Building a Hybrid Toroid Transformer

[Matt Watts]

Progress is slow, but good...

[Matt Watts]

And for those that may be curious...

It's not a dud, but using a push-pull driver is about sixes.  Efficient, but no OU so far.

Next test is going to be driving it with an audio amp.

Stay tuned...

[Grumage]

And for those that may be curious...

It's not a dud, but using a push-pull driver is about sixes.  Efficient, but no OU so far.

Next test is going to be driving it with an audio amp.

Stay tuned...

Dear Matt.

May I be the first to congratulate you on such a fine pair of cold castings.? They look like off the shelf !!

Lets hope your results come out as well.   

Cheers Grum.

[Matt Watts]

May I be the first to congratulate you on such a fine pair of cold castings.? They look like off the shelf !!

Yes I must admit, when I reached into the pan of melted wax with the tongs and pulled the first casting out, I thought I was in the twilight zone--couldn't believe how perfect it looked.  Never saw a large pot core like this before.  My measurements where pretty much spot-on too--the wound inside core dropped right in snug as a glove.

With this particular material, having a thin coating of wax all around it really does give it that factory look-n-feel.  The top part where I made the pour is the only place where things look a little ragged.  I tried sanding a little with emery cloth, but the material is really rather hard.  The more I worked on it, the quicker it became obvious it probably doesn't matter anyway.  So I left it alone and just put everything together.

I did forget to mention...

When you add the hardener to the Devcon steel mixture, it is really runny.  To thicken things up a bit, I added quite a bit of S-280 steel shot to the mix, poured then tapped it down into the mold with the flat head of a nail.  I could see bubbles coming up so I kept working at it until I could tell the mixture was starting to set up--probably about 30 minutes.  I'm sure if one had a vacuum chamber you could get almost all the bubbles out.  I don't have anything like that so I just did the best I could tapping and vibrating the mold.

Anyway, not too shabby.  I'm using a nanocrystaline inner core instead of the megaflux core I initially wound.  Pretty certain there is a lot of room for improvement.  Need to run through a series of tests and see which avenue looks most promising.  Brad being in the lead with this particular device may share some tricks he has discovered.

[Peterae]

Incredible work Matt, looks amazing.

[Matt Watts]

What everyone has been waiting for, power measurements...

I used two precision shunts for current measurements:
  0.01 ohm 1%  on the input    (A channel probe set to amps @ 100X)
  0.20 ohm 1%  on the output  (A channel probe set to amps @    5X)
The scope measurements reflect actual amperage readings.

Sine wave signal produced from BK function generator into a Pyramid PA amp with primary connected to 4 ohm output.

So I calculate:
  Input power =   2.08A    *  9.60V   =  19.97 Watts
  Output power = 0.310A  *  64.0V   =  19.84 Watts

Efficiency = 99%

Not bad for the first real attempt.

Anyone spot obvious measurement errors?

Any suggestions what I should vary for the next trial?

[Centraflow]

Hi Matt

What were your winding ratio's? or did I miss them somewhere!!

Just curious on the voltage output around 7:1, also would be interesting to know what the impedance of the primary coil is as you are driving with a 4 ohm impedance matching from your amp, just wondering how much reflected power there might be if any.

99% is very good, does it change with frequency?

Regards

Mike

[Matt Watts]

What were your winding ratio's? or did I miss them somewhere!!

22:155  or just a tad over 7 to 1, step-up.

So I cannot concur with Brad's 2 to 1 voltage increase having 100 turns each, primary and secondary.  I would have thought in a step-up configuration, Brad's demonstrated effect would be magnified.

Just curious on the voltage output around 7:1, also would be interesting to know what the impedance of the primary coil is as you are driving with a 4 ohm impedance matching from your amp, just wondering how much reflected power there might be if any.

With the relatively low turn count on the primary, I'm pretty certain if I had a two or one ohm output on the amplifier I could achieve better matching.  Hooking it to the eight ohm output shows a clear decrease in power transfer.

99% is very good, does it change with frequency?

It runs rather flat between 550Hz and 2100Hz.  Above that it begins to slowly roll-off.  Below 550Hz I see what looks like clipping of the sine wave, but only on one of the peaks, not both as I would typically see.

[TinMan]

What everyone has been waiting for, power measurements...

