Techies' Corner
Tweaking and Tuning Amplifier Circuits;
Making a Low-Pass Filter

So you've gotten serious about assembling a radio station. You've read what you could find. You've collected a pile of audio sources, a mixer and hopefully a compressor/limiter. You've sent away for a transmitter kit and then taken a deep breath and carefully soldered it together, But now you're stuck. You're afraid to turn the thing on, you're not sure how to tell if its working right, you can't find the real nuts and bolts information anywhere that would make you feel secure in your ability to proceed from here. Besides, the literature you have read warns most alarmingly not to operate without a harmonic (low-pass) filter but none of it tells you how to build one or how to test one you have purchased or inherited. Or perhaps you have built an exciter and an amplifier but your exciter's literature says it puts out 1 watt but your amp only wants 1/4 watt in. What do you do about the other 3/4 of a watt?

Well, good thing you waited! And good thing you found this web page because right now were going to try to cover a couple of harmonic filter and attenuator designs, and hopefully give you some good hints on troubleshooting and tuning up your new transmitter.

I'm going to digress briefly to try to explain this impedance thing you may have heard of. Impedance, measured in ohms, is a special kind of resistance. It is the load into which your amplifier is running. You may have noticed that stereo speakers are rated in ohms, usually 8 or 4 ohms. This is the impedance into which audio amps like to run. Radio amps are built to operate into 50 ohm loads. If you've come this far you have probably heard that the coaxial cables we use are 50 ohms and that you must match your antenna to 50 ohms. Once you come off your circuit board with your RF output in a coaxial cable you are in a 50 ohm environment, which means that your power will travel without much loss until it encounters a different impedance or is used up in a 50 ohm load (antenna or dummy load). If it encounters a different impedance then some portion of your power will reflect from that point back toward your precious amplifier where it will be expressed as damaging heat. If it travels all the way down your transmission line without seeing a mismatch and finds a 50 ohm load at the end then the power will be efficiently converted into another form. If this load is an antenna, matched at 50 ohms, then all your power will be radiated into space as an RF signal, this is good. If this load is a 50 ohm dummy then your power will be radiated as heat, not radio power, but this heat will be safely away from your final transistor. A dummy load is very useful in place of an antenna for testing without transmitting.

Now, there are a few more things you are going to need to get. I know, you've already spent so much on this project, but you really need a couple more tools to move forward. First, get or build the afore-mentioned dummy load, nothing more than a 50 ohm resistor soldered across a coaxial connector or cable; by across I mean from the center conductor to the ground shield. Remember that resistors add in series so two 100 ohmers next to each other (in parallel) will divide to 50 ohms, likewise 4-200 ohmers or 8-400 ohmers and so on. In each case the resultant dummy load will handle twice as many watts as the one before for a given resistor rating (eg. A dummy load built with 4- 200 ohm resistors rated at « watt each will take 2 watts before it starts to melt from expressing your power as heat.) Be sure to get a dummy load going on that will deal with the expected power from your setup. Resistors up to 5 watts are commonly available, 8 of them in parallel will deal with 40 watts, if you're building an amp bigger than that you probably already know where to get a bigger dummy load to test it into.

Another thing you simply must have is an SWR meter. As an aspiring RF tech this is something you cannot do without. SWR stands for "standing wave ratio". It is a way to measure impedance mismatches. If you place an SWR meter in your transmission line between a properly tuned transmitter and a 50 ohm dummy load it will read 1:1, that is perfect. If you were to build a purposefully wrong dummy load, say 100 ohms or 25 ohms, then the SWR meter would read 2:1, get it, a 2 to 1 ratio mismatch. Remember that the reading of an SWR meter is relative to your power output so you must first calibrate your forward power to 100% before reading the reflected power. See? The amount of reflected power only means anything in terms of your overall output, if you are making 10 watts but reflecting 5, the SWR is 3:1, barely usable...if you are making 100 watts but reflecting the same 5 your SWR is some very small ratio 1.05:1, just peachy. Your meter, unless it is of an advanced sort, counts on you to calibrate it to the power level in which it finds itself, simply set it to "forward," fire up the transmitter and twist the knob to make the meter read at the rightmost mark (100% forward) but not beyond, then switch to "reflected" to read your SWR. For our purposes I really like the MFJ 812B SWR meter, actually the folks at MFJ (More Fuckin' Junk) have a lot of really useful tools for cheap, because it is affordable and will give you an idea of relative strengths of very weak RF powers. The 812 is only about $30, look for their ads in any of the ham radio magazines (QST, CQ, or 73).

