The IF stage is implemented using the midband mechanics, as previously discussed in midband mechanics.

The highband Tx and Rx chains connects the external SMA input or output to/from the FPGA and converter device through the RFE containing amplifiers, filters, mixers, switches, and anti-aliasing or anti-imaging filters. The highband chain is automatically enabled by configuring the SDR frequencies within the midband range, shown in Table 3 below.

Table 3: Highband Frequency Bands
Device Min Frequency Max Frequency
Cyan 6000 MHz 18000 MHz

4.1. High-band Transmit Operation

Highband Tx Tuning

Figure 4.1.1: Highband Transmission (Tx) Signal Flow

Highband Tx Mechanics

As the samples progress through the various components in the SDR, the frequencies and bandwidths that we are dealing with change. Now that we have a good understanding of our circuitry, let’s look at what is happening to the frequency at each of these steps, as shown in the figure below:

Figure 4.1.2: Tx Highband Mechanics

  1. Data Port Generated Samples: the left side of the figure shows three waveforms (A,B and C) from the data port which we want to transmit. Prior to the samples getting generated, the user defines a sample rate, which we will call user bandwidth (UBW). The sample rate serves to specify the user bandwidth; an interval [-UBW/2 , UBW/2] which is centred around 0 Hz. Once the samples are generated, they are sent to the SDR for further processing.
  2. Interpolation: after we have generated the user samples, the next step is to perform interpolation used to obtain a larger bandwidth. This new bandwidth specifies a larger interval (also centred around 0 Hz) that is defined by the sample rate of the device (325 MSPS for Crimson TNG, 1 GSPS for Cyan). The user bandwidth is always smaller than the conversion bandwidth. Interpolating the samples to a larger bandwidth is crucial for the next stage where the digital upconversion takes place.
  3. CORDIC Mixing: after interpolating the signal to the conversion bandwidth of the device, the FPGA can proceed to upconvert the samples. Both Crimson TNG and Cyan have CORDIC digital mixers that are capable of both up-conversion and down-conversion (DDC, DUC). Up-conversion is accomplished by mixing the Tx samples with a local oscillator found in the FPGA (set to what is referred to as the NCO frequency). This causes the frequency of all our signals to increase. Using the larger conversion bandwidth that we obtained from interpolation ensures that we can capture more of our mixing products. This also produces image signals via image wrap-around (signal C). The production of images is a fundamental and expected outcome of the Nyquist theorem and so we will still have images of our original signal at each Nyquist zone (i.e. at each multiple of our conversion bandwidth, we are likely to see images of the signal at their corresponding offsets). (The squiggley line on the frequency axis indicates that the signals (A,B and C) are actually shifted to a much higher bandwidth, and the diagram is for illustrative purposes only).
  4. Digital-Analog-Converter: the DAC then converts this digital signal to analog, which attenuates the image signals significantly.
  5. Anti-Aliasing Filter: an AAF is then able to get rid of the image signals, while also decreasing the bandwidth to 80% of the DAC bandwidth.
  6. Complex (IQ) Upconverter/Mixer 1: an analog upconverter/mixer is then used to further increase the frequency of each signal, using a midband frequency LO (IF). (The squiggley line on the frequency axis indicates that the signals (A and B) are actually shifted to a much higher bandwidth, and the diagram is for illustrative purposes only).
  7. Band-pass Filter: this signal then goes through a BPF, which only permits signals from 1.4 - 2.3 GHz.
  8. Passive (Real) Upconverter/Mixer 2: this further upconverts the signal using a highband frequency LO (and is a real valued mixer), and then through a HP filter which rejects signals below 6GHz. (The squiggley line on the frequency axis indicates that the signals (A and B) are actually shifted to a much higher bandwidth, and the diagram is for illustrative purposes only).
  9. RF Gain: the final signals which will be transmitted are amplified. (The squiggley lines are to indicate that the signals (A and B) are actually at a much higher bandwidth, and the diagram is for illustrative purposes only).
  10. SMA edgelaunch: the gained final analog signal can now be transmitted via power amplifier and antenna.

4.2. Highband Receive Operation

Highband Rx Tuning

Figure 4c: Highband Receiver (Rx) Signal Flow

Highband Rx Mechanics

Figure 4d: Rx Highband Frequency Mechanics

The figure above shows what needs to take place in our radio in order to receive an analog signal.