I used two precision shunts for current measurements:
  0.01 ohm 1%  on the input    (A channel probe set to amps @ 100X)
  0.20 ohm 1%  on the output  (A channel probe set to amps @    5X)
The scope measurements reflect actual amperage readings.

Sine wave signal produced from BK function generator into a Pyramid PA amp with primary connected to 4 ohm output.

So I calculate:
  Input power =   2.08A    *  9.60V   =  19.97 Watts
  Output power = 0.310A  *  64.0V   =  19.84 Watts

Efficiency = 99%

Not bad for the first real attempt.

Anyone spot obvious measurement errors?

Any suggestions what I should vary for the next trial?

99% efficient-not bad at all.
I must ask-do you still have the air gap between the pot core and inner secondary?. If so,then i wonder what the efficiency would be if that air gap is eliminated as it is in my HTT ?.

Look at what you have achieved at home on your workbench Matt-the inferior materials used to that of what can be used(E.G-ferrite)-everything going against you,and yet you still achieved to produce a transformer that is 99% efficient.


Brad

[TinselKoala]

Matt, how are you connecting your scope to the current-monitoring shunts? Are you using normal probes on the 1x setting, or what?

[Matt Watts]

99% efficient-not bad at all.

Thanks goes to you Brad.  I had an idea, but needed a mechanic like yourself to show me how to go about it.

I must ask-do you still have the air gap between the pot core and inner secondary?. If so,then i wonder what the efficiency would be if that air gap is eliminated as it is in my HTT ?.

It's a snug fit, but certainly the individual copper wires have no coating around them like your initial prototype.  When I was speaking to my son about the turns ratio not behaving like yours did, he thinks I need to goop-up the inner core with Devcon and mash it altogether, make it a one-piece unit.  I tend to agree.  There is something about how the magnetic field circulates in this device that is unlike a normal transformer.

Look at what you have achieved at home on your workbench Matt-the inferior materials used to that of what can be used(E.G-ferrite)-everything going against you,and yet you still achieved to produce a transformer that is 99% efficient.

Brad

There is definitely room for improvement here.  I feel pretty good about this first test, the power levels and performance achieved.  I'll likely keep this one as a baseline and venture out slowly with minor changes in geometry and materials.  I'm also open to any and all suggestions you have--this device is as close as I have ever come to the allusive goal.

[Matt Watts]

Matt, how are you connecting your scope to the current-monitoring shunts? Are you using normal probes on the 1x setting, or what?

Normal yes, set to 1X on the physical probe, yellow channel.  Both channels are set DC bias.  Common ground used for both probes.

On the higher current input side, the scope channel is set to 100X and connected across the 0.01 ohm shunt.

On the output side, the scope channel is set to 5X and connected across the 0.2 ohm shunt.


Only on the output voltage do I set the blue channel probe to 10X and set the scope channel to match.


For the testing I leave all shunts in-place, both input and output.  I keep the system running stable and just move the probes (since I don't have a fancy four channel scope like yours).  This to me is actually an advantage here because any scope or probe error should be identical between input and output measurements--i.e. if the channel is 2% low, it will be 2% low on both sides.

The load is a 75 watt 110 volt halogen spot light and when driven with a known 20 watt source has identical brightness--checked just for consistency.

Using a digital handheld meter, I can confirm the scope readings are all in the ballpark, with the scope readings likely being more accurate.


This is your ballgame Tinsel; if there is anything you are suspicious of, let me know and I'll recheck it--I get no paycheck for this gizmo either way.  Being shy of the 130% efficiency anyway, might as well be sure the measuring is done right now, especially while I have everything still setup.

Oh, one thing I wanted to ask about, the phasing...  I'm a little surprised the current and voltage are exactly in-phase.  Typically a little offset is noticed.  Is this because the device is actually running at full load or is there some other reason?

[TinMan]

Thanks goes to you Brad.  I had an idea, but needed a mechanic like yourself to show me how to go about it.

It's a snug fit, but certainly the individual copper wires have no coating around them like your initial prototype.  When I was speaking to my son about the turns ratio not behaving like yours did, he thinks I need to goop-up the inner core with Devcon and mash it altogether, make it a one-piece unit.  I tend to agree.  There is something about how the magnetic field circulates in this device that is unlike a normal transformer.

There is definitely room for improvement here.  I feel pretty good about this first test, the power levels and performance achieved.  I'll likely keep this one as a baseline and venture out slowly with minor changes in geometry and materials.  I'm also open to any and all suggestions you have--this device is as close as I have ever come to the allusive goal.

