Re: [TCML] PFC Question (again)

[David Dean <deano@xxxxxxxxxxxxxxxx>]

Hi Bart

On Friday 23 May 2008 10:52:01 pm bartb wrote:
> Hi David,
>
> Thank you very much for this write up! Your link at the top of the page
> regarding PFC pretty much reflects what I learned in school (over 20
> years ago).

Over thirty years for me. Determinants, matrices and slide rules. The closest 
we got to a computer was a teletype machine at the end of the hall. Each 
student in programming class was alloted 15 minutes computer time for the 
trimester. You wrote the program in Fortran, got your cards punched, and then 
hoped it worked. My how the world has changed.

> The memories are still there. However, I haven't worked with 
> it on the job, so it's mostly a reminder (and a good one).
>
> From" my view", if we are to benefit from PFC, it would be between gap
> conduction when the transformer is not shorted out.
O.K.
> I also don't care 
> about resonant cases, but I do care about STR and LTR operation as both
> are very common. For PT's and PIG's, STR operation is the norm.

Probably due to the prevalent use of welders and stuff for ballast.

> For 
> NST's and the like, LTR is common. For an NST using PFC, then the option
> of placement is limited across the input to the transformer.

For LTR the input current is leading already, so adding PFC capacitor would 
not help. In fact it would increase the current draw from the source when the 
gap is not conducting. You would have to use a reactor (inductor) across the 
line to correct the power factor in that case.

My guess is that the leading current when not shorted, lagging when shorted, 
will kind of average out on its own. The key word here is guess.
 
> For PT's 
> and PIG's, it is common to current limit inductively and to PFC before
> the transformer after the ballast. It seems correct to me to connect the
> PFC across the transformer after the limiting is occurring before the
> PIG or PT. Thus, for those transformer types, we have the transformer
> and tank cap to work with in the PFC scheme.
>
The problem though, is that those types of transformers do not give you much 
to work with. You can measure the resistance (at least of the high voltage 
winding) with an Ohmmeter, but it would take a very special meter indeed to 
measure the inductance. It can be done, but it is not as easy as just 
switching the Wavetek 27xt to inductance and sticking the test leads on to 
the high voltage terminals. You will get a reading, but it will not be 
anywhere near the actual value because the core will not be magnetized. And 
while dividing the kVA rating by the output Voltage will give short circuit 
current for a NST or OBIT, a PT or pig is a different animal. There is a way 
to calculate the short circuit current for such transformers based on the 
percent impedance rating, but my 5kVA pig with 1.6 percent impedance 
calculates out to something on the order of 15kA short circuit. Transposing 
that back to the other side (primary and secondary are reversed) gives 250A.

So let us take the case of a pig or PT with STR tank cap. Meaning that the 
capacitive reactance of the tank cap reflected from the HV side of the 
transformer to the LV side is greater than the inductive reactance of the 
ballast. Lets say the ballast reactance is 4 Ohms, and the reflected tank cap 
reactance is 5 Ohms. If we put a capacitor that has a reactance of 20 Ohms 
across the input terminals of the transformer, we will get a power factor of 
unity. 1/(1/5+1/20)= 4. If we move that cap to the input side of the ballast, 
it will still give a power factor of unity in the non shorted condition, and 
help with the shorted condition as well. Not as much as having a 4 Ohm cap 
across the input to the ballast, but if we do that, we overcompensate for the 
non-shorted case. I would guess that the best possible case would be some 
compromise value in between, and that across the input to the ballast. BUT, 
throw in a variable ballast, like a saturable reactor, and all bets are off. 
With a variable ballast that has enough range it should be possible to go 
from STR through resonant to LTR and back again. 

My understanding of the terms STR and LTR may be in error, of course. 
As I understand it, STR means the capacitors value is less than that of a 
resonant size cap. And LTR means that its value is greater. STR more 
reactance, LTR less reactance,  than that of a resonant size cap.
My understanding of resonant size is that value of capacitor that will 
resonate with the inductance that it shares the circuit with, at the 
frequency of interest. Power factor of unity. X(sub-L)=X(sub-C).

> I personally don't use PFC on any coils.

I never have.

> Javatc simply gives a basic PFC 
> ballpark number based solely on the transformer.

If one enters the numbers correctly, It gives precisely the right value for 
the short circuit case. Multiply the resonant cap value times the turns ratio 
squared. Resonance.

> If one was to use 
> PFC's, they would at least have a capacitive range in the vicinity

Call it an absolute maximum value. Range between that and zero.

> (although their mileage will vary with input, transformer, and tank
> cap). I think considering the common transformer sizes used and coil
> dictated cap sizes, PFC's may be a help, but I've never had the need.
>

In the case where the gap is sucking up too much power from the wall, but that 
would be better solved by improving the gap. In the case where someone is 
running STR. Other than that, probably best without.