The other thing you will need is a small collection of short coaxial jumper cables with the appropriate connectors installed on each end, probably PL-259's. These are the most common RF connectors and are readily available because they are used in CB applications. You can buy or steal jumpers at Radio Shack or learn to build your own. I would really like to go over PL-259 soldering now but my editor tells me I'm already getting long and we haven't even started on the meat of the article.

So, on to the meat (or tofu if you choose). An amplifier making its primary power at, say, 100 MHz will inevitably produce harmonics, that is, smaller power levels at 2x, 3x, 4x and so on of your frequency. In this case 200, 300, 400 etc. MHz. (Actually some transmitters will even output measurable harmonics on 1.5x, 2.5x etc. of your frequncy.) A harmonic filter is a device designed to present a 50 ohm impedance to your desired signal at 88-108MHz (FM broadcast) but present a high SWR to your harmonics, to block and filter them out. The actual math is rather gnarly but I have a couple designs here that work pretty well. If you are running under 10 watts, this filter will probably suffice:

[image to be inserted here]
29pF 88nH 51pF 88nH 29pF

If you are running more than 10 watts, consider this filter:

[image to be inserted]
31pF 92nH 54pF 102nH 54pF 92nH 31pF

For the capacitors get silver micas rated at 500 volts, they are avaliable from RF parts (http://www.rfparts.com/) or from really well-stocked electronics stores in big cities. Get as close to these theoretical values as you can but its not that critical within a couple picoFarads. Remember that capacitors add in parallel (the opposite of resistors) so you could, for example, place two 15's next to each other to get 30 or put a 1 next to these to get 31. If 100 or 200 Volt rated components are all you can get its probably OK if you are running less than 20 Watts.

The inductors can be wound out of wire. I like to get silver plated copper wire from the local hippy jewelry supply store because silver is a really good conductor and it makes the filter really pretty, but copper wire works as well for the non-perfectionist. Actually measuring inductance at these small values is rather voodoo, the variables theoretically are the diameter of your wire, the diameter of the coil, the number of turns per inch of the coil stock and the actual number of turns in your coil. If you wind coils on a 1/4 inch x 20 bolt with 22 gage wire, you will have about 500 nanoHenrys per inch. For the 88 nH make about 4.5 turns, for the 92 nH make about 5, for the 102 nH make 5 or a little more but expect to have to squish it. Be sure to leave ample wire leads to connect your coils to the capacitors, you can always cut extra off afterwards. If you wind your coils all the same direction they will interact as if they were one big coil. This degrades filter performance. Wind adjacent coils in opposite directions to minimize this effect. Additionally placing each coil in its own shield box will further defeat this "mutual coupling" but is probably overkill except for the artist. Solder the components as close together as possible, even straight wire does have inductance. Connect your filter to a 50 ohm environment with a connector or coaxial cable by simply soldering the in and out points to center conductors and the grounds to the small piece of copper clad board stock on which you build the filter. I like to keep a couple of test cables around with connectors installed on one end which I temporarily attach to a new filter to check it out.

So how do you know if your filter came out right? Well, think about it, what's important to us is that


The first criteria is measurable with the instruments at hand. Connect the transmitter to the SWR meter to the filter to the dummy load (in that order). Now, if you have previously tuned the output of your transmitter to 50 ohms (by patching the transmitter to the SWR meter to the dummy load and tuning your transmitter for maximum FORWARD power with the variable capacitors in its output) then you know that everything in your coaxial system; transmitter output, feedline, dummy load, everything except your filter is 50 ohms. Since the SWR meter is placed before the filter it will indicate any mismatch encountered there (note that the impedance seen at the output of your filter effects the impedance seen at the input, so it is important that your filter is terminated into the 50 ohm dummy load for this test). Fire up the transmitter, calibrate the meter and measure the SWR into your filter. You can now tune the filter by squishing and spreading the coils. Your filter is tuned when it is at (or very close to) a 1:1 SWR.

OK, OK Whoa! You're saying, you've got me monkeying with voodoo coils hooked to a live amplifier that I haven't even had the guts to fire up yet! OK, you're right, time to backtrack a little and hopefully get a little closer to this "impedance" concept at the same time. Lets make sure your exciter and amplifier are in fact tuned to 50 ohms. Double check all assembly details and power polarity to your exciter (really! Most problems are solved with close inspection and some common sense). Patch the exciter into the SWR meter into the dummy load. Set the SWR meter to forward and place the knob somewhere in the middle. Tune a monitor setup to your intended frequency. Triple check your hookup. Take a deep breath...apply DC to the exciter.

