[Elecraft] Elecraft Digest, Vol 154, Issue 22

[Dave B]

On 14/02/17 00:54, elecraft-request at mailman.qth.net wrote:
> Subject: Re: [Elecraft] K3 with switching power supply??
> Message-ID: <6796bb5c-c5da-c0d2-13ce-0b5bd4ba16f0 at subich.com>
> Content-Type: text/plain; charset=utf-8; format=flowed
>
>
>  > I am on the way to change my old Astron RS35M power supply....
>
> Why would you change out an Astron RS35M?  Unless you need to
> make the K3 station portable where the weight and size of the
> Astron is an issue, the RS35M will be both much more quiet and
> much more reliable than any switcher.
>
> Even if the Astron is getting a little unreliable, an overhaul
> (replacing the diodes, pass transistors and possibly filter cap)
> is likely to be more cost effective than purchasing a switcher.
>
> For repair/overhaul information see: 
> <http://www.repeater-builder.com/astron/astron-repair/astron-repair.html>
>
> or for a more in-depth modification/overhaul that replaces the 723
> regulator see: <http://www.bwcelectronics.com/articles/WP20A190.pdf>
>
>
> 73,
>
>     ... Joe, W4TV
There is absolutely nothing wrong or bad about the 723 regulator.  But
there are a lot of badly designed units using that IC out there.  The
designers did not read (let alone fully understand) the data sheet.  And
then not helped by the physical layout people, who don't appreciate the
effect "inappropriate" wire routing can have on a regulator with a high
loop gain. Don't blame the device for poor product design.

What does go bad over time with most "Ham market" linear PSU's, is not
so much the main cap (often too high a value) diodes or pass transistors
(unless this next issue has happened.)

High value reservoir caps can overly stress the rectifier diodes, and
transformer.   Ripple is to be expected across that, and is what the
regulator is there to remove...

Anyway.  The thermal interface between the pass transistors and heat
sink often deteriorates, not helped by the usual sloppy build quality in
the first place.  (I've even seen pop-rivets used to mount regulator
pass transistors!  BAD BAD BAD!)

Same for the main bridge rectifier if that is heat-sink mounted as
well.    If it is not, consider moving it to the heatsink, or fitting
one to it if there is room.

The cooler you can keep a semiconductor (within reason) the longer it
will last.   Wide variations of temperature are also bad, often for
where the leads are connected to a PCB, if not using flexible wire
resulting in failed solder joints.   Always "kink" the lead with a z
formation for mechanical stress relief, it'll last much longer as a result.

Power transistors with sockets too, can suffer intermittent lead
connections over time, due to thermal cycling and the minute repeated
relative movement between lead and socket, usually damaging the plating
on the semiconductor lead, resulting in bad connections.   Replacing the
socket will not cure that.

Dismantle, clean and re-apply fresh compound or thermal interface pad. 
Use fresh insulating (clamp) washers too if needed.   Most of all, do
not over-tighten those device mounting screws.   No, I can't quote a
torque, you'll have to research the matter, and consult the device
makers data sheet, if you can identify the devices that is...

Re solder any joints that appear a little suspicious around the high
power devices.   Replace any suspect socket connections by removing them
and soldering the wires to the device leg’s directly.  (If the device
legs plating is not badly corroded, if so, replace the device too.)

The same holds true for the pass transistor (and their drivers) emitter
ballast resistors.  Check their condition and connections.

No emitter ballast resistors?   Bad design...

For a any linear PSU, even a small fan wafting air over the heat-sink
will make a world of difference too, resulting in a much cooler PSU that
is much more reliable as a result.   If you don’t like the continuous
sound of that, fit a thermal switch to the heat-sink, say 50'C, to turn
the fan on when it gets hot...   (Or based on PSU output current...)

73.

Dave G0WBX.