looking for a sine wave output device

[LaserTazerPhaser]

Is there a device which can output very low frequencies to =>1Mhz few amps =>12v sine?

Perhaps a wideband pure sine wave inverter but currently those only are specked  for residential frequencies only. I was considering a low ESL large DC link cap for decoupling a linear MOSFET which would output the desired frequency or is there a more efficient method to this?

[LaserTazerPhaser]

Is there some circuit for an application with this already implemented? The linear MOSFET method.

[ogden]

Unless you are a God of control loop designing, usually you need 10x corner frequency than reference bandwidth, and 10x switching frequency than corner frequency, so to have 1MHz reference, your switching frequency needs to be 100MHz, if you are after a class D.

In case you are fine with fixed amplitude and frequency, all components of classic class-d amplifier are not needed. What you need is - precomputed 1-bit DAC bitstream that would come out of class-d modulator which is fed into power (PWM) stage followed by LC filter. I'v done that with great success generating 10KHz sine out of microcontroller that does not have DAC but spare SPI to output 1-bit DAC bitstream. For 1MHz sine it could be doable with 24MSPS in case of steep LC filter (5-th order or 7-th).

[garethw]

From my limited experience with sinusoidal waveforms they are tricky enough to get right over a small frequency range, let alone the range you specified. It can be done, I’m sure, but with a substantial amount of effort and know how.


Sent from my iPhone using Tapatalk

[Dabbot]

Every modern household has at least half a dozen sine wave output devices. 

[Berni]

That is a bit of a lacking question. What loads do you want to drive with it, what frequency range are you looking for, what sort of sine wave purity is needed, is frequency stability a requirement etc..

A very efficient way in making a sine wave is a ZVS oscillator. It can operate at very high output powers of >500W, produce a voltage from milivolts to kilovolts trough the correct transformer winding ratio. The transistors don't operate in the linear region and they even switch at zero crossing to remove switching losses completely making them run very cool. But the operating frequency is dependent on the LC tank resonance so changing the frequency would require switching banks of capacitors.

If you are going to build a AB class high power amplifier that has a flat frequency response past 1MHz with low distortion be prepared for some serious design work and a good number of prototype iterations because it won't be easy. Physically large output stages start developing all sorts of parasitics that can cause the thing to become unstable at those sorts of speeds.

EDIT: Oh and there are also some high power opamps out there such as the OPA541 that will give amps of output with a GBW of 1.6 MHz. If that's not enough then some of the monstrous opamps from Apex Microtechnology might be a thing to look at, but those cost over 100 bucks each.

[LaserTazerPhaser]

Opamp is likely ideal implementation for this since a SOT-227 MOSFET and its heat sinking would almost cost the same amount as an apex part.

Maybe a discrete opamp can do it?

[Berni]

Well these huge Apex opamps are mostly just discrete transistors on a fancy heatsinking ceramic substrate board. Some might be wire bonded transistor dies while others are just normal off the shelf transistors soldered onto a ceramic PCB. So yeah its mostly a box holding a discrete opamp that was likely designed by some wise analog graybeards that know what they are doing.

But opamps still expect to a fairly well behaved restive load on there output. So its more of a question what sort of thing you want to drive with it, funky loads can make big opamps oscillate like no tomorrow.

[LaserTazerPhaser]

Well these huge Apex opamps are mostly just discrete transistors on a fancy heatsinking ceramic substrate board. Some might be wire bonded transistor dies while others are just normal off the shelf transistors soldered onto a ceramic PCB. So yeah its mostly a box holding a discrete opamp that was likely designed by some wise analog graybeards that know what they are doing.

But opamps still expect to a fairly well behaved restive load on there output. So its more of a question what sort of thing you want to drive with it, funky loads can make big opamps oscillate like no tomorrow.

Didn't know load made such an impact on opamp. In this case going with a IXTN600N04T2‎ and a 1500uF DC link cap.

Loads will are inductive, capacative and very resistive and Im going to sweep frequencies which might cause parasitics to ring in which case the opamp might go into a mode were nothing is output. A large benefit with DC link cap is isolation to linear MOSFET from load. All thats needed to protect the cap would be some clamp very much suited would be a gang of MOVs and the cap is self healing unlike the opamp which would be damaged rather quickly.

Would be nice to have an ISO packaged opamp capable of much more than what apex currently has especially driving various non linear loads.

[David Hess]

I did not see anything about efficiency requirements.  I would use a function generator to drive a class-ab linear power amplifier designed like an audio power amplifier but for higher frequency operation.  I have gotten to 500kHz easy using 2MHz Ft output transistors so common 50MHz ring-emitter output transistors or power MOSFETs can certainly get into the MHz range.  Lower power class-ab feedback amplifiers can get to 40MHz.