  1. RF Input/Gain: the first part of the figure shows several pure sine waves (A-F) and their images being picked up by an antenna attached the SMA jack and then amplified.
  2. High Pass Filter: in this stage, the signals pass through a HPF which attenuates all signals below 6GHz.
  3. RF (Real/Passive) Downconverter/Mixer: next the signals reach a mixer (real) which performs frequency down conversion using an high frequency (HF) LO, as shown on the figure above. Note how signal B reflects from the horizontal (0Hz) axis due to this being a real valued mixer.
  4. Band Pass Filter: the down converted signals then pass through a BPF, which permits frequencies between 1.4 - 2.3 GHz while attenuating all other frequencies outside this band.
  5. Complex (IQ) Downconverter/Mixer: these signals are further downconverted by a mixer (complex), using the IF LO, which then separate the signal into IQ pairs, before each component being filtered further and then amplified.
  6. Anti-Aliasing Filter: signals then reach an AAF which are applied to both the I and Q signal, while attenuating signals outside the AAF bandwidth, which is 80% that of the converter bandwidth.
  7. Analog-Digital Converter: this converts the signals from analog to digital domain, where the signals are now able to be manipulated on the FPGA.
  8. Digital Mixing (CORDIC): this decreases the frequency of all the signals by the NCO frequency set on the FPGA. Both Crimson TNG and Cyan have CORDIC digital mixers that are capable of both up-conversion and down-conversion (DDC, DUC). Down-conversion is accomplished by mixing the Rx samples with a local oscillator found in the FPGA (set to what is referred to as the NCO frequency). Note that after this takes place, some of the frequencies might be negative (signals D and C). Signals that are rotated beyond the convertor boundary, wrap around on the otherside of the bandwidth (signals C and D). This operation does not lose any information, it simply rotates the signal in the frequency domain. (The squiggley line on the frequency axis indicates that the signals (C,D, and E) are actually shifted to a much lower bandwidth, and the diagram is for illustrative purposes only).
  9. Decimation: this further downsamples the signals and dictates the final user bandwidth before proceeding to the host computer/application.
  10. Application: the sampled data can now be sent over Ethernet as frames and used for a specific application.

5. Basic Defintions and Worked Examples

Below we discuss how one would use the frequency tuning mechanics discussed above in order to transmit or receive a signal. This is done using the mixers, both real/complex analog and complex (CORDIC) digital mixers,as well as decimation. First, a few basic definitions:

  1. Downconversion is the process of mixing an RF input and LO to obtain an output IF lower than the RF input:

    f_{IF}= \mid f_{LO} -f_{RF} \mid
  2. Upconversion is the process of mixing an LO such that RF output is larger than the RF input:

    f_{RF_1}= f_{LO} - f_{IF}
    f_{RF_2}= f_{LO} + f_{IF}

    Figure 5a: Downconversion and Upconversion

  3. The total offset that we wish to shift a signal (i.e. upconvert or downconvert) is defined as follows for each band:

    Baseband: $$ f_{offset} = f_{NCO, FPGA} $$ Midband: $$ f_{offset} = f_{LO,MB} + f_{NCO} $$ Highband: $$ f_{offset} = f_{LO,HF} + f_{LO,IF} + f_{NCO, FPGA} $$

  4. Decimation refers to the process of reducing the sampling rate (i.e. downsampling) by essentially throwing away samples. The decimation factor (M) is defined as the ratio of the input rate to output rate. In equation form, we have:

    M = \frac {input (SPS)}{output (SPS)}
  5. Interpolation refers to the process of increasing the sampling rate (i.e.upsampling) by essentially inserting zero valued samples. The interpolation factor (L) is defined as the ratio of the output rate to input rate. In equation form, we have:

    L = \frac {output (SPS)}{input (SPS)}

5.1 Rx Tuning

Note

Multiple solutions exist for these examples.

Baseband Rx Example: Suppose we are in baseband mode. There’s a low-pass filter which cuts anything above 120MHz and an ADC with a sample rate of 200MSPS. We want to capture a center frequency of 79 MHz within a 700KHz band at a sample rate of 4MSPS. How much should the f_{NCO, FPGA} downconvert our signal to be centred in the 400KHz band? What is the decimation factor?

Solution: We know that since we are looking for a signal at 79 MHz, it will not be cut off by the LPF. In order to achieve a signal to be centred in a 700KHz bandwidth, we will need to enable the f_{NCO, FPGA} (since this is the only mixing stage in baseband) at the following frequency:

f_{offset}= 79 MHz -400 KHz =78.6 MHz =f_{NCO,FPGA}

Therefore we need to downconvert the signal by 78.6 MHz.

Our decimation factor is simply: $$ M= \frac {200 MSPS}{4 MSPS}=50 $$

5.2 Tx Tuning

Midband Tx Example: Suppose we have a user generated signal of 1000MHz with a sample rate of 32.5MSPS and we wish to transmit this signal at a center frequency of 5025MHz frequency centered at 0. Our f_{LO, MB}=N*100 MHz, where N is a whole number and the DAC has a sample rate of 325MSPS. What should be our interpolation factor, value for N, and f_{NCO,FPGA}?

Solution: First off, in order for our DAC to work properly, we will need to upconvert the signals sample rate using the interpolation factor: $$ L=\frac{325MSPS}{32.5MSPS}=10 $$

Second, our total offset value needed for upconversion to 5025 MHz is:

f_{offset} = 5025 MHz - 1000 MHz = 4025 MHz = f_{LO,MB} + f_{NCO,FPGA}

Therefore, one possible solution is to upconvert the signal using the f_{LO,MB} using a value of N=40 and an f_{FPGA,NCO}=25MHz. This gives us the the 5025MHz center frequency.