by Larry "Tree" Tyree - N6TR / K7RAT
On the fourth of July, 1997, I took delivery of a home-brew amplifier built by David Bertman, AB7B. This diary is intended to document the history of this amplifier as it unfolds while it is in my possession.
The amplifier consists of three major components: RF Deck, HV power supply and Screen/Bias supply. The three units stack on top of each other resulting in a rack about 4.5 feet tall. A wooden platform with casters was provided so the amplifier can be moved. Interconnections between these units are made with cables with octal connectors - except for the HV to the RF deck which uses standard HV connectors.
The amplifier initially was configured for class C operation on 80 to 10 meters and could be driven to a kW output with about 5 watts drive. It was my intent to eventually modify the amplifier to work on six meters for SSB and CW operation.
RF Deck - If you look in the back of the RF deck, you will see the two 4-250As on the left side of the cabinet. In the middle is the plate tuning capacitor, and on the right is a wonderful looking B&W 850 tank coil which includes an integrated bandswitch. Underneath the chassis is the loading capacitor and input circuit and filament transformer (I haven't actually seen this yet as I haven't taken it out of the cabinet yet). A small squirrel cage fan is hanging on the back and there is currently no provision for TR switching for the input/output. Looking at the front panel there are two meters, Grid current and Plate current. There is a rotary switch marked TUNE, PHONE and CW. I am told by David that the TUNE position will disconnect the screen voltage and ground it. This can be used for "QRP" operation.
HV Supply - The HV supply receives the primary power and interfaces to the RF deck and the Screen/Bias supply. Currently the HV transformer runs off of 120 VAC, but it appears there is another tap which hopefully can be used for 240 VAC operation. There are two meters on the front panel - HV voltage (0 to 3KV) and Screen Current.
Screen/Bias Supply - Two separate supplies are contained. The Screen supply uses a 5Y4 rectifier tube and a choke input filter network with a 25K Ohm bleeder resistor. It is hoped that this supply will be stiff enough to use the amplifier on SSB without any additional regulation. The bias supply also uses a choke input filter network, and a VR150 is used to develop the bias voltage. A 1.25 K grid leak resistor is included (I need to remember what a grid leak resistor is for again - I used to know this), and the bias current on the VR150 is set around 8 ma (if I am reading the documentation correctly). Some of the documentation I have seen has a procedure for setting this current which strictly involves watching the color of the tube.
4 July 1997 - The amplifier was moved from David's QTH to its new home in Boring, Oregon. This was done in two separate vehicles with the RF deck and HV supply in my car and the screen/bias supply in David's. David also supplied a spare set of tubes, and a few spare parts for the RF deck. It was decided that the addition of the amplifier really made the N6TR/K7RAT station look much more like a ham radio station. We left the power cord at David's, so we won't be able to power it up today. We take off to McDonald's for lunch.
6 July 1997 - I finally took some time to look over the documentation David provided. I identify four projects to perform on the amplifier:
My six meter station consists of a DownEast microwave transverter, which can output about 20 watts. Currently I am feeding this to a Teletec "brick" which puts out 180 watts. My plan is to drive the pair of 4-250As directly from the transverter.
7 July 1997 - Finally got my hands into the amplifier. I removed the Screen/Bias supply from the cabinet and looked it over. Everything was pretty straightforward. There were two sockets for voltage regulator tubes, but one of them was shorted out. David had put two - 3 position rotary switches on the front panel. One of them could be used to select different bias voltages. It appeared to me that the thing to do was to put a lower voltage VR tube in the unused socket and use the switch to select the proper voltage for AB or Class C operation. A voltage divider might need to be added to allow some adjustment for fine tuning the idling current for AB (I don't know if I should be saying AB1 or AB2 - so I will just say AB). It looks like I should go for about 60 ma per tube of idling current. It looks tempting to replace the tube rectifiers with solid state diodes, but I make the decision to keep them. Same decision with respect to VR tubes versus zeners. I decide to keep as much of the original "glow in the dark" appeal of the amplifier. This philosophy will hopefully be used when making future decisions. The idea is to change as little of the circuit as possible. I think about the fact that "instant on" for the bias might be a good thing to have, but decide to wait and see if this becomes necessary.
8 July 1997 - Dropped by David's to pick up the power cord. He also digs around in his attic and provides me with a OA3 VR tube (105 volts) and a VR150 VR tube (150 volts). One schematic I found in the 1967 handbook uses the OA3 for AB biasing - so maybe that will work out. He also loaned me his box of relays so I could pick and choose any I needed to switch the bias supply (when on receive, the ground of the voltage regulators needs to be lifted so the tubes will be cutoff). I will need to add a new wire to the cable to allow a STANDBY signal to come into the supply. I figure I will make this +12V when on transmit, and use the normally closed contact on a relay to ground the voltage regulator string. I might buy a relay from Radio Shack instead - along with a socket, and see if I can mount that (good chance to use my nibbler which has been resting for about 10 years now). This is my chance to leave my mark on this chassis. I also decided to start this document as I feel the story will be interesting to many people. The joy I have experienced at having this amplifier in my house surprised me. It is like I have made a connection to my childhood and finally own a piece of equipment that was just a dream when I was a kid. I remember looking at similar amplifiers in the handbooks and magazines I had when I was a teenager, and they were never realized. I hope that this document will be interesting to those who had similar experiences.
9 July 1997 - On my way home last night, I stopped at R5D3, a small electronics surplus store. While the name sounds like a character from the Star Wars movies, it comes from the two initials of the couple that own it, and the number of kids they have (or something like that). They had a great selection of voltage regulator tubes and I picked up a few "just to have". I also found two nice ceramic insulators, about 3 inches long, with screws on each end. They are about a half inch in diameter and might come in very handy during the RF deck modifications. I also stopped at Radio Shack and bought a couple of relays, and some hook up wire (solid stuff - 3 colors). After the kids were in bed, I got out the drill and made my first hole in the chassis of the bias/screen supply. I used a nibbler to make a rectangular hole for the relay socket and a file to make it look neat. It came out really well and the relay was mounted. Then I rewired the bias supply to use two tubes instead of one (the socket for the 2nd tube was already there - I guess David had two at one time, and then went to one). I used a VR105 (a.k.a. OC3) to develop the bias for AB1, in series with a VR75 (OA3) to add 75 volts to that to get the bias for class C. One of the rotary switches David had mounted was used to select which of the two voltages goes to the grid leak resistor. All that is left to be done is to work out the details of the control voltage for the relay. It connects the bottom of the VR tube stack to ground when transmitting. When receiving, this leaves everything floating up to full voltage of the bias supply, making sure the tubes are totally cut off and not generating any noise.
I was happy to find the OC3 from my first transmitter as a novice (built from the 1967 handbook 30 years ago) worked just fine and will be used in this amplifier. It was great to be able to tie these two projects together after such a long time. I rescued this tube from the garbage truck at my mom's house just a few years ago.
Defensive design tip #1: I made the 3rd position of the bias voltage select switch work just like the second position, so there wasn't any chance of ending up with no bias. I also want to add a resistor to pull the output "down" in case the switch is left in an open position by mistake, or to handle the brief period of time when the circuit is open when hot switching the bias. I figure a 100K Ohm resistor from the -180 volt output to the grid leak resistor will take care of this. Hot switching is only an issue when you are transmitting, but with little kids running around, and with this switch only a few inches off the floor, I wanted to protect against any potential accidents (note the clever dual meaning of the word "potential").
The bias supply produces -110 volts in the AB1 position, which looks about right when I look at the tube tables. It produces -180 volts in the class C position, which also looks right. The basic power supply produces -322 volts, and the VR tubes are biased through 20K Ohms. This works out to a bias current of 7 milliamps. The VR105 glows the expected blue color, but the VR75 is a much brighter orange (neon looking). I guess a different gas was used for this lower voltage.
10 July 1997 - Spent a little time with the amplifier last night. I added a 200K Ohm - 1/2 watt resistor from the -180 volt position of the bias switch to the grid leak resistor. This should prevent the bias from going away when moving the rotary switch from one position to the next. I also added two wires to control the relay. There are two unused pins on the octal connector that goes to the HV supply, so this is where they will go. The modifications to the screen/bias supply are now complete, so I put it back into its cabinet. The screen voltage came out to 550 volts. That means there is 22 milliamps flowing through the bleeder resistor. Hopefully with the 6 Henry choke and 7 microfarad filter cap, this will be solid enough to use on SSB. If not, perhaps adding more capacitance would help. A pair of 500 microfarad, 350 volt electrolytic caps in series would certainly improve the regulation.