Yes,leave that one as it is,and build an identical one,but where as you fill in the air gap around the inner windings with the devcon putty,and then carry out the same test.

Also remember what power you have not yet calculated,and that is the power dissipated by the coils in the form of heat--and i expect at those power levels,you may have a little more than that 1% you are missing.

[TinMan]

What everyone has been waiting for, power measurements...

I used two precision shunts for current measurements:
  0.01 ohm 1%  on the input    (A channel probe set to amps @ 100X)
  0.20 ohm 1%  on the output  (A channel probe set to amps @    5X)
The scope measurements reflect actual amperage readings.

Sine wave signal produced from BK function generator into a Pyramid PA amp with primary connected to 4 ohm output.

So I calculate:
  Input power =   2.08A    *  9.60V   =  19.97 Watts
  Output power = 0.310A  *  64.0V   =  19.84 Watts

Efficiency = 99%

Not bad for the first real attempt.

Anyone spot obvious measurement errors?

Any suggestions what I should vary for the next trial?

Do these calculations include the CVR's power dissipation ?
If not,then the power dissipated by the .01 ohm CVR is 43.2mW,and 19.2mW by the .2 ohm CVR.
This also dose not include the power being dissipated by the two windings them self-in the form of heat.


Brad.

[Matt Watts]

Do these calculations include the CVR's power dissipation ?

I did not account for the power dissipation of the shunts.

If not,then the power dissipated by the .01 ohm CVR is 43.2mW,and 19.2mW by the .2 ohm CVR.
This also dose not include the power being dissipated by the two windings them self-in the form of heat.

I get the same values for shunt dissipation.  So do I subtract the 43.2mW from the input and add the 19.2mW to the output?

If so, then we have:
  Input 19.97 - 0.0432     = 19.93 Watts
  Output 19.84 + 0.0192  = 19.86 Watts

Efficiency of 99.6%.

You guys will have to help me out a little in this area, +/- 5% has always been good enough in the past.  Not so true when you're on the ragged edge like this.

My thinking is that if you lose power to a shunt on the input, that is power that never makes it to the DUT, so you subtract.
Then power lost to a shunt on the output is power that made it though the DUT, but never made it to the load, so you add.

If that's wrong, please explain it to me better.  If there is an argument, I'll just get some current probes and hopefully eliminate all doubt.


And for the dissipation of the windings themselves...   I guess I would have to put the whole device in a thermal bath and run it over some length of time to see how many BTUs/Hr it releases.  Unless someone knows an easier way...?

[wattsup]

@MW

Niceroni.

First thing that I would like to eliminate is any possible adverse condition that could be caused by the secondary leads of the hybrid that pass through the outer core. Maybe consider sliding a shield over the leads where they pass through the core. This will make sure the output is not bypassing the inner core or creating a blockage from that very potent impress point. This is first just to cover all bases. It's either less, same or more. If it is much much less, then there is a problem to consider. More would be nice but doubtful but if it keeps ticking the same great bobbles of goulash, it's alive.

I would then try a few comparable sized transformers one by one and connect the primary in series with the hybrid primary on the side that is not receiving the pulse, then test it. If that transformer has a secondary, put a small bulb on it as this could be quantified later but for now all you need to see is the effect. You will know if there is anything special.

Then, I would wind the second inner core you ordered from ebay in quad and try it again inside your hybrid core. Could be quad all parallel inside with two exiting leads or all leads coming out (latter is better). Any leads coming out through the outer core needs to be shielded. Maybe a thin 8 conductor shielded cable would be good. So will it produce more or less? If it produces more output, then you know that more secondary sections means less cancellation before the output can exit the inner winding. This is what I am suspecting with your present output. Eben though it is in the 99%, a good portion of the output is cancelled before it can exit due to the single full winding. For the primary first do the same one as before to make the comparison then try it with 4 three turns either in Tesla succession mode or all parallel or 2 opposing parallels then both sets parallel.

The other completely far end of experimenting would be by using the inner core as the primary. Then use a small pick up coil to try and map the outer surface that would then show where to start and stop winds for 1 or 2 output secondaries. This is the one I am interested in the most. Seeing if the growth or gain can occur via an expanding core.

wattsup

[TinMan]

I did not account for the power dissipation of the shunts.

