To replace the STV-4H's in an SX-5570 that came to me in a rather beat up condition, I fashioned up the attached jury rig using some perfboard, 3xUF4007, thermal compound and the testing methods described in this thread.
The forward voltage in the 10's of mA's is noticably lower than the STV-4H (went down by about 300mv), but temperature response is comparable at those currents.
Initial bulb-limited power-up, full power-up and first bias-setting reveal that it's a fairly stable hack, however, could someone offer some reassurance this can work for a long run-time as well?
EDIT:
For those googling into this thread in order to find a quick and easy STV-4H replacement that doesn't cost a lot of money, time and epoxy:
The perfboard-3xUF4007-perfboard-sandwich shown in the picture proved in my case to be a suitable and stable substitute after a few hours of intensive use, but do make sure there's plenty of thermal contact.
I took the following (highly scientific) measurements in my 5570: When properly adjusted, voltage drop across current sharing resistors is in the <10mV ballpark on cold start, goes up to 20mV after 10 minutes of easy listening, and will drop to 18mV after blasting into 4 ohm load for 10 minutes or until heatsink is toasty. For attaining that proper adjustment, of course reference the service manual.