Page 1 of 2 12 LastLast
Results 1 to 10 of 14
  1. #1
    Active User New Member YU1OXB's Avatar
    Join Date
    Jul 2016
    Location
    City of Cacak, West Serbia
    Posts
    13

    MF-1000 Home Made 1KW HF linear amplifier with two GI-7B

    The decision to make this linear amplifier came suddenly when a friend of mine (YT9M – Niksa) asked me to repair his old linear SOMMERKAMPFL-2500 modified to work with 5xPL519, that's when I saw the benefits and advantages of the higher power.

    It's nice to work with a small or medium power (5-10W or even up to 100 watts) but there are a lot of situations where it only brings frustration and annoyance and even the loss of a chance to do the rare country or rare prefix. Just a good antenna and power are solve the problem and there really is no alternative.

    Thinking about how to conceptualize, I decided to work with two small Russian ceramic triodes GI-7B since they are both easy to obtain and relatively cheap, they are physically small but very robust and durable and can provide up to 1KW PEP output power on a quite decent anode voltage which is not over 2500V. There are better and "more capable" tubes but they are more difficult to source and generally significantly more expensive and they demand higher voltages to work with them and that implies more specific and expensive passive elements to reliably work on such voltages which in turn bring more difficulties to source them and they are more expensive.

    All of this initiates a vicious circle leading to a larger, more expensive and more difficult device to work with. Thanks to the goodness of a few close friends, I came (in the form of gifts) in a possession of some key components, so it is very difficult to estimate a total invested funds in this building, but if I've just bought everything as normal material (all!) would certainly not be worth less than 5 - 600 Euros.

    My well intentioned advice is to first get all the key components -tubes, "plate" and "load" air variables, turbine or fans, the main transformer, electrolytics ... - and then very carefully deploy components on the table, seeking the most logical arrangement and layout with the shortest connections (it should not be overdone in order not to make it become so densely packed that components can no longer be approached!). Only when you are positive you have reached the best layout you can think of then you should start thinking about the design and manufacture of the case-housing.

    The reverse way is wrong, and that's what people often do ... they first find a box that they like and then start jostling components into it at all costs just to pack everything inside no matter what!
    Projects like this one demands a lot of will, effort and time, but if you are not sure that you will last to the end, you had better not start. Most of the time goes into thinking and planning the layout of components on a chassis, because the concept and schematic are just classic and there was no need to reinvent the wheel. As such we know similar units were worked out and “chewed” and tested who knows how many times, but the arrangement of components, to be with the shortest lines, in a most logical way ... that is what everyone has to solve in their own way and that aspect of how high power linear amplifier design works, DOES MATTER!

    Many advocate the theory: "... come on, that's HF... it doesn't matter if you run 25 cm wire from the output coil to the band-switch ...people have been working that way many times and it all works ..."but it is not exactly so. Maybe some people do it that way and may be it worked, but later you often hear somebody that “does not know what is going on say this oscillates, "how do I stop it"” and so and so on ...

    I have no intention of theorizing here, but will try, with a help of pictures and important remarks and explanations, to present this project so that other potential builders can repeat it with a greater chance of success. However, there are a few details that are done on my linear differently than usual, so I'll give them in a little more detail.

    As I said, the concept is pretty much classic, and there is not much room to do and say something radically new and different. Lamps are working in GG configuration with excitation in the cathode. At the output there is a more or less standard Pi filter which, to a certain extent, reduces higher harmonics in the output signal and also matches the output impedance of tubes to the standard impedance of coaxial cables i.e.50 ohms.

    An interesting detail is what I've done with a plate choke. In the beginning I started in quite a classical way ... as a body I used a15cm piece of fiberglass old telescopic fishing rod (it SHOULD NOT BE of the newer "carbon" or "graphite" rods ... IT MUST BE plain pure fiberglass) with an outside diameter of 22 mm on which I put *** turns of 0.6mm wire in four unequal sections. I didn't mention number of turns in sections because you will see later that it turned out it doesn't matter ... it may as well be closely wound without sections). Then followed the measurements of serial and parallel resonances until I got serial-resonances not falling close to the amateur bands. At the end it got around 135uH which is already sufficient inductance for correct operation of amplifier ... I have seen amps with a lot smaller chokes. Reactance of the plate choke should be at the lowest operating frequency at least 4 to 5 times higher than the impedance of the point at which it is connected, and this choke at 3.5MHz has a reactance of nearly 3Kohms which is not even a full two times of 1600 ohm of two tubes plate output impedance. Then it occurred to me to try to put inside of the choke former a 12cm long and 10mm thick ferrite rod used for medium-wave antenna from old portable receivers. Logic was: if people are using them in a broadband balloons for antennas in HF range and push 1-2KWthrough them, why could they not be used here.

    The Ferrite rod is positioned in the center of the coil former and to prevent movement, I made two small "toroidal" shaped pieces of hard foam and pulled them tightly over the both ends of the ferrite rod and shoved it all together in the hollow choke former. I did not touch the turns of the choke and the addition of ferrite raised inductance to 980uH and now there is nowhere any resonance throughout the entire HF range and reactance at 3.5MHz is 21.5 Kohms! At full power and on any band plate choke is cold as is the ferrite rod in it.
    The next interesting detail is the plate suppressors, which were(together with plate chokes) the most common pain in the a** when building tube linear. The most common mistake is the use of copper or even silver-plated wire or strip! Suppressor is the component that should thwart the onset of VHF parasitic oscillation, and whenever you want to reduce or disable some oscillations, you are reaching for some form of reduction in Q-factor in the circuit, and by use of copper or (even worse) silver wire or strip you are just doing the opposite – you are increasing the Q-factor and consequently increase the chances of self-oscillations. Probably the majority of people are afraid that if they don't put highest quality wiring, they will lose significant part of the power that will not be transferred from the anode to the output PI section. This is simply not true and the loss, if there is any, is in practice negligible.

    I used 1.2mm thick "constantan" resistant wire, although with the same success one can use Nichrome or Cantal. The difference is that neither the Nichrome not Cantal can be soldered (at least not by our standard way - there are special lots for these materials) while the constantan is perfectly solderable. I just made a loop in the shape of the letter "U" or “hairpin loop” with a total length of 10-12cm before folding. Wires are folded around in the middle over a body with diameter of 20-22 mm. It shows the same results if suppressor is as a classic coil wound with 3 turns on the body diameter 7-8mm but stretched so that length is about 20-25 mm. These things are not at all so critical so they don't have to be copied to the tenth of a millimeter exactly. Parallel to these coils I soldered the 3 carbon composite resistors of 150 ohms / 2W each to get about50 ohms 5-6W. That also is not critical because this value can range from 33 to 75 ohms without any noticeable difference in operation, still I rarely go over 50-56 ohms. I never put a resistor inside suppressors coil and certainly not wind the coil on them. Be sure that the suppressor coils are placed so that they are as far as possible from each other and in particular that they are not parallel.

    TO CONTINUE >>>

    Pictures of front panel...



    Attached Images Attached Images
    To have friends, it is to accept that there are more beautiful, smarter and more capable people than you. Who can not accept that - have no friends!

  2. #2
    Active User New Member YU1OXB's Avatar
    Join Date
    Jul 2016
    Location
    City of Cacak, West Serbia
    Posts
    13
    Schematic Diagram of MF-1000 Amplifier


    Click to enlarge

    Full Size Version in PDF format is in the Files Documentation Section at the top bar for registered members use.

    Attached Images Attached Images
    To have friends, it is to accept that there are more beautiful, smarter and more capable people than you. Who can not accept that - have no friends!

  3. #3
    Active User New Member YU1OXB's Avatar
    Join Date
    Jul 2016
    Location
    City of Cacak, West Serbia
    Posts
    13
    Cooling of the lamps is a story for itself and thinking and weighing what is the most optimal, took a lot of time. I saw so many solutions on the WEB but I got the impression most of them were not designed "to the end" – at least to my taste. Most of these types of lamps need forced cooling and some of them a really GOOOD cooling, and this all leads first to some kind of turbine. This again means that you have to count on pretty high audible noise which is at least undesirable and quickly begins to bother the operator and is difficult to eliminate from transmitting audio signals for voice work. Some resort to replace the original cooler with different coolers of larger area and efficiency but it again requires bigger powerful fan. Quietest solution is a massive Al heatsink with high fins and then you do not need forced cooling, but to make it a good solution, the tubes have to be screwed directly to the heatsink (and you need two separate heatsinks if there are two tubes) and heatsinks must be outside. There is a new problem arising, because then the whole heatsink is on the plate potential, ie. 2000-2500V DC and that situation is not naive.
    I've solved it in a way that to me seems the most appropriate and effective. I used 75mm outer diameter PVC sewer pipe, those ends where one pipe connect with next one. I managed to get two very strong and at the same pretty quiet fans 80x80mm, which are quite different from standard PC needs. They are American production and therefore for 115V AC so that the two connected in series are just right for 230V AC. Each fan is pushing some 34 cfm (cubic feet per minute), which is quite enough for GI-7B. For them, I made two rectangular pedestals that I exactly machined to fit on tops of PVC"chimneys" and push all together over the tubes to the chassis. Original star-like tube heatsinks are not touching chimneys because everywhere in the circle around them is about 5 mm gap to the inner surface of chimney. Protrusion at the beginning of the pipe (in which goes inside rubber seal ring, in normal use of those pipes)served nicely as a support for the three "claws" for clamping and fixing the chimney to the chassis.
    GI-7B's also need partial cooling of the cathodes of tubes but it looks most builders either do not know or do not care about it. Yes, the tube will work without cooled cathode but how reliable and for how long... The manufacturer states that the max. temp. of cathode is 100^C but without additional cooling it quickly exceed that value -confirmed by measurements! That's why I “continued” chimneys from anode part (above chassis) also under the chassis where the cathode section is, and thus I think very effectively forced all the air that fans are moving, must first pass over the cathode than through the three oval holes on the chassis around the grid and then through the anode star-shaped heatsink. There is no“unused”air coming from any side which hasn't cooled the tubes, so I have maximum cooling that those fans can produce. During SSB operation heating is very moderate but during the tests which are mostly done with constant CW signal on the "key-down", heating is the greatest. However, after two-three minutes of key-down signal at full power, temperature immediately above the fans does not exceed 65-70 ^C which is more than satisfactory (to me!).

    Cathode choke, through which the current for heating goes, is made more or less standard and is wound bifilary with 1.5 mm wire closely wound over the entire length of the ferrite rod and it's, in my case, 12 cm rod. Try not to compromise too much here by putting it on a smaller rod like 5-7cm to save space ... it'snot irrelevant! Reactance of the cathode choke for HF is practically connected in parallel to cathode input impedance and can and will affect the parameters and settings if it is not sufficiently high. My choke has measured inductance of 36uH and I think it can be as low as some 25uH which would at 3.5MHz gave reactance of about 550 ohms therefore ideal ... around 10-11x expected input impedance tube(approximately 40-50 ohms for two GI-7B). Any value greater than this is only for the benefit! If you really don't have enough space, feel free to take two or even three ferrite rods of 5-7cm, wrap them together to be as one core and then wind your choke over it. It won't be worse solution than the one long stick.

    However,the main "game" was taking place around the input and output matching. When amplifier in GG topology is in question,EVERYTHING MATTERS AND HAS INFLUENCE so that any change of conditions in the output section - changing power, SWR, tuning- "transfers" on the input impedance and matching. All adjustments-tuning should be done at maximum output power and best alignment of the output filter. So it is not good to rely on the adjustment of the ATU (in transceivers that have them installed),because it is usually factory set that the ATU adjusts with the power level of about 5W and in some radios 50W. Then, after you tune the output perfectly and on full output power, then set of the condition in which ATU stopped as tuned, no longer corresponds to the optimal setting for good proper match to the linear input.

    There is a lot of controversy on the WEB regarding determination of the output impedance of tubes in linear amplifiers and they all defend their ways, some will "swear to god" that this is the only right way. The best test is the practice itself, and in my case nearest to the real result is formula Zpl= Upl / 1.6 x Ipl. The values ​​of the components measured after setting the finished device are such as to indicate the impedance very close to that obtained from formula.In my case, high voltage "idling" value is 2500 V and on full power this fall to 2100 V (power supply was done as voltage doubler – that's why the voltage sags so) and the average max anode current of about 800-840mA, and calculations at the end gives the output impedance 1562 ohms and real is about 1600 ohms for two tubes in parallel. The exact design data for output coil are given on the schematics. For the range of 7 MHz – 30 MHz I used self-supporting air coil made by the thin-walled copper tubing of 5mm outer diameter and wound on the body with diameter of 50 mm. After winding the turns will "relax" slightly but the average diameter (center-to-center of tube) is 57mm and it has 13 turns. The spacing between turns is 8 mm (center-to-center of the tube).
    For 80 m band I decided to do a coil on a toroidal core instead of the classic large air coil. I used two stacked Amidon toroids T200-2 (or one T200A-2), covered them before winding with three layers of Teflon plumber sealing tape and rolled coil over that. Data for it are also on the schematics. The coils made this way occupy significantly less space and are functionally the same as the self-supporting air wound ones. It is also less prone to interact with nearby components and it almost doesn't radiate compared with air wound one.
    Input matching is also pure classic – Pi filter to reduce harmonic content of incoming signal (probably won't do much, for, it is preceded with much better and more effective filter of the output stage of radio) but mostly to match 50 ohms input impedance to the cathode equivalent impedance. As I mentioned above, EVERYTHING MATTERS AND HAS INFLUENCE and cathode input impedance is not a steady figure but very dependable on several things and specially on the output power level (which to a certain extent determines output impedance) and proper output filter tuning to match that impedance to the 50 ohms of transmission line and antenna system. Amateurs won't be amateurs if they don't try to “cut corners” believing that they will save significant money that way – and that's frequently false. When you are doing something commercially, saving of 10-20-30$ on parts and labor might look significant on 1000 or 5000 units but when you are doing just this one unit, I do not see a point … you are doing that for yourself and consciously avoids best solution just to save few bucks?! Will that make you happy – to save 20-30 $ but knowing you have “crippled” unit that is supposed to be your pride?! Ridiculous – at least to me! I saw many attempts to make wide-band solutions but they are always more-less big compromises regarding quality of matching and transfer of energy from radio to the linear.
    In my project, I've done separate input matching Pi configured filters for almost every band. Where you can save few bucks is to use the same filter for two nearby bands and that is possible only with upper bands. It turns out that 10 m filter is more than adequate for 12 m band too, with SWR raised just up to 1:1,2 and that is perfectly acceptable for any radio.
    Attached Images Attached Images
    Last edited by YU1OXB; 07-25-2016 at 05:59 AM.
    To have friends, it is to accept that there are more beautiful, smarter and more capable people than you. Who can not accept that - have no friends!

  4. #4
    Active User New Member YU1OXB's Avatar
    Join Date
    Jul 2016
    Location
    City of Cacak, West Serbia
    Posts
    13
    Also, 15 m band filter is usable for 18m band with SWR deteriorating to 1,4 – 1,5 compared to 15 m where the needle is practically not moving. All modern transceivers are perfectly happy with 1,5 SWR on their output and will deliver full power under such condition. For 20 m, 30 m, 40 m and 80 m I've done separate filters for each band and tuned them to almost perfect SWR. It is a tedious job but well worth the effort and time. It is important to do all this with the same piece of coax between radio and QRO because that has influence on input tuning and SWR figure. Once you have your input filters matched you actually both match impedance and compensate for various existing reactance in cathode circuitry as well as in coax between radio and QRO. So first estimate what is the most convenient layout-setup of units on your desk and make linking piece of coax to reach comfortably from transceiver to the QRO and than do all your tuning with that piece of coax and do not change it ever. Try not to make it too long... 1 to 1,5 m (3,5 - 5 feet) should be all you need... the shorter – the better. Switching is done with two small 12 V DC relays for each band filter and I have small 12 position wafer on the extension of my main band switch shaft. One can have separate small switch for that purpose, but then you have to (do not forget!!!) turn two switches every time you change the band of operation. That's why my solution is making less headache... when you turn main band-switch you have your input matching properly and simultaneously switched too.

    There are two analog meters on my front panel, one is cross needle POWER/SWR meter connected with "Sontheimer bridge" type of sensor at the output of the unit, and “control meter” wired to measure 4 different parameters I found interesting to monitor. To my opinion, grid current is the most valuable parameter to monitor in linear amplifiers, for, it will better than any other show you when is your amplifier properly tuned. By watching grid current you are looking for a peak = maximal needle deflection when you adjust PLATE capacitor, and that is exactly opposite from adjustment using plate current as indication when you are looking for dip in current. To adjust LOAD capacitor using grid current as indication, look simultaneously for maximum power on POWER/SWR meter and slow reduction of grid current. Once you reached max power always “open” (reduce capacity) your LOAD cap just slightly more and you should notice very slight reduction of power (probably not more than 30-50W or so) and also reduction of grid current, and then you loaded your tubes properly. Grid current will also show you if you are overloading your amp with excessive drive from your transceiver. You will notice that as a significant and abrupt raise of grid current which is not followed with proportional jump in output power. Reduce your drive to just under the point when that start happening, having on mind never to go above maximal allowable grid current of your tubes.

