Unique Linear GG Amplifier Input Circuit
The ideal input circuit needs to be able to handle 100 watts of RF, present a 50 ohm impedance to the exciter over the desired frequencies, and prevent the driving power from being wasted in the filament transformer.
The conventional solution is to use a ferrite loaded filament choke and band switched individual low Q tuned circuits – one for each desired band.
Recently there have been several articles offering an alternate solution using a homemade coaxial line composed of copper tubing into which a length of #12 copper wire is inserted as the center conductor. The copper tubinged coax is then wound into a 1 inch diameter coil. At one end the shield and center conductor are wired to the filament transformer. At the other end the center conductor and shield are wired to the tube filament.
A 1000pf variable capacitor is wired across the copper tubing and a 2.5k ohm 2 watt resistor is wired across the capacitor to lower the Q of the circuit. The input RF is applied to the third turn from the filament pins through a .01mf ceramic capacitor.
With the tubing coiled to 15 turns around a 1 inch diameter form and the variable capacitor value at 1000pf we can tune 80 through 10 meters. The tap at 3 turns from the pins at the tube socket provide a 50 ohm impedance to the exciter over the same frequency range.
This alternate method requires the addition of a variable capacitor to the input circuit and burdens the operator with an additional adjustment. It also eliminates the need for seperate individual circuits, an input bandswitch, and a ferrite cored filament choke.
This alternate method is both effective and inexpensive.
The most troublesome part of the alternate method is the uninsulated copper tubing used to create the homemade coax shield. This brings up an obvious question, why not use real coax. Well, for two reasons. Most coax you can wind onto a 1 inch diameter form will have too much voltage drop across it at the current levels required by the filaments.
So the first step is to reduce the filament current. As an example, assume the use of four 811 tubes. If you wire all the filaments in parallel the power requirement will be 16 amps at 6.3 volts. If you wire two tubes in series and parallel their seriesed connections, the power requirement will be 12.6 volts but the current will only be 8 amps. Wire all four filaments in series and you can power the filaments with 4 amps at 25.2 volts.
With the filament current reduced we can now use RG8X to build our input coil. This further reduces the cost of this input circuit solution by eliminating the copper tubing.
Double Bazooka Dipole
There is a lot of information about this antenna on the internet. Lots of offers to sell. Lots of construction information with formulas to calculate lengths for various bands. Very little information on why it works, how it works, and where best to use it.
The bazooka is a dipole made up of two quarter wave transmission lines connected in series and wire tails connected to the ends of the transmission lines to resonate the antenna to the desired frequency.
As the transmitter frequency is varied the antenna impedance (real and reactive) also begins to vary. The impedance of the quarter wave sections will also vary but in the opposite direction of the antenna impedance. So the variation in quarter wave impedance offsets the variation in antenna impedance and we have an antenna that maintains an overall impedance over a larger span of the band. We have a broadbanded antenna. About twice as broad as a normal dipole for the same band.
Now, where is more bandwidth really needed? Although more bandwidth would be welcome everywhere it is most useful on 80 meters and 2 meters. Lets take a look at the 80 meter version.
First, the quarter wave sections. The two quarter wave sections are constructed from coax. Most of the construction articles on the internet assume that solid dielectric RG58 will be used. This coax has a velocity factor of .66 and is not suitable for high power levels. We could substitute RG8 but it is expensive and heavy. So lets use RG8X instead. But wait. RG8X has a velocity factor of .84 so the formulas designed for RG58 will not work. They will not get us the quarter wave sections we need and without those quarter wave sections we do not get a broader banded dipole. We still get a dipole, it will just not be broader bended.
Something else to notice is that quarter wave sections for 80 will be quarter wave only for 80 ( and odd multiples of the 80 meter frequency they are cut for i.e x3, x5 etc.). This makes the bazooka a single band antenna. Those claims of multiband bazooka dipoles do not tell the whole story. Yes, you can have a multiband antenna that incorporates a bazooka dipole but it will be a broadband bazooka only on one band. Acually, that is probably okay if the bazooka effect is on 80 meters. You don’t need a broadbamd antenna on 40. A normal dipole is plenty good on 40 meters.
If we want to construct a bazooka dipole for 80 from RG8X, we can use the formula for a half wave wire dipole and apply the .84 velocity factor. 468/Fmhz X .84. This comes out to a little over 100 feet. Good deals on RG8X are available for lengths of 100 feet ($40 on Amazon). 100 feet of RG8X will make two quarter wave sections for 3.932mhz. This should allow 1.5:1 SWR from 3.78 to 4.082 mhz or the entire 75 meter band.
