Matt’s Power Supply Upgrade

[Matt_Gibson]

Hi everyone,

I decided to start a thread here to document my progress towards an upgraded power supply. Right now, I’m using a car amp and a function generator to power up a homemade ferrite transformer that I bought off one of our members (Steve Haid).

I am going to replace the car amp with a half bridge as well as wind two new secondaries to match my 6 stage full wave CW multiplier.

Here is a picture of the cm300dy-12nf based half bridge and two newly 3D printed coil bobbins.

The plan for the half bridge is to drive some gate drive ICs (inverting and non inverting) ucc37322 and ucc37321 via my function generator before developing a pcb and incorporating a PWM IC.

The transformer will be wound with 24 awg magnet wire, kapton tape in between layers, and a hand crank coil winding machine. I haven’t settled on number of turns, yet.

-Matt

[Matt_Gibson]

Getting closer to winding my new transformer and have a question about insulation: I was planning to use some kapton tape (7.5kV breakdown voltage) in between each layer but see that my Essex wire has something called “ultra shield” coating. Looking this up, it looks to have a breakdown voltage of over 10kV. Would it be overkill to add the tape? Could/Should I just skip the tape and wind each layer right on top of the other?

-Matt

[Matt_Gibson]

The windings have begun! What a royal pain in the ass this process has been…

This is coil #1 and has 470 turns on it. I need more kapton tape (been adding in between layers) to finish at somewhere close to 700 turns (12 layers).

I’ve been keeping each layer at 58 turns rather than staggering or leaving space on the sides. I figure the wire insulation strength ought to be enough.

Here’s a picture of it so far:

[Mark Rowley]

Where is the first leg of your winding? Does it enter through a hole in close proximity to the core or does it drop down along the inside of the bobbin?

Mark Rowley

[Matt_Gibson]

It enters through a hole near the core.

Edit: Should clarify that I have a square ferrite so there’s space for the wire to run without being jammed up against the core.

Here is a picture of my existing transformer that shows how the other end of the coil will be run.

-Matt

[Mark Rowley]

You may have arcing issues with it being that close. I've found that running the first leg down the inside of the bobbin and covered with kapton tape worked well OR if your bobbin is thick enough, drilling a hole down the edge of the spool and feeding in that way to the bottom (best method).

You may also want to wind your coil in epoxy resin. I too wanted to avoid epoxy but ultimately went that way due to internal coil failures and lower than anticipated output voltage. Oil helped but epoxy resin was a total game changer.

Mark Rowley

[Matt_Gibson]

I’ll be using some kapton tape for good measure, but don’t anticipate any issues since I don’t have any with my current transformer. I won’t be operating over 6kV as I’m using a 6 stage, full wave, multiplier.

-Matt

[Matt_Gibson]

Here is a visual of what I’m planning to do to accomplish a center tap transformer for my full wave multiplier.

[Finn Hammer]

Matt,

Regarding your proposed arrangement of the coils, did you by chance read the transformer FAQ?

viewtopic.php?t=14502


Btw, did I hear you complaining about winding 470 turns? Praise yourself lucky you are working on a switcher!
For reference, below here is my first high voltage transformer on an iron core. Carries 10.000 (ten thousand) turns on each secondary coil, and put out 28kVpp, and the secondaries are arranged as detailed in the aforementioned FAQ.

Core is 50 x 50 mm (2"x2")

I made this transformer 25 years ago, and submerged in oil, it has been powering a Museum Tesla Coil til this day.


https://www.youtube.com/watch?v=H1RzN43Ql1I


Good luck!

Cheers, Finn Hammer

[Matt_Gibson]

Hi Finn,

I read the FAQ, did I get my coil orientation backwards?

10k turns per coil is quite the feat! I’m working with a less than ideal winding setup, but have the most trouble with getting the kapton tape layers down nice and smooth. The tape is slightly wider than the bobbins so it drags as I apply it…

I just need more practice. Thankfully I have to do another coil after this one is complete

-Matt

[Finn Hammer]

Matt,

Read the FAQ again, part 2.