    Don't squeeze every possible watt from your unit! That looks important just to our minds but in practice you will neither sound stronger nor will you make S-meter to show more on the corespondent Rx if you have 1100W compared to 900 or 1000W. One S-unit is nominally 6dB and that means to go better 1 S-unit you would have to raise your power 4 times (6dB = 4 x Power) hence, 100 or 200W difference means NOTHING, but can bring a significant stress and heating of your components jeopardizing overal reliability, as well as much bigger chances to become annoyance to everyone on the band.

    POWER/SWR meter sensing element has been done as a Sontheimer bridge which is probably better known as “Tandem Match”. It consist of two toroidal transformers with ferrite cores and for this power level I used 2 stacked FT-114-61 toroids for each transformer. Instead of two stacked cores you can use one FT-114A-61 with same results. Wrap cores with some Teflon tape prior to winding, and then put 30 turns of #22 AVG wire on each transformer evenly spread over about 80% of circumference. This type of bridge is very reliable, precise and frequency independent if done properly with proper screening between transformers and around them to reduce capacitive coupling and erroneous readings. It will safely operate even slightly over 2KW mark.
    Last edited by YU1OXB; 07-25-2016 at 06:07 AM.
    To have friends, it is to accept that there are more beautiful, smarter and more capable people than you. Who can not accept that - have no friends!

  5. #5
    Active User Contributing Member KD8MJR's Avatar
    Join Date
    Aug 2014
    Posts
    332
    Blog Entries
    1
    Excellent work my friend. Thank you so much for sharing this with us.
    Please keep us updated and if possible could you post a larger picture of the schematic. It is too small to make out the details.

    Thanks do much

    73s
    Rob

  6. #6
    Active User New Member YU1OXB's Avatar
    Join Date
    Jul 2016
    Location
    City of Cacak, West Serbia
    Posts
    13
    I am still learning about the way this forum operate, Larger PDF files will be in Files section under the name MF-1000 Amplifier
    Last edited by YU1OXB; 07-27-2016 at 05:18 AM.
    To have friends, it is to accept that there are more beautiful, smarter and more capable people than you. Who can not accept that - have no friends!

  7. #7
    Active User New Member YU1OXB's Avatar
    Join Date
    Jul 2016
    Location
    City of Cacak, West Serbia
    Posts
    13
    Some additional files for potential builders...

    PDF files of PCB-s are ready but for precise printing purposes they are in PDF format with 4000dpi resolution and 1:1 scale. Unfortunately, forum do not accept such type of files-extensions and I don't know how to make them available here. Other formats would not be precise enough for decent printing.

    Components layout (components side view)



    Top layer (components side view)



    Bottom layer (solder side view)
    Last edited by YU1OXB; 07-27-2016 at 07:37 AM.
    To have friends, it is to accept that there are more beautiful, smarter and more capable people than you. Who can not accept that - have no friends!

  8. #8
    Active User New Member YU1OXB's Avatar
    Join Date
    Jul 2016
    Location
    City of Cacak, West Serbia
    Posts
    13
    and some more....

    HV supply board layout


    Linear MAINS layout



    Linear IN-OUT layout



    Linear Input Matching board layout



    >>> Again, PDF files for those PCB-s are ready but forum have to find way to store them and provide download capabilities.
    Attached Images Attached Images
    To have friends, it is to accept that there are more beautiful, smarter and more capable people than you. Who can not accept that - have no friends!

  9. #9
    Active User New Member YU1OXB's Avatar
    Join Date
    Jul 2016
    Location
    City of Cacak, West Serbia
    Posts
    13




    Attached Images Attached Images
    To have friends, it is to accept that there are more beautiful, smarter and more capable people than you. Who can not accept that - have no friends!

  10. #10
    Active User New Member YU1OXB's Avatar
    Join Date
    Jul 2016
    Location
    City of Cacak, West Serbia
    Posts
    13
    Two short videos of first power on and tuning of finished amp.

    First part - https://youtu.be/svZPtKFAijQ
    Second part - https://youtu.be/wWOcFjz8l6U
    To have friends, it is to accept that there are more beautiful, smarter and more capable people than you. Who can not accept that - have no friends!

Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •