Wireless Mains Adapter-Frequency etc.?

[boiss]

Ok Al
Here is the cct, hand drawn, I don't have any CAD software. I found a datasheet for the Fet too, I guess from it that I am running on the limit re. drain source volts with over 60V.
Another oddity is that on the Rx side the power transfer is best with half wave rectification. I have used schottky diodes both as a single diode rect or 4 arranged as a bridge. The best result obtained with half wave rectification was just just over 10W into 50 Ohms at 20 or 30 KHz (not sure which it was) but everything was very hot. With full wave rect max power was around 6W. Waveform shapes did not change just the amplitude. I guess that this has to do with impedance matching in some way.
thanks
Mike

I just measured the current on the Hacker, quiescent around 20mA and at about 25% on the volume control (reasonable Volume in a small room)
the current peaks around 100mA.

[Skywave]

Thanks Mike. Never mind about CAD: whenever I have sent my D-I-Y ccts. in to any forum, they're always been scanned copies of hand-drawn ccts. What matters is clarity of drawing and the detail therein: your drawing meets those requirements. √

Looking at the waveforms that you sent in, the gate waveform seems to show signs of parasitic oscillation. At the rising edge of that waveform you can see some 'jaggedness' to that waveform: a sure sign of parasitic osc. This will reduce the power efficiency of the FETs, since they will not be acting as pure switches - which is what is required: they will spend some time in a 'linear mode' - thus dissipating heat.
Apparently, MOSFETS are prone to parasitic oscillation. I cannot give you any ready-made solutions to prevent that osc., (my experience with power FETs is very limited), but improved decoupling may help and perhaps low-value resistors (say 4.7 ohms as a starting value) in each drain may also help. Of course, the physical layout will be important: check the 0v. earthing arrangements: ideally, single-point earthing and substantial R.F. earth / 0v. conductors.

As regards half-wave and full-wave rectification, since with half-wave rect. one half of the waveform isn't 'doing anything', then the input impedance of that cct. will be higher than that for a full-wave rect. And that implies that wire of thicker diameter for the coil sourcing that current will help. Of course, at the same time, the power demand from the driving source will be higher, which, in turn, requires that the impedance of that source needs to be lower than that for a half-wave rect. scheme. And that, in turn, could place an even higher demand on that driving source for good R.F. dynamic stability: i.e. freedom from parasitic oscillations.
Which brings me neatly back to where I came in.

HTH,
Al.

[Nick]

Al is right - in an ideal world, you want the FETs to have binary operation - either ON or OFF.

If they operate in their linear region, they are effectively resistors and will heat up due to the current flowing through them. The minimum resistance they can show is the RDSon figure from the data sheet - that is their drain-source resistance when fully on. To get low RDSon, you generally need a large die size and modern designs which can result in RDSon figures in the low milliohms. The IRF2807 is an very nice FET - pretty good for this application - lowish Qg (total gate charge), low RDSon (which leads to decent current rating for the case size), decent Vdss (perfectly adequate for this application).

I wouldn't expect particularly high efficiency out of this circuit - perhaps 50 or 60% - so assume maybe 40 - 50% of the power you put in is being lost as heat. Stuff will warm up.

Regarding the parasitic oscillation, try using a ferrite bead over the gate leads, close to the FETs - the parasitics are at RF frequencies and the bead will help damp those. Also, constructional techniques make a big difference - do you have a photo of your layout? I would also consider using a low value resistor (between 4.7 & 100 ohms or thereabouts) in the gate leads close to the FET - equivalent to a grid-stopper in valves... I used to design SMPSs and care-and-feeding of FETs is important...

[pwdrive]

Bung a couple of 807's in and up the frequency......? (I might be joking)

Lawrence.

[Skywave]

If all else fails, a re-design might be fruitful: a separate oscillator driving a push-pull class-C power amp. 'All-in-one' power oscillators - such as this - must always involve some compromise.

Al. / June 13 //

[pwdrive]

Not an expert on these matters but it looks from the circuit that it's an Osc plus PA using the same devices, I would have thought that it would be diffcult to obtain an ideal switching rise time if inductance is a factor?
Maybe use the MOSFET's as PA only and drive them from a seperate switching source ie: multivib. or whatever where the rise time can be tailored to suit the switching requirements of the MOSFET's?

Lawrence. EDIT: Al beat me to it....

[Nick]

Not an expert on these matters but it looks from the circuit that it's an Osc plus PA using the same devices, I would have thought that it would be diffcult to obtain an ideal switching rise time if inductance is a factor?
Maybe use the MOSFET's as PA only and drive them from a seperate switching source ie: multivib. or whatever where the rise time can be tailored to suit the switching requirements of the MOSFET's?

Its a perfectly acceptable circuit for low power if you aren't worried too much about efficiency - normally, a Royer oscillator is used (bit more efficient), but we aren't aiming for 100% power transfer - its simple, compact and does the job...

My guess is that with a decent layout, control of the parasitics and a couple of small TO220 heatsinks, this will be perfectly fine. Personally, I'd have used the Royer configuration, but this one is ok.

If you don't have suitable ferrite beads, I could bung some in the post should you wish...

[pwdrive]

Does the Royer rely on core saturation?

Lawrence.

[Nick]

Does the Royer rely on core saturation?

Indeed it does - it relies on the core saturating to make it switch, i.e. air-cored coils will not work... you need a core in there...

[pwdrive]

Does the Royer rely on core saturation?

Indeed it does - it relies on the core saturating to make it switch, i.e. air-cored coils will not work... you need a core in there...

Not related to this thread so I will be brief, there was a guy who tried the switching core saturation idea for a direct conversion receiver (switching mixer) He had some success with it I believe

Ok back to the thread.

Lawrence.