Cheers, Finn Hammer

[Matt_Gibson]

Ah, I see. 180deg/flip one coil. This will let me keep the low voltage (grounded end) near the core and the high voltage ends away.

I should have both coils finished tonight and will probably test it out.

-Matt

[Matt_Gibson]

Got my windings finished and transformer hooked back up. Decided to go with 506 turns instead of my 760 planned after an accident on coil 2 at winding 506

Hooked up to my car amp and was able to do some testing/fusing. I see I’m only able to get 30v rms out of my car amp with this new transformer configuration. I guess it doesn’t like the extra load from having two coils now. Before I could get 55v rms. 30v rms on the primary (6 turns) at 20khz got me to 45kV per my metering. Current was 8mA during this test.

Here is my calculation: 506/6*30*1.414*2*6 = just under 43kVdc. So my metering seems pretty close, no?

I guess I’ll have to complete my new power supply half bridge to really test this out since my car amp is being so stubborn.

I’ll add some pictures later. Got to pretty it up.

-Matt

[Richard Hull]

Great work Matt! I am sure you will be happy with the result. Teething problems are normal. Car amps and all audio amps are ok with rated peak powers, but the makers claim 1000 watts 10hz-30khz with ease knowing that no speaker system will ever be attached to run their specs to full CW, RMS value.

This is why H bridges are the way to go. They pass what they pass or blow up trying to do it with a minimum of components compared to a professional 1000 watt audio amp. The low parts count allows the H bridge to be readily repaired and or updated by the user.

Richard Hull

[Matt_Gibson]

Here’s a picture of it a little more tidy:

I’ll be going half bridge to start and probably going H-bridge afterwords.

I need to play with the primary turns. My 12v dc power supply cuts off quickly now that I reduced turns from 7 to 5.

-Matt

[Richard Hull]

Get ready for corona losses, if not arcs from those flying thin wire leads! Lots of KV at 25khz wants to crawl over everything including flat plastic surfaces. (Dielectrics)


Richard Hull

[Mark Rowley]

Very clean work! Gotta agree with Richard, this is going to be a spectacular coronal light show. I know there's oil haters out there but your design would do best being totally submerged. Even under oil I'd replace those thin leads with some decent 30kV rated wire.

Mark Rowley

[Rex Allers]

I think the concern about the arcing or oil is misplaced.

A number of posts back Matt said,
"I won’t be operating over 6kV as I’m using a 6 stage, full wave, multiplier."

Then later when he got 45 kV out he said,
"Here is my calculation: 506/6*30*1.414*2*6 = just under 43kVdc."

So I think that last 6 in the calculation is his 6 stages of multiplier. Out of the coils around 7 kV. I think he should be fine without oil.

Maybe the multiplier needs oil but I think the registered concerns were thinking 40-some kV was coming out of the transformer coils.

[Matt_Gibson]

Rex is correct. I am using this to feed a 6 stage full wave multiplier that IS under oil. I’ll be keeping this transformer under 6kV so as to not over stress my multiplier capacitors.

My previous configuration with the single coil didn’t have any corona issues (that I was aware of anyways). I’ll take a look in the dark to see if there are any obvious signs of trouble.

Now I need to settle on a gate drive method for my cm300 half bridge. Anyone have any objections to using doubled up ucc37322 gate drive ICs? This ought to be good for 18A of current to the cm300.

-Matt

[Finn Hammer]

Matt.
Those UCC373xx chips were the golden standard driving full bridges in the early years of Solid state tesla coiling.
TO247 or minibrick form factor for the switches.
As coils grew bigger with "brick" style igbt's, the need for more powerfull gate drivers emerged.
Your proposal may work, and I like the path you are on, trying stuff out, and learning by doing, so my advice is: try and see. They may get hot and fry, if they do, time to build the upgraded totem pole version with TO220 cased mosfets.
Good luck.
Cheers, Finn Hammer

[Matt_Gibson]

Finn,

You’re right, these were my go to back in my drsstc days, but I never drove anything larger than those 40n60 mini bricks that were so popular. The cm300dy is a much larger beast with much higher gate charge (1200nC) but I’m only going to be running at 35-40khz and much lower current than a drsstc, so hoping that keeps them happy.