Take that 100 feet of RG8X, Short each end of the coax, center to shield, find the center at 50 feet and cut the outter covering and shield being careful NOT to cut the center conductor, fray the cut ends of shield to make two pigtails for connection to the feedline to the rig.
(The 100 foot roll of rg8X from Amazon comes with coax connectors attached to both ends. You can avoid cutting these off by terminating them into female coax connectors and shorting them at the new connector ends.)
The part of the bazooka that does the radiating is the shield of the coax we used to build the quarter wave sections. The quarter wave sections are a quarter wave due to their relationship to the center conductor and the insulation in between that conductor and the shield. The shields are too short to be a quarter wave alone. They are too short by the velocity factor of .84. To make this a resonant dipole we have to add wire to the ends of the coax. The overall length of the antenna should be a half wavelength or 468/Fmhz. Some articles show 300 ohm twinlead added to the ends. This may help the broadbandedness but ordinarry wire will do too.
The bazooka has an additional advantage over a standard dipole. It is quieter. No static buildup because the antenna is a short circuit to DC and can be grounded at the feedline. Use a 1:1 balun at the feedline to prevent radiation from the feedline.
Universal Boatanchor Power Supply
Collins, Heathkit, and Drake all produced transmitters and transceivers that require external power supplies. These supplies were usually housed in speaker cabinets but could also be used as stand alones. Collins also had provision to house its 516-F2 supply on a shelf inside the 30S-1 amplifier.
Heathkit provided the 23B supply and Drake provided the AC3 and AC4 supplies. All three manufacturers supplies were designed to power tube style transmitters and transceivers in the 100 watt class.
Filament voltages of 6.3 and 12.6 volts at up to 8 amps, low voltage B+ around 300 volts at under 200 milliamps, high voltage B+ of 600 to 800 volts at up to 250 milliamps, And -40 to -80 volts bias at 10 to 20 milliamps.
By using two transformers, solid state rectifiers, and modern filter capacitors, I have been able to incorporate all the components needed to supply the various voltages an currents required by the various brands of transmitters and transceivers into the 516-F2 foot print. This 7X9X2 inch chassis is equipped with three output cables each terminated in the required connector for the target equipment. Eleven pin female for Heathkit, 11 pin female for Collins, and special 12 pin female for Drake. Each cable is terminated inside the supply to provide the voltages needed by the various equipment supported.
The only compromise is the high voltage. Both heathkit and collins provide 800 volts. Drake provides 650 volts. My supply provides 570 volts. This means, that at 65 percent linear operation, the finals will only deliver 74 watts output with this supply. Both heathkit and collins offer 104 watts output under these conditions with 800 volts to the finals. Drake will deliver 84 watts output.
The lower output with this power supply should not effect the ability to drive a linear amp to full output and it allows the finals in the exciters to loaf along, extending their useful life.
I hate doing plumbing work but I hate to have a running toilet even more. It wastes water and makes noise.
I always blamed poor quality fill valves and recently discovered that the fill valves are not causing the problem. They are merely functioning as they should to keep the water level in the tank at a flushing level. The real culprit is the flapper valve.
Flapper valves are made of rubber and chemicals in the water designed to kill germs also kill the rubber in the flapper valves. Over time, the rubber in the flapper valves becomes hard, brittle and no longer is able to keep water from flowing past it.
A recent toilet problem was solved by replacing the flapper.
Selecting Lawn Grass
Take a walk through your neighborhood. Look for cracks in the sidewalks. Do you see grass growing in the sidewalk cracks? That is the kind of grass you want for your lawn.
In my area that happens to be Bermuda.
Kentucky blue grass is fine if you are in Kentucky. I live in Texas and I think grass should be green.
St. Augustine grass does not like the sun, or the heat, or lack of water. It does not like the area where I live so I don’t use it. Grass abuse is a terrible thing.
When you decide on the grass you want, buy it in seed form. If you can’t grow it from seed, (assuming it propogates from seed), it will probably not survive anyway. Grass that propogates from seed is particularly low maintenance. Just let it go to seed a few times a season to replenish any weak spots in the lawn.
Sod is expensive to buy and install. Sod guarantees a finished lawn almost immediately and is the best way of establishing non-native grasses quickly. That might last a year or two but eventually you will find out why those grasses are not native.
