Isolation Transformer Follies - How to Use?

[jjasilli]

I cannot seem to find any practical information on how to actually put an isolation transformer to use.  Here are my questions:

1. What about grounding: 

           a) on the primary (input side) 

           b) on the secondary (output side) - can the secondary just be left ungrounded

2. What specifically do you plug into the output of the isolation transformer: 

          a) only the device under test?

          b) only the test equipment?

          c) both into the same isolation transformer?

Thanks for your help, guys. 

[PRR]

> put an isolation transformer to use

Why?

If you have a death-trap transformer-less radio or amp, feed its wall-plug from the iso-former. All the rest of your gear is presumably "safe".

Grounding is unclear. Things may work with no ground at all; that's how we usually did it. In the unlikely event of a short inside the iso-former, grounding its core may blow a fuse rather than pass a shock. The radio/amp "may" work better with its chassis grounded.

If you need an iso-former to reduce crap from the wall-power, things get confusing. In the best case you have two internal shields to block capacitive coupling of high-frequency hash. One goes to wall ground, one goes to amp ground, usually. And in this case, line-noise reduction, you may need an iso-former big enough for everything on your bench. A hospital operating-room may all be fed from one iso as big as a trash-can.

[RicharD]

Assuming this is a small 120V isolation transformer, you do not bond either the primary or the secondary.  You bond the case only.  Bonding the primary would either put the neutral and ground in parallel or put the hot and ground in parallel.  Neutral and ground in parallel is uncool.  It creates the possibility of current traveling on the ground wire instead of the neutral wire.  Think ground loop in amplifier terms, similar thang.  Hot and ground in parallel throws sparks, trips breakers, and is basically bad for your nerves.  Now you can bond one side or the other of the secondary.  Heck you can take a couple big ol honkin 25 watt 1K resistors and do a virtual tap like you might do with filaments.  PRR asked the right question, "Why?"

What do you plug into it?  Whatever it is that you wanna isolate.

-Richard

[jjasilli]

Sorry, I should have been more clear. This is for Testing purposes.  I keep reading that the device under test should be isolated from the test equipment. I bought this puppy on an impulse, not realizing it's half the size of a shoebox and weighs 70 lbs!  Then I stumbled upon the problems of wiring it up & then using it:  http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&ssPageName=STRK:MEWNX:IT&item=260349552645

specs:
http://www.signaltransformer.com/ProdPage-PowerIsolation.asp?pPart=DU-3
http://www.signaltransformer.com/Data/Datasheets/DU-SU.pdf

[RicharD]

>I keep reading that the device under test should be isolated from the test equipment.

I've never read any such thang other than isolating your oscilloscope from high potential, but that's what an isolated scope probe does.

[jjasilli]

I guess part of the problem is that I don't understand the problem.  I think one source of danger is if a chassis becomes live.  This could happen due to poor design, or an inadvertent short-circuit of high voltage within that chassis. If that piece of equipment under test is connected, say, to a signal generator and a to listening amp, then then all 3 chasses could be live and dangerous to touch.  If the person doing the testing touches any one of them he could be shocked.  With isolation, only the one short-circuited chassis is dangerous. ???

[RicharD]

I'm of the school, that the chassis should always be earthed.  In fact one of my bench receptacles has long 6-32 plate screw so I have a clip lead point for earthing ungrounded chassis.  Usually the first thang I do is ground the chassis.  I'm a master electrician, ground is my friend.  Shock prevention is fairly simple and absolute: wear rubber shoes, touch only 1 thing at a time.  An isolation won't help you if you have the chassis in 1 hand and B+ in the other, but with proper shoes, you can touch B+ all you want (not recommended) as long as you are not grounded or completing a circuit.  Birds sit on 346KV wire all day long. 

If you want to add safety to your bench, GFCI receptacle brother!

[jjasilli]

Good points.  Thanks!

[PRR]

3KVA(3000VA)?!?!?!?!?!??!!??!!?

I had a whole-house drew less than that.

> I keep reading that the device under test should be isolated from the test equipment.

There are so many things you can read.

Some days I think the Internet "IS" the "million monkeys typing Shakespeare" paradox.

[jhadhar65]

If you decide you don't need it at some point, let me know before you peddle it.  Some of the stuff I mess with actually needs to be worked on using an iso.  Of course, the best solution is to add one to the device permanently, but there's not always room for that.  In those cases, I want to run it through an iso, which I don't happen to have yet.  I end up poking the chassis with a cheap DMM a lot.

