Primary circuit is now single-phase. Transformer is a 8-shaped core. Top coil is in straight phase, bottom coil is reversed 180 degrees. Center segment is phase shifted 90 degrees using a 120uf 400v motor-run capacitor at each end, this preserves the figure-8 flux pattern of the transformer core. Note that since the original configuration was Y, that means the applied voltage across each leg of the Y was about 250v or so. This means that each leg needs to be jumpered as if it's a high-voltage supply. What isn't shown, is the fan-motor- leave it alone, it will be fine wired just-as-it-is.
All these connections will require pulling the input power terminal strip (carefully) and removing all the original crossconnect jumpers, putting everything back, and reinstalling crossconnects from the front (user) side.
Secondary circuit has been changed from delta, with full-wave rectification, to Y with full-wave rectification. This was incredibly easy- lift three large wires on the top side of control contactor W, wrap them in tape, slide some plastic tubing over 'em and tape 'em down, then make an E-shaped shorting bar, and tie it to the freshly-removed terminals.
All functionality remains, and you get full-snot output. Welds are nice and stable.
Here's a piece of scrap-crap that I did my test-welding on. You'll see other welds- some from testing other machines, and many from previous tests on this welder.
The shiny ones, however, are new.
Prior to converting the output to Y, I was able to get a satisfactory weld, but only at full-crank on the output tap. This yielded 15v/125A at the arc. Converted to Y, it now hustles along at a little over 2/3rds with 25v/190A at the arc.
There are notes of others who have converted theirs to single by simply eliminating the center winding, and running at reduced output. There's nothing wrong with doing it the simple way, but I wasn't satisfied with 65%, as I knew that there was plenty of potential to be put to work here.
I haven't taken a measure-reading of wire speed, nor have I put it in a penetration-depth or duty-cycle test yet, but I have found that, as converted, it will clearly keep up with the single-phase MillerMatic 250 in our company shop. Over the next few months, I'll be using this CP200 off-and-on making a variety of things, and will be able to tell after putting it to hard work, wether the capacitors and wiring structure impose any limitations, but based on it's performance so far, it appears that it's capable of well over 95% of it's original design spec.
I'll be doing a similar conversion on an SRH-333 3-phase welding supply, so look for a thread there, too.
I welcome comments and questions here, but if you post questions or PMs to the thread, please be patient, as my job (yeah, real work) has changed quite a bit over the last eight months, and has already cut into my forum-response time. If you need fast response, Email me at dave(at)kampnet dot net too.
Well, it is similar, in principle, to the way an RPC or static converter works, however, it's not the same. ;-}
The reason WHY I did it this way, and WHY it works, all goes back to the physical construction of the transformer. It's a figure-8 core. In it's original 3-phase operation, you had magnetic flux flowing in a figure-8 pattern. Had I set all of 'em up for same polarity, or one in one direction, and the other two reversed, I would have lots of inefficiency- core losses and reactance limiting current, hence, no efficiency.
By putting that center coil between two capacitors, I've set it out-of-phase by 90-degrees, and if you finger-trace the current flows, you'll see that it still makes a figure-8 pattern through the core.
Now, it's not perfect... phase shift is a function of frequency and reactance, and while capacitance is constant, inductance is not (because the secondary is variable)... so the phase shift isn't precise, but it's good enough to work well.
On an AC induction motor, you've got one more set of 'spinning' windings on a 'spinning' core, and it won't be that forgiving, so I wouldn't bet on it working quite as well. Work? Yes, but not as well.
But someone who comes up with a clever idea, could modify this setup and wind up with a variation-on-a-theme that DOES work... it's all about waking up in the middle of the night with a crazy idea.
One thing you SHOULD be aware of, Cal- the values of the caps work for the CP-200... Peter and I calculated the starting point, and I found the correct value by trial-and-error. Fortunately, it wasn't a highly-sensitive situation, you can fudge by quite a bit, and probably not notice much... but...
IF you do this with a different machine, keep in mind that capacitance required is a function of current flow, frequency, and voltage. Peter explains it very well with the following:
I = 2 * π * F * C * V
where I is in Amperes, π is 3.14, F is 60, C is in Farads (NOT
microfarads) and V is in Volts.
