Transformer Saturation

[dknguyen]

THe transformer's peak power (or current) draw is what causes a transformer to saturate right? Not the average power?

The transformers I am looking at for the ultrasonic driver are 1/4W transformers and no other information is realy given to help figure out saturation other than that they can handle a maximum of 250mA DC. THe transformer will need to output a peak 200Vp@62.5mA = 12.5W. Now the transformer has an input of 12Vp and efficiency of 50% with 16:1 ratio so the peak input power would be like 24W.

THe average power however works out to be 0.2W output or 0.4W input, near the power rating of the transformer and I can reduce the duty cycle by 10 to greatly reduce average power. But since the peak ratings are so far beyond what the transformer is rated for it would probably still saturate right? Regardless of how much I lower the duty cycle to reduce the average power since the peak powers still stay the same (I can't do anything about those).

 

[mvs sarma]

THe transformer's peak power (or current) draw is what causes a transformer to saturate right? Not the average power?

The transformers I am looking at for the ultrasonic driver are 1/4W transformers and no other information is realy given to help figure out saturation other than that they can handle a maximum of 250mA DC. THe transformer will need to output a peak 200Vp@62.5mA = 12.5W. Now the transformer has an input of 12Vp and efficiency of 50% with 16:1 ratio so the peak input power would be like 24W.

THe average power however works out to be 0.2W output or 0.4W input, near the power rating of the transformer and I can reduce the duty cycle by 10 to greatly reduce average power. But since the peak ratings are so far beyond what the transformer is rated for it would probably still saturate right? Regardless of how much I lower the duty cycle to reduce the average power since the peak powers still stay the same (I can't do anything about those).

Perhaps you may create reluctance Air gap while packing thre core and thus prevent saturation-- this would of course reduce inductance of windings slightly. This has to be compensated in advance.
generally power and driver transformers of smps are wound like that and as you know they work slightly above ultrasonic range.

 

[Hero999]

50% efficiency is pretty poor for even a small transformer working at that frequency. There must be a way to improve that whether by using a higher inductance core so you can use less turns of thicker wire or by using a core with a tighter hysteresis loop to cut down on core losses.

 

[Boncuk]

Transformer

Hi,

a transformer usually saturates as a result of remanence (rest magnetism). Assuming the device was switched off and the magnetic field was already built up to the top value, a new start might just go into the same direction and then the magnetic field increases and will finally saturate. If that happens a transformer is just a resistive load (dead short).

It is very unlikely that an iron core saturates, but toroidal transformers tend to saturate, because the core is iron powder (easier to magnetize but with the disadvantage of remanence).

I your specific case you are obviously using a transformer built for mains voltage and frequency of 50 or 60Hz, which gives you poor performance. If you want to use a transformer as an ultrasonic driver (in the range of 40KHz) you better make your own design using something like a ring core. That has an insulating plastic core inside which you can wind easily. The ring core comes in two halves and is mounted as one unit with an air gap for higher frequencies. You might also look for a suitable transformer being used in a secondary switching power supply.

Another alternative is an audio transformer of the 100V-ELA-technique.

Sorry, if I missed the correct terms (a primary switching power supply uses mains and a secondary uses high voltage DC and chops it). That's just a language barrier (I'm german).

 

[dknguyen]

I your specific case you are obviously using a transformer built for mains voltage and frequency of 50 or 60Hz, which gives you poor performance. If you want to use a transformer as an ultrasonic driver (in the range of 40KHz) you better make your own design using something like a ring core. That has an insulating plastic core inside which you can wind easily. The ring core comes in two halves and is mounted as one unit with an air gap for higher frequencies. You might also look for a suitable transformer being used in a secondary switching power supply.

It's actually an RF transformer and not a mains transofmer. It's 3dB bandwidth is right at my frequency of interest which is where the 50% efficiency comes from:
**broken link removed**

My operating frequency range is actually 50kHz-100kHz so I don't think audio transformers or 60Hz transformers would work at all.

I'm just realy trying to find a way to not have to wind my own transformer because I've been reading up on it and there are a lot of unknown variables, and a lot of uncertainty as to the final product, and I'm still having trouble finding where to actually get cores. It's a large dose of math and EM theory and equations that I've never come across before.

I've also been trying to find SMPS transformers since they seem to be of my frequency range but I can't seem to find any, or they don't have the turns ratio, sufficient insulation, or frequency response (if it's even given) that I want.

 

[Speakerguy]

Rod Elliot has a series of three articles on transformers at his website:

**broken link removed**

It is very enlightening. It should hold the answer to your question.

 

[Boncuk]

Tranformer

Hi dknguyen,

you might try Wuertz electronics, (**broken link removed**). They offer a wide range of transformers for a lot of purposes. (e.g. flybacks, switching power supplies etc.)

If you want to make your own design the free software "WE-Flex-designer" might be a good idea too. You just have to put in your specific values and have the program make suggestions.

Regards

Hans

 

[mvs sarma]

television spare shops sell horizontal driver tranformer. this is designed at 15.625KHz. and uses ferrite material. you may try using it at 45KHz- it might function . if needed you can get some datasheet to rewind the same to meet your needs. it should not be difficult. But I have some doubt can t you manage with out any transformer? what is the make and model number of the transducer you are tryng to use?

 

[felis]

My operating frequency range is actually 50kHz-100kHz so I don't think audio transformers or 60Hz transformers would work at all.

That's normal range for SMPS transformers - even iron powder core transformers would work and there is plenty of them for all kinds of currents. You need to figure out peak current through primary winding, winding inductance and turns ratio. Then look what's available (Coiltronics, Sumida) and recalculate - peak current depends on inductance so you can play with values here. Also you said your input is 12v - is it possible to change that to a higher voltage like 30-60v? It's easier to design this way and turns ratios of 1:3-1:4 are more common than 1:16.