[jjasilli]

PRR:  There's plenty of stuff to read about theory.  I find little to nothing about practical implementation.  It seems that the device under test, NOT the test equipment, is connected to the isolation transformer.  But what to do about chassis ground for: a) the device under test vs. b) the test equipment?

As to current capacity, if I got this right: 3000KVA is the functional equivalent of 1800 Watts.  Given a 120VAC wall supply that = 15 amps.  So this puppy can handle the full current of one supply line in my house.  Also I have a 7.5 Amp variac.  So I wanted at least that for an isolation tranny.  This one was pretty cheap and exceeds my variac's capacity.  Some amps, like big Ampegs, have 6 Amp fuses, blah blah. . . so I've got plenty of "headroom" here for cool operation without strain on the test equipment. 

jhadhar65: I'll definitely keep you in mind.  But if I don't use this puppy as test equipment, my alternate plan is maybe to use it as the heart of a diy power conditioning circuit for my home theater system.  That has 3X Stromberg Carlson PA 100 watt, KT-88, power amps drawing almost 200 watts ea of wall power.  That's 600 watts (5 Amps) right there without the TV set, home theatre preamp, etc., etc.  I could go crazy  ::) and feed that through the variac, used as a "manual voltage regulator".  Even without brownouts it seems my houose voltage varies from about 117 - 127 VAC.  On the other hand, I could resell the iso tranny! :D

[jjasilli]

Finally blundered into my answer:  http://www.bkprecision.com/products/docs/manuals/TR110_manual.pdf

I get it now.  So the iso tranny is primarily used to test live-chassis equipment.  A mistake could make any chassis go live, but as Buttery says, it should be grounded anyway.  That makes the iso tranny redundant, though there's nothing wrong with redundancy where safety is concerned.  However, I now believe I'll use my new iso tranny to build an AC power linr conditioner, with power line filtering & isolation for my home theater sytem. 

[jhadhar65]

>So the iso tranny is primarily used to test live-chassis equipment.

Oh, yes.  Sorry, I didn't realize that's what you were asking.  Usually, these are devices without a power transformer.

I've started tinkering with tube radios now and transformerless designs are quite common with those.  One side of the line - presumably the neutral - is connected straight to chassis.  Neutral was also presumed to be grounded as part of the building wiring.  That actually works fine if all the presumptions could be guaranteed, which they can't.  The line plugs used were non-polarized, so the user had no way of knowing which one was supposed to go to neutral.  To make matters worse, the wall plug couldn't be counted on to be compliant with 'left neutral, right live', so there was always a 50/50 chance of connecting the metal chassis straight to the live side.  The radios (and amps) actually functioned fine like that, though, other than being a serious shock hazard.

The manufacturers recognized the problem, so they isolated the chassis inside wooden, plastic, or bakelite cases and knobs so that the user couldn't come in contact with any metal parts during normal operation.  This didn't help the exposed heads of metal chassis and back screws located on the back panel and under the case, though. Adding insult to injury, the on-off switch was often part of the volume pot and, to keep noise down, this was usually in the neutral line instead of the live side.  Leo did this, too.  UL testers would have a coronary these days over stuff like that.

There were tubes manufactured back in the day that were intended to run from straight from line voltage, too.  I built an AM transmitter sometime back that I use with my radios to broadcast old radio shows I download or stream.  I used a 117M7 for that one.  As you can tell from the nomenclature, the heaters are designed to run from the line at 117V.  I used a small iso transformer with it, but it didn't have a high enough current rating for the heaters, too, so I still run those from the line.  A heater cathode short could be an issue, but I'm aware of it and I'm the only one who messes with it.  I don't actually ground the heaters at all on that one.

[jjasilli]

To confuse matters more, it apppears that some new consumer devices like TV's are back to live chassis, because the consumer is protected by outer plastic housings.  Not exactly a guitar amp concern, but good to know. 

[jhadhar65]

>...it apppears that some new consumer devices like TV's are back to live chassis...

No way!  How are they getting them through UL?  Those guys are relentless when it comes to earth ground.  Do you have a link to something?

[jjasilli]

Do you have a link to something?  It's in the BK Precision iso tranny manual, on-line.  See a few posts up in this thread.

[jhadhar65]

That link never worked for me, but no biggie.  I believe you, of course, just surprised that sort of thing is still going on in modern day production and UL allowing it.  Heck, you can't even tie the safety ground to a PT bolt like Hoffman does without getting their shorts in a wad.  It has to be at least 2" away from anything else or they reject it out of hand.  I'd think new production hot chassis would give them a rash for months.