If you are looking for Bermuda seed, then buy Bermuda seed. The bad should read Bermuda Seed on its face. Grass Mix, Sun and Shade, Quick Grow Grass Patch, these are all examples of products to avoid. If it does not say Bermuda, it is not Bermuda. In fact , generic platitudes are an indication the vendor does not want you to know what is in the bag because if you knew, you probably would not buy it.
Just because it is not approved for use in parts of California does not mean it is bad grass. California is run by people of low intellect and low morals. Being banned there may be a good thing.
Expect to pay at least $3 a pound for real bermuda grass seed.
Scotts products are usually overpriced and underperforming.
The HB stands for Home Brew.
In its current configuration this amp is equipped with four 811As and puts out 550 watts on 75 meters.
The total investment was about $250 and four weeks.
Purchased parts include:
Plate tuning cap
Antenna load cap
High voltage diodes
HV filter capacitors
HV filter equalizing resistors
Parasitic suppressor kit
All other parts were scrounged or donated. If all parts were purchased at current (2013) rates cost would exceed $700 and the project would not be worth doing.
It is built into a much modified tower computer case. An old computer power supply serves to exhaust air form the case and act as a source of 12 vdc at 15 amps and 5 vdc at 20 amps. The 12 vdc is used for powering relays but it can also be accessed off the rear panel to power other equipment. Like a transceiver to drive this amp.
A second computer power supply case has been gutted of all parts except the fan. It is used to house the filter capacitors and rectifiers for the Plate supply. The plate supply uses the plate transformer out of a Heathkit SB-220 and can be configured as a doubler or a bridge. For the four 811As we only need 1500 volts so the configuration is as a bridge.
The tubes used in this amp are mounted to a removable plate. Additional plates are available for other tubes as they become available. This design can use 811As, 572s, one or two 3-500s, 4-125s and even 4-400s. The basic circuit was taken from: How To Build The DX-811A All-Band Linear Amplifier. CQ Magazine September 1982, BY F.T. MARCELLINO, W3BYM.
A very simple soft start circuit using 120 vac relays is employed. One for the filament supply and one for the plate supply.
Controls include Plate tuning, Antenna loading, Band switch, ans five toggle switches with functions as follows:
Right to left
1. SSB/CW engages primary tap on plate supply transformer. Transformer secondary delivers 1000 vac in SSB and 1200 vac in CW.
2. Turns on ATX PC supply. Provides 12 vdc at 15 anps and 5 vdc at 20 amps. Also powers primary supply relays and three cooling fans. Nothing works until this power supply is turned on/ You can tell iy is on if you hear fan noise.
3. Filament supply ON/OFF
4. Plate supply ON/OFF. Plate supply will not come on unless filament supply has been turned on.
5. Transmit/Standby Interrupts amp keying line to transceiver. Keys at 12 vdc.
Transmit ready light comes on when all conditions for transmit are met.
Tuning controls are equipped with 3:1 velvet vernier drives.
Band switch has 6 positions. Correlation to frequency through tuning chart. Covers 3.7 mhz to 30.5 mhz. Frequency coverage below 4.2 mhz requires addition of a fixed 120pf capacitance to plate tuning capacitor.
Meter on the left is for grid current, measures 0 to 240 ma. Drive should be adjusted to 120 ma which correlates to .4 on the meter scale.
Meter on the right is for plate current. Reads directly from 0 to 1 amp. Typically runs at 550 ma when amp is running properly.
Stand Alone Soft Start
This would be neat if it could be made to work. The trouble is that in order for this to work, an on/off switch has to be located at the soft start relay and the on/off switch in the equipment has to be on and configured in a way that allows the impedance condition of the primary of the power transformer be reflected to the soft start circuit. Otherwise the soft start relay will engage immediately taking the current limiting resistor out of the circuit before the current surge has abated.
The short version of this story is ‘stand alone soft start does not work’.
If we can agree that a soft start feature is good, we might also agree that an external soft start might also have merit.
All soft start devices I have seen are added internally to the target equipment. This requires digging into the device, finding room for the circuitry, coming up with an acceptable mechanical installation and deciding if two soft start mechanisms are needed, one for filament and one for high voltage.
