See below,
--- In ham_amplifiers@..., "pentalab" <jim.thomson@...>
wrote:
### I looked at the 8k/3 k ultra manual I have [same manual does
both amps]. On the 8k they use 5 x 20k-100w resistors. Diss is
a whopping 300 W CCS..... presumably so the resonant choke filter
will work. On the 8 k.. they used a 6000 V xfmr. .9 x 6000 =
5400 vdc no load.
### here's the kicker. IF just one bleeder opens up... the OCV
plate V shoots UP to a whopping 6000 x 1.414 = 8484 V !!!
Now the poor 3CX-3000A7 [rated for 5 kv] won't like it.... nor
anything else in the amp. You are lucking you didn't blow up ur
3k ultra.
### I gave up on choke input HV supplies yrs ago. The
resonant choke idea although a good one... is not so simple to
implement.
## My feeling is one is better off to use a C input filter. The
8 k ultra's plate xfmr is flaky at best. They could have used just a
3800 vac xfmr..... which would require LESS turns on the sec
winding.... which in turn would mean the xfmr coulda used bigger
ga wire in the 1st place.
### Dahl does a roaring business selling a 127 lb 10 kva
replacement xfmr + mating choke for the 8 k ultra. Stock, the
8k is only good for 2 kw out CCS rtty/ 3.5 kw pep cw/ssb....
then in real fine print on the schematic it sez..."200 w in... 5kw+
out".
### IMO... why put a ton of weight.. in the form of a choke in a
HV .... better to put the added weight into the plate xfmr in the
1st place... but that's just my opinion.
### I had some 3 k owners e-mail me a while back.... they were in
the 10m contest [where both cw/ssb modes are used the entire
weekend]. The poor fellow had forgotten to change the plate V
from High... to low... when switching from ssb... to cw. The
HV supply bit the dust shortly afterwards. [smoked plate xfmr]
## Henry has gone through so many plate xfmr manufacturer's, I have
lost count. The last supplier for the 3k/8k was ECA. And just
b4 Henry went "outa business"... they dumped ECA as a plate
xfmr supplier !
## I know several fellows who bought the Henry 16 uf - 7.5 kv
rated oil caps.... used em in 4-1000 amps.... and blew up the caps
!
With the proliferation of the newer low esr lytics out there
[loads of em on the surplus market].. for a new hb project, they
can't be beat. The latest bunch I got are 50 miliohm and 10 A
CCS ripple current rating.... small things too.... 2500 uf @ 450
vdc... and just 2" diam x 4" long.
## It just seems nuts to me to suck 300 W ccs of bleeder...
then another 150 ma @ 5200 Vdc on idle [8k] on top of that.
### The 8k drops from 5400 vdc to 4700 vdc with a typ load. It
weighs a whopping 350 lbs too ! Since the alpha 87-A will
do 1.5 kw CCS... and the 8k will do just 2 kw CCS... I'd say
Henry just lost it... in the weight dept.
### BTW... a Dahl 10 KVA CCS xfmr weighs.. "just 127 lbs"
Later......Jim VE7RF
Jim,
My guess for the reason Henry used a choke input power supply was that
the transformer can be smaller than one for a capacitor input by a
good bit. Most transformer manufacturers give for a choke input, the
rms current is 1.2 x DC current, and for a cap input it is 1.65
(Hammond only) to 2 X, with most saying 1.8 X. The difference is the
rms voltage of the choke input transformer has to be higher than that
of a cap input to match the same DC voltage.
A resonant choke has to resonante at the peak current draw, so the cap
and chokes (Lcrit) critical value has to match that figure. The only
problem, there's no resonance at idle current. You still get the
filtering of the choke, it's just not resonant to where the ac portion
would be killed down like it is at the peak current. So, Lcit has to
be selected at the peak current. See a copy below from a conversation
between Peter G3RZP and I from the Amps archives.
On transformer regulation, the voltage sag I've found on most all
transformers in better built amps runs about 12% to 14% (I always
figure 13%). I've never seen the 10% value listed in most books.
Heathkit had worse regulation than this in a few models starting at
15%, and one at 30% if I recall. I've never heard of 9% regulation
anywhere. I guess you could reach that, but you'd have to have a
wheelborrow full of caps, and a lot bigger transformer than need be.
You have to watch at sizing transformers by weight as that always
don't hold true since it has to do with the design of the core. If you
have a core with a larger window, one with a smaller, and both with
the same weight, the smaller would have more core area and thus be
capable of more watts out. This holds true especially in C-cores. I
always did wish Bill Orr would have never said something about using
weight for transformers in his book.
The problem with a bunch of capacitance is when something shorts,
there's a lot of joules of power released at once which can destroy
more stuff than need be. A ripple factor of 10% or less is all one
really needs. If one sticks with the published formulas by the
transformer manufacturers, you can't really go wrong as they did the
tests to come up with them.
Quote from Amps;
Because the critical choke inductance in a choke input filter varies
inversely with the current, the swinging choke can have less air gap
i.e.
less reluctance than a non swinging choke. This means that it needs
less
wire for any given inductance, and so can be smaller. As the air gap
is
reduced, so the flux density in the iron goes up - less reluctance -
so
the number of ampere turns needed to start reducing permeability
reduces.
Now look at the numbers. Suppose we have a 2kV supply with a bleeder
current of say 50mA, which is still 100 watts of heat.
Now Terman gives Lcrit>R/1130 = 40,000/1130 = 35.4 Henries for a
60Hz
single phase full wave rectifier.
At 1 amp, however, we only need 1.77H.
So you can see why a swinging choke is used.
Interesting question: if the transient performance is so bad, why were
swinging chokes considered OK for Class AB and B audio amplifiers used
as
modulators in AM tx's?
Anyway, back to the point.
At 120Hz, the reactance of 35.4H is 26.7K
If we used a 20H choke, the reactance is 15.08K. Now shunt that with
0.08
microfarad. The result is
1[1/(-jXc) + 1/(jXl)] equivalent to 221.2 Henries, so we've made a
small
choke look like a much bigger one.
the resonant frequency is 125.8Hz. ( f = 1/ 2pi. rt LC)
Now this is all rather idealised insofar as no account has been taken
of
the choke resistance, which will modify the resonant frequency
(resonance
being defined as the frequency at which the circuit dealt with as a
one
port looks purely resistive) and the effective inductance. But you can
see
that adding the 'tuning' capacitor makes the choke look like a much
bigger
choke. Now if a swinging choke is used, as the current increases, the
choke
inductance drops, but that doesn't matter as long as the circuit still
offers more than critical inductance.
My suspicion is that a swinging choke provides worse transient
regulation
because when the load decreases, until the domains within the magnetic
material can start de-aligning, i.e. getting further away from
saturation,
the choke critical inductance is too low, so the volts start to swing
up.
In any case, you have a parallel tuned circuit with varying current,
so
there will be a biggish transient. I would have thought that a fairly
big
output capacitor would cure this from the viewpoint of the load, but
the
rectifiers could see a large PIV. I've never used a swinging choke in
a
tuned choke system, though. Nevertheless, all chokes with iron cores
change
inductance to some extent as the current varies - some more than
others.
As Will said, tolerances mean that very careful choice of components
is
needed if you are to be certain that the choke is tuned HF: if you
tune it
LF, it will look like a lossy capacitor and the whole thing will
revert to
a capacitor input filter. If you get the wrong resonant condition, you
can
get enormous voltages built up too, so this is not an exercise for the
guy
who isn't experienced in working on HV circuits. Tom, W4JI, has a
frightening story about that, which you'll find somehwere in the Amps
archives. I find a Variac very useful at this early stage of the game:
later, when you want to get realistic loads on the PSU, the Variac
isn't
such a good idea - they're not renowned for their regulation.
73
Peter G3RZP
End quote.
Best,
Will
