Here is a brief update on the noise figure discussion …
Earlier in the year I corresponded with Elonics about the noise figure. They told me that although the best case noise figure is under 4dB, the worst case noise figure is significantly higher. Further inquiries elicited the information that the maximum noise figure is around 14dB and occurs in the VHF range. This explains completely why there is 20dB of relatively low noise gain in front of the Elonics tuner chip …
For the purposes of this discussion it will be sufficient to treat noise figure as a relative figure of merit. The conventional wisdom concerning noise figures can then be summarized as follows:
Noise figure is expressed in dB. Smaller numbers correspond to lower noise and a greater ability to detect weak signals. It is not possible to have a noise figure less than 0 dB. In the context of amplifier noise figures, 8 dB is extremely poor, 5 dB is poor, 3 dB is reasonable, 2 dB is good, 1.5 dB is very good and anything below 1 dB is truly excellent. Mixer stages are generally noisier than amplifier stages. Each stage of a receiver contributes to the overall noise figure of a receiver, but the relative contribution of each stage is reduced by the cumulative gain of the preceding stages.
Two consequences of this conventional wisdom are as follows. If the first stage of amplification has a reasonably high gain, it will effectively determine the overall noise figure of a receiver; and the use of a stage of low noise amplification before a noisy mixer will greatly reduce the impact of mixer noise on the overall noise figure. Both considerations support the use of a stage of low noise amplification with reasonably high gain prior to the first mixer. How much low noise gain is needed? Most engineers would say somewhere between 10dB and 20dB.
With this background, consider the FCD Pro. The LNA of the E4000 has a nominal noise figure of <4dB. This is acceptable for consumer TV applications. Depending on the overall configuration of the receiving station (including the antenna, and perhaps a preamplifier) it may also be acceptable in the context of the Fun Cube satellite project. However, when the FCD Pro marketed as having general broad-band capabilities there are applications, including ham radio applications, for which a noise figure of ~4dB is unacceptably high.
The designers of the FCD Pro seem to have responded to this noise figure problem by using a low cost Si MMIC to provide lower noise amplification in front of the E4000. The BGA2717 MMIC has a nominal noise figure of 2.3 dB at a frequency of 1 GHz. This is acceptable. It is significantly better than ~4dB, and people who need a noise figure below 2 dB would probably expect to use a special-purpose preamplifier that uses a more advanced (and more expensive) technology, such as pHEMT technology.
The noise figure trap is that lowering the noise figure from ~4dB to ~2.3 dB increases the gain in front of the mixer by up to 24 dB. This results in the default gain of the FCD Pro from antenna to mixer output reaching close to 60dB (with up to +24dB from the BGA2717 amplifier, + 20 dB from the E4000 LNA and + 15 dB from the mixer.) The gain can be brought back down towards 20 dB if the gain of the E4000′s LNA and mixers are set at their minimum values – but lowering the gain seems counter-intuitive to casual users. Faced with a seemingly deaf receiver, their instinct will be to increase the gain of the E4000′s LNA to 30 dB, not to reduce it to -5dB!
There can be situations in which the uncontrolled gain of the BGA2717 alone is high enough to cause problems. In this case it may be helpful to put an attenuator in front of the FCD Pro. Other things being equal, overload problems get worse when a preamplifier is used in front of the FCD Pro. With another 20 dB of gain from a preamplifier, the total gain between the antenna and the mixer can be around 80 db. This corresponds to a voltage gain of around 10,000, which is much too high for a wideband system with puny little mixers.
The heart of the FCD Pro is the E4000 low power multi-band tuner IC. This IC was designed by a fabless IC design company called Elonics. News releases indicate that the E4000 is manufactured in a modern 90nm mixed-signal CMOS process offered by IBM. Appendix A of An Introduction to the Funcube Dongle provides a basic information sheet for the E4000. Information sheets for the E4000 and other products can also be accessed at the Elonics website. In order to obtain additional technical information, it is necessary to sign a non-disclosure agreement (NDA) with the company. The discussion in this post is based solely on the content of the information sheet.
The E4000 has a zero-IF architecture. This means that it translates a portion of a received RF spectrum directly to baseband frequencies without using an intermediate frequency. Very simple direct conversion receivers that use a single mixer do not reject image frequencies. A well-known phasing technique solves this problem. The basic idea is to use two mixers fed by oscillator signals that are 90 degrees out phase. The resulting baseband signals are referred to as the I (in-phase) and Q (quadrature) signals. Adding and subtracting the I and Q baseband signals allows separation of the upper and lower sidebands and the rejection of unwanted image frequencies. The extent of the image rejection depends on how well the amplitudes and phases of the I and Q paths track.
The E4000 uses this phasing technique. It generates the required oscillator signals on-chip using a PLL-based frequency synthesizer. The information sheet states that the E4000 works over the entire frequency range between 64 MHz and 1700 MHz. However, according to the technical FAQ page of the FCD Pro, ‘there is a gap in coverage between about 1,100MHz and 1,270MHz where the design of the local oscillator VCO, PLL and divider chain in the tuner chip do not provide seamless coverage.’ The same source reports that the FCD Pro has been found to operate at frequencies as low as 51.5 MHz, and at frequencies that exceed 2 GHz.
