MODIFYING THE CLIPPERTON 'L' LINEAR AMPLIFIER FOR BETTER 160 METER OPERATION
The stock Dentron Clipperton L linear is a great little amplifier....but NOT on 160 meters as it comes from the factory. There are several reasons for this and this page looks at the deficiencies in the stock amplifier as well as the modifications which will turn it into a very good performer on topband.
WHAT'S WRONG WITH THE STOCK CLIPPERTON L (Part 2)?
In short, design flaws! This amp was designed using a similar layout and similar components in the output stage to several other Dentron products. When I first obtained mine, I found that it worked well on 80 through 20 meters (I am not inclined to use an amp on 15 or 10 meters very often). Running at 1800V DC in the CW position at 700 ma, with any kind of efficiency the power output should be about 750 watts or more. On the higher bands, this was obtainable. However, output on 160 meters was quite low (500-550 watts), plate current tended to be high for this output level (over 750 ma), and the output began to drop after the key was down for just a few seconds. Apart from these problems, the amp worked well and seemed the perfect candidate for modifications.
SLOWLY DECLINING POWER OUTPUT
I presented this problem to readers of the Topband Reflector and quickly received information from a number of sources, Carl, KM1H and Tom, W8JI being two of them. Both suggested that the problem was being caused by the doorknob caps-these are used as coupling and loading in the Pi output circuit. These caps are not NPO types (which have very low drift) but N-3300 or N-4700, and as they heat up with RF the capacitance changes, causing the output to drop.
Replace the doorknob caps. In my case, I replaced C-11 and C-12 (across the plate tune variable) with HEC N750 units, 200 pf each. While I was at it, I added an HEC N750 200 pf doorknob and a pair of 400 pf @ 7.5KV mica caps in parallel with C14/15/16 (switched in on 160 meters). I found over time, when operating at the low end of the band, that even this added capacitace was not enough, and so I added a further 500 pf Centralab 858 series doorknob rated at 5 KV in parallel with the others on 160M. Addition of all these extra caps has the advantage of distributing the circulating current among all units and thus reducing the chance that any one will heat and change in value.
This shows the added caps and the method of mounting. The aluminum
plates fasten to existing bolts.
LOW POWER OUTPUT RELATIVE TO PLATE CURRENT
Adding that extra capacitance in the Pi output enabled me to obtain a somewhat more power output, but not enough. The whole Pi network seemed out of whack on 160 meters...power output highest at maximum load capacitance with no headroom for peaking. According to W8JI, all Dentron amps had essentially the same problem, which stemmed from attempting to employ ONE plate tank circuit in ALL their amplifiers no matter what the physical configuration.
In the case of the Clipperton L, the output components are too close to the top cover. Good engineering practice dictates that the distance between the coil and any metallic material be at least half the diameter of the coil. In the Clipperton L this distance is about 1/4 the diameter of the coil or less! Proof of this comes by operating the amp with the cover off (CAREFUL!) Output rises almost 100 watts for the same plate current with the cover off.
You can cut out a section of the top cover 6" square immediately above the plate coil. Obtain a piece of 3/16" fiberglass sheet and replace the removed section of aluminum. I did this as documented in the photos below. I first marked off the area above both plate coils with a piece of paper and transferred it to the inside of the top cover. I cut a piece of 3/16" plexiglass to a size 3/8" larger all the way around the proposed cutout. I then proceeded to cut the hole and finish the edges. Then, I drilled four holes for short machine screws to fasten the plexiglass. Finally, I painted the whole cover with a satin black paint to match the rest of the amplifier. After re-installing the cover and measuring the power output, I found that it was now about 90 watts higher than before while drawing the same plate current.
NOTE: If you carry out this modification be certain to use short machine screws to hold the plexiglass in place and extend the cutout in such a fashion as to place these screws as far away from components as possible.
The finished top cover with plexiglass installed. View through the unpainted plexiglass showing
extent of the cutout. Note the added mica caps.
