Intermod Resistant Preamplifiers

You decide to improve your 33 cm receive system by adding a new preamplifier. Whether it is a home brew or kit preamplifier, you tweak it for best noise figure and maximum gain possible on your test bench. It is then integrated into your existing system. You tune to the beacon frequency and confirm that it is louder than before but also detect other signals that are 10, or 20 kHz away along with other random burst of noise and garbage that were not there with your old system..--- What happened??

This scenario may not be the same for everybody, and is not exclusive to 903 MHz, but the cause and result are the same. Somewhere in your receive system, something is being overloaded by signals that may or may not be in your receivers pass band! The cause may be the local Cellular Telephone system , the club repeater, or your daughters cordless phone! If you understand the problem and the parameters of your receive system, all or most of this problem can be eliminated.

Intermod Resistant Preamplifiers is the topic, but we need to establish a basic receive converter design to work with to understand how certain changes will help your system. The four block diagrams (A -D) are standard receive converter line ups. The noise figures and gains are calculated using the standard cascade noise and gain analysis equation which states:

NF (total) = + + + ….

Every system is made up of the same components and is shown to have a specific characteristic. System "B" has the best noise figure but if out of band products exist, they can be amplified by 30 dB before passing through the filter. System "C" would take care of that problem but has the worst noise figure of the four. System "D" is the most robust of the four (strong signal handling) but lacks in system noise figure. It could be tweaked for better performance if some of the parameters of the basic blocks were altered but for simplicity of this discussion, it was left alone. The most common is system "A" so lets use it for the example.

System "A's" parameters would be welcomed on just about any Tropo station. The conversion gain and noise figure are more than adequate on any band. So what the problem? Nothing, unless it experiences some intermod problems. So, how about optimizing the same system for strong signal handling while keeping its relatively good system noise figure and conversion gain. If we assume that we are using a high level mixer, (+14dBm P1dB) and a relatively "stiff " MMIC as a preamp after the filter, (+17dBm P1dB) system "A" would start to compress with approximately +4dBm into the filter! That's a big signal on receive!----------- If your receive system is optimized, your low noise preamp should be able to produce more than +4dBm output before compressing to take full advantage of the rest of the receive system.

What output power is a low noise preamplifier capable of producing when biased for low noise? If we take a standard preamp design at any frequency with a Hewlett Packard ATF10136 in it, and bias it at 2 Volts, Ids @ 25 mA, (Optimum Noise figure specification) it's P1dB is approximately -2dBm. The gain of the preamp will effect the total P1dB of the system (more gain will compress sooner) but the P1dB of the preamp will not change unless it is re-biased. After examining the ATF10136 data sheet, and looking at the P1dB specification, it is noted that the FET is capable of producing +20dBm output. (4V, Ids @ 70mA) This is 16dB more than needed! But what would the noise figure be if it was biased for power?

If the FET is biased for maximum power, it will degrade the noise figure. But depending on your situation, it might not be all that bad. At 50 thru 222 MHz., less than 1.0 dB noise figure is not a problem to achieve. At 432, 1.0 dB or less can be done if you are careful. At 903 and 1296 less than 1.3dB are possible. But please note this is at maximum power biasing!!! If you do not need the highest output power possible (like our example in system "A") then don't bias it up that high! What I have found is that 50-55 mA Ids with 3.5 V on the drain will produce a FET with a P1dB of approximately +13dBm and will not degrade the noise figure below 222 MHz. and only effect 903 and 1296 by approximately .1 - .2dB. Since there is plenty of head room in the system "A" example, the biasing could be lowered if the change in noise figure was unacceptable. The schematic below will show the biasing of a ATF10136 FET. (resistors and regulator changes) Other FET's could be substituted if their specifications prove equal or better. For lower frequency preamplifiers with L-C circuits on the input, the inductor may need to be increased for best tuning. Series inductor circuits (903 and 1296 MHz) may need may need to be shortened .

After making the bias changes (resistors and regulator) be sure to check all connections with an ohm meter. Maximum Drain source voltage for the ATF 10136 is 5 VDC. (+5ds) If the source resistors are open, 9 volts from the regulator will be present (no voltage drop across the 100 ohm resistor ) on the Drain of the FET. It will die! Also be sure that the 100 ohm resistor is a ½ watt version and preferable a carbon composition. If you are using 1206 size chip resistors, use 3 , 300's ohm in parallel. 1206 chips are 1/8 watt dissipation. 3 will bring you up to about 3/8 of a watt (375 mW) dissipation, where 300 mW is required.

Now I know we are only talking about a 6 dB increase in signal level before P1dB occurs. Can this help in a intermod prone area? If the rule of thumb is used that states that after 1 dB gain compression is achieved, every 1 dB more of signal will produce a 3dB decrease of third order IMD. Then, in this example, most intermod interference could be reduced by 12 dB!--------------------------- I'll take it!

Other FET's are available and can be used. Using a PHEMT would work in the "A" system. Examine the data for the ATF36077. It can be biased to produce +5dBm output P1dB. That just makes it! If you decide to "Go for it", and want to use a different FET, be sure to read the data sheet for the bias conditions. The data may not spec the bias for maximum P1dB but if the maximum power dissipation is 350 mW or higher, the FET can be biased up way past its low noise specifications. Just watch the dissipation level.

With higher level mixers becoming more and more the norm, they can now be coupled with low noise FET's that will produce +27dBm!!! At these levels, standard MMIC's will give up and fall flat on their face, so high power BI-polar or power FET's will need to be used as second stages before or after the mixers. If you plan to upgrade your system in the future, you might want to start with a high power preamplifier. The limitations will be in the transverter, not the preamp! That's a good place to start but,----what will your poor transceiver do!!!!

Other modifications to the input of the microwave preamplifiers (903 MHz. and up of W6PO and WB5LUA designs) may be accomplished to further help eliminate out of band amplification. A modification makes use of a shunt inductor on the input circuit to roll off low frequency gain. Please see the article in 1997 Microwave Update Proceedings titled "Eliminate Low Frequency Intermod of LNA's"

Review some of the data sheet attached to get familiar with the way manufactures specifications work. Most of the higher power FET's do not have low noise parameters but all of the low noise FET's have P1dB specifications. So Experiment and have fun!

Down East Microwave Inc. 954 Rt. 519, Frenchtown, NJ 08825 USA
Tel. 908-996-3584(Voice) 908-996-3702(Fax)