[jhadhar65]

I did a quick search and found this:

Electrical equipment that does not utilize a three-prong AC cord does not have an earth grounded chassis.  In this case UL, CSA, and EU standards require that hazardous live voltages must have double or reinforced insulation (two layers of adequate insulation).  This is also referred to as Class II insulation.  Transformers of Class II products also have to meet the insulation requirements.  Parts that are not insulated must meet specific “creepage and clearance” requirements (which is to say there must be adequate space between conductive parts and hazardous live voltages – the air is the insulation).

I wasn't thinking about two-prong plugs.

[jjasilli]

That link never worked for me, but no biggie.

I changed the bk precision link; it seems to work now.  Thanks for that Klipsch link!

[PRR]

> That actually works fine if all the presumptions could be guaranteed, which they can't.

The KEY assumption in a transformerless radio, as you say is that the user can NOT touch the circuit. Radio waves come in, sound waves come out, through plastic case. Pot-guts and switches really should be insulated from case. Screw-heads on back were just stupid, and the last of the bunch really were dead on the outside. Double Insulation rules also came in, but maybe later. Creep Distance was more a Euro thing; not such a problem on US 110V, but UL now harmonizes with EUR and anyway nobody makes a special high-creep product just for the US market, most products are "global".

> UL testers would have a coronary these days over stuff like that.

This stuff was all UL. But UL works slowly, and sometimes a new product slips through a UL mis-assumption and floods the market before UL can update their standards. And by the rules of the game, if UL does not have a rule to objectively refuse a design, they have to go with their own rules and approve it. (Lately UL has been more proactive about borderline cases.)

The difference between 50B5 and 50C5 is a UL rule-change. The B had plate next to heater pin, the C moved it 'cuz even at 115V, it was just too close.

Transformerless gitar amps violate this all-plastic assumption. Standard gitars have one side of the signal very exposed to fingers. What it is: old motor varnishes were not the best, stuff leaked, there was a standard for maximum leakage. Your new Waring blender might be 0.007mA leakage, or it might be 0.7mA when old and damp, but it was not allowed to leak 700mA. I forget what the limit was, a couple mA. OK, that's sorta-safe, though quite tingly. However the gitar-amp makers realized that the signal current was much less than a mA, so if they "leaked" a whole mA, they could get good signal into the first tube without violating the limit. Yeah, but the limit assumed that most samples most of the time would be much-less than the maximum, now we have a class of products that pushed the max every time. Rare accidents became not so rare. The limit was reduced, some other requirements added, and also pretty soon a crap transistor amp with PT got cheaper than a crap tube amp without PT, so the deathtraps vanished.

> some new consumer devices like TV's are back to live chassis

Most large TVs have always been hot-chassis. The power demand is so high, that skipping the PT is worth a lot of trouble with case and I/O connections. Anyway, there's 25,000V in there, you can NOT let idiots muck-around inside. And you "can" get radio, sound, and light in/out of a plastic/glass box without any copper-connection. (ANT screws go through an RF transformer which can easily isolate line voltage.) This may be changing with switchers (though un-isolated switchers are common). It may be changing with LCDs, because of all the voltages a LCD TV needs, ~~150V is not one of them. It needs ~~3V and ~~900V, and at that kinda step-up/down, you may as well use a transformer.

[bnchwrmr]

PRR: 
As to current capacity, if I got this right: 3000KVA is the functional equivalent of 1800 Watts. 

No, 3000VA, or 3KVA, is functionally equivalent to 3000 Watts (3KW)
Volt-Amps is used in place of watts when the AC line current is out of
phase with the voltage going into the rated device. Light bulbs are
rated in watts because their voltages and currents are in phase, and
they do not return energy back into the line, as inductive loads do.

[jhadhar65]

(Everything he said.)

Very good information!   I'm just now getting back to this post thread.  This is one of the reasons this place is so cool - you just can't get insight like this anywhere but here.  Thanks for delving into this, PRR.

[jjasilli]

bnchwrmr:  I've been reading here and there on the web that when evaluating VA or KVA ratings for use with reactive devices, like computers & amps, to "convert"  that power consumption rating to simple watts, multiply the VA (or KVA) X 0.6.  That's how I converted 3000VA to 1800 Watts.  If you could shed more light on this it would be appreciated; or perhaps I should insteasd ask for your reactance.   ;)

[PRR]

> with reactive devices, like computers & amps, to "convert"  that power consumption rating to simple watts, multiply the VA (or KVA) X 0.6.

