TU56/TC11 restoration - VARIAC question
eric at brouhaha.com
Thu Mar 17 16:28:25 CST 2005
> For re-forming the cap'sin my unit, my plan is to (hopefully this
> weekend), disconnect the screw terminals and use a programmable power
> supply [DC] applied direcly to the terminals. Current will be limited and
> the voltage slowly increased. If I can get the lab computer working, then
> it will be a nice slow smooth increase over a 12-24hr period. If I resort
> to adjusting the supplies manually I will end up making larger steps and
> watching the current..not as clean but still better than a full "bang".
As long as there is an adjustable current limit, all you have to do is
set the voltage to the rated capacitor voltage (or perhaps 10-20% higher,
but some people dispute that, even though the manufacturer forms the
capacitor to a much higher voltage), and set the current limit such that
the capacitor cannot dissipate enough power to be damaged. This can be
computed based on the rated leakage current. There is no need to
programmatically control the supply voltage; the current limit will
cause the voltage to ramp up at whatever rate maintains that current.
Once the voltage reaches the voltage limit, watch the actual current
being drawn. It should decline to below the rated leakage current.
If the capacitor is in good condition, the initial rampup may occur
fairly quickly, i.e., a few minutes. If it needs a lot of reforming,
it may take hours. I wouldn't normally expect a capacitor that took
more than a few hours to succeed at reforming.
We used voltage programming on the PDP-1 project for two reasons,
neither of which is strictly necessary.
1) The allowable reformation current is higher at lower voltages,
e.g., 2V @ 30mA is the same power as 4V @ 15mA. However, setting
the current limit as appropriate at a higher voltage, it will
still reform properly unless the capacitor is in very poor
condition, in which case it's probably not recoverable anyhow.
2) We wanted to log a lot of data on the reformation process.
Although it's not clear that we've learned too much new or useful
information from the logs.
More information about the cctalk