On Tue, 2007-03-06 at 19:32 -0500, Bill Acito wrote:
> Jerry,
>
> So what causes the pass transistor to fail? (my crowbar is after the pass,
> in the cable... added after the first fail took out half the ICs on my
> Omni back in 1980 (under warranty, thankfully)... if my crowbar trips, it
> is because the pass has failed and it is protecting the rig, not to protect
> the pass transistor).
The pass transistor failing is the desired reason for tripping the
crowbar. Often the crowbar is sensitive to RF, sometimes the regulator
is sensitive to RF. One time I built out a regulator with a couple
transistors and a LM309, and it oscillated at 50 MHz, then I damped that
and it oscillated at 200 KHz. After killing that oscillation it lasted
until abused with overload.
>
> To that end, I seem to get a fail every couple of years. Is that because the
> 252 was "under designed" i.e. not enough headroom in the 2N5301? too small a
> heat sink for the application?
The 2N5302 gives more voltage headroom, not more current or thermal
headroom.
>
> If so, what is the best way to beef up the design? up the device? add a fan?
Some control chips naturally do a good job of current limiting, some
don't. Some are more RF sensitive than others, some also are sensitive
to their package heat and IF mounted on the same heat sink as the pass
transistor(s) can help protect the pass transistors.
>
> (and yes, obviously, I could just go out and plunk down for an Astron 35...
> this rig has sentimental value, and I like running it as the original
> set-up)
I'm no fan of Astron supplies, the ones I've been around use a first
generation regulator chip that is often improperly applied, doesn't have
thermal shut down and is often squirrelly. Then they minimized the pass
transistor emitter resistors so they don't share current as evenly as
they should and they run a high unregulated voltage which demands the
crowbar protection and contributes to poor efficiency.
>
> Bill
> W1PA
>
Crowbars often are sensitive to transients and RF and when applied at
the output of the pass transistor, they short the regulator and if the
pass transistor isn't at fault, its easily overloaded, so its junctions
melt and it becomes an expensive three terminal connection. E.g. a dead
short. That makes it hard to detect which went first, the pass
transistor or the crowbar.
If it is the pass transistor that failed, its probably due to heat or a
transient over voltage. If overvoltage, the 2N5302 would help. When the
pass transistor fails (and the only hazardous failure is failing
shorted) it doesn't matter which side of that transistor the crowbar
shorts. But if the crowbar has triggered falsely, its a whole lot more
healthy for the pass transistor if the crowbar connects to the input of
the regulator, essentially across the filter capacitor. The SCR gets the
same energy connected there as it would get with the pass transistor
shorted. It makes life easier for the SCR if there is a fuse between the
filter capacitor and the regulator AND the SCR is connected to the
regulator size of that fuse. And that fuse must be a fast acting fuse.
No slow blow fuse there.
More heat sink or more air helps keep the transistor cooler for normal
operation, it takes a lot more heat sink or more fan than you will live
with to keep the transistor cool when the crowbar has shorted the output
side of the regulator. And more cooling that the TEC likely will supply.
What I'd do is make sure the crowbar didn't trip below 15 volts, make
sure the SCR is hooked at the input of the regulator, right after that
fast fuse. I'd add more RF bypassing on the crowbar gate circuit, with
disc ceramic capacitors with zero length leads. I'd lower the gate shunt
resistor value to increase the current required to trigger the SCR (and
if necessary increase the power rating of the zener diode). I'd test the
supply with a variac on its input and lower the AC line voltage until it
barely stayed in regulation, then I'd rig a bucking transformer to lower
my line voltage to that value. I'd mount the 2N5302 to a heat sink with
fins twice as deep as present now, with minimal thermal grease (aluminum
oxide or silver). Be sure to orient those fins so air flows up through
them naturally. I'd likely want to use a TL431C programmable zener for
the crowbar trigger so it had a sharp knee compared to the soft knee of
the curve of the standard zener. It doesn't like too much C
unfortunately, else the op amp in the package will oscillate.
A resistor in series with the regulator input can limit thee peak
current and if chosen carefully for the maximum load can reduce the
power dissipation in the regulator, but if its too large it will cause
loss of voltage regulation more rapidly than lowering the line voltage.
Or I'd get 80 amps worth of N type MOSFETs and convert the circuit to my
latest favorite with Schottky rectifiers and lower the transformer
voltage to the point that a crowbar isn't needed. Needs a 2N2907 and a
TL431C to act as the reference and regulator control chip. But then its
not the original, not even close.
--
73, Jerry, K0CQ,
All content copyright Dr. Gerald N. Johnson, electrical engineer
_______________________________________________
TenTec mailing list
TenTec@contesting.com
http://lists.contesting.com/mailman/listinfo/tentec
|