The ALC circuit in the 781 is too slow to respond. This results in the following phenomenon: If you set the power level to 30W, for example, there is a leading pulse that can be more than 200W, particularly if the drive control is advanced. This results in a serious problem if you are using an amplifier. It makes it very difficult to limit drive to an amplifier properly.
This behavior is clearly seen on a scope and to some extent on a good peak-reading meter. However I suspect many hams are unaware of the problem as it does not show up on RMS meters due to the short duration of the spike.
I have used three 781's over the years ( ser# 19??, 21?? and now 2589) and they all exhibit the same behavior. This behavior is not exhibited by other Icom radios I have checked such as the IC751A, IC735 or IC765.
When I had my previous 781, the tubes in my amplifier, a Ten-Tec Titan using a pair of 3CX800A7's, developed shorted grids, even though they were only a few years old and lightly used. I suspected the 781 caused the failure, but sold it before the amplifier was repaired. I recently bought another used 781 from Burghardt radio, serial # 2589, and within 3 weeks of obtaining the 781, my tubes (less than 2 years old) are shorted again. I do not believe this is a coincidence! I run the amp conservatively, and I doubt the tubes have 50 hours transmit time on them.
I believe the problem is vaporization of the gold plating from the grid of the 3CX800 causing the tubes to become gassy, which causes arcing.
I spoke to Technical Support at Icom USA and was informed that the 781's just work that way and nothing can be done. That is a ridiculous answer! Of course it can be fixed, the attack time in the ALC circuit has to be shortened.
This turned out to be easy to accomplish....
Locate the following components connected to the base of Q34 in the ALC amp, R107, R108, C47 & C48. These determine the attack and decay time of the ALC amp.
Values as per Service manual:
Values as per Production ser # 2589:
Note that C47 and C48 have been swapped... a step in the right direction.
Values that work:
C48 .022uf - .1uf
.1 reduces the overshoot from 200%+ to 20%, .022 eliminates it almost completely. However, as the value of this capacitor is reduced, I'm concerned that IM products in the audio may start to increase due to syllabic compression of the RF envelope. I had no way to verify this.
Perhaps a project for a better equipped workshop! I would think .047 would be fine, though I'm using .022 without any noticeable degradation of audio quality.