If this really the first time firing the thing up, watch the circuit,not the meter. Pop? Flash? Smoke? No? Good. Did the monitor suddenly loose the static and become silent with your first carrier? WOW! If not shut off and try to deduce what's wrong. Check any literature provided with your kit. Check the manufacturers website. Email their techs. Most of all go over the board for reversed components and solder bridges. I repeat, it is rare for components to be bad, most problems are just trivial oversights.

Assuming that your board is locked on frequency and not smoking, check the SWR meter. If the needle is crammed over to the right, throttle back the sensitivity with the knob (counter-clockwise) till it comes off the peg. If the needle is still at zero, twist the knob to the right until you get a reading. If you can't get a reading but you do have your carrier in the monitor you have a problem toward the end of your RF chain. Get a reading in the middle with the knob and then twist the variable caps in your exciter's final to deflect the needle to the right. Most exciters, the venerable and problematic, but great sounding Dunifer 1 watt, the Veronica, the Max 1, have two variables in their final. Turn one. If the power drops turn it the other way. You will find that there is a point where your power peaks. Leave that cap there and try the other one. You will probably find a point on it where you get even more power; in fact you might have to turn the sensitivity of your meter down to read the peak. The capacitors interact so bounce back and forth a few times to get the absolute most out of your exciter. Listen to the monitor while you do this; sometimes the last amplifier stage can sort of wobble out of control and make a lot of power but also horrible noise and interference. Avoid this, even if it means running slightly less power. Congratulations, you have just tuned your first amplifier.

Note that some circuits use transistors that don't need to be adjusted to 50 ohms, the BGY 133 20 watt transistor is a good example. It operates naturally into 50 ohms, very handy. Its that impedance thing again. Remember that we said that as long as your power sees 50 ohms as it goes along then it will be propagated without much loss? Well that's true, but by building special networks of capacitors and inductors we can kind of fold ourselves into another impedance in an efficient manner. This is called matching, and how good our match is determines how much of our power gets through into the new impedance. Most transistors operate at some other impedance and need to be matched to your 50 ohm environment. Since the best power transfer occurs in a matched impedance we know that your exciter is now tuned to 50 ohms.

If you have an amplifier on hand place it in the line just after the exciter, make sure the dummy load is hooked up after the SWR meter, and try tuning it up in the same fashion. Some amps just have variables on the output, others need their input circuits tuned as well. If you happen to have two SWR meters, one placed between the exciter and the amp can help you adjust the input match, but really just tuning for maximum output power is fine. Be sure to listen to the monitor. Another trick is to place a television near the setup and scan the channels for interference produced by your station, then tweak with the amps to minimize it.

Now, back to the filter. Place it in line between your RF source and the SWR meter with the dummy load on the end. Turn the radio on and calibrate your forward power to full scale, then switch to reflected power to read the SWR. If its in the red turn the radio off and try squishing the coils closer together. Check the SWR. Still bad? Shut down and try pulling the outer two coils farther apart than they were at first. Try different amounts of squishing, but remain as symmetrical as possible, that is, keep the outer two coils squished the same. At some point you should get an SWR that's not in the red, then you can leave the transmitter on while you mess with the coils with a couple of toothpicks to get the SWR as low as possible. To test your filter try tuning in one of your harmonics on a scanner. Fire up without the filter and see how far away you get the harmonic. Now put the filter back in and verify that the harmonic is weaker. If you have access to some sort of VHF transceiver you can hook it up to your SWR meter and dummy load, transmit, and set your forward power reading to full scale, then put the filter in between the transceiver and the SWR meter and see how much your filter defeats the VHF frequency. If you had access to the tools you could plot the response of your filter on a graph, it would look something like this...

[image to be place here]

See how it lets your desired signal through with just a little loss but stomps the heck out of the harmonics? Pretty trick.

OK, what if you have to get rid of excess power between amplifier stages. Sure, you could just mistune the stage, but its a bad idea because then the excess power is reflected from the mismatch back into the source amplifier. You need to convert the excess power into heat somewhere other than your delicate transistors, a sort of inline dummy load. The circuit you need is called an attenuator. It is easy to build and presents your exciter with the proper impedance while dissipating some of your power as heat . Build them out of resistors rated for more than the total power of the amp you are attenuating. For example if you are knocking 1 watt down to « watt, 2 watt resistors would be good. The first of these T-sections will cut your power about in half, the other will cut about a quarter.

[images]

10, 150,10

5, 300,5

You can place one after another for even more attenuation, for example in the not uncommon situation of wanting to drive a BGY 133's 1/4 watt input with a robust 1 watt PLL you could cut it in half, then in half again.