Tuesday, February 03, 2009

RFFE1 Why you may (or may not) need it

RFFE stands for "Radio Frequency Front End" and the "1" stands for the first version (0-62.5 MHz coverage).

When I was designing the QS1R board, I had to decide whether to include bandpass filtering and RF amplification on the board. In fact the initial prototypes "RevA" included RF amplification on the QS1R board. Unlike another DDC based direct sampling receiver "Perseus", QS1R was designed to be more than a SW receiver. In addition to a SW receiver, QS1R was meant to facilitate experimentation in the RF spectrum up to at least 300 MHz. I finally settled on a 55 MHz low pass filter (which can be bypassed) and no active components in front of the ADC on QS1R. Any active devices, bandpass filters, or attenuation would be added by a separate board such as the RFFE1.

The antenna that I use for my QS1R is a center fed, non-resonant dipole at 50 feet. The total wire length is about 240 feet with only about 100 feet of that running horizontally. It is fed with 450 ohm ladder line connected in the house to a 4:1 balun. The last 10 feet of feed line is coax to the QS1R from the balun. I do not use a tuner with this antenna for receiving. In my neighborhood, all of the utilities are under ground so the noise level is typical or maybe a bit less than the noise level of a suburban area. With this antenna and the QS1R, I find that band noise is always greater than the QS1R's noise floor until about 21 MHz. I arbitrarily have assumed (possibly incorrectly) that my antenna installation is average. Any amplification in front of the QS1R's ADC (below 21 MHz) is only a detriment in this case. The QS1R typically overloads at +9 to +10 dBm so amplification would degrade the strong signal handling capability of the receiver. The largest signal level that I have observed at my location has been about 0 dBm in the 0 - 62.5 MHz region. Contrary to what some may believe in the above case for the QS1R, bandpass filtering will improve nothing and, because BPFs typically have some loss, degrade the sensitivity slightly. As long as the atmospheric or band noise is greater than the noise floor and the signals levels in the 0 - 62.5 MHz region is below +9 dBm, bandpass filtering and/or amplification does not gain anything.

By providing RF front end add-on boards such as the RFFE1, a front end (if even desired) can be tailored to the specific application of QS1R and will not impose a significant cost increase to the QS1R board for those who do not need the RFFE(1) boards. The QS1R is capable of under sampling up to ~ 500 MHz, so it will be possible to design RFFE(x) boards that will allow coverage up to ~ 500 MHz. For example, RFFE2 would provide low noise amplification, bandpass filtering, and attenuation of the range of 125 to 187.5 MHz ( including 2 meter ham band coverage). There is some interest by individuals in FM band Dxing which could be provided by a RFFE(x) board.

The RFFE1 board is designed to fit within the existing QS1R Aluminum enclosure by replacing the front and rear end plates with new end plates (supplied with RFFE1). RFFE1 has switchable attenuation, two stages of switchable low noise amplification, and switchable filtering. It is controlled via the QS1R's I2C bus and derives its DC power from QS1R. The RF amplifiers provide up to 26 dB of gain (noise figure ~ 2 dB). With the attenuation switched in, the QS1R can handle signal levels up to +29 to +30 dBm without overload of the ADC. Each RF amplifier provides ~ 13 dB gain. With both amplifiers switched in and attenuation switched off, the QS1R can handle signal levels up to -17 dBm (or -4 dBm with one RF amp switched in) without overloading the ADC. The QS1R's IP3 is > +50 dBm so the RF amplification was designed to not degrade the IP3 excessively. With the amplifiers switched in, the IP3 is ~ +32 to + 33 dBm. Contrary to what some have speculated elsewhere, the RFFE1 does not use the LTC6400-20 ADC driver that HPSDR Mercury uses for a preamp. In HPSDR Mercury, the 20 dB amplification provided by the LTC6400-20 is always present - the preamp is bypassed by switching 20 dB of attenuation in front of the LTC6400 ADC driver providing a net gain of 0 dB. In the RFFE1, the two RF amplifiers are individually switchable in and out of line and have their own impedance matching and filtering networks for enhanced performance.

With any amplification in front of the QS1R's ADC and without filtering, the whole 0 - 62.5 MHz region is amplified equally - so any signal within this range could possibly overload the ADC if its signal strength amplified by the gain of the RF amp exceeds +9 to +10 dBm. This is where band pass filtering is an advantage. It is not easy to design the filtering so that adequate attenuation is provided out of band as well as not degrading the IP3 performance of the receiver. Luckily we are not dealing with an analog radio, so the assumptions about exactly what is needed in a conventional receiver does not necessarily apply to direct sampling receivers such as the QS1R. I spent a very large amount of time in experimentation and testing to come up with a filtering system for RFFE1 that does not degrade IP3 performance. I was surprised in what I found and eventually decided on, but that is a subject for a future blog entry ;-). I did a lot of modeling of various filter networks to determine the RF current in each inductor given the physical and electrical characteristics of each type of inductor. This helped me chose the best filter arrangement and inductor type for the RFFE1 without degrading IP3 as much as other competing receiver's bandpass filters degrade IP3.

So who needs a RFFE1? Here are my guesses:

1. Anyone who wants to use QS1R above ~ 18 MHz with an average antenna system. This includes 10 and 6 meters.

2. If your antenna is less efficient and you have determined that the noise floor of QS1R is above the band noise at some point below 18 MHz. An easy way to determine this is by disconnecting your antenna while watching the noise floor or s-meter in SDRMAX(II/III). If the noise floor increases by at least 6 dB when you reconnect your antenna then you have adequate sensitivity at that frequency.

3. If your antenna does not provide a 25 - 75 ohm impedance to QS1R. Antennas that deviate far from 50 ohms will manifest as severely reduced sensitivity with QS1R. In my case with my non-resonant 450 ohm ladder line fed dipole, a 4:1 balun is all that is required for me to have adequate sensitivity up to about 21 MHz.

4. Signal strengths in your location in the 0 - 62.5 MHz range exceeds +9 dBm at the antenna input to the QS1R receiver. In this case the switchable filtering and/or attenuation provided by RFFE1 will eliminate the overload. QS1R and SDRMAXII provides the nice capability of viewing 50 MHz of spectrum at once - it is easy to see the maximum signal levels your are receiving in that range.

5. RFFE1 has two switchable antenna inputs as well as additional static/surge protection. RFFE1 also provides a connector for external receiver mute input, two protected general purpose IO lines, as well as an external connection to the QS1R I2C bus for external relay control and/or switching. While these options are not necessary for the functioning of QS1R, they are nice to have in some cases.

Who does not need a RFFE1?

1. If your operation is below ~18 MHz and your antenna system is adequate (see 2 and 3 above) and you have no overload problems (you don't see your "clip" LED on the front of QS1R illuminating) then you won't gain much with RFFE1. This generally includes using QS1R with transverters and for an IF receiver with other receivers/transceivers.

Exact performance numbers for RFFE1 will be posted in the near future as well as pricing. Hint: The price of QS1R and RFFE1 together will not exceed the price of competing direct sampling receivers!