Winding your own HV Transformer

In Transformer Corner Part 3, I looked at how to choose materials for a custom HV transformer. One way was to pull stuff from the junk-box – I did this in my early Prototype. The much, much better way was to use an off-the-shelf core with documented specs.

Let’s look at winding up the transformer. It’s amazingly easy to get a workable result!

Here’s what we’re aiming for!

The Junk-box materials – NOT great…

So what did I do originally at the time of the Prototype? The junk-box core and bobbin I used did not have a nice spec. I had no clue how much inductance I would get per turn. I had no ‘design’. What on earth did I do then?

Simple! I arbitrarily picked some nice-looking wire and wrapped it around and around and around. Every few winds, I measured the winding on my inductance bridge until it reached 80 µH. I noted the number of turns and calculated the secondary turns count.

I then started winding the secondary, trying to keep track of the number of turns… but, Oh, blast! Oh, bother! I forgot where I was!

I wound up working wistfully winding winds with wire without wealizing where I’d wound to.

Finally, after starting from scratch several times, I completed the secondary winding. And, the thing actually worked!  But – only barely! The transformer doesn’t have much power, so the image on the screen is a little dim. And it gets very, very hot…

NOT ideal!

Winding the BETTER transformer

Let’s do it properly now. Here’s the nice design we came up with in Part 3:

• Core: TDK PC44-EPC19
• Bobbin: TDK BEPC19-1111
• Primary and Secondary wire gauge: 0.25 mm
• Primary turns: 9½
• Secondary turns: 133

Winding it up exactly per the design should just work. But, I still like to use my trusty equipment to confirm things along the way, and make adjustments! So here we go:

1. Scrape the enamel off the end of the wire.
   

   Some wire enamel melts off with a soldering iron – but mine doesn’t; have to scrape it

2. Fix the primary wire around one of the bobbin’s pins. I wouldn’t solder it on at this stage, because you may end up starting again later with a different gauge wire, or if you break the thing while winding. Check the wire is actually electrically connected to the pin, otherwise you need to strip more enamel off.
3. Wind the primary around the bobbin. I’d keep the primary bunched up on one end of the bobbin, to leave room for the secondary on the other end – ensuring beyond doubt that they are electrically isolated! (You could also let the primary cover the whole bobbin, and cover it with some insulation before winding the secondary.)
4. Measure the primary inductance when you’re at the right number of turns. You’ll have to scrape a little bit of enamel off the wire to do this, so be careful not to accidentally cut the wire. If the inductance is too low, add a turn or two. Too high? take a turn off. In my case, I measured 77 µH at just 7 turns, instead of 9½! Do I trust the specs and wind an extra turn or two? I should!! But no, I love my antique, likely-misaligned-or-even-broken equipment. I’m going to go with 7 turns and see what happens…
   

   My trusty (?) Yokogawa HP 4260A Universal Bridge – from the university junk pile

5. Check the primary resistance. I don’t (yet) have a milliohmmeter, so the best I could do was to set my georgeous Fluke 867B to measure relative to a shorted probe state. I got 0.08 ohms. This is probably just right for my modified turns count of 7 turns. And it was what we wanted in the first place anyway, so the feel-good factor is there. Yay!
   

   Make sure primary resistance is within bounds

6. Close off the primary by fixing it to another pin on the bobbin.
7. Wind the secondary in the same way. If your primary turns count deviated from the calculation, remember to change your secondary turns. In my case, I wound only 7 turns so my secondary will become 7 x 14 = 98 turns. I actually wound 100 turns just because I felt 98 was an unlucky number.
8. Check the secondary inductance against the spec. As shown below, I measured 9,260 µH, against a calculated value of 0.940 x 10,000 (square of turns) = 9,400 µH. Close – but remember, the spec’s value is plus-or-minus 25%! So this check is far too inaccurate to really be a good indication of success. But again, it feels nice, and it gives me a chance to show off Grandpa’s vintage calculator.
   

   .. featuring a stunning Texas Instruments TI-1250

9. Close off the secondary, and we’re all done!!
   

   A hand-wound HV transfomer!

Testing it out

Now comes the fun part. Wire up the circuit and give it a go!

The circuit, from Part 2

Test circuit – with a voltage doubler to get 300V

In my test circuit, I included a voltage doubler, to get 300V out. If you do this, you’ll need to adjust the resistor voltage divider network that goes back into the regulator’s feedback pin.

Also, these supplies like to be loaded somewhat, so I put an 82K resistor across the output, which means just over 1W load.

IMPORTANT: The circuit has 2 diodes back-to-back across the primary. These are to cap off high voltages that get induced in the primary. Do NOT make any mistakes with these diodes – like, putting them the wrong way around, or sticking them in the wrong row so they are not connected… Just in the making of this blog, I fried no less than 3 regulators – and they ain’t cheap!

LESS (?) IMPORTANT: I doubt I’d ever get sued but just in case I’ll stick this in. What you’re about to power up can most definitely kill you. Yep, 300V at 12.5 mA is enough. So be very careful, and power down before touching the circuit. The electrolytic capacitors will store a nasty charge. Discharge them with a screwdriver whenever in doubt.

So what did I end up with? A beautiful 300V output at 1W!

Beautiful! A clean, steady 300V output

A word on Efficiency

The transformer runs very cool. The regulator runs cool (only warm to the touch, not scalding like in my Prototype!).

But how efficient is this thing? I measured the input current at 0.2A. 9V x 0.2A = 1.8W power consumption, so 1 / 1.8 = 56% efficiency. That is really pretty awful. WHY? There are 3 reasons.

1. I only wound 7 turns. I think my ‘trusty’ YHP inductance bridge is a bit off tune. Maybe stray inductance from the connecting leads contributed. I think I should have wound 9 turns as the calculation said.
2. I used 0.25 mm wire. The calculation said I should have used something thicker, between #28 or #29 gauge. This also would have brought down the resistance, so even at 9 turns I could achieve 0.08 ohms.
3. I used a voltage doubler. In Part 2, WEBENCH calculated values for two quite critical components, the resistor Rcomp and capacitor Ccomp, that form the ‘compensation network’. These values were computed partly based on the ‘ESR’ characteristic of the output filter capacitor Cout1. But since I used a voltage doubler instead of a single capacitor, this characteristic is totally messed up!! Efficiency can be improved by tweaking these values.

And I should add that I’m in too much of a hurry to get this post off. I’m too lazy to spend more time rewinding the thing. I’ll leave further experimentation to the reader!

Practice makes Perfect

I know… this series of posts made it all seem relatively easy… But did I really get it right the first time? NO!

Some mistakes here! Trial-and-error will give you a feel for the decisions involved

In closing… About Production model Oscilloclocks

Production model Oscilloclocks, like the Model 1, do NOT use my home-made transformer. I’m way too fastidious about quality to allow that! No, folks owning an Oscilloclock can all feel safe, knowing that their transformer is designed and wound by magnetics industry professionals, in the USA. These babies are highly efficient, with very little heat loss…

Custom switching transformer – made in U.S.A. to strict specifications

Production model Oscilloclocks use professionally-wound transformers