Auto-Tune Linear Amplifier


Why

 

I begun to chase DX also on the low bands and realized that five hundred watts is not really enough. The years using a TenTec solid state amplifier had made me really like automatic band switching so to buy a manually tuned bigger amplifier was not an option. Commercially built auto-tune amplifiers are good but they are expensive. For example, an ACOM 2000A costs about 5500 euros and you can get a decent car with that amount of money. Thus, I built a new amplifier myself.

Picture 1: Generic view to the tank circuit.


Generic Info

I chose the basic concept that was introduced in the local amateur radio magazine called Radioamatööri (in issue 11/1994). The amplifier is based on the Russian GU-84 tube. That article suggested using nowadays hard-to-find stablizer tubes for screen power supply. Solid state is better for this task anyway, so I chose to make the screen and control grid power supplies with the G3SEK circuit, instead. That gives protection to many fault conditions, too.

This article describes the only thing that really differentiates an auto-tune amplifier from a manually tuned amplifier, namely the tank circuit. The pictures show only the main principles of each module. Sorry for the poor quality of some pictures, I didn't scan the hand-written papers properly. If one wants to make a similar type of an amplifier, one needs to clarify many details oneself. For an experienced builder, that should not really cause too much problems.


Continuous Or Steppwise Tuning?

There are two basic ways to automate a tank circuit: continous and steppwise tuning. The former uses capacitors or coils that are motor-driven. Surplus part suppliers sell both but the prices are high. A motor-driven vacuum capacitor can cost as much as thousand dollars. Alternatively, one can use an ordinary capacitor and make the mechanics oneself, though that is a bit difficult for an ordinary amateur. The advantage of continuous tuning is that one can fine-tune the tank circuit to precisely correct values.

Steppwise tuning can be done by combining coil and capacitor values using relays. The advantage is that the mechanical design is simple and tuning is fast. When one changes the frequency band the amplifier is tuned right away instead of, say, some dozen seconds. The disadvantage is that one may not be able to fine-tune the amplifier for all possible situations due to gaps in the possible value steps one can take. This can be caused by choosing incorrect component values right at the start or due to some component values being different to the (otherwise good) plans.

After a lot of thinking I chose the steppwise option.

I did consider also if I should combine the above-mentioned principles. There are commercial amplifiers that have separate tune and load variable capacitors for each band, and then the correct one is chosen with releys. The problem in this approach is that when one buils an amplifier that is supposed to give a kilowatt out, the price of the variable capacitors becomes extensive. Once, I was also pondering about a solution in which some base capacitance would have been set by variable vacuum capacitors and then the rest would have been done with additional capacitors switched by relays. The idea was that switching between CW and SSB were to be done by tuning the vacuum capacitors. As e.g. Jaska OH1MA has made auto-tune amplifiers that only use fixed value components, I also thought that there is no need for any continuously tunable parts. Mainly, this seems to have been the correct decision, though some of the fixed capacitors would perhaps have had some good use for small variable capacitors in parallel to change the actual values to be closer to the specifications.


The Tank Coil

To make a coil for an auto-tune amplifier is no big deal. The band switch is just replaced by relays. Note that relay control cables have to be protected well from RF. One should use diodes in reverse direction and some small capacitors over the relay coils and one should route the cables through areas with little RF, if possible.

The relay type I used for the tank coil was Russian V2V. To mount such is easy, just drill a 31 mm hole and a couple of smaller holes for the screws and you are almost ready. I should have planned and measured the relay positions a bit better as I had to make a few changes afterwards. The control cables go to the other side of the shassis which is good for RF protection. Russian V1V relays are cheaper but I find them less comfortable to work with. The operating voltage for the V2V relays is 27 volts though at first I run them with 24 volts. Unfortunately there was one relay that did not connect with 24 volts but really needed 27 volts as specified. Better to test before assembly.

Originally, I had a great idea to make the tank circuit totally modular so that each part is easy to disassemble, if needed, see picture 1. The load capacitor module is at the top, the tune capacitor module is at the bottom, and the coil module is in the middle. In principle, all was well, but due how some other components of the amplifier were placed, to disassemble the coil and put it back to it's place while debugging the amplifier was, in fact, surpricingly laborous. Well, it could have been worse. I had used D25 connectors to connect the control cables and that saved a lot of soldering when disassembling the coil.

In his amplifier Jaska OH1MA has planned component placement better and thus service work is easier (picture 2).

Picture 2: Master piece of art by OH1MA.

