toroidal transformer questions

[LDSisHere]

I am building a project that uses a toroidal transformer with one input and two outputs at a 1:1 ratio. The instructions for the project say that you can use 24awg wire from phone wire or Cat5 for this purpose. I have already learned that using a powered iron toroid will not work, it has to be ferrite.

I want to know what the best technique to wrap a toroidal transformer is. In my attached pictures you can see my various attempts. The easiest way for me was to twist the three wires together before wrapping it around the toroid but I am not sure how good this is to do from an electrical standpoint. I also have some wire that has the strands joined together but the problem with this wire is that the insulation is thicker than the other wire. When I tried using just three wires paralleled they would not stay tightly together. Of these three options which would be the best choice or does someone have a better idea?

Also what effect would using a larger toroid have? What would using more turns do to the transformer? The plans call for 14 turns, which works, but the chip that drives the transformer gets hot after a short period so I would like to make it us less power, any suggestions on how to accomplish this would be very much appreciated.

[redwire]

Re: toroidal transformer questions

What is the transformer for? A DC-DC converter, audio etc. Pics? For most apps, the primary is wound first with the secondary on top. For audio applications, sometimes windings are bifilar wound (as you have done) but it increases mutual capacitance between all windings.

If your driver IC is heating up, you need more turns or increase the frequency or your core is small or the windings are not phased properly.
Where did the 14 turns number come from? You can do some math to verify, if you know the ferrite core type/size. Less primary turns = lower flux

A larger core will have a lower flux for the same number of turns, so you should compensate buy adding secondary turns. Slightly higher losses because each turn needs more wire.

[LDSisHere]

Re: toroidal transformer questions

This is for a plasma speaker.

http://www.instructables.com/id/A-re...speaker/#step1

I have the two following toroids.

http://www.digikey.com/product-searc...ds=495-3862-ND
http://www.digikey.com/product-searc...ds=240-2139-ND

Right now the speaker will work, but I think it can be better. The distortion seems to get worse as the driver chip gets hotter.

[redwire]

Re: toroidal transformer questions

The SG3525 seems to be setup for about 31kHz PWM.
The Laird core is for EMI filtering 300kHz-30MHz so it's not what I would use. These cores are lossy and for HF; it prob. saturates.
The Epcos core is best, doing some math it will run about 140mT with 14 turns; lots of headroom before saturation. Junkbox PC power supplies are also good source for cores.
If the chip is roasting with the Epcos core, check the phasing (dots) and your windings match (one winding is reversed).
You can add a small resistor 10R-22R? feeding pin 13 to limit the IC's output current, as long as you still get over 10V gate drive for the big MOSFETS.

I don't like the missing output cap on the LM7812; I would have a 100-470uF to GND there for stability. Try that.

[LDSisHere]

Re: toroidal transformer questions

Redwire, thanks for taking the time to respond, you have been very helpful.

The Laird core is for EMI filtering 300kHz-30MHz so it's not what I would use. These cores are lossy and for HF; it prob. saturates.

I did not try this toroid on my speaker, I am glad I did not waste my time trying.

The Epcos core is best, doing some math it will run about 140mT with 14 turns; lots of headroom before saturation.

This is the core that I am using and the one in my first picture where I twisted the wires prior to wrapping the toroid. Would there be any benefit to wrapping the toroid with the wires parallel?

Junkbox PC power supplies are also good source for cores.

All the toroids I found in PC supplies seem to be the powdered iron core type which I originally tried. When using these the SG3525 chip would get hot enough to blister skin in just a few seconds.

If the chip is roasting with the Epcos core, check the phasing (dots) and your windings match (one winding is reversed).

The phasing is good because the speaker will work and the FETs have not fried. The one winding is reversed on the PCB.

You can add a small resistor 10R-22R? feeding pin 13 to limit the IC's output current, as long as you still get over 10V gate drive for the big MOSFETS.

I will give this a shot when I get a chance. The datasheet for this chip did not have much information I found useful, usually they at least give a description of what each pin does.

I don't like the missing output cap on the LM7812; I would have a 100-470uF to GND there for stability. Try that.

I am actually using a LM7815 instead of the LM7812. I did add a 4.7uF and 100 nF cap on the LM7815 output when I built this speaker, I could easily put in a bigger cap if it will make the chip run more stable.


I was also thinking of using a driver chip between the SG3525 and the transformer. I have an IXDN602PI I could use for this and I do not think I would have to worry about it overheating.