If I’m not mistaken, the need for driving the cm300dy and larger bricks with so much current (50A or more) was to allow for much higher tank currents, right? I think that this is what pushed everyone away from the ucc chips.

-Matt

[Finn Hammer]

Matt,
Your guess is as good as mine, and time alone will tell.
Cheers , Finn Hammer

[Matt_Gibson]

Time for an update:

Decided to build a new full wave multiplier using the same capacitors as Finn Hammer (1600vdc Vishay MMKP383) except mine are a little bigger at 100nF. It’s a 7 stage (left room for an 8th stage) complete with 37Mohm (Vishay VR37) bleeder/balancing resistors (suggested on another site). This will replace my existing full wave multiplier (6 stages). I chose to do this because my existing multiplier capacitors were failing at voltages less than their rated voltage. I came across another person that mentioned actually testing these (WIMA FKP1 6kV) capacitors and finding them fail at voltages under their ratings. What a frustrating waste of money? This will fit in my existing housing, under oil.

I am experimenting with a new transformer configuration that utilizes two ferrite x ray transformers connected in parallel for more current capacity since I don’t have any information on them. Wire size is small so I don’t want to push them. This will be run under oil as well.

Finally, I’ve made a lot of headway with my new driver. It’s using the popular UD2.7 drsstc driver board that was inspired by Steve Ward and created by Loneoceans. I modified it to take input from my function generator (30khz) and drive a cm300dy IGBT half bridge. Testing with my existing setup, so far so good.

[Rex Allers]

Matt,
Re: the Xray transformers you have installed for your current design. I ordered one of those. Was it you that shared the ebay listing for the several of these that the seller had? I looked for that post here but can't find it now.

Anyway I got mine in mid Aug. I haven't made any attempt to actually use it but I have made a number of measurements to work out how they might be used. I took it apart to see how it is constructed, where the coils are and basics of windings. I measured the inductances of coils and have calculated the number of turns in primary and secondary and resulting turns ratio. Pretty sure I know the ferrite core used too.

I started to write up what I learned about these transformers but got side tracked again. I need to see if I can find all my notes and get back to writing it up. I'll share here when I have it done.

[Matt_Gibson]

Rex, that would be great! Yes it was me that had posted that link to them.

-Matt

[Matt_Gibson]

Time for an update! I’ve been busy with alot of other projects but decided to get the fusor back online with a power supply upgrade.

I’m using two of those x ray transformers (driven at 40khz) in parallel and under oil to drive the new 7 stage multiplier. Since it’s been offline for awhile, I’ve been taking it somewhat easy before letting it rip:

Running at 10microns, I have pushed it up to 45kV and 20mA. After a few minutes, the fusor is pretty hot and the cathode is white, so I let it cool down. Power supply doesn’t show much stress other than the warm heatsink that the cm300dy is mounted on.

Over the next few days, I’ll keep pushing voltage higher and higher to see just what it can do.

-Matt

[Matt_Gibson]

Down to one transformer now. Had an internal flashover in one of the coils. Upon inspection, the secondary punched through the coil form and arced to the grounded shield that is between secondary and primary. Primary was undamaged. I was pushing close to 1kW (around 60kV and 17mA or so).

[Richard Hull]

Yes, 60kv has a way of transporting itself about the world. This is a level where super human effort is required to have things run smoothly. With real power behind 60kv, burning stuff up is the order of the day. All the best in keeping that last transformer working. HV power work can be such a pain.

