Hafler P3000 Test

John Andrews has popularized the use of the Hafler P3000 audio amplifier as a linear rf amplifier for use on 136 kHz. The P3000 has a -3 dB bandwidth of 300 kHz and is rated for 400-watts output in bridged mode making it an appealing way to go if you're planning to transmit modes that require linear amplification. Although no longer in production, P3000's can often be found on ebay, but expect to pay $250 for a unit in good condition...unless you get lucky!

Recommended input and output matching circuits can be found at www.w1tag.com/pix/Hafler_Amp.jpg and this is the arrangement used for the tests. The amplifier tested was factory stock with no modifications or adjustments. Harmonic and two-tone IMD tests were performed at three power levels - 200, 250 and 300 watts in an effort to determine the highest practical output level consistent with a clean transmitted signal.

The test setup for measuring harmonic content and two-tone IMD is shown below. Harmonic measurements were made with and without the recommended low-pass filter in line.

Harmonic tests:

Harmonic testing at 200 watts - without low-pass filter on top and with low-pass filter on the bottom. Reference line of the plot is +60 dBm (1 kW) - the 200-watt signal registers +53 dBm. Note that the addition of the low-pass filter lowers the transmitted signal level approximately 0.4 dB - the insertion loss of the filter.

Harmonic testing at 250 watts - without low-pass filter on top and with low pass filter on the bottom. Reference line of the plot is +60 dBm (1 kW) - the 250-watt signal registers +54 dBm.

Harmonic testing at 300 watts - without low-pass filter on top and with low-pass filter on the bottom. Reference line of the plot is +60 dBm (1 kW) - the 300-watt signal registers +54.8 dBm.

Even at the 300-watt level the amplifier is probably clean enough to run without a low pass filter - especially considering the narrow bandwidth antennas typically in use. That said, it's possible that an antenna can have unintended resonances at harmonic frequencies where it might be a better radiator than at 136 kHz. The addition of the low-pass filter offers a degree of confidence that the operation is not interfering with other services.

IMD tests:

Two-tone IMD test at the 200-watt power level. Each tone is adjusted to 6 dB below the PEP level used in the harmonic test (+53 dBm - 6 dB = +47 dBm). IMD levels should be referenced to +53 dBm.

Two-tone IMD test at the 250-watt power level. Each tone is adjusted to 6 dB below the PEP level used in the harmonic test (+54 dBm - 6 dB = +48 dBm). IMD levels should be referenced to +54 dBm.

Two-tone IMD test at the 300-watt power level. Each tone is adjusted to 6 dB below the PEP level used in the key down harmonic test (+54.8 dBm - 6 dB = +48.8 dBm). IMD levels should be referenced to +54.8 dBm.

At the 300-watt power level the IMD products are nearly 45 dB down.

Additional info:

There's always the temptation to see what an amplifier can do in "full tilt boogie" mode. This amplifier topped out at at a respectable 450 watts, however at power levels above 300 watts the harmonic content and IMD rise quickly. Since we're investigating the P3000 as a linear amplifier the 300-watt level is probably an appropriate upper limit. While efficiency of the P3000 wasn't measured, it's doubtful that it exceeds 50%. A high-power, high-efficiency class D or E amplifier would be better suited for modes that don't require a linear amplifier.

Testing was done without fans which resulted in heatsinks temperatures reaching 140 degrees F during the relatively short testing periods. For extended operation at these power levels a fan for each heatsink is advisable.

Input power required to drive the amplifier to specified output is as follows: 200-watt output / +6.7 dBm input, 250-watt output / +7.5 dBm input, 300-watt output / +8.5 dBm and 450-watt output / +10 dBm input.

During an initial close in look two minor spurs approximately 35 kHz each side of the carrier and 65 dB down were noted. Breaking the ground connection between the exciter and pin 1 as shown in John's diagram cured the problem. This may have been an anomoly in this particular setup, but may be worth checking in your installation.

Jay Rusgrove, W1VD
11/2005