Out of view, lower right under the cover is the
500 pf doorknob loading cap for 160M.
Even after completing this modification the plate tuning capacitor still was fully meshed when loading on 1825 khz. Obviously, more inductance and capacitance was needed in the plate circuit,
I added 5 turns on the end of the 160 meter tank coil closest to the panel. I also added another 100 pf HEC N750 doorknob cap to the plate tuning in parallel with C-11 and C-12. After doing these modifications, plate tuning at resonance was considerably less than maximum at 1825 and somewhat less than maximum at 1800. There was no change on 80 or 40 meters. Power output was up to over 900 watts with plate current at about 700 ma-an efficiency of over 60%. There was no sign of instability or power drift.
This shows the modified L1. Note that the 80M winding (top) is
spaced the width of the wire using string while the 160M end is
wound almost close-spaced.
Now, the 572 tubes can easily work at 250 ma each...but I'm not inclined to drive mine any more than about 180 ma per tube . At 900 watts nominal output this is some 2.5 Db down from full legal output power in the USA, or less than ½ S-unit. You can make up the difference with a good antenna and superior operating technique!
Others are not afraid to drive the 572Bs somewhat harder, viz this e-mail I received from Brian, K8BHZ in November 2008 outlining performance of his Clipperton L on 160M:
" Just went down and took some readings on the Clipperton. I always use the SSB position, and read: 2200V / 840mA 1250W for an efficiency of 67.6%. I then put it in the CW position and read: 1500V / 775mA / 850W for an efficiency of 73.1%. I'm using a Bird wattmeter & the internal meter of the amp. I noted that our CW voltages are noticeably different. I presume you are reading the B+ with the key down, as it drops quite a bit. Otherwise the meter & dropping resistors may be different by tolerance. I think the SSB efficiency drop is probably mostly due to your coil proximity & turns observations. I am going to repeat the above tests with the cover off to see."
OTHER SUGGESTED MODIFICATIONS
I added another 1000 pf doorknob plate blocking cap in parallel with C-9. Several sources suggested that the original value was too low.
Carl, KM1H suggests changing the plate choke to a B&W #802 to avoid blown parasitic suppressors on 15 and 10 meters. He also suggests adding a .01 disc ceramic cap to the B+ end of the choke (do this even if you don't change the choke).
I removed the resistor in series with the fan to improve the cooling but found the noise too high. The final solution was to add a small 12V relay to short out the dropping resistor on TX-you can use a plastic case unit, parallel the contacts and epoxy it to the chassis.
KM1H, AG6K and others have suggested improving the protection of all amps, by including in the B+ lead BEFORE the RF choke, a 10 ohm 20 watt wire-wound resistor. In addition to helping suppress parasitics, IF the "big boom" occurs, the resistor will blow before your expensive tubes!
So that's it! I use my Clipperton L primarily on the four lower bands but have, on occasion, used it on 15 and 10 meters. I have not found any adverse effects of the modifications on operation on the higher bands (but then I limit my power to about 400 watts or so). I know that my amp is a much better performer on 160 meters now and look forward to many more years of faithful service. By not pushing the 572Bs too hard this should be entirely possible.
HINTS and KINKS
Much has been written about the merits of various brands of 572B tubes and general opinion seems to suggest that the CETRON tubes made in the USA are pretty hard to beat. The problem is that these are no longer manufactured, though they are often available NOS at a premium price. That leaves other North American brands from yesteryear, Russian and Chinese tubes. I have had no experience in the use of other than Cetron 572Bs; however, here is one recommendation for a particular brand of Chinese tubes which appears to be every bit as good as the Cetron. This comes from Brian, K8BHZ:
I have had limited experience with 811A tubes ordered from RF Parts and have found that their "RF Parts" branded tubes are very good. Perhaps this also holds true for their 572B tubes, especially the "Taylor" brand which are, their catalogue says ,"virtually an exact copy of the 572B/T160L original US tubes".
