RF amplifier - why typically 4:1 transformer on input

[Al (astral highway)]

I've noticed that in many, many class E MOSFET power amplifier circuits, there is a 4:1 matching transformer on the input. I just pulled this one out of the air, there are countless, from HF to UHF+

I'm surprised, given the high input impedance of a MOSFET gate, but I see it over and over again. I'm familiar with the turns ratio formula for RF transformers.

Can anyone help me or understand this ? Thank you

[ukcol]

Perhaps to allow for gate/source capacitance and gate/drain capacitance? The latter being made worse by the miller effect.

[G8HQP Dave]

A MOSFET gate has high impedance at low frequencies. At higher frequencies a power MOSFET gate looks like a capacitance in series with a resistance and significant power may be needed to push the required voltage onto the gate. A simple way of thinking about this is that the capacitance comes from the gate electrode to the FET channel, but the resistance comes from the channel itself - like a capacitor with one plate made from resistive material rather than a good conductor.

Gate-drain capacitance and Miller effect adds an extra problem, as this can vary with bias and signal level.

[G6Tanuki]

Also consider the effect - in the circuit shown - of the 25 Ohm gate-to-ground resistor R1

With RF amps it's desirable-if-not-obligatory that each stage presents a stable and consistent input-impedance to the stage-that-drives-it.

Over a significantly-wide bandwidth.

Failure to get this right can cause the driver-stage to oscillate, distort, or both.

So it's quite-normal to see apparently-rather-lossy circuits at the inputs of RF amps in order that the non-linearities and complex input impedances of the active device(s) are "swamped" and so don't compromise the operation of the preceding stage.

A 3dB Pi-section attenuator's another favourite between stages.

It may seem counter-intuitive but these days the problem's pretty much *never* one of merely getting enough gain; it's getting the gain you need coupled with the linearity and stability and repeatability-of-performance in the finished product.

"Passive Grid" is another technique you're likely to come across in amateur valve HF linear amplifiers: basically the input-circuit is merely a 50-Ohm resistor between grid and cathode of the valve. It presents a nice stable impedance to the transmitter driving it, and means that the amp is unlikely to suffer with parasitic oscillations. Again it may be counter-intuitive to have a 100-Watt transceiver feeding into what amounts to a dummy-load with a whiff of the RF fed to the big valve, but it does give reliable, repeatable, linear and stable results.

[Argus25]

G8HQP Dave has answered the question well. Power mosfets have substantial input (gate) capacitance, the only way to generate a good signal level (charge that capacitance to a voltage quickly) at the gate, is to have the voltage sourced from a very low impedance. By analogy you will see gates of power mosfets in switch-mode power supplies driven from low impedance or low source resistance driver IC's with very low series gate resistances, less than 10 ohms usually, to get the fets to switch fast enough. So Mosfets are only high input imedance devices at low frequencies as pointed out.

If you are interested in great broadband RF transformer cores for your amplifiers & transmitter projects, great ones are available from CCI (Communications concepts Inc). A good compact one is the RF400 series and I use the type 77 material for medium wave frequencies. These make great input & output transformers for transmitter amplifiers (like pantry transmitters), the already have the tube inside for the single turn, but you can add other turns easily for other primaries or secondaries to easily & quickly make matching transformers. They are pcb mount. Have a look at the range of types:

http://www.communication-concepts.com/rf-transformers/

CCI also make great amplifier modules & kits.

[Al (astral highway)]

Originally Posted by G6Tanuki

So it's quite-normal to see apparently-rather-lossy circuits at the inputs of RF amps in order that the non-linearities and complex input impedances of the active device(s) are "swamped" and so don't compromise the operation of the preceding stage.

Ahah, that makes a lot of sense. I've seen the 3dB attenuator pad around and wondered what that was about, too.

Thanks!

Originally Posted by G8HQP Dave

A MOSFET gate has high impedance at low frequencies. At higher frequencies a power MOSFET gate looks like a capacitance in series with a resistance and significant power may be needed to push the required voltage onto the gate.

Thanks, Dave. Very helpful to be able to visualise this!

[Al (astral highway)]

Originally Posted by Argus25

If you are interested in great broadband RF transformer cores for your amplifiers & transmitter projects, great ones are available from CCI (Communications concepts Inc). A good compact one is the RF400 series and I use the type 77 material for medium wave frequencies

Neat explanation, thank you Argus25. Also, this is a great link and I will def be investigating.

MOSFETs seem to have such a wide range of very different operating conditions, hard-switched at impressively high frequencies by a square wave and high-current driver IC with no fixed bias, pushing Amps through the gate; or as linear amps in e.g, class E, where they can burn up in an instant if the gate voltage isn't set correctly!

[Argus25]

Originally Posted by astral highway

MOSFETs seem to have such a wide range of very different operating conditions

One issue apart from the input capacitance is they don't like excessive gate-source voltage, so to prevent this sometimes you will see zeners in the gate circuits to keep the voltage below 15V for many. Some types have the zener built in, off hand a VN10kM, so these can kick around in junk boxes with BJT's without damage, but other types its risky to leave the gates floating.

One thing I find amusing is the response that some give to practically any circuit designed with transistors, something like; "I could build a better one with mosfets". However many a time the transistor is far superior and easier to use & bias in various RF and other circuits and doesn't have the high input threshold voltage and input capacitance issues of the mosfet (though logic level ones are a bit better for threshold voltage). So mosfets excel in some applications and transistors in others. Mosfets make better switches than transistors because of the low D-S on resistance, some in the milli-ohm range, so at high currents they have a much lower D-S voltage drop than the C-E drop of any power transistor. For most of my RF work I stick to transistors.

I have a 1500Khz mini push pull transistor class AB power amplifier circuit using those RF400 transformer cores for an AM Xmitter, I will look for it for you.

[Al (astral highway)]

[QUOTE=Argus25;952228]

Originally Posted by astral highway

I have a 1500Khz mini push pull transistor class AB power amplifier circuit using those RF400 transformer cores for an AM Xmitter, I will look for it for you.

That would be great, thank you!

[Argus25]

Hi,
will post more info on this soon, just going on a trip..

[Argus25]

Astral Highway,
As you can see by the circuit, it's almost a dead ringer for a standard audio amplifier design but with the ferrite transformers and higher operating frequency. There is enough bias to just overcome the crossover distortion.

The transistors I used there were some I had in my RF parts box. But there are plenty of TO-220 sizsed silicon npn's that will work at these frequencies. If you look at the AN779L amplifier (sold by CCI) you will see some MRF prefix TO-220 package transistors that were made for the CB market.

Of course the power out depends on the input but you can use a lot less output than this little amp is capable of for say a pantry transmitter.

I found that the modulation was better done (for linearity for music)at low energy levels (and do it using a MC1496 modulator IC) and passing the signal to a broadband amp, than the common way of amplitude modulation the output stages with a power am modulator.

With the ferrite transformers it's dead easy to alter the turns ratio and change the load impedance while still getting the best out of the amplifier.

[Radio Wrangler]

Communictions Concepts Inc have long been a good source of RF power kits for amateur radio etc, often based on Helge Granberg's applications notes designs from Motorola.

I don't think they make ferrite cores, but just re-sell them. Bill Amidon in Hollywood is well known in amateur radio circles for "Amidon Cores" but he's a re-seller.

Most probably the ferrite cores come from Fair-Rite inc (Wallkill Noo Joisy) and the dust iron cores from Micrometals inc of Anaheim California.

As for me, the day job involves designing 300W 1GHz power amplifiers. My Mosfets look like a couple of nanoHenries of bond wire feeding about 3 ohms of resistance feeding almost a hundred pF. the bond wire dominates at that frequency, I need to maximise the voltage across the capacitive component and make nice enough broadband impedances to keep everything stable. With devices costing several hundred pounds each in small quantities... well... you need to take care. Modern devices can now give over 20dB of real, useable gain under these conditions.

David

[G8HQP Dave]

as linear amps in e.g, class E

Class E is not a linear amp; it is a high efficiency switcher.

[Andrew2]

Indeed. My mate G4LUL uses a class E final on his topband transceiver and it requires high level modulation because it is not a linear amplifier. It is certainly efficient - he runs about 5 watts carrier (20 watts pep under full mod) and it stays spookily cool!

[MrBungle]

It's quite surprising the current you can switch with them. If RDS(on) is low you can sling ridiculous amounts of current. It wasn't unusual to have RC PWM speed controllers switch 120A or so in a little box an inch square with a tiny heat sink

Up to a few MHz you can PWM a mosfet to generate an RF waveform of any modulation type. Very efficient.

[MrBungle]

Interesting. Class D RF operation. Wonder what the harmonics would be like though.

[dave cox]

The buck convertors surrounding a modern high performance CPU use some pretty fancy mosfets to push 100+ amps into the CPU! The losses in a bipolar version would be interesting !

dc

Wonder what the harmonics would be like though.

Horrible! But if you PWM at a much greater frequency than the wanted RF it is easy to filter out. Of course you only need the Niquist frequency (2 X wanted) then the filter would be somewhat spectacular.

[Al (astral highway)]

Originally Posted by Argus25

http://www.communication-concepts.com/rf-transformers/

CCI also make great amplifier modules & kits.

Thank you, Argus25. That's brilliant.

Thanks also for the circuit diagram of the 1.4MHz RF amplifier; I like the design and will investigate.