A circuit equivalent of a DPDT switch?

AnalogKid said:

Because a 2N7001 cannot handle 10 amps. And, we still do not know about the power source(s) for the circuit, which might rule out using MOSFETs.

Click to expand...

I know that. But the circuit layout from an H-bridge design using 2N7001s can be adapted to use mosfets. By ME. Do YOU see the difference?

danadak said:

Overlap in H Bridge or 1/2 H bridge "normally" handeld by dead time setting in
a PWM -

Click to expand...

I don't need (or want) PWM. Just on/off bursts of square waves at a well controlled (tunable) frequency. The tuned LC circuit (xformer secondary + piezo) takes care of knocking the corners off the square wave to give as reasonable approximation of a sinewave as the piezo needs.

But I am aware I may need to do something to ensure that the on-states do not overlap. I'm just not sure what that something is yet.

danadak said:

Regards, Osc, its 145 Khz ....what resolution do you need ? How do you want to set it, with a pot or a V ?

Click to expand...

Typical piezo specs. tend to list their resonant frequencies as a value with a tolerance: eg. 142kHz+/-4kHz; but it gets more complicated once you start looking into it.

When you look at the frequency response curve -- when you can pursuade the supplier/manufacturer to let you -- you see something like this:


Fm is the resonant frequency, which is the point of highest physical displacement, but also the greatest power draw. (Lowest impedance.) Fn is the 'anti-resonant' frequency, which produces slightly less physical displacement, but considerably lower power requirement. Both are very sharply defined. I'm hoping to target the latter to reduce power requirements for a battery powered device.

In addition, different chips of the same size in different piezo materials (eg.PZT-5 or PZT-8) have different resonant frequencies. Those of the same size and same material from different suppliers again differ (somewhat) in the resonant frequencies. And as the tolorances show, even the same device from the same manufacturer vary from batch to batch and even within batches.

In general, the 10x5x2mm rings I am targeting have resonant frequencies in the 100-200kHz range, so until I find a supplier that will: a) sell me a couple at a reasonable price; b) provide full specifications for what they are selling -- the two requirements seem mutually exclusive! -- I need to keep my options open.

The ideal situation would be an oscillator chip that I set close to the required frequency by an external component and then tune. The tuning could be some kind of feedback loop detecting the impedance -- perhaps via small tap or additional coil on the secondary side of the transformer. Or it could be done programmically using a µCpu via a DAC.

Thanks for your help, Buk

rjenkinsgb said:

Look at an IR2153, a combined oscillator and push-pull FET driver:

Click to expand...

Somehow I looked right passed this post and only just now saw it as I was reviewing the thread. Thank you for the link and oscillator component specs.

Now I know such an animal exists, it makes sense to combine the oscilator and the gate driver. Though I am puzzled by the need for a 'gate driver'. (Yes, I know I am showing my ignorance of the subject here!)

My 'understanding' of mosfets says that a small voltage -- here VGS(th) Max. 2V -- (and negligable current) applied to the gate turns the mosfet on; so why do I need a 'driver' to do this? (And why would I need a "600v rated gate driver" -- something from one of the infineon vids -- to drive a device that turns on at 2V?

Gate drivers are used for moderate frequency, in the 50 - 500 Khz range, switching
of mosfet so they do not spend a lot of time in linear region dissipating a lot of power.

In your case seems a driver not needed as the frequency of change is very low, ie.
turn on and walk away so not much time spent in active region.

But you do need to fulfill Vgs on at Rdson rating, eg. HV like 10V, to get the MOSFET
hard on so I squared x Rdson losses low.

Regards, Dana.


**broken link removed**

Buk said:

My 'understanding' of mosfets says that a small voltage -- here VGS(th) Max. 2V -- (and negligable current) applied to the gate turns the mosfet on; so why do I need a 'driver' to do this? (And why would I need a "600v rated gate driver" -- something from one of the infineon vids -- to drive a device that turns on at 2V?

Click to expand...

VGS(th) is the threshold voltage, the point at which it just barely starts to conduct.
To switch fully on so you get the rated RDS(on) value, the gate must be at somewhere typically from 3V to 10V, depending on the device. You need to look at the individual data sheets.

And, the gate structure in a power MOSFET is typically a "mesh" pattern within the device; it reminds me of the grid structure in a valve, using electrostatic control.
It has a large area and large capacitance - often several nanofarads.



To get the FET to switch on or off quickly so the losses are as low as possible, that capacitance has to be charged and discharged rapidly, so the gate drive has to handle high current spikes as each on or off transition. Once the gate capacitance is fully charged or discharged, the gate current is again zero or near that.

That small device you link to has input capacitance of 805 and 1400pF for the two devices. This higher power one picked at random has 4100pF gate capacitance:


The 600V or whatever is presumably relating to a "high side" driver such as an IR2113?
Motor drivers & inverters etc. often run on a DC bus which is pretty much just rectified mains, eg. 380 or 415V in.
The "high side" is the MOSFET that switches between the +600V and output, the upper half of one of the "half bridge" sections.

A high side driver contains an isolation and coupling system so the gate driver for the high side FET control can be offset from the logic inputs by any amount up to it's rated voltage.

The isolation voltage rating is a maximum; using one of those on eg. a 24V system is fine.

Buk said:

I am puzzled though. What sets/regulates the frequency?