"CLIPPERTON L MAKEOVER"
For those interested in totally reworking the Clipperton L to make it a better overall performer on all bands, Cecil, K5DL offers the following:
"Inspired by the work of others, I rehabilitated a nice Clipperton L a few years back, adding many of the tips offered. (and a few of my own). I completely re-wound a new set of tank coils from #6 and #12 copper. Silver plated all of it for whatever efficiency it would add.
Re-worked the tank taps with copper strap I sheared from some wide copper grounding sheet. Silver plated all that as well. An Ameritron AL-572 Plate Choke for good measure topped with their parasitic suppression board and ceramic plate caps. Added the LDG raw tuner board, that was featured in an article I read a few years back as well, as a tuned input for the amp so it would play well with my FT-1000D.
The board fit nicely in the power supply compartment of the "L". Re-capped the HV supply with screw terminal modern caps. End result is a nice amp that keeps the input match proper for the solid state rigs and gives me 1KW indicated out in the CW position on all bands with no more than 120W drive...160M included. (I never use the SSB position)."
And another "makeover", this one from Todd, K1TM:
"My father and I purchased a Clipperton new 30 years ago. It never performed on 160m, but we did not work there much. The amp finally died several years ago and my Dad handed it over to me and purchased an AL-80B. In the last year, I have gone through it and brought it up to a level commensurate with the AL-572. 1100 Watts PEP, stable on all bands (including WARC), and full QSK.
160m was the last problem I tackled. After looking at the schematic, doing some probing and measuring, and doing the math, I came up with a plan. Here is what I observed:
1.) The plate choke had overheated when trying to use the amp on the WARC bands and there were resonances in the bands. (The Ameritron WARC choke is a good replacement at a little over $20)
2.) The plate blocking capacitor at 1000pf was way under sized. Typically, a pair of .001uf at 7500v are used. (They can be purchased from Ameritron for $2.88 each). 1000pf at 1.8mhz is 88 ohms of reactance and should never have been used.
3.) The plate supply bypass capacitor at the plate choke base is only 500pf. Therefore, with a 2000V plate voltage and about 2000 ohms of reactance through the plate choke there is 1A of RF that the capacitor must bypass to ground. The 500pf capacitor at 1.8mhz has a reactance of 176 ohms. 1A multiplied by 176 ohms yields 176 volts of RF being dumped into the power supply capacitors! One of the .001uf at 7500V caps used to improve the plate DC blocking capacitor can be used at the base of the coil to ground ( in parallel with the existing 500 pf.-ed) I used the tube base screw that is next to the choke as a ground point to obtain very short lead lengths. A .01uf yields about 12 volts of rf across it on 160m which is normal.
4.) One last change was made before retuning the tank circuit. The stability of the amp on 12m and 10m needed to be improved. The original parasitic suppressors were not up to the task. The simplest fix is to build (or purchase the AL-572 board) a new suppressor using a modern 3 part suppressor. Also, the simple spring clips used to attach the tube anodes to the suppressors need to go. The Ameritron board comes with new ceramic anode clips attached and is the most cost effective way to go. I also replaced the simple plated wire that connected the plate blocking capacitor to the tube anodes with a good low impedance conductor.
At this point, I now had a stable base using sound engineering principles that could be re-tuned. Parts cost is around $60 purchasing everything from Ameritron.
5.) With the fixes above, I observed that the 160m output had dropped even lower. The plate dipped, but the loading capacitance was maxed out on 160m and 40 meters. As you ( referring to VE3NH) pointed out, C14 to 16 are used in parallel on 160m, where C14 is also acitve on 40 meters. So, the first task was to add 170pf doorknob cap in parallel with C14 to bring 40 meters back to normal operation. I checked 80m and it still looked good (C14 is also active on 80m) and 160m was slightly improved. I then reused the 1000pf plate DC blocking door knob by placing it in parallel with C16. This brought 160m back in line with the other bands. I could still use more loading capacitance below 1.85 mhz, but the delta in power did not justify the expense or time. A good doorknob cap runs $24 these days new. So, 160m mid band, I get the same output as all the other bands with slightly lower output below 1.85 mhz (50 to 75 watts). I did not need to cut the case or add turns to the coil to achieve full power.