> I've thought about this PFC value in the past with this exact scenario.
> I've contemplated simply removing the output from Javatc (heck, I don't
> use it). But some have found it handy, and so I've kept it in the program.
>

Maybe label it "maximum" PFC value. And for self-ballasting transformers only.
I do not know. 

One thing though, I think where some folks might get in trouble, is when they 
want to design a Tesla coil around a transformer (PT or pig),  and they 
insert some number into the box called "rated mA". For example, lets take my 
5kVA pig. If I use 5000/14400 or 347mA, it gives me 4997VA, 41499 Ohms, 
resonant cap 60nF, PFC 230uF, static LTR 95.9nF, and  SRSG LTR 166.7nF.

If I use the actual measured short circuit current with the ballast (15/60 
NST, dual primaries in series for 240V operation, both secondaries shorted) 
in place, of 85mA, it gives me 1224VA, 169412 Ohms, resonant cap 15.7nF, and 
PFC of 56uF. Also static LTR 23.5nF, and SRSG LTR 40.8nF.

Considerable difference. 
I run a sync rotary gap, and usually with my 16nF MMC. That is resonant mode. 
Has run many hours over the years with no Terry filter, just a horn gap made 
of two spoons hammered out of some sheet copper directly across the rotary. 
My input VA gives me a Freau factor close to 1.9, so my power factor must be 
pretty good. I also have a pair of radar pulse caps that each have two 48nF 
sections. One cap, the two 48nF 33kV sections in series for 24nF at 66kV, 
puts me in the LTR range. Suggested for static gap. Both caps in parallel 
makes 48nF at 66kV, that is just over the SRSG LTR value. The radar caps are 
kind of lossy, though, and last time I ran like that I fried my ballast, so 
have not fired the coil in over three years. 

Anyway, I think there may be more folks running resonant or LTR with their 
pigs or PTs than are aware. Just a guess. 

> Javatc has born an interesting scenario that I am privy to (and no one
> else is on the TCML). For example, if I take only 2 coilers and have
> them decide what they use and don't use, etc. I will get very different
> answers. But when you compound that with 100 coilers over the years, you
> get amazingly different points of view of what they want in a program.
> Some things you can do well, other things are iffy. PFC is one of those
> iffy things. There's no way to make everyone happy, so every output
> requested by coilers becomes a judgment call of which I am ultimately
> responsible for. I get ridiculed every week for something, but I also
> have the past request from other coilers that helps me balance out the
> ridicule.
>
> Only those who have tried to program Tesla Coil info can relate. Those
> who haven't written and had coilers use it are arm chair cowboys at
> best. They are experts at finding an error and inflating it into mega
> proportions. Two years ago, I was begging for someone else to write a
> program for coilers (I was just bogged down with personal opinions), but
> no one did. I even emailed my core code to a couple prospects who were
> going to take it and write something great for all of us. Nothing came
> of it. So here I am today still defending the program. I don't defend
> the PFC thing however as it is simply a ballpark value and
> experimentation will tell the tale.
>
> I wonder if users would rather not have PFC listed? Deleting stuff is
> easy. I have no problem getting rid of it.
>

Ultimately, it IS you, and not the users, who will have to make that decision.

I will say, however, that I am quite impressed with JavaTC. Over the years I 
have avoided it because of the name, "Java"TC. I have nothing against the 
language, and have done some simple programing in it before. A decade ago. It 
is the VM that is such a resource hog that I do not like. Then the other day 
(or week) when Kris was working on that mini twin thing, and my numbers from 
Ed Sonderman's spread sheet did not match yours from JavaTC. The old tried 
and true spreadsheet had never failed me before, so I thought "I'll have a 
look." Then I saw the website which said JavaScript Tesla Coil Designer, and 
so that was cool. JavaScript runs on the client, I do stuff with it all the 
time. Turns out there are some Java classes back there anyway, so I have to 
close the browser window that displays the view out the back door from the 
security cam, to keep the browser from becoming unresponsive. Anyway, I put 
the numbers for my Medicine coil in there just to see how well it handled a 
small weird coil. To my suprise, it found the sec. operating freq. right on 
the money. 680kHz, right smack on top of KONO AM in San Antonio. It also got 
the primary frequency as being too low, obvious in operation due to the ring 
of short (1.5") streamers around the toroid when operating. Looks more like a 
crown of thorns than a gas burner. If I put an eXacto knife, or screwdriver 
up to it, however, I can pull a single streamer out to 6.5" with ease. It 
also said the tank cap was too large and the gap would never fire. That is 
true too. It has a half wave rectifier and takes about three cycles to pump 
the cap up to firing voltage. Runs at about 20 PPS. Very weird coil. Also 
drew a very nice picture of the coil, to scale.

So, as far as getting someone else to write a TC program, you are probably out 
of luck. You have set the bar too high. Just one of the costs of excellence.

> Take care,
> Bart

later
deano
_______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla