The Solidly Built Drake L4B

Build Your Own Power Supply – One KW or Legal Limit – Your Choice

Tom Sowden K0GKD

I am a sucker for a deal and it sometimes gets me into trouble. Having always admired the Drake line of amateur equipment when I saw a Drake L4B amplifier on the popular web portal at what looked like a very good price I had to make a bid. The next thing I knew I was the winning bidder. Not so fast – does this baby come with a built in power supply? I quickly realized I had not checked out the devil in the details – it was a solo deal without power. Any fool could have seen that the relatively small amplifier could not possibly encompass a power supply.

Okay, no problem – so I will just have to build my own. Not that I need another amplifier – I am the proud owner of two home brew solid-state amplifiers that output 600 watts and one kilowatt, respectively. I had rebuilt several Heathkit SB220’s but this was my first power supply design from the ground up with an unfamiliar model.

I am what some of my friends call a “parrot” – I am good at repeating what I learn from others. I guess in one sense we are all squawkers as one of the wonderful things about the hobby is sharing ideas with our brethren. I knew the perfect mentor – my friend W0YXM (Dick). Dick is a tube and amplifier guru of the first order. Dick has forgotten more then I will ever know about tube rigs!

When the L4B arrived and I carefully unpacked it and had a look. Wow, this rig is very well built with a huge variable capacitors and a big tank coil. No question that it is a much beefier amplifier than the Heathkit SB-220 (not that there is anything negative about the old boat anchor!). The parched and worn manual was included so I started to study the power supply design that Drake engineered.

My persona is such that I have to understand what is happening in any circuit before I can make changes. This project was no different than others that I have taken on in this respect – know the basics and you can work your way through the trouble spots. At first blush the Drake power supply schematic (Figure 1) looked conventional with a voltage doubling circuit that is common with most commercial tube type amps. For those not versed with this dated technology it is important to follow the electrons around. Basically two banks of four capacitors are hooked up in series and are shunted with equalizing resistors. During the positive half of the duty cycle the first string of rectifier diodes conduct (shown in red – note the return path through the remaining transformer secondary lead) and charge the first four capacitors). The voltage divides more or less evenly across each individual electrolytic. When the duty cycle turns negative the other set of diodes conducts (shown in green) charging the second bank of four electrolytic capacitors. Now you have two sets of four capacitors that are charged up, in most designs to about 1,500VDC. Since the two sets of four are hooked together in series they combine to result in a total DC voltage of 3,000. A pretty slick trick to enable transformers with secondary voltage ratings that are one half of what would otherwise be required.

Figure 1

Engineers usually chose the secondary of the transformer output voltage for these types of applications to be about 1100 to 1200ACVRMS. Because the values are “root mean square” ratings they do not allow for peak voltage on the high end of the 60-cycle sine wave. A good rule of thumb is to multiply the RMS voltage rating by 1.3 to 1.4 to get the peak value. The electrolytic capacitors will charge nearly to this peak level or about 1,500VDC, assuming the secondary voltage is as noted. Since this voltage divides evenly across each of the eight capacitors they require a rating of at least 400 volts (1,500/4). A good safety factor would augur for selecting higher ratings like 450VDC (a common rating).

In voltage doubling circuits one needs to divide the high voltage (HV) transformer amperage rating by two. For example, lets assume you need a continuous current rating of one amp – that is key down, not intermittent. This will require a transformer rated at two amps. While this was the benchmark that I used in searching for a HV transformer for the project to get to the “legal limit”, I ended up using a one that was rated considerably less at 1.3 amps. I chose it more on the basis of costs and availability. I considered that this would give me ample power output with the L4B. After all 650 milliamps at about 3,000VDC will easily result in over a full kilowatt out the door, which in my book is more then enough power.

In the Drake design I did not like the combination voltage divider/bleeder resistor circuit (R9, R10 and R11). The L4B engineers used the tap between R10 and R11 to pickoff 100VDC to act as the shut-off bias in the amp during the receive mode. My first thought was “what happens if R11 fails?” The scary answer is that you put 3,000+VDC on the cathodes of the 3-500Z/s. In further examination of the schematic routing the B minus to the amplifier through pin 6 to the amplifier raised further concerns. Under this setup the B minus is terminated in the amplifier through a one-ohm resistor (for metering), which is okay unless the resistor fails and leaves a negative 3,000 -VDC potential without a home.

In thinking these concerns I decided to make some modifications that would deal with these potential problems. First by putting a one or two thousand ohm five-watt resistor between pins six of the power plug to ground the B minus would always find a ground path in the unlikely event that the one-ohm meter resistor inside of the L4B would fail. This resistance in parallel with the one-ohm resistor would not alter the meter accuracy enough to matter.

In my many ham projects I did not have much experience with bias issues of grounded grid triode amplifiers. I needed to educate myself so that I would better comprehend the flow of electrons from the tube’s anodes through their cathodes. The 3-500Z, like so many triodes of this design, are “zero” bias tubes. Basically this means that one does not have to provide positive bias on the cathodes in the no drive transmit mode to keep the idle current at a reasonable level (this assumes the high voltage is below a certain level – I believe the 3-500Z’s are limited in a zero bias state to 3,500 VDC). However, this is a subjective thing as to what is a “reasonable level”, and it has more to do with the plate voltage than anything else. I knew from some of the L4B blogs on the Internet that some folks were not too happy with the high idle currents typical of this amplifier.

In the L4B design one set of the relay contacts is used to place the 100VDC obtained from the power supply voltage divider circuit (R9,10 & 11) through the center tap of the filament transformer to the cathodes of the two tubes. In the standby mode this voltage provides enough reverse bias to completely shutoff any current flowing in the two 3-500Z’s. In the transmit mode the relay contact removes the 100VDC bias and grounds the center tap of the filament transformer via the one-ohm meter resistor. Now the positive electrons coming off of the anodes have a home, and the tubes begin to conduct. In engineering jargon this is the zero bias setup, where the tubes bias themselves in a “no excitation” state. The amount of current the 3-500Z’s will draw is dependent on the plate voltage. At about 3,000VDC they will typically conduct about 250 milliamps of idle current – too much in my book.

To reduce this to a more efficient level I needed to incorporate some form of restricting bias. It helps to think about the electron flow to make this happen. Consider that the electrons want to surge from the anode through the cathode in search of the B- via the filament center tap and one ohm resistor. In order to restrict this flow some form of automatic reverse bias would be needed. My first thought was to use a circuit similar to the one Heathkit engineers designed into the SB-220. They used a 5.5V Zener diode in series with the cathode return path through the center tap of the filament transformer. The avalanche breakdown of the diode would keep about the same voltages as its rating (5.5V) on the cathodes of the 3-500Zs in the transmit mode, reducing the idle current to about 200ma. I would still need a shutoff positive voltage in the standby mode. In discussing this with Dick (W0YVM) he raised a lot of interesting alternatives. After about a one hour QSO on 40M I begin to understand where he was going. His first comment was “you do not need a separate supply to bias the tubes to a shutoff state in the standby mode”. Instead he suggested placing a 47K resistor in series with the filament center tap that would provide a near complete cutoff of current in the idle state. The resistor would need to be shorted out in transmit, which could be done with the third set of terminals on the T/R relay (these terminals were used in the original design to place the 100VDC on the cathodes in the receive mode). He liked the idea of the Zener diode, similar to the Heathkit SB-220 model, to provide better current regulation in the transmit mode, and suggest a slightly higher value of 8.5v.

After some thought I came up with the design noted in Figure 2 below, and emailed a copy to Dick for his comments.