Solve for C
C = (18.6 / 3) / 2 * π * 60 * 240
C = 6.2 / 90477.87 F
C = 68.53 µF
π is Pi = 3.14.
SO... plug in the appropriate data to get in the ballpark. Notice that I used 60... I actually tried 70uF and 67uF, and found no noticeable difference to 60uF... there's enough reactance in the coil at 60hz to put it 'in the range' well enough to work fine... but just so you don't try it on something substantially larger or smaller, and wonder why it doesn't work well... now you know.
Yes, 'truely correct' for only one condition, but it really doesn't need to be perfect to yield reasonable performance. When you have a spinning shaft of an induction motor, load is mechanical, and slip affect orientation of the rotor's field with respect to motor windings' field.
In this case, there is no rotor, frequency is constant. The only thing that changes, is coil current. Really, all I'm doing, is putting enough phase shift into the center core, so that the alternating flux patterns flow in a figure-8. If the center winding is a little out-of-phase from 90 degrees, it won't cause a big problem... just so long as it maintains relative sequence.
In the same respect, an induction motor likes relative sequence, but when you put static load on the armature, you get slip, which has a much larger effect than just my welding transformer.
Finally, the transformer's load is the secondary... which is now Y and rectified, so it doesn't really care about sequence, as long as there IS some sequence.
I've got an SRH-333 that I'll be doing a similar conversion on, and it will require more capacitance. When I set that one up, I'll have a pretty good idea how sensitive the machine is to different capacitance, and I'll make a similar post then.
Kevin- I'll be around this weekend- make some nice, dry project-doin' weather, and we'll roll it out into the driveway and make some metal turn red!
I think the only thing that would munch the duty cycle, would be if the capacitors got really hot. They're really close to the rectifier's cooling air path, so my bet is that it'll be a whole lot stouter than I'll ever work it with 0.030" wire.
I've got some Johnson vacuum variables in my ham-radio-equipment warehouse (the dairy barn), but they're a bit too low in the range to be of use in this situation.
Generally, the formula will get you close enough. In the CP's case, Peter calculated 68uf for these circumstances (I screwed up the math, should'a used a spreadsheet). I had a pair of 60's, and some smaller 10's, so put 'em all in, and tried it. Then I yanked the 10's, and had no perceptable difference, Swapped in some 30's, and it wasn't so happy (pretty growly, and down on OCV). Since the 60's were pretty big cans, and sufficient for operation, I stuck with'em, and left the extras out... less connections, less bleeder resistors... less things to go wrong.
A big THANK YOU to you, Josh, for giving me an economical opportunity to attempt this in the first place. Had you not offered it at a reasonable price, I probably wouldn't have had the opportunity to try, nor would I be able to own a machine of this capability on my 'hobby' budget.
By the way - I did find the problem that'd been plagueing you- one of the conductors going to one of the transformer windings had cracked, and was making intermittant connection, and THIS is why you'd get reasonable performance for a while, then it'd drop one leg's worth of power. With the output being delta, it didn't stop the machine, but it crippled it's output, making it seem like the feed rate was wrong... and about the time you got it set right, it'd establish a fair connection again. I found it fairly quickly, but that's because I had to pull the covers and big terminal strip... it is impossible to see without doing so. I also gave it a good blowing-out with compressed air, and sprayed the transformer's shorting-bar with contact cleaner... aside from that, it's a ride-like-stolen situation... it's one mean metal-burnin' machine. After comparing it's performance to the company's MillerMatic 250, it's clear that this machine is definately worthy of it's ratings.
It'd probably be a waste of resources, Cal. Better way to make the VFD run full-snot on single, is to build a single-phase full-wave rectifier, with caps and inductor to smooth out the DC, and a soft-start circuit ('cause the capacitor will look like a dead-short on powerup) and feed the DC direct to the VFD's buss.
The technique I used here was really based on the welding transformer's core design, and while it'll work fine on other welding transformers, it really isn't going to show high levels of performance in a VFD application, at least, not compared to making a single-phase to DC circuit.
This welder, see... is either sitting there doing nothing, or it's carrying substantial secondary and primary current. there's no real 'in between'. That was the basis for the concept.
That being said, if you're willing to try it, give it a shot. Use the formula above to put you in the ballpark, but don't be surprised if it gets really, really unfriendly as the VFD passes through tertiary resonant points to the L/C/R combination's FO.