[Amps] switching power supply for tube amplifier

[Pete Lancashire]

Like Peter, I use to work in a similar environment, Component Engineering
at the old Tektronix in the 70's and 80's. And at home have a few more
tubes in use, I would guess around 150-200, in a Tek 547, 575/175, and a
couple older HP and Tek instruments.

My failure rate at home has been a bit higher, but I would say less
then 5, mostly from the ones being pushed to their noise limits.

With semi's when I had the opportunity to see what the failures where
at Tek < 20 yrs ago, and today in my home lab, nothing much has
changed. Manufacturing quality control with wire bonding and contamination
being the top 'winers' and the belief plastic packages will play nice
for > 10-15 years, they just don't. Add things like bean counters
where all you have to prove is a 3 to 5 years life
and you can ship that diode, transistor, etc.

Another is power supply design, you can not always blame the piece of Si,
add poor cooling, crappy assembly, under rated caps, no over load/short
protection (deleted by the bean counters) etc. etc.

But on the flip side, who will pay for military grade true
hermetic packaging, or for that bazillon watt transistor
taking up more run then the whole power supply does today ?
A price we may have to pay ?

Btw in my 30+ y/o 547 every HV diode has at least once in its life
failed, only two of its two tubes when noisy.

Now .. caps thats another story

-pete




> I find it interesting that over the last 20 years, I've had more solid
> state failures than tube failures.
> Ok, I have only 11 valves (tubes)  in the main HF station and an awful lot
> more semiconductors.. Of the 11 tubes
> 1 was bought in 1936 (6L6G)
> 2 were  made in about 1944 (6SJ7, VR150)
> 2 were made in the late 1940s (5R4G, 6X5)
> 2 were made in the early 1960's ( QY4-250 , similar to 4-250A)
> 4 were made in the early 1980s  - 3 off 6146B, 1 off12BY7A
> Those last 4 have worked hard in several contests and many pile ups.
> Since 1985, failures on tubes - nil.
> Semiconductor failures - 17.  Such things as IC's that died for no
> apparent reason -  although 4 of them were caused by metal corrosion,
> according to the failure analysis guys at work - I work for an IC
> company..
> Latest was last week, where a ULN 2803A octal Darlington driver rated at
> 500 ma, switching 150mA, decided to have a saturation voltage of 9 volts.
> I've had such things as an JFET lose gm, for no apparent reason.
> 3 cases of second breakdown in transistors, although everything says that
> they were well within limits.
> There can be advantages in having a semiconductor failure analysis lab
> handy, staffed by people who owe you a favour or two!
> So I wouldn't really want a switcheing PSU at HV.
> 73
> Peter G3RZP
> _______________________________________________
> Amps mailing list
> Amps at contesting.com
> http://lists.contesting.com/mailman/listinfo/amps
>