I got bold and hooked up the power cord so I could see the filaments light up. After a minute or two, I built up enough nerve to turn on the HV switch. After a 20 second (or so) delay, the HV came on. I was interested in seeing how close to 3KV the supply was. The supply came on okay, but there was some arcing going on in the HV supply somewhere. The meter would flick down with each one, and the frequency was about 1 or 2 per second. Perhaps my high line voltage has pushed the amp to the point where one of the filter caps is breaking down. I removed the HV lead to the amp, and it was still there, so it is in the HV section.
I let it run for a few minutes to see if it would work itself out and stop, but it didn't. Looks like the next step will be to take out the HV chassis and figure out where it is coming from. This wouldn't be a fun project if there weren't a few things to overcome like this. I view it as an opportunity to become familiar with the hardware and possibly make some improvements. The idea of redoing the HV section to allow higher plate voltage has always been in the back of my mind, and perhaps this is the harbinger of that effort.
I didn't bother energizing the bias relay to see what the idling current was like.
15 July 1997 - After a busy weekend being a part of the W1AW/7 effort, I finally got rested enough to work on the HV supply (being sleep deprived and HV don't mix). I removed the chassis from the cabinet and turned it on. The sparking was occurring in two places: on the octal connector that goes to the amplifier, and at the base of the "screen protection" relay. This relay is powered from a resistor in the bottom of the bleeder circuit to make sure the HV is turned on before the screen voltage is passed on to the RF deck. At first, I thought perhaps the resistor used to develop the voltage wasn't down towards the "ground" side of the bleeders, but this was done correctly. What I did find is that the negative side of the B plus doesn't go to ground. Instead, it goes through a 50 Ohm resistor, then a one amp fuse and then ground. Checking the fuse found that it was open, so the HV circuit was floating and somewhere there was enough leakage to have a high voltage (with respect to chassis) build up on the B- side of the supply. When the zap occurred, the HV meter flicked went down, but probably that was a result of the ground potential shifting with the energy of the zap being absorbed by the chassis. The other arc confirms that major amounts of current were being conducted, as it was a bypass capacitor to one of the control signals going to the RF deck. The bypass cap had broken connection (but was still very close) and was arcing some of the time (when the other arc had a big discharge).
Today I will be thinking about how to prevent this type of failure when the fuse gets blown. Perhaps a high value of resistance between the B- to ground would prevent the build up - but prevent much current flow if something shorts out in the RF deck.
Defensive Design Tip #2: When using fuses, make sure the circuit behaves in a controlled manner if the fuse is blown.
18 July 1997 - I made a posting to the "amps" reflector (amps@contesting.com) and asked about the best way to resolve the HV problem. I received many answers, and after sifting through them, I have decided to do the following:
I believe that will address all of the concerns I have with the supply and keep much of the existing circuitry.
I thought long and hard about where to put the fuse on the B+ side of the circuit. I wanted a blown fuse to turn off the screen voltage, so the existing bleeder circuit had to be after the fuse (since the current in the bleeder is used to turn on the Screen voltage relay). I also didn't want to discharge the capacitors through a HV short, so I wanted the fuse after the capacitors. This required adding a new set of bleeder resistors directly across the capacitors. I will probably go with a fairly high value here - maybe 100K per capacitor (there are two of them).
Now, if the fuse blows, my HV meter will go to zero indicating something is wrong. Currently, I have no indication of a blown fuse, except the arcing.
21 July 1997 - Modifications made to the HV supply. There is now a fuse in the positive HV section (made out of a strand from zip cord), after the caps, but before the bleeder resistor. Need to buy some resistors for bleeding the caps in case the fuse is blown - couldn't find anything in my junk box that was appropriate.
During the modification, I found the filter section to be wired differently than the schematic. The schematic showed 8 microfarads of filter cap, with the negative side connected after the inductor. However, it was wired up as a pi circuit, with 4 microfarads before the inductor and the other after. This was probably done to increase plate voltage. I wired back up with all 8 microfarads on the output, and did notice a reduction in the no load voltage (maybe 300 volts less). However, I expect the regulation and AC rejection to be better and intend to leave it that way for now. If I decide I want more voltage, I have another plate transformer that will provide more voltage. It looks like it will fit with a few modifications to the chassis. The filter caps are rated at 4KV (as are the tubes).
I also need some DC for the TR relays (12 volts). I have a 24 VDC supply (power one) in my junk box that probably can be turned down to 12 volts (or about 16 which is close enough). I was happy to find the supply can mount on top of the existing HV choke which has holes exactly in the right place.
So the next step is to mount this supply, get 120 VAC to it (when the HV is turned on), add a PTT jack to the back panel, and then wire up the connectors to the other chassis' so they will be able to control their relays (the bias cutoff relay in the screen/bias supply and the TR relays in the RF deck). When this is done, I should be able to fire up the amp an see how it works.
10 December 1997 - Yes, a long gap. Other projects and life in general were taking their tool on this project, but now that "winter" has set in, there is more time for inside projects. Last night's progress was focused on installing a DC power supply for the three relays: the one in the bias supply (already installed) and the two that will be used in the RF deck (RF input and RF output). An old regulated supply that has been in my junk box for about 20 years was cranked down from 24 volts, down to about 15 volts and was mounted on top of the HV choke in the power supply. A phono jack was mounted on the rear panel of the power supply and the two unused pins on the octal plug for the bias supply interface were used to pass the positive voltage and return for the bias relay. Next comes routing the same signals up to the RF deck and installing the two relays to switch the RF.
The primary power for the relay supply comes from the HV transformer primary, so it will be impossible to activate the relays until the HV has come up in the power sequencing. With one or two more evenings spent on this project, we should be ready to play with the amp on HF before making the six meter modifications.
16 December 1997 - Over the weekend, the two relays were installed in the RF deck. One was mounted next to the RF Output connector and the other near the RF Input connector. A length of RG-58 was used to connect the two relays in the bypass position (normally closed). The previous owner did the switching with some external coax relays. After finding a wire that I had broken in the power supply (preventing the HV from turning on), the relays were working and the amplifier was powered up.
I saw expected results with the two bias settings, hardly any plate current in the Class C position, and about 200 ma in the Class AB1 position. I saw some higher values at first with some grid current, but it turned out to be some instability without a load. On 40 meters, I was able to produce about 600 watts out in Class C, and slightly higher in AB1. These values are probably lower than expected due to the lower plate voltage resulting from my change in the power supply configuration. After I have things working on six meters, I might look into increasing the high voltage by using another power xfmr I have laying around. I did have some funny results on SSB, which might be related to RF in the shack. There can be some improvements made to the shielding of the RF deck (there are a few gaps that need some screws added). I probably won't work on those issues too much until it is working on six meters.
Now my attention turns to changing the RF deck over to 50 MHz. Eliminating the bandswitch. for both the input and output will make it easier to get rid of the extra loading. Some of the old handbooks have examples of 50 MHz amplifiers for which I can copy the details of the input and output circuits.
22 December 1997 - Over the weekend, the amplifier went through many changes. The existing loading capacitor assembly was removed (two very nice E. F. Johnson 500 pf caps ganged together). The B&W 850 tank coil/bandswitch assembly was removed (make me an offer - I even have the original box and the switch contacts look perfect). I also took out the existing parasitic chokes and the old input circuit.
I installed a new tank coil, (2 and a half turns of copper bar, about 1/8" thick, 3/8" wide, with a coil diamter of about 2 inches), a new loading capacitor (something close to 150 pf), the new KM1H "six meter approved" parasitic chokes, and attempted to make a new input transformmer.
The output section seems able to resonante on six meters. I can create an impedance down around 20 Ohms when looking back into the pi network with my MFJ SWR Analyzer. The main tuning cap is near minimum capacitance. It might be that I will need to replace this eventually. I have a nice 100 pf vacuum variable (CCCP surplus no less - out of a "jumming transmitter") which would be wonderful in this application (minimum capacitance of 4 pf).
However, the input circuit isn't working. I am not able to get more than a few watts out of the amp on six meters. There wasn't any plate current to speak of (other than the idling current of about 200 ma when I am in the AB1 mode). Obviously I need to make a better input circuit. My existing one is similar to what was used on 10 meters, with a few changes (obviously not enough changes). There is a winding, about 3/4" in diamter on the input side with two turns, and then about 10 windings for the output with the tuning cap in parallel.
Once I get the input cirtcuit working, I can evaluate the performance of the output circuit and see if I need to do some more work.
Getting close...
Comments to tree@contesting.com and more information about N6TR/K7RAT can be found at http://web.jzap.com/n6tr.