All the connections are made using copper straps. I found that the copper plates that are used to cover the floors in front of fire places are just the right material. One needs to drill 8.5-9 mm holes to the copper straps for the V2V relay connection screws. The thin copper plate should be drilled so that it is placed between two pieces of wood. Otherwise it gets difficult. The copper straps are easier to install correctly when one bends them a bit between the components (see picture 2).


The Capacitors

The tune and load capacitor modules are made using fixed capacitors whose values increase so that each bigger value is, in the ideal case, exactly twise the corresponding lower value. If for example the smallest value is exactly 1 pF then the next value is 2 pF and the next one is 4 pF etc. All the necessary tune and load capacitor values are achieved by combining these small values. Supposing that our capacitor values were e.g. 1, 2, 4, 8, 16, 32, and 64 pF then e.g. 97 pF would be achieved by choosing 64, 32, and 1 pF capacitors.

I did not have capacitors with exactly the above-mentioned criteria. Besides to my great plans, the actual capacitor values are not exactly similar to the nominal values anyway. Consequently, tuning the amplifier is not always totally optimal, but close enough for most practical purposes.

Picture 3: Tune capacitor module.

The capacitors I used for the tune capacitor module (picture 3) have the following nominal values: 1.5 pF, 3 pF, 6 pF, 13 pF, 25 pF, 47 pF, 100 pF, 200 pF, 390 pF (2200 pf and 470 pF in series), 790 pF (680 pF in series with two 220 pF capacitors that are in parallel). The very smallest capacitors are home brewn of copper plate. Also Jaska OH1MA starts tune capacitors with 1.5 pF.

The loac capacitors are 7 pf (22 pF and 10 pF in series), 11 pF (in series two 22 pF capacitors), 25 pF, 47 pF, 88 pF (four 22 pF capacitors in parallel), 170 pF (100 pF and 70 pF in parallel), 340 pF (two 680 pF capacitors in series), 680 pF, 1260 pF (two 680 pF capacitors in parallel), and 3200 pF (this was originally 2200 pF according to the binary principle, but it was later found that for topband a bigger value is optimum). Jaska OH1MA begins with 12.5 pF.


Band Selection

Band selection system is made for Yaesu FT-1000 though the circuit can easily be modified for other radios. In addition, there is a possibility for manual band selection.

The system is based on a product called Unified Microsystems BCD-10 Automatic Band Decoder. The BCD-10 reads the Band Data Port of Yaesu and based on that triggers a relay that corresponds to the band in use. BCD-10 can't handle much current so in fact it only triggers a small relay that controls the relays in the RF unit of the amplifier.

Manual band selection is done by breaking the BCD-10 to Yaesu connection with small relays and inputting self-created five volt control signals to BCD-10 through diodes (picture 4). The picture shows only the principle of the circuit related to the D pin of the decoder. For each band there are diodes according to the Yaesu BCD control logic. For example the 7 MHz band is represented by the following logical values: A=1, B=1, C=0, D=0. Thus when the manual selection knob is in the 7 MHz position we connect diodes to the decoder pins A and B but pins C and D are left disconnected. The resistors at the bottom corner of the picture are related to creating a suitable voltage for BCD-10.That part could have been done more elegantly.

 

Picture 4: manual control logic (principle only).

My Yaesu did not provide enough current for several decoders. I use them to automate antenna selection of my station. So in addition to the amplifier I also had to make a relay box that repeats BCD coding of Yaesu providing more current.

Many amateur radio bands are quite wide. A tube amplifier cannot really deal with that having only one capacitor value setting per band. Thus I had to have a possibility for more. I chose the following band segments to have separate capacitor values:

28 LOW MID HIGH

24 ALL

21 LOW HIGH

18 ALL

14 LOW HIGH

10 ALL

7 LOW MID HIGH

3,5 LOW MID HIGH

1,8 LOW HIGH

These have been quite sufficient in practise.

A switch that controls the relays R1 to R5 chooses the segment within each band. Those relays pass the RF relay control voltages to some dip switches from the relays that BCD-10 controls. Finally, it is the dip switches that decide which relays in the RF unit are triggered (picture 5). Complicated? Sorry about the deficiencies of the circuit diagram. One should, however, be able to decipher the idea by thinking about the following question: when you have 9 relays, of which only one provides 27V max 2.5A at any given time (band info), how do you connect additional relays to divide that voltage to maximally three alternative routes (corresponding to LOW, MID, HIGH segments within each band).

 

Picture 5: Band segment selection.


Dip Switches

Jaska OH1MA controls the RF relays using a micro processor. A fairly simple computer program chooses everything (frequency band, segment within a given band, the inductor relay, and the capacitor relays within each segment). It is a good solution. In fact, it is much better solution than mine. There are quite cheap but yet powerful micro processors available, such that you can even program with such a simple language as Basic. I chose the dip switches just based on them being free for me (thanks to my favourite radio club). 

The main idea is in picture 6. The control signal of each band segment comes to a veroboard having rows of dip switches. Each row of dip switches corresponds to a band segment. The dip switches  choose if the relay that corresponds to the dip switch in question triggers or not. The diodes are essential to guarantee that a band segment does not put voltage to another band segment. Let us imagine, for example, what happens if the first and third dip switch of the 28 HIGH segment are connected and for example the first and fifth dip switch of the 28 MID segment are also connected. If we have 28 HIGH selected then without diodes also the 28 MID segment would get voltage. Thus we would trigger relays 1, 3, and 5 even though it should only be relays 1 and 3. It was this types of problems that I had to find when a few diodes were shorted.

 

Picture 6: The diodes of the dip switches.

The control unit of the amplifier employing the logic described above is shown in picture 7. In addition to what has been shown here, there are also some indicator LEDs that show for example which of the band segments is selected. These LEDs are a great help when trying to correct several types of soldering errors and when finding blown diodes.

Picture 7: Control unit of the amplifier.


Commissioning

To commission this kind of an automatic amplifier is quite laborous. The principles are the same as when tuning a manual amplifier. Some tune for maximum power, some seach for a dip in anode current. The main difference is that one cannot just press the cw key and start turning capacitor knobs. Each change of tune or load capacitor values is done by switching on or off one or more of the dip swithches when the transmitter is not on. When one is at a border of some big elementary capacitor value then this may mean touching quite many dip switches. It is a good idea to tune the amplifier first into a dummy load and only when the ballpark values are found do the final touches with real antennas. The good part of the story is that when you have finished, you don't have to touch the knobs again. It's automatic.


Miscellaneous

The Russian V1V relays take about 110 mA and V2V relays take about 150 mA. There can be roughly ten of both types triggering the same time. Thus one needs about 2.5 A for the relays. I installed the transformer to the control unit mainly because of having limited space in the RF deck. This amplifier has external anode power supply. At first, the amplifier provided 1 kW out with 1800 volts at anode. When the Finnish power limit was increased to 1.5 kW I upgraded the amplifier simply by building a new 2700V anode power supply. 

It is a good idea to verify the functionality of all RF deck components before installation. I had one suspicious fixed vacuum capacitor and one relay that did not quite work with my original control voltage. One should be careful with the diodes. It took me a long time to find the problem when a particular diode looked exactly like all the others, except that it was a zener. Debugging blown diodes was a story of it's own. There are differences in D connectors. There are good ones and then there are such that loose pins when soldering. Free D connectors became very expensive, if I value my time in any way.

It would be good idea to plan beforehand both the electronics and the mechanics. Less surprises so. Using printed boards instead of veroboards would have saved time. Soldering diodes for dip switches is really, really laborous when one uses a veroboard. And finally, you usually need a much bigger RF deck than what you at first imagine.


Conclusions

So, was it worth doing? That is a good question. This amplifier is not as good looking, small-sized, and quiet as an ACOM. But it does provide the necessary power. The basic structure is simple and reliable. If there ever are any problems, I can fix them myself.

To reach the original 1 kW target cost about 2000 euros. Better deals at hamfests could have saved hundreds of euros. The most expensive item group was the vacuum relays, they cost about 500 euros. In addition there are spare parts worth about 700 euros. I have actually needed one of the spares. Despite of having being built like a tank and being easily able to provide double the power I use, the grid of my original tube was mechanically broken after a few years. It is that nasty tingle tingle sound that you hear while shaking a broken tube that finally revealed me why I occasionally did not get any power out despite of having high anode current. Luckily, I had bought quite many spare tubes with reasonable prices, so this is no big deal. Despite of some recent price increases, price to performance ratio of the Russian tubes is still superior to anything else on the market. Having said that, I do admit that nowadays I would like to have something more - I would like to get one of the modern solid state legal limit auto-tune amplifiers. And having said that, I also admit that the price to performance ratio of those is still so bad that I will, for years, easily find more sensible things to spend my money on. To build this amplifier took about one and a half person working months and about a year in calendar time. An experienced amplifier builder could make a similar one in half the time.

Special thanks are due to Jaska OH1MA, Juha OH7JT, Maukka OH2BYS, and Markku OH2RA.Their help made this possible.

To main page