[Amps] power supply transformer

[Robert B. Bonner]

Yes Manfred we have already been past the "All American Five about a week
ago.

That design would never be UL approved in 2007.

It was also the absolutely favorite weapon to murder your spouse from 1947
until 1976.  Just drop the radio into the bathtub while "Joe" or "Betsy" was
screaming at you through their bubble bath.

I personally killed my first wife this way.  June 14, 1976 in OMAHA on our 1
year anniversary.  No one ever suspected.  The cops did take away my radio,
and I have had to put up with the next one for 31 years since.

PS. I highly recommend unscrewing your 15 amp FUSE and placing a penny
behind it before entering the bathroom, adds a few more amps worth of
capability.  After the dirty deed, screw in a left over blown one.. Then
dial 911 about 15 minutes later.

Well I was watching TV then the lights went out officer.  After cycling
these 5 new fuses into the socket I realized I had another problem and went
looking for it..  Betsy had sort of frazzeled look about her. At first I
thought was normal but for some reason she wasn't yelling like usual.

BOB DD

-----Original Message-----
From: amps-bounces at contesting.com [mailto:amps-bounces at contesting.com] On
Behalf Of Manfred Mornhinweg
Sent: Monday, February 05, 2007 7:12 AM
To: amps at contesting.com
Subject: Re: [Amps] power supply transformer

Hi Paul, Peter C., Peter V., Tom, and all!

As usual, I come late, but I will still add my comments to this thread.

> I had never seen a 'real' design, besides a switched mode power
> supply that did not have an isolation transformer on the mains before
> actual discreet components,

Google for the "all-american five" design for tube type radios, which
was used in countless millions of radios roughly from 1935 to 1960.
There you have a very real and hugely successful design, which ties not
only the entire circuit, but even the metal chassis directly to one side
of the power line, usually without the benefit of a polarized plug, and
without a protective earth connection! Earth ground was used ONLY as a
counterpole for the antenna! The safety of these radios rested entirely
on the principle of keeping the chassis and the entire circuit inside a
wooden or plastic box, that insulated it from the user. These radios
were somewhat dangerous, because if a knob loosened and came off, the
shaft below might be connected directly to the hot side of the AC line!

A 110 or 120V shock is rarely damaging. With 220 or 240V the situation
is worse. This design apparently was not used as original design in 220V
countries, but here in Chile, where we do have 220V, lots of radios were
sold which were basically built for the USA market and then "modified"
for 220V by the addition of a very long, resistive AC cord! Power plugs
aren't polarized here either, so you could well end up with 220V on the
chassis and knob shafts of such a radio! I have several of those in my
collection, so I know!

And then, take the millions of tube-type TVs. Most of them also have the
internal chassis tied to the AC line, no transformer, no earth ground,
and a plastic case around everything. Motorola TVs with that design were
very common in Chile, and surely they were sold in the USA too! I
repaired many of them, and later, when black&white TVs fell out of
fashion, I gutted several for parts.

So, transformerless designs have a long and proud pedigree!

> As far as technicians safety, well, you have to have a minimal
> understanding of what you are working on, and know where and where
> not to put your fingers.

That's right. And for added comfort while servicing such equipment, my
workshop has an isolation transformer which I can connect between the
line and such equipment during servicing.

> I'm not sure that I understand what the input and output RF 
> transformers have to do with providing insulation from the power 
> supply.  Also, when you say that the "transistor-to-heatsink" also 
> must comply with this, what does that mean exactly?

Well, Peter C. already answered this question. Just to reinforce on it:
You have to think of the entire line-connected circuit as an isolated
circuit, not connected to ground at all. So, everything coming in and
out of that circuit has to be galvanically isolated. If you think of a
typical amplifier stage with two MOSFETs, You have the following
electrical inputs and outputs:

1) RF drive input. Since there is anyway a broadband impedance matching
transformer, it's easy to use this transformer to provide galvanic
insulation. But it forces you to use a traditional transformer, not a
transmissio-line design which interconnects windings.

2) RF output. Usually there too is a transformer, so the same as for the
input applies. The whole low pass filter circuit can be grounded, since
it is outside the insulated cell.

3) Drain power input: Since you have choosen direct line rectification,
you do it and so you have the amplifier circuits connected to the line.

4) Bias input: You use a dedicated small auxiliary power supply that is
used for nothing else.

5) Mechanical connections: You need to couple the MOSFETs to a heat
sink, you need to mount the printed circuit board, and you need thermal
contact to a temperature sensor. All these must be done with
consideration for electrical insulation.

So, you end up with a line-connected amplifier stage, that electrically
connects to the outside world ONLY via the power line. All other
interfaces are magnetic (input, output and bias transformers), or
mechanic (heatsink mounting, PCB mounted with insulating screws). No
other galvanic connection between the amplifier cell and the outside
world exists. Implemented in this way, it is safe for the user, while
the technician needs to be aware of line voltage in the amplifier.

> As far as winding my own, I thought of that option.   As I have 
> expressed to others privately (now publicly), I don't have a feel for
>  where to acquire parts quite yet.  My limited expierence usually
> take me to DigiKey or Jameco, but they are not good sources for these
>  types of components.

I can't advice you about wire and lamination sources in the USA, but
surely there must be some, if even down here in Chile I can find them!
In the town I live, there are two places that carry a small selection of
enameled wire and insulation material, but no laminations. In Santiago,
I know of a place that has everything: Wires in all sizes from #00 to
#46, laminations in a wide assortment of quality and size, bobbins, and
insulation material. Anyway, I rarely buy laminations. Usually I recyle
them form old, burned transformers which I get for free, and buy only
the new wire. Recycling old enamel wire is no good, because it comes out
kinked, stretched, with worn insulation, and often with lots of potting
varnish sticking to it.

> Although I was first licensed in 1982, I have not been an active 
> operator for at least the last 14 years.  In fact, I about 4 months 
> ago, was putting my remaining radio equipment up for sale on ebay 
> when my 12 year old daughter started asking me questions about 
> it...she's now KB1OGL, and we're putting together a new and improved 
> station!

Congratulations to your daughter! She has saved a part of ham radio!

Peter C.:

> You also have a problem with bypass capacitors to the the input and
output grounds, since any capacitor that goes to ground from one side of
the line can lead to enough current flow through it to trip the GFCI.

This is not entirely right. Since the amplifier circuit is coupled to
the input and output via RF transformers, there is no need for any
bypass capacitors! The entire block formed by the amplifier cell and the
line rectifier/filter can have all required bypass caps within itself.
Then the power line is filtered at the inlet, and ONLY THERE do we need
two small bypass caps between the two sides of the AC line and ground.
These caps are typically 10nF, which provides a decent RF bypass but
conducts less than 1mA at line voltage and frequency, which is well
within the acceptance range of a GFCI.

> All in all, I'd say it's more trouble than it's worth.

That depends on your exact situation! Skipping the big power transformer
indeed requires some thinking and some slight added complexity, and
limits the designer in some regards, but will make the amp VERY MUCH
lighter, somewhat smaller, improve power supply regulation, and also
reduce the cost, assuming you have to buy the parts.

> The suggestion that a couple of 48volt high current SMPS in series
could do the job sounds attractive.

If you can find them. I have never come across any. These are location
differences, I suppose. Anyway, I don't have a 120VAC line to rectify,
and my 220V line is a bit high for present-day MOSFETs, at least for
linear service!

> I'm not convinced that, at the end of the day, at the 1000 - 1500
> watt
level, it isn't easier to use tubes, in spite of the high voltage
problem....

A tube circuit is simpler indeed, if you make a conventional one. But
the high Q required to match the plate to the antenna forces us to
manually tune the amp for every major frequency change. I do see this as
a disadvantage, as I have often found that if another ham invites me
over to a different band, I prefer shutting down my amp instead of
having to tune it up first!

Also, almost all tubes require cooling fans, which tend to be noisy to
very noisy. Most tubes don't adapt well to conduction cooling. This is
another reason why I'm looking into solid state amps now.

