As I do my research here to help you understand the EKL thingy's 
I've come across several different issues regarding how to Drive the Gate - with two main components Voltage and the immediate required pulse Current it contains, that if combined correctly, can offer a solution to solve a third.
We talked about Gate Capacitance, a value that the maker arrives to thru testing. This value is important to know because it sets an upper limit to the Maximum Useable Frequency as well as the Safe Operational Area the device can be used in.
The Gate Capacitance in itself can provide a working RANGE to know of it's Speed, as a factor of Rate but how well does it follow a signal arriving to it's Gate? MOSFETS are non-linear - yes, but there are factors the Gate exhibits that can be used as a guide and ability to perform linearly - in your favor.
[ATTACH=full]43151[/ATTACH]
In this post, I'll be talking more about Region "B".
This region is often called Miller Plateau...
A region many MOSFET Builders act upon as a linear curve of switching.
[ATTACH=full]43152[/ATTACH]
But you also need to consider how the Gate is designed - so that if the spacing between the PLATE of the Gate, and the Substrate underneath it - can affect not just the Speed it can switch, it's also HOW QUICKLY and to what negative effect the fields the Device handles will negatively affect the pulse.
The SURFACE Area the Gate uses to exhibit the field - offers us a way to see inside the package without having to know the physical attributes of Die design.
We have Clue thru a value called Gate Charge. It's a factor of "how much charge in Coulombs" the Gate will show, exhibit - when we apply an RF, AC or Audio field to it. It will take effort to fill in that Gate Plate. So the amount of current the part needs to push the Voltage onto the Gate is an Instantaneous value - once the Charge Change is acheived, the Gate field becomes a static event until you turn off the device.
In this post, you are using the Miller Plateau as this effort of Linear operation - so the Gate Charge is used as a working value in Frequency of Xc and the IDS as a XL to help us determine working resistive Range values for On and Off and the Diode is used as a Directional Flow for the Devices ON state.
So the ON resistor value needs to be of a good ohmic value as to allow the Gate the best time ON in duration DURING the Miller Plateau event and the OFF resistor value is so you can remove the Gate Charge easily enough to allow the Fall Rate to be equal or better than the Slope you get of the Fall side of the Frequency of interest.
This loop can generate a lot of current thru the Gate, so care must be taken to allow some drive current against the Voltage needed to turn on the Gate - hence the Devices' Gate Capacitance needs to properly considered so the circuit can be tuned to make it effective as well as efficient for the Gate to follow the RF signal in this Linear region.
This in turn takes effort to ACCEPT a charge and also works against a condition of a secondary event of another field, that once the Device starts to Turn On, works as a counter-EMF field (think of this as the pulse and the effort it takes to overcome the inertia of being held off, now turning on and carrying thru the action) Much like how a garden hose nozzle hooked up to a faucet and as you turn on the water - nothing happens for a moment - this is that Plateau action; the moment of pressure building but not yet filled the space that air is being displaced from - for the water to arrive/hit the nozzle and then develop pressure or recoil force as the water shoots out the nozzle.
- Not all of the device turns on at the same time - it Propagates - and also produces a field UNDERNEATH the Gate, that interacts with the Gate's own field - as if it were Reactive (READ: Impedance-Related) event - being out of phase with the incoming charge - absorbing the Gate's Field in the process - while the device goes from it's Off state, into On-State, the transition is a linear state, this is the Miller Plateau - the "hold" of voltage that the device "takes up" as the Device goes from High-Impedance State to a Low-Impedance state.
- In this transition region, the device is not a true switch, it is taking in Current, IDS at the same time a voltage drop VDS is occurring
- - these two events are affecting the Gates ability to keep accepting a charge until the device is fully on - this is a Linear region we can use.
- If we treat this as a resistive event, we'd simply choose the time on and time off and install the part and be done with it. We cannot...
- Because of the RISE and FALL times - the Gate capacitance and the Charge affects of the switching pulse affect how quickly or fast the switch the MOSFET is - to work for RF.it also works against us as a Frequency Response - to Impedance effort. Not just Resistive, it's capacitive and inductive.
- We have to treat this as installing a BJT that is designed to RF, so not just any MOSFET will do.
- We need to consider the Gate Capacitance and the Amount of Charge it will take as a factor of time so we need to remember our Timing being the Realm of 37nS
- - so we should use values of capacitance that offer the LOWEST Xc reactive element
- - because XL will be working against us as the counter EMF effect. That effect, looks like an Inductive event from the Transition of no Current Flowing thru the Device to Full Current capacity.
- So I told you the above to show you this...
Article can be found here : EiceDRIVER™ - Gate resistor for power devices (infineon.com)
[ATTACH=full]43153[/ATTACH]
IGBT or MOSFET - note the use of RGoff RGon and DGoff - look Familiar?
We're doing it in our EKL devices
- but when you have no Driver to provide power,
You use what is the most power present
- your RF Input.
The article is based upon 2006 Data, but still applies to the effort were trying to accomplish.
So, go back to the Graphic showing those values in the Screengrab...
[ATTACH=full]43158[/ATTACH]
In the above enterprise Example:
- R values are low, but in this application, the operator focused on making the MOST output from the IRF520 and attain large values of RF which did include 2nd harmonics
- This approach gives you the greater Faster, Switching times (being low resistance they provide faster response to the flow of current within this circuit)
- YOU DO NOT WANT THAT RESULT
- Shortens the life of the parts...Gate Current rises and can damage the part. In this example they maximized the Drivers output to attain the values - which needed to remain constant - this was not done with a Swing Kit or had that in mind, just demonstrate how to convert BJT to MOSFET.
- In our example, you may want to raise the ohmic values to lessen the Gate current to improve the Switching times.
- Yes, extra Resistance may cause a longer On time, use discretion with the chosen values of Resistance used to keep the On-Off-On times to the Frequency of interest.
- Due to the Gate current possible, the values should equate to a TOTAL Ohmic values of about 3K (3,000 ohms) ( RGON + RGOFF ) as Maximum so you don't lose the RF power you're trying to send to the Gate, from becoming lost in the circulating currents and rectification this additional circuit provides nor push the ON voltage too high and wash out the RF signal you want to amplify.
Another aspect to consider...
The Gate itself and the capacitance is exhibits affects the Admittance of the output from the Previous stage. So yes, the input capacitance needs to be considered...why?
Due to the effects of Series Capacitance - which lessens the ability to transfer RF energy over to the Gate, To Alleviate the condition you can apply a Resistive element to allow the Input Capacitance Admittance a chance to transfer energy into the opposing plate and not lose energy in the Transfer of charge - RF signal - into the circuit.
Again adding a Resistive elements can have negative impacts for driving the Gate.
- - including the On and OFF times
- - an accumulative event affecting the tuning of the Gate capacitance to allow the Miller Plateau to achieve its' full effect.
- - so Selection of a value of both Capacitance for the power flowing INTO the circuit as well as Resistance for power flowing THRU the circuit into the Gate
- - for if its' not considered, the Admittance output of the preceding stage can generate a mis-match in the coupling between the stages. an admittance issue that can damage the preceding stage.
- So if a Resistive element is not required, but Gate appears to lag behind, the effort may not be caused by the RGoff being not low enough, for if RGoff is TOO low, Gate performance suffers in both On time and lack of linearity in Range of amplification
- - the Linear region you simply are taking away too much RF to obtain an ON Gate voltage.
- This condition can occur from the Range of RF levels entering in thru the Input Capacitor is too low in RF power
- - the amount of RF the previous stage passes thru this cap is not enough to sustain the On event.
- Try adding a low-ohmic value of resistance in Series with the Input Capacitor BEFORE the EKL derived device (EKL for Gate).
- The additional Series resistance helps limit RF current and can offset an inherited Resonance problem of stage coupling if you are having issues with Gate charge and it's latent delay effects where;
- the device doesn't turn off correctly or quickly enough, generating a power dissipation problem
- the device simply is staying on too long and RGoff resistance would be too low for the Gate to remain properly On during the cycle because of RF input Dynamic Range of level change is too great.
- Unable to Trigger ON for the length of the transmit, like SSB modes.
[ATTACH=full]43165[/ATTACH]
Another Method.
Use the DC power added as a low trickle voltage, to provide enough power in the EKL derivative design to help offer linearity in Range - that the RF Signal develops from the Previous stage, so the RF power can Trigger and sustain the On cycle event
IN a current Radio design...
[ATTACH=full]43160[/ATTACH]
There are several approaches to Biasing that can be utilized in various ways.
The above is from a radio of current production that has SSB modes - uses two MOSFETs' one as a Driver the other as the Final so the effort to make both work linearly can help us in making the Divider circuit work.
In the examples the Driver and Final use Voltage exclusively to power their respective Gates, using a 4.7K resistor as a "buffer" to keep RF at the Gate, and not flowing onto the Bias circuit. Again, a Isolation of the Gate from external inputs so the Gate can see a DC voltage and not thru the Intrinsic Diode Junction effects. They both use a Variable located on the Ground Rail side - to offer better means to control the Rise of the Gates ON voltage by holding the Ground Reference to the overall effects of the Composite the Variable resistor provides - all input of the Potentiometer are used which then reduces the over Reactive effects in the variable value the Potentiometer provides.
