KL7UW

144-MHz 1500W 8877 Power Amp

This page describes the construction of a 1500-watt 2m amplifier using an 8877. The design follows the classic W6PO design which is described in the Eimac Notes. I am using the Triode Board built by G3SEK (2023 appears no longer avaialble). I purchased the Plate lines and output tuning capacitors as a kit from Harry, WA4OFS(sk). Here are some photos of the amplifier under construction: (click to see larger image) Block diagrams of the amplifier showing control, bias, and power wiring: (click to see larger image) QRO HV power supply acquired:

  • Schematic
  • Front and top views of the surplus HV Power Supply
Here are photos of the assembled PA Control Panel (before wiring):
  • Front view of Control Panel.
  • Close up view of Meters.
  • Backside view-1.
  • Backside view-2.
  • High Voltage Divider to provide HV detection & metering.
  • High Voltage switch.
  • High Voltage Connectors.
  • Front Panel view of RF Assembly.
  • Side view showing terminal strip and filament transformer.
  • Close up view of Filament Transformer.
  • Input cavity of RF Assembly.
  • Detail showing input coils and tuning cap.
  • Rear view.
  • View of other side.
  • Top view.
  • Top view two.
  • Top view showing RF circuits.
  • Close up of HV RF Choke, bypass cap. and MHV connector.
  • Workbench testing setup.
  • Amplifier with new cover and fan.
  • Side view.
  • rear view of rack







  • Finished control panel (open).
  • Finished control panel (covered)
  • Amplifier installed in cabinet.
  • Amplifier showing operation at 600w.
  • New 7-foot open rack














