Output impedance will go down as transconductance goes up.
The transconductance of a 5998 is generally higher than a 6080 at equal operating points. There is some transconductance variation based on the operating point, which suggests a range of 7,000uMho to 21,000uMho (approximately) for the 5998 at typical operating points. Given that the Speedball fixes the current at about 30mA per section of the 5998, then the operating point of the driver stage roughly sets the cathode voltage of the output stage. (75V on the driver plate means the cathode of the output stage will be above that, but not by a whole ton)
If the driver stage plate voltage is set high, then the result will be less available plate-to-cathode voltage on the 5998, which will mean less grid to cathode voltage for that same 30mA of current. This would actually increase the transconductance, but at the expense of potentially limiting the amount of drive voltage that you can apply to the 5998. In other words, you could lower the output impedance, but you'd also lower the maximum voltage output of the circuit in the process. This, unfortunately, would also raise the voltage across the big C4S load under the 5998, and the power dissipation would increase as that voltage crept up, which is not acceptable with the current design.
-PB
