Tags : :
After 5 years now, of trying to improve the Stephens 821 series transport, we've finally got enough empirically born data to share w/ current owners. All of these ideas have been tested on various machines and we have a very high level of confidence that they will improve any Stephens deck that is "otherwise" healthy. SO many of them have been monkeyed with over the years that we see many that need to be taken back to a more original state before considering these mods.
The biggest buggaboo with John's 821 servo system was/is flutter. He tried numerous band-aids over the years, but never really quantified what it takes, more globally, machine to machine, to fix things and make them reliable at 15/30/60. We are still furthering the design of NEW servo/converter cards that give more useful improvements, but the following will go a long ways to making the original systems work a LOT better.
We went through all of the various modifications proposed on all of the schematics
regarding stability and came up with a 'fix' that involved some minor
modifications to the original boards. Here they are:
- Change 4uF capacitor C713 to a 2.2uF capacitor in parallel with a 22uF
capacitor and a 10k resistor in series with it. This slows down the
long-term output of the 60 Hz phase-locked loop but still allows fast
corrections to pass. It's sort of like the dual time constant of the
- Change the 430k resistor R2004 in the Q2 remote control down to 180k. This
allows the PLL to adjust the tape speed over a wider range. Stability would
be impaired except that we fixed that by changing the time constant above.
This allows the speed pots to not need to be tweaked every other day. It
seems to work very well, except that when it looses lock the speed wanders
around more than it would otherwise. But if the pots are adjusted right it
seems to not matter much.
- Add a 10uF capacitor and 10k resistor between the collector of Q711. This
is a 'tweak' that is along the lines of the 68k resistor added to some of
the later converter boards, except this goes a bit further. One could
use a trim pot here and perhaps a few different capacitor values that
would be selected by a jumper or switch for tuning to that particular
machine, if it was required.
- Use the 'modded' values of 220k and 0.047uF for the velocity lead network
C701/R701. This 'mod' is already present on some converter cards. This
should probably be a trimpot. Note that many of the 'revised' boards do not
have the diode network D704,D705,C725 installed.
- Change C716 and C718 to .22uF from .47uF. This allows more of the speed
pickup signal through to the differential amplifier, but since we have
reduced the high-frequency gain of that amplifier this tends to speed the
feedback loop and stabilizes the speed control.
- In addition, one mod done to the machines is to remove the damping capacitor
from the tape lifter solenoid and replace it with a diode, cathode to
chassis ground. This prevents the tape from wobbling back-and-forth when
stop is pressed. And the servo boards need to have Polyswitch circuit
breakers added so they don't burn up.
With these mods, the speed response is very acceptable, even with a
somewhat sticky roll of tape. As well, it seems to exhibit much less of the
instability when going from rewind to play. We believe the source of that
instability is actually mechanical - maybe the tilt of the guides is off
and it takes about 10 seconds of normal playback to get it back to normal.
It could also be the brushes in the servomotors being a bit cranky.
As stated earlier, there are other issues that should be addressed on
a Stephens machine. The guides need to be NOT too smooth...which causes stiction
(800 grit sandpaper, properly used to roughen up the guides a bit), the speed pots may
need to be replaced (often bad) and power supply cap upgrades are all worth looking
Hoping this help the lucky few that own these machines.