Richard Hull

[Matt_Gibson]

I decided to empty the oil, clean the enclosure (had debris in the bottom) and refill with fresh oil. I don’t have any way of vacuum potting the oil. How vital is it? Maybe that was the cause of the failure? The clearances are pretty tight so a trapped bubble is definitely possible. It’s been many months since I last ran the fusor, so I was hoping that soaking this long might have been good enough.

Starting to look back at my old plans of winding my own using the 3D print bobbin…

I think that I’ll focus on figuring out the popping sounds that I hear coming from the multiplier while this last transformer is still working

-Matt

[Matt_Gibson]

So I decided to wind the new coil last night using a 3D printed bobbin. I wound the secondary such that there are two separate coils, center tapped and 180deg out of phase, on the bobbin. Each coil is made up of 5 sections wound with 90 turns each of 27awg Essex ultrashield wire. So, 450 turns for each coil.

I used my inductance meter (10khz signal) on the primary and my scope on the secondary to “verify” that I did this correctly. Here is the scope shot of the center tap and one of the coils.

The core is a very large TDK n27 type core.

Here is my math for what I expect:
Primary Turns = 10
Frequency = 40khz
Vin = 120vdc
Ae = 8.4cm

This gives me a B = 179mT and a step up of 45 times.
120Vin should give me 5400v out.

Edit: This will be submerged in oil. Also, I’ve got the flu so I’ve probably omitted something important

[Richard Hull]

Beautiful work! Isn't it nice to have just a few really valuable electronic instruments to verify your phase arrangement. I am sure you knew about the wire wind direction to get this. I would have wound only two opposite pies about half way and clip lead scope/oscillator test the phase. I'm the guy who would cry at a fully wound failure to phase properly.

What we are sure we know often bites us in th' butt at the end of a job. Best to test and go, test and go, inch-worming to a successful outcome.

Richard Hull

[Matt_Gibson]

Hmmm…Didn’t think about that! That’s what happens when you tinker under the Flu

I had re-read Finn Hammers nice transformer guide, so knew enough that I needed to be 180deg out of phase.

Hopefully this one is the last one. It should be able to handle more than my breaker can support.

-Matt

[Matt_Gibson]

Placed the new transformer in its enclosure and soaking in fresh mineral oil. Still have some more connections to make up and bubbles to get rid of before testing it. Any alternatives to vacuum degassing? Not sure I can find a suitable container big enough to fit this in…

-Matt

[Dennis P Brown]

I can't offer any method that is as effective as vacuum degassing, then oil back filling. The very next best method is elevated temp to dry/remove trace water vapor in the dry assembly. Then putting under warm oil. Then, if available, placing this in a sonic cleaner tank to better remove remaining air.

Otherwise, soaking in as warm oil as safe is the only other alternative.

From the look of your very thick spacing, I am sure your x-former will operate fine if allowed to soak in oil for a while.

Your overall transformer work is very professional. Hope you experience no further issues.

[Matt_Gibson]

I’ve been slowly stirring the oil every few hours to loosen bubbles. It’s looking like all of the obvious ones are about gone now, so it’s just trapped air in windings that has me worried. Hopefully me keeping things in the 5-6kV range will help?

I’m thinking I’ll let it soak a few days and then try it out at low voltage and higher current to warm up the windings. Maybe that’ll help?

-Matt

[Peter Schmelcher]

Matt I dry 3d filament spools and other things in a pressure cooker. I had to reverse or defeat the pressure safety vent devices and print an adapter to an automotive ac vacuum hose thread. Don't need to run the vacuum pump continuously because I installed a diaphragm isolation value in the hose. System lacks a window to see what is going on inside. Could use a USB camera and some LEDs instead of adding a window.
-Peter

[Emma Black]

Matt this new transformer setup looks the business! I hope it performs well. You also have my sympathies, we also have the flu here, get well soon.

You could try adding something that will vibrate and agitate the oil while soaking to help free up the air bubbles, like the things you get for cement.