Figure 2

From the above schematic you can follow along as the B+ from the anodes finds their home. Basically the highly charged positive electrons emitting from the tube’s anodes arrive at the cathode and need to find continuity to the B minus. The first obstacle theses guys run into are the 47K (5 watt) resistor, then the 8.5V Zener, the one amp fuse, and finally the B minus routed to ground through the one-ohm meter resistor. Initially there is no current flowing until the Zener starts to conduct, which immediately draws current until the voltage reaches its threshold of 8.5VDC. The current flowing through the Zener diode (to the center tap of the filament transformer and home through the one-ohm resistor and the B minus source) causes a significant voltage drop across the 47K resistor parked in front of the Zener. This shows up as a positive voltage differential on the cathodes, and is sufficient to stop the current flowing from the anode to the B minus source in standby mode (there is a very small amount of current but it can not be measured with the L4B amp meter). This little trick does the same job as the original design, which placed a positive 100VDC on the cathodes’ to keep the current in check.

Sine the 47K resistor needs to be bypassed in the transmit mode it is shorted out by the T/R relay. Now, during transmit, the positive electrons only see the Zener diode (note: the Zener must be isolated from ground using a Mylar washer and feed through insulator on the mounting!) and it is, by design, providing just enough positive differential on the cathodes’ (8.5VDC) to reduce the idle current to a nominal value of around 100miliamps. This is a much better situation then the zero bias state that allowed the idle current to rise to 250 milliamps or more depending on the plate voltage. The one amp fuse takes care of runaway currents that can destroy components and tubes in the event of a cathode to grid short or some other problem. The 10K bypass resistors on the Zener diode, and the fuse, are there to insure continuity in the event either one fails. This is very important, as you do not want the highly charged positive electrons to not have a home under any condition!

Rewiring the T/R relay was easy, and the two wires hooked up to the relay that route the connection to the B minus are simply hooked up to the one amp fuse as shown on Figure 1. You can test the continuity to ground, as it will show one ohm as it is routed through the one-ohm meter resistor. Test out the T/R relay by hand to make sure the 40K resistor is bypassed in the transmit position.

The bias changes – zener diode, fuse, and 47K wire wound R1

Selecting the transformer for the new power supply was made easy by the availability of a suitable and economical solution from Ameritron. Their unit used in their AL80B has the specifications noted below:

Circuit type: Full wave voltage doubler

No load voltage: 3100 V

Full load voltage: 2700 V

Full load current: .6 A (voltage doubling cirtuit)

Regulation: >13%

Transformer: 26 lb. E-I lamination grain oriented

Since the specifications were based on a voltage doubling circuit the transformer had to have a rating of 1.2 amps. In checking the schematic of the AL80B I knew the transformer had both a 120 and 240VAC primary – perfect! The f.o.b. cost of $129 seemed very reasonable. I purchased the unit directly from them and paid a little extra for freight.

Figure 3

In looking at the schematic of the transformer in the AL80B manual I was pleasantly surprised to see that there were two extra sets of secondary windings – one for the filaments, and the other for a 12/24V source. Since the L4B has its own filament transformer I would not need the filament windings (the AL80B has only one 3-500Z, not two, and would not have the amperage rating to supply two tubes). I could use the center tap of the 24V secondary to provide 12vdc to fire a surge delay relay, as will be explained later.

Selection of the capacitors was part of my discussions with Dick. I had eight 900ufd 450VDC electrolytic units in my parts supply bin so I decided to use them. Dick did not like this idea (he had a fit – too many joules of kinetic energy!) at all as he felt the net capacitance of over 110ufd (900 divided by 8) was too high. He suggested I would be better off finding ones that would average down to 30ufd or 240ufd capacitors. When I told him that I had rebuilt a HK SB220 with similar values and operated the amplifier for several years without difficulty he acquiesced providing I would incorporate a safety resistor in series with the B plus. This would avoid a massive discharge of the considerable joules of stored power if the B plus were to short out.

With this amount of capacitance a “soft start” method was required due to the large current inrush that would occur when the power is applied. I found a suitable 12vdc relay in my junk box that would make a good surge delay device.