Testing of 2m-8877:
  • Filament voltage has been applied for several hours over several days. Filament in-rush circuit results in 2-second delay of full filament voltage (4.9vac).
  • 8877 temperature barely detectable with filaments powered indicating blower is working well
  • Repair of the Triode Board and re-calibration of metering/bias is done
  • Application of HV was done in stages up to 3.1 kV with no problems
  • Setting of bias is adjustable up to 150 mA anode standby current
  • Application of low power (5w) RF shows a high SWR preventing further testing
  • Input coils were modified: L1-6T, closed-spaced; L2-3T, wide-spaced
  • Testing the input impedance shows the grid circuit is resonant at 144 MHz with SWR=1.7
  • Use of a MFJ-921 antenna tuner matches the FT-847 so that it may drive the 8877
  • Reset bias for 8.29v, resting Ip< 10 mA
  • With 27w input, getting 600w output: Ep= 2400v, Ip= 410 mA, Ig= 16 mA, 984w dc input with 60.9% efficiency and 13.6-dB gain
  • With 30w drive, 650w output is possible, but with more plate current
  • I found that the 8877 will drive without the tuner by retuning the input, but takes 30w drive and Ip= 420 Ma
  • Reconditioning a surplus CAI HVPS will permit running the 8877 at full output
I discovered that using a galvanized steel screen for the cover of the amplifier enclosure was not the best choice. Apparently, a magnetic metal is subject to higher circulating RF currents. First, I mounted the 1/4 wavelength copper lines too high such there was indadequate clearance between the anode straps and the top cover (3/16-inch). This resulted in a HV flashover (arc) to occur resulting in a very small spot to appear on the screen directly above the anode straps (carrying high RF currents). I deformed the straps to increase the air gap between the anode straps and the enclosure cover. My solution resulted in RF heating in the area directly above the straps (see photo). Two large dark circles appeared in the mesh directly above the anode straps. My solution was to use the original sheet aluminum cover and make a large vent by drilling a pattern of 3/8-inch holes too form a vent. To this I add a six-inch radial fan above the vent to aid air flow out of the anode cavity. The cover was raised by using 3/4 square channel as a spacer around the upper lip of the enclosure. This resulted in roughly 1-inch separation of the anode straps to the cover. This worked out well, with no HV arcing or heating evident. I have now run about ten hours time on the amplifier (March 3, 2009), with it preforming very reliably. Oct. 18, 2010 - Testing the CAI HVPS:
  • After removing C106 which shorted, I have the following results:
  • Unloaded the HVPS runs at 2900vrms and output voltage is 2780v
  • Bias=8.29v, Ip = 100 mA
  • With 42w RF drive:
  • Ep = 2560v
  • Ip = 660 mA
  • Ig = 50 mA
  • Po = 800w
  • Gain = 12.8 dB
  • HVPS max load = 1908w
I will try adjusting the ouput load capacitor and more drive to get more RF output. Atempting to drive the 8877 with 50w using a solid-state amplifier resulted in a flashover on the -HV line. Dec. 7, 2010 - Testing the CAI HVPS: The HV power supply has been modified to capacitive input by removal of the filter choke an redesign of the bleeder; slow-start circuit on the primary of the HV transformer and relocation of HV surge and fuse to the HVPS. The Bias control transistor and a mosfet switch were found to be shorted and replaced in the Control panel. Schematic Dec. 8, 2010 - Testing the CAI HVPS:
  • Unloaded the HVPS runs at 3077vrms and peak output voltage is 4350v
  • No RF drive:
  • Bias=12v, Ip = 100 mA
  • Ep = 4220v
  • Bias= 5v, Ip = 250 mA
  • Ep = 4100v
  • HVPS dc load = 1025w
Dec. 9-12, 2010 - Testing the 2m-8877 Amplifier:
    Several days trying to optimize tuning of the 2m-8877 amplifier ended up with a failure of the coax jumper from the amplifier to the main Bird power meter. A bad connector shorted out. Subsequent inspection of the jumper showed extensive corrosion in the cable from water intrusion (this must have occured prior to being installed in the radio shack). The cable has been replaced with a new run of LMR-400 with new N-connectors (loss=0.27 dB which will result in 86w being dissipated in the coax when the amp runs at 1400w). Final measurements that will be used as the published parameters for the 2m-8877:
  • No RF drive:
  • Bias=15.5v, Ip = 85 mA
  • Ep = 4220v
  • Drive = 50w
  • Ep = 3690v
  • Ip = 720 mA
  • Ig = 42 mA
  • Po = 1436w (meter indicates 1350w)
  • Gain = 14.3 dB
  • DC Load = 2657w
  • Efficiency = 54%
  • Anode Dissip = 1221w
Dec. 14, 2010 - HV Power Supply Fails: The power supply ran for two hours in the morning doing meteor scatter at 1300w. At 1810 utc the mains breaker tripped shutting down the HVPS. Attempts to power up the HVPS continued tripping the 20 amp breaker. Investigation revealed that the transformer was arcing either from the case or across the 240vac line. The tranformer input shows a short. Half of the secondary is open. the other side of the secondary shows leakage to ground (800K indicated with ohm meter). A replacement HV transformer has been being shipped from Oregon and in process of installation. Here is the revised HVPS Schematic. QRT on 2m EME, temporarily. Rebuilt HVPS, Feb. 22, 2011:
  • Back of the modified HVPS shows +HV, -HV, 240vac, and 120vac connections.
  • Top view of the modified HVPS showing filter cap and 240v relay in center with HV bridge to right.
  • Inside view shows bridge rectifier, slow start circuit, and HV surge resistor and fuse (left to right).
  • Inside view shows -HV to ground resistors and surge diodes on the insulated bleeder bracket.
  • Old transformer next to "New" HV transformer (190-lbs.), installed (in its shipping box) next to HVPS
Testing, Feb. 25, 2011:
  • unloaded the HVPS outputs 4000 vdc
  • No RF drive:
  • Bias=15.5v, Ip = 65 mA
  • Ep = 3930v
  • Drive = 50w
  • Ep = 3690v
  • Ip = 730 mA
  • Ig = 34 mA
  • Po = 1436w (meter indicates 1350w)
  • Gain = 14.6 dB
  • DC Load = 2693w
  • Efficiency = 53%
  • Anode Dissip = 1257w
Testing, July, 2 2011:
  • Filament voltage raised from 4.8 to 5.06
  • unloaded the HVPS outputs 3960 vdc
  • No RF drive:
  • Bias=15.5v, Ip = 60 mA
  • Ep = 3920v
  • Drive = 50w
  • Ep = 3640v
  • Ip = 740 mA
  • Ig = 38 mA
  • Po = 1320w (meter indicates 1250w)
  • Gain = 14.2 dB
  • DC Load = 2694w
  • Efficiency = 49%
  • Anode Dissip = 1374w
New Blower, November, 2013: Amplifier and HVPS Sold in 2018.
Return to 2-meter EME page.