We had a JH16 at the Webster College recording studio that had very unstable bias levels. The bias would drift all over the place; let it record for five minutes and the bias would be several dB different from when you started. We finally cleaned it up, sort of, by brute-force replacing the transistors in the bias card with TO220s and hefty heatsinks.
Not long after, we bought a 440-8, and its bias was rock-solid. Lower flutter, too, although that may have been the individual JH16's fault. That 440-8 was a joy to use and to work on. One weird thing, thoough: it was a 440C, and the repro head resonance control couldn't go low enough, which made it impossible to get truly flat response out of the thing. There was always a slight dip in the midrange. Still beat the pants off the MCI. Your mileage, of course, may vary.
Peace,
Paul