All of these concerns disappear when we build an external soft start device. The external device is built into a cast aluminum enclosure. The proper power socket is mounted onto the enclosure and the current limiting resistor and soft start relay is mounted inside the box. A line cord and strain relief exits the box. The line cord is plugged into the power connection on the wall. The equipment is plugged into the the soft start box.
For it to operate properly, the equipment needs to be left turned on. It needs to be on such that the turn-on load will be reflected back to the soft start box. This means there can be no relay switching of power inside the equipment. The transformer primaries must be a continuous circuit to the power connector of the equipment. Otherwise the soft start relays will engage immediately bypassing the current limiting resistors and the equipment will see the full startup current surge.
This also requires that the equipment to be protected be energized by a switch mounted in the startup box. One switch per circuit. These switches need to be rated at the current limit level. For instance, if the current limiting resistors are 100 ohms and the line is 120 vac, the startup current would be limited to 1.2 amps and a 2 amp toggle switch would be adequate as an on/off switch.
These considerations may not be realisable in all aplications but where they can be used we have a portable and universal soft start feature.
Generating RF Power
You can work the world with 50 watts and a dipole. Yes, you can also hike from San Fransisco to New York. You have to be insane to do things because you can. Look what I can do, look what I can do, those are the ravings of a two year old. Adults set relisable goals and use adequate resources to get there.
Work the world is not a realistic goal, how about lets talk to folks in 100 countries. We don’t want to spend the rest of our lives on this one goal so put a one to two month time limit on the effort. Get at least a three element beam so you can hear the DX. Run at least 500 watts to the antenna so the DX can hear you.
That brings us to the subject, where do we get 500 watts? The most you are going to get out of a transceiver or transmitter is 200 watts. We need an amplifier to get to 500 watts.
Note that the specification for 500 watts is the reading we get on a watt meter two minutes after we go key down, constant carrier, in CW mode, driving a load with an SWR of less that 1.5:1. We don’t want to play games with PEP input or PEP output and have no respect for those who do.
This is 2013 not 1960 so we might want to see what sort of solid state solutions are available. Indeed, a 500 watt solid state amp is available at a price of $3000. It uses 8- MPF-150 transistors. These transistors are rated at 150 watts each and cost $90 each. There is nothing particularly difficult about building your own but the transistors to do that are going to cost over $700 all by themselves. It appears that the solid state solution is more of a problem than solution.
If I had $700 I would buy a used tube amp. A used tube amp in that price range will get you 1000 watt OUTPUT!. Unless you have a mobile application and are wealthy we might want to forget solid state amps for now.
Going back to 1960 we see that we can buy 811 tubes for $5 each. Four of those will give us 500 watts output. If we drive them to 150 percent maximum ratings, like Collins does with the 30L1, we can get 800 watts output. It will shorten the life of the tubes and only add about 2db over the 500 watt situation, but the tubes were cheap at $5 each.
Today, 811s are still being produced but the price has increased to $20 each. Well, everything has increased in price since 1960, some things have increased by more than 5x. The apparent increase in tube prices merely reflect the loss of buying power of the dollar.
Is there a better deal for our money? Yes, the 572 at $40 is a better deal than two 811s at the same price. Looking at plate dissipation alone, two 811s are rated at 120 watts for the pair. At 65 percent efficiency for linear operation we find that we can get 184 watts out of two 811s without exceeding maximum plate dissipation.
A single 572 is rated at 160 watts maximum plate dissipation an can deliver 240 watts under the same conditions. We can get 64 watts more out of a 572 and only have to light up one filament It is interesting to note that 4 x 240 is a lot closer to 1000 than 4 x 184. Makes you wonder why the 30L1 was not designed around 572s because they seem far better suited to their evident goal, no need to abuse a set of 811s.
There is an easy way and a hard way. First the hard way otherwise you would not appreciate the easier way.
The hard way requires lots of arithmetic and knowing the internal resistance of the meter.
Secure a voltage source. A regulated supply is fine or get a battery. Lets say you found a 9 volt battery and it is actually 9 volts. Now lets assume our meter is a 1 ma current meter. Actually all meters are current meters. Volt meters are current meters with series resistances.
Now, using ohms law E=IR we find we need a 9k ohm resistor in series with the 9 volt battery to make the 1 ma meter read full scale. So hook it up and verify that the meter reads full scale.
Now, select a potentiometer. A 10k would be a good choice. Wire the wiper of the pot to one of the fixed terminals and connect the pot across the meter. Adjust the pot until the meter reads half scale. Remove the pot without disturbing its setting and measure its resistance. That resistance reading is equal to the internal resistance of the meter coil. Why? Because it shares the current equally with the meter. Two equal and parallel resistances take an equal and half share of the current previously conducted by only one of the resistances.
Now you can use that value to calculate the shunt you need.
Now for the easy way. An example will be used to explain the process. I had two meters. They were both 1 ma meters. One needed a shunt allowing it to read grid current as high as 200 ma. The other needed to read plate current as high as 1 amp.
These meters were to be used to measure plate and grid current in a linear amplifier I had built around four 811As. This amp’s relay supply was being provided by an old PC supply that was delivering +5and +12 volts. The PC supply was being used in normal fashion and was at home in the computer case that was being used to house the new amp.
I had just finished setting the band switch taps on the tank coil and had removed the 50 ohm resistor across the output of the pi network. This 50 ohm resistor was now employed in series with the regulated 5 volt supply to ground to generated a current flow of 100 ma.
I had previously purchased a bag of 50 1.8 ohm metal film resistors at a very low price and decided to use these to build the meter shunts I needed. I verified the 100 ma current flow with my $3 Harbor Freight multimeter. Then I placed one of the 1.8 ohm resistors in series with the 50 ohm current limiting resistor. I placed the meter across the 1.8 ohm resistor and noted the meter deflection. Too high. So I added a second 1.8 ohm resistor in parallel with the first. The deflection was now 17 divisions on the meter scale. 100/17= 5.88. Each division was was now indicating 5.88 ma for a full scale of 294 ma for the 50 divisions. My goal was to measure 120 ma which would be about 20 divisions or 0.4 on the meter scale. Close enough. The additional range would come in handy for use with other finals. I had purposely set this amp up to use other tubes as they became available.
For the plate shunt I kept adding 1.8 ohm resistors until the 100 ma current source deflected the meter to 5 divisions. 100 ma X 10 = 1000 ma or 1 amp. 50 divisions full scale divided by 5 = 10. Thus full scale translates to 1 amp on the 1 ma meter.
Done. Accuracy so – so. Ten percent would be fine.
Installing meter parts to tiny single hole solder lugs is an unpleasant task. That is why I mount the parts to circuit boards that can be installed to the meter terminals with two nuts. Wires soldered directly to the printed circuit board pads.
Select a fairly thick cicuit board scrap and drill two holes to fit the meter studs. Use a dremel tool to cut away a strip of copper on both sides creating isolated pads. Now solder, surface mount style, all the required parts across the isolating gap.
Once done, solder the meter leads to the pads and mount the assembly to the meter.
Using Obsolete Computer Hardware for Ham Equipment
I am suggesting the use of old computer cases to house communications equipment. Primarily for the housing of linear amplifiers. In a recent project I used a mid sized tower case to house the parts of a four tube 811a amp.
A small tower case measures 7 x 15 x 13 inches. The case I found measured 9 x 15 x 17 inches and was just barely large enough to house the required parts. To prepare the case I cut the cover into three parts using a metal cutting blade on a table saw. The sides of the original cover were used for the top and bottom plates of the resulting cabinet. What used to be the top of the case became the front of the amp. The bottom of the case became the rear wall of the amp. What used to be the front of the computer case became the right side of the amp after the plastic computer front was removed.
A shelf used originally to support the mother board was mounted to the bottom of the five inch hard drive enclosure and supported about two inches above what ended up being the new bottom of the case. This shelf was cut out to receive the upper half of a filament transformer which further stabilized the shelf. The plate transformer was a spare for a Heathkit SB-220 and fit nicely into the five inch drive enclosure.
This particular tower case had provision for two standard computer power supplies. I think it was an either or sort of deal instead of providing for two supplies at the same time. In my project I used both locations. One to house an old Dell supply to provide relay power, 5 volts dc, and 12 volts at up to 15 amps for powering an exciter or other lower power equipment. The second supply location was used to house an empty PC power supply case that was used to house the rectifiers and the filters for the HV supply. Both PC power supply cases have fans to exhaust air from the case. A third fan blows air into the case over the tubes.
I am not too hopeful that all the parts could be fit into a more normal sized case. The more normal dimensions are 7 x 15 x 13. In my case the 7 inch dimension would become the height. Seven inches is not enough to accomodate an 811 standing up. Of course you don’t have to lay the case down on its side. There should be no reason not to use it in its original ‘tower’ c0nfiguration.