I decided to experiment with DC on the 12AX7 filament tonight, ala the Eico HF-87 method. Nothing to it really. A couple cuts and a jumper was all that was needed.
Background
The HF-87 used one set of power tubes to act as a current source for the 12AX7. Two EL34s biased at close to 70mA each would be a 140mA (and greater) current source, connected to pin 4 of the 12AX7 and pin 5 went to ground. The cathode resistor was appropriately lesser in this channel than the other (165 ohms versus 235 ohms). Some may rip on this method, but doggone it, it's brilliant. The Eico bias sheet shows that each half had 5.2VDC on the filaments at idle, and up to 5.8 at max ouput.
Procedure
I wasn't crazy about biasing EL34s that hard, so I stayed with the KT88s so that I could bias them where I could get a better filament voltage. I have my test board as a fixed bias unit, so I decide to go for a mixed-fixed config. I lifted the ground ends of each of my 10 ohm bias resistors and tied them together. I then cut the AC filament traces to the 12AX7 (remember, we are just using half a tube in the Ike). I scraped and tinned part of the filament trace connecting to pin 5 and connected a 5-inch shielded wire between it and the 10 ohm resistor junction. I scraped a little solder mask away at the ground plane near pin 9, then tinned it and jumpered pin 9 to ground. I powered up, reset my bias (about 72mA each) and put in a CD. Simple! My filament voltage at idle was 5.8VDC.
Epilogue
Is it worth it? If you have sensitive speakers, sure. Very neat trick, to say the least.
Shannon