[eccerr0r]

Re: toroidal transformer questions

This is weird, you shouldn't be dissipating much power in the toroid transformer unless the core is extremely lossy at the frequency you're working at, or your drive is too high. And due to the nature of a transformer/inductors the gates can't be stuck on, which will help against frying them (current too high or gate drive too low are usually what kills mosfets.)
Is there a reason why you're using 7815? I'd be worried that the output of the PWM is too high (sorry did not read the datasheet for the pwm on its output driver and how close it gets to the rails), resulting in the transformer voltage output getting very high, and it won't help drive the gate because the zener diodes will just cut you off at ~12.7 V anyway. Anything higher will simply heat the zeners, 22R resistors, and that transformer.

I don't know if there's a reason why the original designer chose a toroid. I hate winding them and try to use split core bobbin transformer with just because they're a heck of a lot easier to wind.

[redwire]

Re: toroidal transformer questions

Yeah, the chip shouldn't get that hot. The SG3525 is good for 100mA output max.
I found if you try take that much or more current, the IC goes unstable due to the single GND pin it has. So a driver/buffer after it really helps (like the IXDN602PI). That IC gnd point (pin 12) is critical, it should run to the star-ground.
You could try run it with no power to the big mosfets, just to see if they're loading down the toroid too much.

[LDSisHere]

Re: toroidal transformer questions

Is there a reason why you're using 7815?

The instructions said that if want to drive it with more than 24V then you should use a 7815. Currently I am only using 19.5V@7.75A but I eventually want to drive it with approximately 30V so I went with a 7815.

Anything higher will simply heat the zeners, 22R resistors, and that transformer.

The toroid does not heat up at all during operation and I do not think the resistors or zeners will be overloaded.

I don't know if there's a reason why the original designer chose a toroid. I hate winding them and try to use split core bobbin transformer with just because they're a heck of a lot easier to wind.

If I were to guess I would say it was because that is what he had laying around, but I agree that winding the thing is not much fun.

Yeah, the chip shouldn't get that hot. The SG3525 is good for 100mA output max.
I found if you try take that much or more current, the IC goes unstable due to the single GND pin it has. So a driver/buffer after it really helps (like the IXDN602PI). That IC gnd point (pin 12) is critical, it should run to the star-ground.

I have drawn up a new PCB that uses a driver chip between the SG3525 and toroid. It just may be a while before I have time to etch it and move the components over to it.

You could try run it with no power to the big mosfets, just to see if they're loading down the toroid too much.

I may give this a shot to see what happens, I will just have to cut a trace on the PCB.


I originally wanted to make this because it looked neat and relatively simple but it is actually a lot of trouble to make it work. It is becoming way more of an educational experience than I anticipated but that is not necessarily a bad thing but it is tiresome. I am attaching a picture of the speaker as it is now in case that helps any.

[redwire]

Re: toroidal transformer questions

That looks great, a good build. I can't see anything wrong there.

Grasping at another straw, the gate-drive transformer (output) is clamped by the zeners, and with a 1:1:1 ratio assume say 12V in and 12V out. If you run a 7815, the zeners might be conducting and contributing to the SG3525 heating up. The IRFP250's max. is 20V on the gate but 10V is enough for full on.
An LM317 can take >40V in max. vs. 35V for the LM7812/15

[budm]

Re: toroidal transformer questions

I would add diode in parallel with the Gate's resistor, Cathode to the left, Anode to the Gate, they will help discharge the Gate/Source cap faster so it will not be on when the other Mosfet is turn ON.

[LDSisHere]

Re: toroidal transformer questions

That looks great, a good build. I can't see anything wrong there.

Grasping at another straw, the gate-drive transformer (output) is clamped by the zeners, and with a 1:1:1 ratio assume say 12V in and 12V out. If you run a 7815, the zeners might be conducting and contributing to the SG3525 heating up. The IRFP250's max. is 20V on the gate but 10V is enough for full on.
An LM317 can take >40V in max. vs. 35V for the LM7812/15

I may try a LM7812 just to see if it makes a difference. It will be a while before I can get a more powerful supply than what I am currently using.

I would add diode in parallel with the Gate's resistor, Cathode to the left, Anode to the Gate, they will help discharge the Gate/Source cap faster so it will not be on when the other Mosfet is turn ON.

You have no idea how appreciative I am for this information. I do not yet know how much it will help with this particular project but I have been trying to figure out for quite some time what the purpose was for having a diode parallel with a diode. I understood that it would allow the resistor to be "bypassed" in one direction but not the other, just not why it was needed. Now I get it, rapid discharge of gate capacitance.

[eccerr0r]

Re: toroidal transformer questions

The other thing is I wonder why not simply add more windings on the secondary of the gate drive transformer... that would increase the drive voltage and not require a higher input voltage. Granted either way, you don't want the zeners to conduct if avoidable. May need to grab a scope to see how high the gate drive gets and ensure it fully turns the transistor on.