I believe that the E4000 is very well suited for low-cost, low-power, medium performance, high-volume, TV tuner applications. It is low cost because it uses silicon technology, and low power because it uses a high performance mixed-signal CMOS technology that uses a low supply voltage of 1.5 V. The ability to perform digital and analog operations on the same chip seems to have been exploited effectively. I like the ability to control the gain of individual stages using a simple data bus. It would be interesting to know more about the topology and the electrical performance of the RF mixers, and the implementation and phase noise characteristics of the PLL-based frequency synthesizer. The mixers may be prone to overload, since a 1.5V supply does not allow much headroom.
What is not to like about the E4000? Not much … but there is always something. The E4000 includes a broadband low-noise amplifier (LNA) whose noise figure is <4dB. This is quite good, considering that it is achieved using silicon CMOS, and is probably adequate for consumer applications. However, it is a long way from being state-of-the-art. I will discuss the implications in the next post.
The FCD Pro is not really a ‘plug and play USB Software Defined Radio.’ OK, it is plug and play with respect to the drivers that are used by the operating system; it uses a USB interface; and it is a software defined radio. However, the user has to locate, download and install some additional programs in order for the FCD Pro to do anything – and at that point its utility may be limited significantly by a high susceptibility to overload. This susceptibility is due to too much broadband RF gain and too little RF filtering. Overloading leads to desensitization, which manifests itself as apparent deafness. It is ironic that having too much gain causes receivers to become insensitive …
The most appropriate solutions for an overload problem depend on the local electromagnetic environment, the specific applications of interest, and the characteristics of the antennas that are being used to receive signals of interest. The gain problem can often be addressed adequately by lowering the gains of the LNA, mixer and baseband amplifier stages and (if this was not done during the initial setup) by lowering the ‘microphone’ level that is set in the host computer. In the most severe cases of overloading an external attenuator may be required. The filtering problem can be addressed using a passive filter or a preamp – but losses associated with passive filters degrade signal to noise ratios, and the additional gain associated with a preamplifier will tend to exacerbate the gain problem. Relatively expert users can achieve excellent results – but it is very easy for less expert users to experience disappointing results.
I view the FCD Pro as an innovative foundation for software defined radio systems. I am perfectly willing to select antennas, adjust gain parameters, and augment RF filtering, in whatever ways are appropriate for applications that I am interested in …
The inherent sensitivity of the unit seems to be very good. After reducing the gain sufficiently to avoid obvious manifestations of overload I can hear my local NOAA station on 162.4 MHz without any antenna, and I can pick up several commercial FM stations using an unwound paper clip as an antenna. FM stations sound distorted, because the receiver bandwidth is a less than the modulation bandwidth, but they provide useful sources of signals when making gain adjustments.
The FCD Pro is a USB dongle that contains the hardware and firmware of an innovative software-defined receiver that works at VHF, UHF and low microwave frequencies. The product created a lot of excitement when it was first announced. The supply was limited to batches of 100 units at a time, and each new batch sold out within minutes.
The supply of units eventually caught up to demand. In August 2011 I purchased a unit out of inventory. Two days later a package arrived at my QTH in Arizona. It contained unit #2748 in an anti-static envelope inside a little plastic box, with a sheet of paper informing me that the unit is ‘a plug and play USB Software Defined Radio (SDR) that is compatible with many operating systems and software.’ The sheet of paper specified a web site from which to download software for setting the center frequency, and advised me to ‘keep an eye on the site for upgraded firmware.’ I immediately recognized a case of missing manual syndrome! Installing and using the FCD Pro was going to require some time and effort …
I was not surprised by the fragmentary nature of the documentation. Talented engineers invariably try to avoid the tedious and time-consuming task of documenting a product in a way that will serve the needs of a broad range of users. In the case of the FCD Pro there is sufficient documentation to meet the needs of motivated experimenters who have an idea of what they are doing – but there is not enough documentation for people who want to plug-and-play. I decided to experiment as soon as I had some time …
Several weeks elapsed before I got around to setting up the unit. I went to the suggested web site, downloaded a few documents and followed the instructions. The process was not difficult – but nor was it entirely problem-free. As I became familiar with the unit I started to see why some people are delighted by its capabilities while others become frustrated with it. The FCD Pro is innovative and provides excellent value for money – but it is not a turn-key appliance. Users need to understand the limitations of the units and how to work around them. This blog offers my perspectives on these topics.
I do not claim to be an expert on the FCD Pro, software defined radio architectures, or other topics that I will be discussing, My goal is to help other people to get started with the FCD Pro and to use it effectively. If the material proves useful, and if I become qualified to do so, I will even write the first draft of the missing user manual. Please use the comments to point out my errors and misconceptions, and to offer alternative perspectives on any of the topics that I cover. Welcome to my journey of discovery!
 | drtune on Hello FCD Pro! |
 | Noise Figure on The Noise Figure Trap |
