Just get an HF amateur transceiver with MARS modification. Make sure it has AM mode (Most do).
The problem with that, Blakkzakk, is twofold:
1) The output power is only 25 to maybe 40W carrier (the ALC will be kicking in, limiting the AM peak envelope to 100W and even though the 'carrier' may measure 40W you're being fooled on actual 'talk' power vis-a-vis energy appearing in the sidebands once ALC begins to operate) with a nominally rated "100W SSB" radio. The PAs in these are rated for 100W peak (70.7 V RMS) into 50 Ohms so the power is 'allocated' one might say between carrier and sidebands in AM mode (it is 'allocated', but I want to avoid using vector math to describe and explain what is actually taking place at the moment for brevity's sake).
2) The audio passband *WILL* be limited without substantial modification, i.e. removal and replacement of components on several stages and on a last 10 years vintage radio this will be no small task!
Some of the older, tube-based xmitters were simple and mods are easily made ... making them easy choices for someone not put-off by tube-level voltages.
Even the venerable old Yaesu FT-101E (mostly discrete parts with leaded resistors and caps; no SMT!) has limited audio BW without making changes!
Yaesu FT-101E high-frequency Audio Frequency response test,
AM-mode High Frequency Transmit response at -3, -6, -10 dB points:
. . .
http://www.youtube.com/watch?v=nQdc2ai6FqkThe vid above shows the -3, -6 and -10 dB high frequency cutoff points in AM mode in transmit due to the audio response characteristics in a Yaesu FT-101E series transceiver's PB-1315 "AF UNIT" plug-in circuit board. The audio signal generator signal is input via the 'Patch' input in the rear, which is in parallel with the front panel mic input.
The modulation is adjusted for 100% then the HP Modulation monitor is put in the dB relative-mode .. this aids in finding the -3 dB (70.7 % modulation), -6 dB (50% modulation) and -10 dB (10% modulation) frequency response points as the audio signal generator is brought higher in frequency.
Yaesu FT-101E, low-frequency Audio Frequency response test,
AM-mode Low Frequency Transmit response -3 and -6 dB points
. . . .
http://www.youtube.com/watch?v=rgy8bJzqCgITabulated Results:
. . . . . . . . . . . Spice . . . . . Actual
. . . . . . . . . . . Modeled. . . .Measured
Point . . . . . . . Hz . . . . . . .Hz
--------- . . . ---------- -----------
Peak . . . . . . . 881 . . . . . . . 1000
-3 dB . . . . . . 279 . . . . . . . 363 Hz
-3 dB . . . . . 2730 . . . . . . 2637 Hz
-6 dB . . . . . . .192 . . . . . . . 261
-6 dB . . . . . . 3950 . . . . . . 3781
-10 dB . . . . . . 136 . . . not measured
-10 dB . . . . . 5710 . . . . . . . 5538
As one can see from the results, the measured upper and lower 3 dB points are 363 and 2637 Hz respectively.
Nominal voice BW is usually spec'd to be 300 - 3000 Hz, and this particular FT-101E doesn't even meet that! Not very wide and certainly not 'music' quality ...