This brings the expense side up to $80 and yields a stable well tuned tank that is easy on the power supply caps. It also eliminates RFI by getting all that rf out of the supply.
Some other mods I did: replaced the relay with a QSK system using vacuum relays that provides a low voltage low current keying interface to the rig and changed the bias circuit so that the 47k resistor is in series with the zener diode and then a contact pair in the qsk relay shorts the resistor when transmitting (the bias is off between dits and dahs). The advantage is bias no longer has the discontinuity that the old relay system induced by making and breaking both the zener and resistor connections. This saves power and helps the amp to run cooler and cleaner.
I rebuilt the power supply with new diodes and caps. I added an extra series capacitor in each leg of the doubler to increase the working voltage while still getting twice the capacitance of the original caps. The new lower esr capacitors work the diodes harder so I resized them appropriately. I replaced the blower with a high volume unit and I added a 12v blower over the power supply to force in air, which is only on when I run rtty and digital modes. The amp runs much cooler now.
The ALC needs work to be usable, but would be a waste of effort as most rigs are limited to 100 watts and the amp is not pushing close to the 1500 watt limit. "
When asked for details about power output and plate current, Todd adds "It takes 80 watts of drive power to make 900watts (or 10.5 db gain). With 66uf total equivalent series capacitance in the supply, I see 2150 volts @.65 amps to make 900 watts on a Bird 43 meter into a dummy load for about 64% efficiency. My line voltage is 114vac tonight. (I don't general drive it higher than 900 watts personally). Set like this, I'll see 950w to 1000 watts on my PEP meter as the instantaneous stiffer supply voltage supports instantaneous voice peaks better than key down. Driven with 100watts, I see 1100watts PEP. I'm using matched ceramic based Svetlana 572's from RF parts. "
Note that Todd opted to employ much more filtering in his rebuilt power supply. The original 12uf is way too low, and the power supply is just not husky enough to prevent voltage sag. I suspect that his use of 66 uf in the filter is one reason that he is able to achieve 1100 watts output without resorting to the methods I employed to increase power out.
CLIPPERTON "L" MODELS
The "original" Clipperton L was supplied without a tuned input. Later models included the tuned input board (an extra wafer on the band switch selects the appropriate toroid L/C circuit) and a different configuration of the final coil for 80/160M. For owners of an amp without the tuned input board, if drive on 160 and 80 seems to require too much power then perhaps changing the filament choke is in order. The original is only some 5" in length and its inductance on the lower bands is marginal. Changing to a B&W FC-30 or FC-25 or an RF Parts model FILRFC-30 might be in order. The six to seven inch length will fit with a bit of effort and drive requirement will be somewhat less. Changing the choke in an amp which has the tuned input board is not necessary.
POWER SUPPLY RETROFIT
The original filter caps provided just 12.5 mfd total filtering, which causes the power supply to be less stiff than it should be. In CW mode, the indicated B+ voltage in my amp drops from 1925 at idle to below 1800 under load. I have never measured the actual voltage, but I suspect that the resistors in the metering circuit have changed value and the voltage may be other than what is indicated. For these reasons, plus the fact that old electrolytics have a good chance of failing, it is adviseable to update the power supply board.
For those wishing a drop-in board, Harbach Electronics in Polk, OH (Jeff, W8CQ) has a nice unit with 180 mfd electrolytics giving a total of 22 mfd total filtering. It also has 1% metering resistors and capacitor equalizing resistors. For those planning to install individual components, several suppliers offer 180-270 mfd electrolytics at from 450 to 600 volts in various configurations.
While a step-start circuit is not required to preserve the filaments in the tubes, it is certainly desireable to save wear and tear on power supply components. Harbach Electronics offers a "universal" step-start board or you can build your own using a relay and a couple of resistors.
Links and References