 

[dknguyen]

That's normal range for SMPS transformers - even iron powder core transformers would work and there is plenty of them for all kinds of currents. You need to figure out peak current through primary winding, winding inductance and turns ratio. Then look what's available (Coiltronics, Sumida) and recalculate - peak current depends on inductance so you can play with values here. Also you said your input is 12v - is it possible to change that to a higher voltage like 30-60v? It's easier to design this way and turns ratios of 1:3-1:4 are more common than 1:16.

THe peak current on the secondary winding is 62.5mA so the primary current peak is ideally x16 = 1A. THe output is 400Vpp (but it's centered on zero so the insulation requirement is only 200Vp). This peak current was calculated using the transducer capacitance and operating frequency that would produce the most current draw.

The problem with output voltages of 30V-60V is they far exceed the battery supplies, so what happens is then I need to get a "high"-voltage amplifier, in which case I might as well get a 200V or 400VDC converter and then build the whole thing without transformers- but then I run into problems of finding 50kHz-100kHz switches capable of withstanding that voltage. Also, if I use 12V I can use a bunch of simple gate drivers for the H-bridge since it is on the same scale as both the gate voltage and source-drain voltage.

On the other hand, if I do get a 60V DC converter I can still use MOSFET switches just fine, but then I'm stuck with using transformers AND step-up DC converters. I am trying to avoid having that many parts right now.

Regardless I still haven't come by an 200-400V insulated SMPS transformers but I will look at your sources.

Also, I have found out the bandwidth limitations of a transformer on the low-end are core size (since lower freqnuency is higher flux density) and on the high-end are the interwinding capacitances and core losses. Doesn't this mean I can buy a bunch of large cores and then wind them and *hope* the interwinding capacitance is small enough to not affect functionality and then test it out on a frequency generator? I'm not sure how I myself would minimize interwinding capacitance though? (Trying to side-step the whole equations stuff as much as possible). 50kHz-100kHz is really low though, I don't think interwinding capacitance would play a role at all since those are not realy high frequency relative to RF where all the problem seem to come in.

if needed you can get some datasheet to rewind the same to meet your needs. it should not be difficult. But I have some doubt can t you manage with out any transformer? what is the make and model number of the transducer you are tryng to use?

THe transducer is this:
https://www.electro-tech-online.com/custompdfs/2007/12/60020environmental20spec.pdf

It's just realy nice to use a transformer since I can keep everything else low voltage which would make the circuit simpler and allow me to have more flexibility in parts.

From my understanding it shouldn't be technically difficult to make such a transformer since my requirements are so undemanding, I just can't seem to find the part. I don't even want it to be wideband really, since it would be preferable for the transformer to burn off the higher and lower frequencies outside of 50-100kHz to produce a more sinusoid output from my square wave input to reduce the stray noise sent out by the transducer.

 

[felis]

in which case I might as well get a 200V or 400VDC converter and then build the whole thing without transformers- but then I run into problems of finding 50kHz-100kHz switches capable of withstanding that voltage.

What's your application? Here -> http://www.discovercircuits.com/DJ-Circuits/300vpp.htm is an example of pretty fast 300Vpp switch - I used to scale this up to 600Vpp 100ma without too much difficulty.

 

[dknguyen]

Hmmm. THat could work. I'm so used to looking at motor drivers that I tend to forget you don't always need a boot-strap capacitor to turn a high-side switch on if the source voltage far exceeds the gate voltage (since current levels are high enough NMOS is better for high-side for motors...or at least I never use PMOS for high-side unless it's a tiny tiny motor or power circuit).

Gotta start thinking about simpler, less high power drives...

 

[felis]

Also, take a look at Supertex high-voltage interface ICs and MOSFETs. They have complementary MOSFET pairs with zeners and stuff built in; IIRC they make them up to +-200V which might work for you as well.

 

[dknguyen]

Well it looks like Zetex is good for the high voltage MOSFETs. THe first place I go for transistors is International Rectifier but they don't make any MOSFETs that work at such a high voltage that can switch fast enough (aka. a power level that low).

THe second place I look is at Diodes inc. but they only have flyback diodes that will work at that voltage (though, since I am driving a capacitive load rather than an inductive load, I don't think they are even required? I am so used to making these circuits to drive motors).

 

[Hero999]

It seems like you're making this more complicated than it needs to be. I would wind the primary in bifilar style to run of 12V, use two MOSFETs to switch the coils alternately and wind the secondary to give 200V. To minimise the effects of interwinding capacitance, make sure you winde the secondary very neatly, lay the turns flatly and build them up layer by layer. You're also better off using more narrow layers rather than fewer wider layer. If you're worried about the insulation between the primary and secondary, then bond the side of the secondary that's nearest to the primary to 0V.

Do you need to vary the output frequency?

Is the output frequency that critical?

What about a resonant Royer converter, like those used to power CCFLs?
https://www.electro-tech-online.com/custompdfs/2007/12/ANP005_AP2001.pdf

 

[dknguyen]

I do need to vary the output frequency (otherwise I'd just use piezo transducers which are driven by 12V). The frequency is important...whether it's critical or not depends on whether I ever try to do doppler anything with it.

 

[felis]

Well it looks like Zetex is good for the high voltage MOSFETs.

There are plenty of high-voltage p-channel MOSFETs - since your currents are small, you won't worry much about Rds. One thing you need to know about this circuit - when it dies it takes everything with it; use opto-isolation at least when breadboarding the thing.