And if someone asks me to QRO, and my amp is off to avoid the fan noise,
I have to wait a full minute for the filaments to come up! If the other
guy doesn't understand my QRP cries that he should wait a minute, he
might be gone before my amp comes online to continue the QSO! This too
makes me want a solid state, instant-on amplifier.

The high voltage requirement really isn't a problem in my view, but the
things listed above are. And then there is the cost! Building a legal
limit amp with tubes, from parts bought new, easily and perhaps widely
exceeds 1000 dollars. The solid state amp I'm planning, also built from
all new parts, will cost less than half that.

Now for Peter V.:

> Direct rectified power supplies will not work with a GFCI,

Strange... I run more than 15 devices with direct rectified power
supplies at my home, and my GFCI never has tripped on them! Instead, one
time it tripped when my sister splashed a bit of water on an outlet, so
it certainly is in good condition.

> as the neutral and safety ground must be always bonded together.

NOOOOOOOOOO!!!!!!!!    That's forbidden!!!!  Anyone doing that while
designing electronic equipment is not only going against safety codes,
but also doing a very stupid thing! ONLY at ONE point of the network is
the neutral tied to ground. The exact point varies with the country.
Here in Chile it's at the distribution transformer. It might be nearer
the home, but certainly before the GFCI and certainly NOT in any
equipment used inside the home!

> On a correctly built direct rectified power supply there will be
> ALWAYS neutral/ground hardwired to chassis ground.

No, no, NO!!!  Where did you get this wrong and dangerous information?

> This can be forced by a two-pole mains relay which only gets
> energized
with
> correct polarity on the wall plug. The "ground" side of the amp then
> will never be at 120V above ground.

According to existent electrical codes, such an approach does NOT
provide sufficient safety. Instead, a device should remain safe even if
the neutral gets lifted away from earth ground due to some failure
outside your home.

To round it up again: Any metal parts exposed on electrical and
electronic equipment must either be grounded via the power plug third
pin, or doubly insulated. The neutral must never be connected to ground
in the equipment. Grounding of equipment is optional: You are not
required to use grounding if either the metal parts are doubly
insulated, or if there are no metal parts. Direct line rectification is
allowed, but the circuit must then float, instead of being grounded.
Small RF bypass capacitors between the floating circuit and ground can
be used.

Peter C.:

> The problem comes when you connect your
> external RF ground to the mains ground - in the event of a ruptured
> neutral, you can get the load current flowing (or attempting to flow)
> down your RF ground. 

This is one of the cases where the GFCI comes in. It should trip before 
this current down your RF ground can reach a dangerous level. Earth 
ground connections provide best safety ONLY when coupled with GFCIs!

> Different countries have different approaches on
> wiring, especially in Europe. The UK uses a ring main system, which
> is why we need fused plugs 

I would like to learn about that system, as I have always wondered about 
those fuses in UK line plugs! If you can tell be, by direct mail because 
it might be too off-topic for the list, I would be grateful!

> The GFCI or RCD
> (Residual Current Device) as we know it, are have the potential of
> being something of a pain, in that they trip from RF

These things come in widely varying quality levels. Since it is quite 
hard to make a device that will pass 40 amperes of differential mode 
current but trip on 0.03 amperes of common mode current, the lower 
qualities tend to suffer from the problems you describe. I have had good 
results with Ticino and ABB ones, but have had the trouble you describe 
with some cheaper brands.

And now a final note for those who are interested in my amplifier 
project: I have now started the phase of detailed design, which goes 
hand in hand with component selection, from the catalogs of Digikey, 
Mouser, Jameco, RF Parts, Amidon, and others. I would be very grateful 
if someone could point me to a source for ferrite double-E cores in 
sizes larger than what Amidon carries, perhaps also some large two-hole 
cores made of RF ferrite. I couldn't find any sellers so far, even if 
the parts themselves do exist!

73,
Manfred.

----------------------------
Visit my hobby website!
http://ludens.cl
----------------------------
_______________________________________________
Amps mailing list
Amps at contesting.com
http://lists.contesting.com/mailman/listinfo/amps