Cap-input DC power supplies, as used in PCs and audio amps, are NOT "reactive".

I do not object to a 0.6 multiplier. A 1,000VA transformer is unlikely to deliver much over 600 Watts DC without significant sag or heat.

[bnchwrmr]

Had to go back to my school text book for correction (for me)
and clarification concerning TRUE power, in Watts, and APPARENT
power in Voltamps. Blew about 30 years of dust and cobwebs from
my brain.

The formula for true power is P = V x I x cos(angle), and is
correct for all types of circuits. The cosine is "1" for totally
resistive loads, and "0" for purely reactive loads (whether
inductive or capacitive). The formula for apparent power is P =
V x I, and ignores the fact that the voltage and current are
out of phase. IN A CIRCUIT CONTAINING BOTH RESISTANCE AND
REACTANCE, THE APPARENT POWER DOES NOT EQUAL THE TRUE POWER.
My mistake because of the cosine term.

It is APPARENT power when the voltage and current are measured
separately. and this is what is seen by the power company, and
determines the Voltamp rating of the device. Measuring the
voltage and current is easily done, and along with a wattmeter
(to measure true power), we can use these readings for some
clarification in an example.

   Given:
            P = 813W (measured by Wattmeter)
            V = 220V (measured by Voltmeter)
            I = 6A (measured by Ammeter)
   P(apparent):
              = 220V x 6A = 1320VA
   Cos(angle) = P / P(apparent)
              = 813W / 1320VA = 0.616
   Phase Angle       
              = 52 degrees (from cosine table)
Only when the inductive voltage leads the resistive voltage by
about 52 degrees, does the true power drop to about .616 times
the Voltamp rating.

I was approaching this from the AC line input viewpoint only.
When you add in the losses from the line to the B+ point, in
a guitar amp, computer, TV ,etc., the true and useable power
delivered drops even further, and as Paul Harvey used to say,
"That's the REST of the story!". I guess that what was everyone
else was referring to. I haven't done any research online.

That's my story, and I'm st-t-ticking to it!  :-D

All the Best    bnchwrmr

[PRR]

> is correct for all types of circuits.

No, for all linear circuits.

Rectifiers are, generally, non-linear.

The common capacitor-input rectifier/filter scheme used to turn AC to DC for electronics has voltage and current IN-phase (zero "phase angle") but the current wave is NOT sine. It is narrow peaks.

Simple "COS(angle)" formulas are easy to apply to pure or pretty-good Sines. This covers many practical AC problems. Resistive heaters. Motors. Phase-tuning capacitors. Fault limiting inductors. Even a few "DC" cases: unfiltered raw DC for electroplating, and choke-input DC filters, can be considered "linear enough".

The narrow pulse waveform of a capacitor-input rectifier "could" be handled with COS() functions, IF you broke the pulse down into Fourier sine series and applied COS to all the partials, then combined. But no simple "phase angle" can come of that, since the partials are at different frequencies.

Anyway, before cheap computers, most non-linear analysis was too-too-too-tedious. Vacuum tube amplifiers are nonlinear parts usually used in a "linear" way, so we take the overall trend and call it "linear", "constant", then rough-check to see if the actual non-linearity is acceptable.

The cap-input rectifier filter is such a common case that approximate cheat-sheets were widely published, and around 1938 Schade did a very detailed approximation which is "perfect" for any practical use.

The high peak current in a cap-input filter heats the winding more than you would expect from the DC current you get out. There's no single conversion constant. Today you can have an idiot CPU run calculations for you. If you can get all the necessary specs. Including thermal.

However it usually works out that you need AC VA somewhat bigger than DC Watts. But rarely "twice", unless you need low-low sag. So 1,000 AC VA for 600 DC Watts is usually going to work out OK.

[bnchwrmr]

I was approaching this from the AC line input viewpoint only.
When you add in the losses from the line to the B+ point, in
a guitar amp, computer, TV ,etc., the true and useable power
delivered drops even further,

Yes, in my origional "simple" assumption, I saw only a mostly
effective AC load on that isolation transformer. but whatever
load will be put on it will be affected by what circuitry is
beyond, and probably most will be an AC to DC conversion type
with the intermittent current draw as you illustrated, with
the accompanying inefficiencies, whether a guitar amp, or an
oscilloscope, etc. It is amazing how much power loss has to
be tolerated to get the power we need.