Click to expand...

mostly the inductance of the primary. it varies somewhat with the load, but if you want to get it within a certain range, i've seen versions of this circuit with small value capacitors across the sides of the primary. it's going to be somewhat of an experimental process. if you're wanting a particular frequency, and want it really stable, you can eliminate the base feedback winding and drive the transistor bases alternately using either a transistor based multivibrator circuit, or drive the bases with TTL logic (i.e. from D flip-flop running from a stable oscillator at twice your desired frequency.

unclejed613 said:

if you're wanting a particular frequency, and want it really stable, you can eliminate the base feedback winding and drive the transistor bases alternately using either a transistor based multivibrator circuit, or drive the bases with TTL logic (i.e. from D flip-flop running from a stable oscillator at twice your desired frequency.

Click to expand...

Thanks. This is the route I have been aiming for from the beginning. piezo chips only really work at (one of) their resonant frequencies and they are very sharply defined. I'm thinking of using a tunable IC oscillator. The exact frequency I need will depend upon which chip I eventually buy. For the size/power/polarity I need, around a150kHz is likely, but I'm not buying the chips until I can find a supplier that will a) provide full specifications for the chips they sell; b) doesn't charge $35,000/kg for shipping. (eg. $69.97 for <2g!)

Right now I'm trying to find a useable, free simulator that has a transformer emulation that works properly at 145kHz. I love the simplicity and useability of falstad, but its transformers do weird things above about 1kHz.

Buk said:

[looking at a two-transistor inverter/DC converter]

That looks much simpler than I've been playing with: https://tinyurl.com/yfxpqc4o

I am puzzled though. What sets/regulates the frequency?

Click to expand...

The transformer material that provides the positive base feedback is... saturating.
When the magnetism reaches the material limit, that base drive goes away,
because only the dB/dt in the winding is productive, and B just hit the wall.
Instead of ONE transistor turned off, TWO of 'em turn off, and as the
field in the transformer collapses (no current, no electromagnetism) that
diminution of the field turns on the opposite transistor.

So, it's the size and shape of the transformer core that sets the frequency,
with some dependence also on the input voltage.

whit3rd said:

So, it's the size and shape of the transformer core that sets the frequency,
with some dependence also on the input voltage.

Click to expand...

Do you know of any information explaining the math behind the frequency of this kind of circuit?

Do you know of a working, fully specified circuit description?

If you are still looking at an inductor / transformer driving the piezo, the resonant frequency of an L-C circuit is:

1 / (2 * pi * square root (L* C) )

rjenkinsgb said:

If you are still looking at an inductor / transformer driving the piezo, the resonant frequency of an L-C circuit is:

1 / (2 * pi * square root (L* C) )

Click to expand...

I've found that equation all over the web; but it does not help much with a flyback converter that has a two wire inductor (transformer) and the driven element (piezo) that presents as both series capacitance, and parallel capacitance, resistance and inductance:

I need to target the second resonance (fa) minima with the inductance of the secondary coil; but at the same time the transformer has to provide a 1:17 uplift in the voltage. And of course the core has to be big enough that it can store enough energy for that uplift without saturating, but small enough to be responsive at 145kHz.

There's a balancing act in there that I have yet to find a learning resource to help me with.

The responses are no different to a crystal oscillator type crystal, looking at that.
With those, you can pick the series resonant (low impedance) or shunt resonant (high impedance) peak, depending on the oscillator type.

Some info here:

And a useful practical description in the last answer here:

Or you can just buy a commercial drive transformer:

rjenkinsgb said:

Or you can just buy a commercial drive transformer:
https://www.benthowave.com/products/Specs/BII-6000Datasheet.pdf

Click to expand...

Those transformers are 4 to 8 times the volume and 10x the weight of my entire device and roughly 150x the volume I have alloted for my transformer

I've read several papers on sonar applications, but they use huge piezo chips (eg 50mmØ x 20mm discs) and are targetted at creating ultrasound beams into large volumes of water eg. 9m x 9m x 9m, and strong enough that waves reflected back from the extremes of those volumes can be detected.

My application uses a 10mmOD x 5mmID x 2mm ring and is creating concentric waves into water contained in a 1mm diameter bore tube. 50Wrms at most and possibly substantially lower.

Despite the fixed frequency operation, it has more in common with a piezo speaker than those kW power transducer applications.

rjenkinsgb said:

Some info here:
https://www.maximintegrated.com/en/design/technical-documents/tutorials/5/5265.html

Click to expand...

Thank you for this. The explanation of the typical values and effects of the two parts of the device characteristics is very helpful.

I do wonder how much changes between the crystal use where the chip is driving the circuit frequency; and my application where the circuit has to drive the chip frequency?

I've tried using the falstad built-in crystal component to simulate my requirements but I get the most bizarre results. Of course, I don't know enough to know if that's due to the simulation or the values I am using. Eg. A simple LRC using a capacitor. Hold the top button closed for a millisecond (simulated) and when you let go, the circuit resonates at the required 145kHz:


But substitute in a crystal with the series capacitance set to the same 350pF and then despite that the res.f (bottom right) still equals 145.045kHz:

the circuit actually resonates 788kHz. (Although that seems to be modulating a 145kHz carrier?)

Buk said:

Do you know of any information explaining the math behind the frequency of this kind of circuit?

Do you know of a working, fully specified circuit description?

Click to expand...

Pico manufactures the cores for this saturated-transformer usage, and gives good tabulated info,
and charges high prices for the tiny transformers that they specify.

<https://www.picoelectronics.com/node/13238> is a start

The description of core nonlinearity is sparse. The design procedure is
contingent on keeping losses low (high frequency and saturation
will burn the transformer up; saturation is necessarily heat-producing).