I get the same values for shunt dissipation.  So do I subtract the 43.2mW from the input and add the 19.2mW to the output?

If so, then we have:
  Input 19.97 - 0.0432     = 19.93 Watts
  Output 19.84 + 0.0192  = 19.86 Watts

Efficiency of 99.6%.

You guys will have to help me out a little in this area, +/- 5% has always been good enough in the past.  Not so true when you're on the ragged edge like this.

My thinking is that if you lose power to a shunt on the input, that is power that never makes it to the DUT, so you subtract.
Then power lost to a shunt on the output is power that made it though the DUT, but never made it to the load, so you add.

If that's wrong, please explain it to me better.  If there is an argument, I'll just get some current probes and hopefully eliminate all doubt.


And for the dissipation of the windings themselves...   I guess I would have to put the whole device in a thermal bath and run it over some length of time to see how many BTUs/Hr it releases.  Unless someone knows an easier way...?

If you did not account for the CVR on the input as being part of the input,then you dont subtract it from the input.
On the output side-if you did not account for the CVR power dissipation,then you must add that to the total output power being dissipated. There is a way to calculate the power being dissipated across the two coil's,but i am not well versed in this area yet--so maybe ION,Poynt or one of the EE guys could help there.

Yes,+/- 5% dose not cut it when the measurements are this close. Even higher up in the wattage ranges like say 100 watts ,the +/- 5% can mean a difference of 5 watts either way.

Brad

[ion]

Matt

If you get to the point where you wish to do a thermal analysis of the entire transformer plus load, you can get some ideas for differential test techniques on my bench here:

http://www.overunityresearch.com/index.php?board=40.0

I prefer differential testing over baths as the results are real time and can be very exact.

But to each his level of comfort.

[TinMan]

Matt

If you get to the point where you wish to do a thermal analysis of the entire transformer plus load, you can get some ideas for differential test techniques on my bench here:

http://www.overunityresearch.com/index.php?board=40.0

I prefer differential testing over baths as the results are real time and can be very exact.

But to each his level of comfort.

ION

Is there not a way to calculate the power dissipated by each coil by knowing the coils resistance,and the voltage across it?. I think i remember one of you guys telling me how on one of the other HTT threads,whether it was here or at OU.com-im not sure,but i have forgot how it was done that way.

Brad

[ion]

ION

Is there not a way to calculate the power dissipated by each coil by knowing the coils resistance,and the voltage across it?. I think i remember one of you guys telling me how on one of the other HTT threads,whether it was here or at OU.com-im not sure,but i have forgot how it was done that way.

Brad

It can be done, but it's like trying to hit a moving target, since copper has a positive temperature coefficient of resistance vs temperature, the measurements need to be carefully made considering the cold vs hot temperature of the coil.

The thermal methods I like to use take all these variables into account. All copper losses, eddy losses and thermal power delivered to the load are summed in a final temperature rise and compared to a known input power to a separate control resistor's temperature in a null balance arrangement.

The various implementations of thermal methods are described on my bench in a few threads. If there is more interest I can detail it here.

[Matt Watts]

Some food for thought...

I opened up the two outer core halves and noticed they were not fitting together tightly.  Apparently the inner stem was just a few thousands taller than the outside edge.  Hmmm...  So I proceeded to place a piece of emery cloth on a solid piece of glass and lap the surfaces so they would be equal in height.  Took a few hours but now each core halve sits perfectly flat and when assembled there is no perceivable gap.  Thinking this would improve the performance I retested.  Any guesses what happened...?

Ninety one percent efficiency now--basically, I killed whatever effect might have been happening and now have pretty much a conventional transformer.

So it appears an outside gap may be necessary.  I have no idea how much, but I will see if I can find or make some steel washers to fit between the inner stems and try to at least get it back to where it was, possibly better.

Certainly did not expect this to happen, but after watching these videos and thinking a while, I may have some idea what is happening.
https://www.youtube.com/watch?v=lHT9xI01sqw
https://www.youtube.com/watch?v=L3w_0y8VYAE

[wattsup]

@MW

Make sure you wound the outer primary exactly like is was before relative to the leads of the inner coil.
You have images of the original so compare them. Or remove the primary wind, wind a three turn primary that can slide around as you pulse it and scope the secondary for the highest output. Then start your primary there and try again.

Bummer man, but not the end.

wattsup