Very Low Power Multichannel Telemetry - Part 1
by Michael Sciarra

Background

Today, with higher frequency inexpensive monolithic ICs flying off manufacturers' shelves, it is comparatively easy to make RF links. That is, compared to 5 years ago. For example, if you really get lazy and cost is less important than time to market, Links Systems has a complete RF solution that is FCC type approved when used as supplied.

Bandwidth

Digital encoding schemes generally use a great deal more bandwidth when transmitting. This gets worse when the system is required to have high data rates, or transmit high-frequency sub-carrier information. High-bandwidth digital encoding schemes equate to higher power requirements, to say nothing about gobbling up bandwidth. In some portable equipment, the trade off of a digital modulation scheme vs. an analog one is obvious. If you tell the client that you can give him 30 days operation on one AAA using analog technology, or 6 days employing "digital," most will opt for the analog. I am not saying that an analog approach is the best way for all RF links coming to market. What I am saying is that for specific applications that do not require more than 5 or 6 channels of data on one carrier, and do not require great speed, i.e. 19,000 bps with 10 users on at the same time, an analog scheme could provide a much lower power and cost-effective solution.

Wide-bandwidth systems used to claim that they were almost immune from competing transmitter interference since the carrier was "spread" over such a large area of the spectrum. Typically this is about 3 to 6 MHz in spread-spectrum systems. Frequency-hopping or direct-spread systems all use large amounts of bandwidth. Direct-synthesis systems have the edge over hoppers with regard to the system speed capability, with hoppers typically only able to keep up with about 5 Mbps max. Somewhere I heard of a practical way to overcome this limitation, but that is a research project for a later article.

The immunity claim arose because most transmitters were narrow-band, and therefore the spectral energy disturbance to a wide-band spread system was very small.

There is no free lunch. Spread systems are everywhere—cordless phones, security equipment, wireless LANs—you can actually buy three 2 GHz color cameras for about $150 plus receiver, plus remote, plus encoder etc. (see www.x10.com). Let's not forget about Bluetooth "A" and "B." When that gets going, it is my belief some surprises are in store for the folks subscribing to the immunity syndrome of spread-spectrum digital transmission. I have personally watched my spectrum analyzers noise floor rise about 5 to 10 dB at times, thinking it was out of adjustment only to discover that I was looking close in at around 1 MHz/div and right in the 900 MHz cordless phone band (902–928 MHz). Granted, my lab in Fairfield, CT, is in a fairly dense neighborhood.

The situation is now reversed for us analog guys. With most users now on wide-band modulation schemes, our narrow-band signals can claim to be bullet-proof from CDMA, WCDMA, TDMA, etc., systems. Of course, the receivers must also be narrow-band to take advantage of this feature.

Design Example

Here are some basic guidelines to keep in mind when you design a multiparameter data-transmission system.

1. Establish the frequency bandwidth required by your sensor. For example, if you know that you are going to see a spread of 0 to 1 V out of the sensor, and you want to resolve 1 mV, then it would make sense to have at least a 1000 Hz bandwidth. In reality, it is not so simple because there is the VCO conversion gain, which may NOT be linear, and the maximum sensor rate of change. If the sensor changes at rates higher than about 470 Hz, that 1000 Hz bandwidth may not be enough. The rule is to make the band about 1.2 times larger than the maximum frequency of the signal being resolved. In general, I make it about 1.5 times. This seems to work well, leaving some room at the edges for drift or component changes.
   
   
2. Make sure that there is NO possibility of overlap of your bands. Allowing approximately 500 to 1000 Hz of dead space between your sub-carriers usually precludes sub-carrier interference, provided number "3" below can be achieved.
   
   
3. Put brick walls on your sub-carrier VCOs. This can be done with VCOs that provide band stops. Most VCOs can be made to do this by ascertaining their VCO limits and making sure that your voltage inputs achieve that limit. For example, the 4046 provides this by way of simple resistors at pins 11, and 12—one for the center frequency and the other for any desired offset. What this means is that if you want the VCO to go from 500 to 1000 Hz, you select the resistors for just that bandwidth. For 1500 to 2000 Hz, you establish another set that satisfies the requirement.
   
   
4. Select a good PLL for the receiver decoding and design the loop to exclude harmonics. Make sure the input characteristics of the PLL are well defined—some are not—and that the VCO stability and loop characteristics are suitable for your design.
   
   
5. Try to select a minimal range of input voltage to the receiver PLL that produces a good lock with no phase slip. Do not let the input voltage exceed that limit. This will tend to control the loop's response to harmonics.
   
   
6. Condition your received signal before you present it to the PLL. Use an active filter of at least fourth order. Linear Technology has a nice product line of filters, as does National. Traditionally, you will find MF10 types configured as fourth-order bandpass to filter the individual telemetry sub-carrier before it hits the PLL. Offerings from National include the MF 8, which is essentially two separate second-order filters, making it easy to implement the fourth order required. The LTI LTC1064-7, when combined to form a bandpass function, enables an eighth-order sub-carrier isolation. Because of the much steeper out-of-band attenuation, this would allow you to reduce the "dead band" required with a fourth order circuit.
   
   
7. Finally, the mixer circuitry that takes all the VCO outputs and combines them for modulation of the transmitter needs special attention. You cannot just add them capriciously and expect the transmitter RF to be clean and easily decoded. Just as in commercial FM, some care must be taken to make all the levels equal when they are decoded at the receiver end. The varactor's linearity, the frequency response of the circuitry being used to modulate the transmitter, the "pull" ability of the crystal, if one is used, and other factors are in play. The bottom line is to actually look at the modulated waveform and the received waveform simultaneously, and then adjust the respective sub-carrier levels so that clean, equal detection levels are observed. In addition, the transmitted signal's spectrum analysis should show no clipping or spurs caused by overmodulation of the main carrier.

Wireless Archive

		Click here to get your listing up.