[budm]

Re: toroidal transformer questions

By the way, did you check to see how much current the IC is drawing?

[kaboom]

Re: toroidal transformer questions

The other thing is I wonder why not simply add more windings on the secondary of the gate drive transformer... that would increase the drive voltage and not require a higher input voltage. Granted either way, you don't want the zeners to conduct if avoidable. May need to grab a scope to see how high the gate drive gets and ensure it fully turns the transistor on.

That would demand more current from the 3525.

LDS:
Get that scope out and check the drive.

You may need a damper resistor, or even a snubber, across the primary of the GDT. Improper secondary phasing will only blow the MOSFETs- it should not cause the 3525 to overheat. Speaking of which, where's the filtering for it?

Put a 100u cap across both input and output of the 78xx regulator. Don't couple the audio directly into the RT pin. Use the error amp! Connect audio via a 100k resistor and a .1u cap to pin 1.

Also put a .47 or larger film cap across both of the 10,000u lytics. Have your return to the power supply connected at the bottom of the half bridge, not at the 3525.

[budm]

Re: toroidal transformer questions

Kaboom brought up the good point about the missing filter caps, I took a quick look at the diagram without looking at all the details the first time, the regulator may be Oscillating like crazy right now, also how the circuit grounds are applied to your board lay out is also very important. I wonder the person that put up the SCH actually made the real working board.

[kaboom]

Re: toroidal transformer questions

Kaboom brought up the good point about the missing filter caps, I took a quick look at the diagram without looking at all the details the first time, the regulator may be Oscillating like crazy right now,

I'd bet it's probably almost as bad as a Deer, when the AUX supply gets noisy. Noisy enough and the output stages in that 3525 can "stick," a sort of semi-shoot through.

also how the circuit grounds are applied to your board lay out is also very important.

This can never be understated! Years ago, I had a TDA2002 amplifier "module"- aka "prebuilt kit." This amp always seemed to desense AM BCB RCVRs nearby. One day, I measured the output power. With the power supply I had, it should've been an easy 10 watts. Here it was, giving less than three. I had a freq counter hooked up, suspecting the usual ultrasonic oscillation. Turns out it was oscillating at about 1.2MHz, hum-modulated on a radio from the diodes in the power supply.

The routing of GNDs/returns was completely wrong, taking it too close to the input, but also taking speaker return too close to the input! And the NFB return as well. The NFB was taken at the IC, but the snubber/zobel was after the output electrolytic.

The easy way to fix the ground was to simply bond the tab, heatsink, and trace together, taking the current loop(s) completely away from the input. I don't remember if I moved the NFB point and zobel, since this thing was fixed back in 2005 and I gave it away.

I wonder the person that put up the SCH actually made the real working board.

Sounds like the "expert" kipkay, doesn't it?

It seems like this with most of these "projects," doesn't it?

[LDSisHere]

Re: toroidal transformer questions

Thanks for all the input on this project. I have not had time to try any of the suggestions yet but I will post the results when I do.

[LDSisHere]

Re: toroidal transformer questions

Well I finally had so time to mess with this speaker again. I put the diodes in parallel with the resistors as Budm suggested and replaced the 7815 with a 7812. The chip barely gets warm now.

After these changes I then had another problem. The bottom arc point like the top one is a solid piece of copper that I got out of some romex. While I was messing with it I noticed the sound quality and the arcing were acting flaky. I then noticed smoke coming from around the bottom arc point. Then there was a blob of solder on my table. The solder holding this pin got so hot it desoldered itself from the PCB.

Kaboom, you may be correct in that this project may need a complete redesign.

This project is really getting on my nerves.

[LLLlllou]

Re: toroidal transformer questions

er these changes I then had another problem. The bottom arc point like the top one is a solid piece of copper that I got out of some romex. While I was messing with it I noticed the sound quality and the arcing were acting flaky. I then noticed smoke coming from around the bottom arc point. Then there was a blob of solder on my table. The solder holding this pin got so hot it desoldered itself from the PCB.

That's to be expected, the plasma is at several thousand degrees. The actual arcing point should be attached to some nice ceramic or porcelain stand offs, to get the arc away from the pc board and anything that doesn't like heat. Something like this:

http://www.ebay.com/itm/HIGH-VOLTAGE...item27d9799d18

You might also want to cover anything conductive that's carrying the HV with some silicone RTV, and keep all conductors rounded off with no points that could arc.

[LDSisHere]

Re: toroidal transformer questions

I knew the plasma itself got hot but I did not realize how that would transfer since the wire sending the arc does not get hot at least where I touch it when I adjust it. Thanks for the information on the stand offs, I will see if I have anything around the house that would work for this application.