PIEZO Transformers

No "real" transformer in the cards?

Regards, Joerg

a

No sure I understand your reply.?

Hello,

Can you let us know your first name?

I just wondered why it has to be a piezo transformer. Couldn't you do a regular flyback converter? That would be the customary (and probably cheaper) way to generate such voltages.

Regards, Joerg

I have a flyback design, but it's not as efficient as I would like, also I have been reading about Piezo and it would appear to be more efficient and perhaps occupy a smaller space.. cost is not an issue, performance is..

Hello,

You should be able to reach an efficiency similar to piezo with flyback. Anyway, if you are looking for piezo transformers try Fuji-Piezo or other companies like this one:

Regards, Joerg

Stray capacitance is the killer. During the switch on part of the cycle you have to charge the capacitance. This usually leads to a large current spike just at the turn on time. How about giving this a try:

L1 small V1 ------+--)))))----+----- ! ! ! -------- --- ! )( ^ D1 ! )( ! ! )( +---!!------ )( ! C1 ! --------- ) !!- ) !!- Q1 )L2 !!- ! ! GND GND

When Q1 switches on, the L1 slows the increase in current breifly. The energy that would normally end up in heat ends up in L1. When the stray capacitance is charged, the Q1 current decreases. At that time, the L1 inductor flysback doubling the voltage at V1. The energy from L1 gets divided between going into the transformer and being put back onto the input power line. At the switch off of Q1, all the remaining energy from L1 gets put back into the power supply.

You may really be better off with a resonant converter. Stray capacitance matters a lot less in them.

--
--
kensmith@rahul.net   forging knowledge

Hallo Ken,

Resonant converters are great. But in low power apps like this one I fared pretty good with regular converters, mostly made from logic and some discretes and without specialty PWM chips. Efficiencies of 90% were not a problem but often that is traded off against a somewhat wimpier FET for cost reasons.

In this case you almost have to do the discrete thing because PWM chips will typically either gobble up lots of quiescent power or be pretty expensive and hard to obtain.

Regards, Joerg

Thanks for the link..

BTW: I tried TDK and Murata.. but they are not interested unless you need to place a $1,000,000 order.... sad same old story..

In article , Joerg wrote: [...]

You forgot to say "or both".

I'd also include and can be buggy.

At low frequencies, the LM339 makes a faily nice PWM controller, if you don't want a constant frequency with load.

--
--
kensmith@rahul.net   forging knowledge

Hello Ken,

Oh yes.

Usually I am even cheaper than that. I might toss the converter a few logic gates. If I feel generous maybe even a flip-flop.

Regards, Joerg

Hello Doc,

With piezo transformers you will most likely not find any low volume supplier. Unless there is a standard part that somehow ended up in the low volume distributor chains. Unlikely though.

We usually had the same situation with PZT5-H. Since we didn't need several containers full of it the solution boiled down to a hefty NRE charge.

If this was a low volume application and I was the guy tasked to design it I'd focus on the usual magnetics.

Regards, Joerg

There is a shop on ebay called steminc selling piezo transformers and other piezo stuff. I have never dealt with them since it says they only ship to the US and I am not there.

Chris

You are probably right.... Piezo looked like an interesting solution.... I need to take a fresh look at the existing design..

thanks .

Hello Doc,

There is a lot of information about flyback and other converters in the app notes in the old Unitrode data book. Now most are available via the TI web site since they bought them.

Regards, Joerg

In article , Joerg wrote: [... me ...]

Last time I checked the LM339 was about as cheap as a logic gate package and cheaper than two.

Its usually the inductive elements that sets the price of the whole thing.

--
--
kensmith@rahul.net   forging knowledge

I have found Morgan Ceramics in the UK helpful for samples/small quantities In message , Joerg writes

--
dd

Hello Ken,

Yes, you can have them for about a dime at large qty TSSOP. But the concern is quiescent current. They draw about a milliamp whereas CMOS logic goes to zilch when you starve the converter under light or no load. Cost: If you use Schmitts you usually get a six-pack for less than

10c and only need two or three of them for the converter. The remainder is "for rent".

That's what I thought, too. Until I had to devise a custom inductor for a switcher in the early 90's and saw a quote on it from Taiwan. Less than a catalog inductor costs over here.

The usual trick is to find an inductor that is mass produced for some telco or other consumer app and try to design the switcher around that. Kind of an odd design flow but if cost is paramount it's the only way.

Regards, Joerg

[... LM339 ...]

I wasn't thinking of that. Most of my stuff draws way-way more than 1mA so the chips current really doesn't matter much.

BTW: You can also make a booster with an LM555. If you hook a zener onto pin-5, you can do it with just the 555 and a transistor no added op-amps needed. As far as idle power goes, its worse than the LM339.

[...]

Don't tell my boss I said this "cost is no object".

Priorities:

1A - will it still be working in 20 years time 1B - will it out perform the other guys 2 - can production actually build them on demand 3 - will we still be able to make one in 10 years time 4 - cost

--
--
kensmith@rahul.net   forging knowledge

Hello Ken,

It wouldn't be a concern if it never has to idle. In my case some of the stuff I design does not have a power switch so things have to drop to a few uA when not in use. That rules out most classic circuitry.

So far I have never used a 555. Never really seen an advantage in doing that since a Hex Schmitt or something like that can do it for less.

I won't ;-)

Sure he doesn't read usenet? Many of these threads are ported onto sites on the web. Once I searched up and down the web because I couldn't get an answer here in the group. The only thing I found was umpteen web copies of my own post.

That all calls for a solution that can live with jelly bean parts. More than a decade ago a slick sales guy tried to convince me to use "modern logic". After I balked at the cost he said that my CD4000 logic is going to be dead in just a few years anyway. I am still designing in CD4000 and the mfgs even ported most of it to newer and smaller TSSOP packages. They wouldn't ever do this in a situation of declining sales. So, I guess that sales guys was very wrong.

Regards, Joerg