Figure 4

Lets review how this simple soft start relay works. When the power switch (not shown) is first turned on the primary of the power transformer sees continuity through the 20 ohm 25 watt wire wound resistor and starts to conduct. Because the high voltage capacitor bank is charging rapidly they demand a lot of current, which is suppressed by the 20-ohm resistor. Since the transformer now has a source of energy the 12V windings on the secondary provide power (through the rectifier) for activation of the surge relay. However, it does not immediately close until the 2,200ufd electrolytic capacitor across the activation coil charges. This takes about one second after which the relay engages bypassing the 20-ohm resistor, providing a direct path to the 220VAC source. The combined time elapse allows the capacitor bank to charge to a level that will not destroy the rectifier diodes with excessive current.

The new power supply

Let’s take a look at the L4B schematic to observe the changes required:

Figure 5

In looking at the on/off and the SSB/CW switches I had a thought. First, by bringing up one leg of the 240VAC source through Pin 1 the filament transformer and fan could be activated without activating the high voltage. To do this I wired the filament transformer for 120VAC and grounded the lead going to Pin 4. Now when the ON/OFF switch is closed one side of the 240VAC is grounded through the transformer and the fan providing 120VAC for their activation. This also feeds the 24V windings on the filament transformer for the relay power supply. Disconnecting the wire from the CW terminal of the SSB/CW switch, and rerouting it directly to Pin 2 provided the return path for the activation of the transformer in the power supply. With these changes I can warm up the filaments with the fan activated, and then close the other switch bringing the high voltage on line.

One drawback from the changes and elimination of the 100VDC is the ALC circuit depended on this voltage for its use. Normally I chose not to use ALC in amplifiers as keeping the drive level down and watching the exciter ALC in my experience allows amplifiers to remain clean. Those wanting this feature could probably work up a different circuit using the 24VDC relay source.

The schematic for the new power supply is shown below, and basically is not much different from the original Drake unit except for the changes noted previously.

Figure 6

I could not find a female plug that would fit on the original Drake eight pin male connectors so I had to provide a male/female substitute. Only four connections were needed (Figure 5). Two to route one side of the 240VAC to the power switch and for the return line off of the CW/SSB switch, one line for the B minus, and one line for the common ground. (Note: as an added safety factor I grounded the amplifier and the power supply separately.) It is a good idea to check continuity on all of the connections prior to the “on” test.

Most of us do not have a volt ohmmeter that will test voltages above 1,000VDC. This makes it hard to check out the power supply prior to hooking it up to the amplifier. One safe method is to use a low voltage transformer on the primary – like a 12VAC filament transformer. This will allow you to check out the basic flow safely – you can ball park the results by multiplying 20 times the output reading to interpolate how the final voltage will appear. If you don’t get the output you expect you can play around with the various taps on the secondary of the Ameritron transformer – I used the red/yellow and red wires which provided 3,200VDC (no load) with a 240VAC source.

Once you have completed the power supply, checked out the continuity, and made sure the B plus line is not showing a low path to ground you can go for the real thing. If you trip your circuit breaker you may have to adjust the time delay on your surge circuit, as this is the likely culprit. If everything is working okay you will see the high voltage on the L4B meter more or less where you calculated it would fall – 3,000VDC plus or minus. You should see no current in the standby mode.

Once you get the driving source hooked up correctly you can key the amp and view the idle voltage. It should be around 100 milliamps depending on the plate voltage, and this will be the sure sign that the Zener is doing its job. I like to set the power output to about 60 watts, as this is sufficient to drive the 3-500Z’s to a full kilowatt. With the driving source in the tune mode you should adjust the tuning capacitor for maximum output, and then adjust the load capacitor to achieve maximum output (this will require another adjustment of the tuning capacitor). Once you get to the full KW you can adjust the load capacitor clockwise to where the power just drops off nominally. This should allow the plate current to grid current ratio to reach its optimum of about 3.66 to one. Running the amp at these settings will extend the life of the L4B – possibly beyond your lifetime.

The L4B with these changes and its new power supply has worked marvelously, and consistently puts out 1.2KW on SSB on voice peaks. Everything runs cool including the Ameritron transformer. This dude will likely go a long, long time in my shack hooked up to my Kenwood TS-940S. Since the power supply is on coasters it was easy to slide it under the operating table – never to be seen sans any problems!

The power supply on a plywood board with coasters