And I don't even know who Schade is.....
My schooling was for an Electronic Engineering Tech, and in no
waqy am I a full blown engineer. Thanks for your in-depth
response.
All the Best   bnchwrmr 

[bnchwrmr]

Better clarification.........

Why is that isolation transformer rated in Volt-Amps in the
first place??

The manufacturer of that isolation transformer gave their
device an Apparent Power rating in Volt-Amps, so that the
maximum current the device could pass safely would not be
exceeded. The VA rating is the maximum equivalent potential
power that could be delivered through the device with total
disregard of any actual losses, because it sets the maximum
limits of available input current at the rated input voltage,
in the primary, no matter what type of load is placed on the
isolation transformer's secondary.

Of course diodes are non-linear, what's that got to do with
the origional Volt-Amp rating?? Whether the load on the iso's
secondary is linear or non-linear, resistive or reactive, or
anywhere in between, simply determines the type of reflected
load, and the type of reflected load sets the phase angle of
the current delivered into the primary of the isolation
transformer versus the input voltage phase angle across the
input at that point in time.

The in-phas voltage and current in the guitar amp power supply
example would be seen generally as a reflected changing mostly
resistive load, having a constant resistive load from the tube
filaments, and a non-continous resistive load from the B+ and
Bias circuits, during the time of the input AC cycle when they
draw current, whereas a different kind of load may have some
reactance, to a greater or lesser degree, and thus moving the
input current phase angle further away from 0 degrees. And the
primary doesn't care that power is wasted as heat in the load
on the secondary, or what the final output true power amount
will end up being, it just needs to be able to deliver that
power.

The source impedance of the power line, and even an isolation
transformer rated at 3000 VA should be able to handle the puny
current spikes from a small transformerless Gibson amp , or an
All American Five radio without too much variation in the input
phase angle, especially after the B+ filter caps have charged up. 

The textbook author is correct when he states this phase angle,
as a cosine function, is true for any input primary current
versus input voltage (thus the rating in Volt-Amps and not in
Watts) no matter what the reflected load, whether anywhere
between 0 and 90 degrees, even with changing current demands
through even just one cycle of the AC line voltage, a la the
tube amp causing the phase angle variation during every AC cycle
because of the rectifiers turning on and off. .

No matter what losses are incurred by the type of load on the
secondary, the Volt-Amp rating also always remains true for the
device, for it is just a definition by design for the maximum
safe delivery of energy to a load.
 
All the Best   bnchwrmr

[PRR]

> Of course diodes are non-linear, what's that got to do with the origional Volt-Amp rating?? ... reflected load sets the phase angle

"Phase Angle" has meaning when voltage and current waveforms are both sine-like.

In a LINEAR circuit, phase angle gives the ratio of real to reactive current. Reactive current is also called "imaginary", because there is no power in it. But in any practical system, there are Real power losses associated with reactive energy. This is an important fact in LINEAR AC systems, and particularly large motors. Big factories can have large phase angle, which heats the utility system without doing real work. Factory power is surcharged for power factor much different than unity.

The problem with NON-LINEAR diode rectifiers is that the current is not even roughly "sine". It is a narrow spike. In fact the "phase angle" is nearly zero: current peak is near voltage peak. But by RMS concepts, the heating value is much higher than you would expect from the average current output. Also the DC sag is strongly affected by resistance versus high-current peaks, so sag is bad compared to a linear load pulling a sine-like current. Hence the VA rating has to be larger than the DC Watts you want.

> an isolation transformer rated at 3000 VA should be able to handle the puny current spikes from a small transformerless Gibson amp , or an All American Five radio

Yes, of course. Don't know why this was in doubt.

But the 3KVA lump costs a lot. Could he use something less to light his AA5?

I ran numbers on a "Champ-like" amplifier. About 340V DC, 48mA DC.

The transformer current is 410mA peak, zero most of the time. Current flows in two 1.5mS bumps per cycle, or 3mS/16.7mS or less than 20% of the time. Although the DC current is 48mA, the RMS current in the winding is 126mA. The winding has to be designed for that 126mA, not the 48mA DC current.

If you accept the 400mA peak, you can confirm the RMS on a napkin. Over 80% of the time, current is zero. We square that and multiply by 80%, still zero. The pulses are roughly half-sine, and we know that 0.4A peak sine is 0.28A RMS for that half-sine. Square that and multiply by 20%. Add the 80% of zero. Now take square-root. I get 0.125A on my abacus, and I know where the 1% error went.

Yes, in vacuum-tube systems, the load is both dead-resistance (heaters) and peaky DC. And for many practical vacuum-tube systems, the resistance load is "similar" to the DC load. An AA5 is mostly heater load. A Champ amp may be more DC load than heater load. But rarely wildly different.

[jjasilli]

But the 3KVA lump costs a lot. Could he use something less to light his AA5?

I got this puppy for $49 + $28 shipping.  Actually I didn't expect to win the auction.  I figured I would at least learn something, and that I did.  Anyway I really didn't pay for the large capacity of this thing, as much smaller iso trannies typically go for much more money, and this is a good quality unit. 

It now appears that I don't really need this thing on my bench.  But I have really dirty AC power here in Brooklyn, NY, so I will use it as the heart of a DIY AC power line filter - for my home theater / hi-fi system.  It consists of 2 complete trannies in 1 package, which can be wired in series or parallel. I plan to wire it in parallel, with 2 separate filtering circuits on the secondaries, one for source equipment; the other for my 3x vintage tube power amp mono blocs.  The latter draw 175 watts of AC supply ea, totalling 525 watts.  One side of this iso tranny is good for 900 watts so I don't have to worry about voltage sag or over-heating.  The remaining half of this iso tranny @ 900 watts is overkill for the rest of the system, but I didn't really pay for it; and it's all in one package - so what the heck.

[bnchwrmr]

jjasilli

I am glad you are going to keep your $71 bargain. If you didn't have
a use for it now, in the future you might. How many of us regret not
keeping something we once had that we could use now?

PRR

Quote:
> an isolation transformer rated at 3000 VA should be able to handle the puny
current spikes from a small transformerless Gibson amp, or an All American Five
radio

"Yes, of course, don't know why this was in doubt."

There was no doubt whatsoever. I was not asking if it could, but stating
that it should, because it was certainly able. "Could" being implied.

Quote:
Reactive current is also called "imaginary", because there is no power in it.

I agree with you! And you are right, this is true only for linear sinewave
function analysis.

Quote:
But in any practical system, there are Real power losses associated with reactive energy.

I disagree with that statement in what it seems to say.
Real power losses are related to resistive losses. This is implied in your statement: "Big factories can have large phase angle, which heats the utility system without real work."

I agree with you on this.
The DC resistance of the power transmission lines consume power as heat, and the DC resistance of the motor windings as well, and is that real (or true) power consumed. The inductance of the motor windings, uses the same current and then returns it back into the power transmission lines, thus being seen as the "without real work" part of your statement. The "big phase angle" is between the greater inductive voltage amplitude vector vs. the much smaller resistive voltage amplitutde vector. The large inductive voltage multiplied by the system current would seem to demand more power than what is actually being consumed, or rather what appears to be needed (apparent power), whereas the smaller resistive
voltage multiplied by the in-phase system current, is the Real power in Watts,
and we could improve things with capacitors [n parallel with the motors, as you know. There are not two currents here with any single motor, the DC component is in series with the motor winding and thus it's inductance.
In an RMS system, reactances store energy, then release it, Thus appearing to
use power. That's why I disagree with that statement.

In the broader view here, I think we've agreed on the main things. Yes, I do
agree with you that rectifiers remove the RMS factor and "force" the filter
capacitors to have in phase voltage and current. My mental stumble here seems
to have been how this is seen by the primary. To me, this would be seen as a reflected changing resistive load, changing during those "20% of the cycle" time when the current spikes. By your Champ and AA5 examples, you seem to be saying the same thing.
And some "lily-guilder" (and I like your term, it's pretty funny) could argue that "Even DC has a phase angle...it's always zero!" (Yeah, go away and paint some flowers.)
I agree that "Hence the VA rating has to be greater than the DC Watts you want." I was saying the same in a different way about the VA rating being a maximum limit of what could be delivered vs. what would be under real usage conditions. You have numbers to illustrate my statement about what is reflected back into the primary when the rectifiers turned on and off,

Thanks for the graphic and the numbers. I will save them for reference and
thoughtful perusal. I might ask for expansion about other statements later
down the road, if you don't mind. 

How much real wattage will jjasilli demand from that iso? The VA rating will
just be a guidline. And "Could he use something less to light his AA5?" We all know the answer. And fortunately, not so expensive a "lump" vs full pprice.

But, we are guilty of batting around different balls here  Wasn't "how to use
this for efficient isolation" the origional question?

All the Best and thanks again for your time and making me think.
bnchwrmr