EMRFD Message Archive 9771

Message Date From Subject
9771 2014-02-18 22:00:29 rftdugan EMRFD 2.11 power amplifier
I've been experimenting with high frequency RF amplifiers along the lines of that in EMRFD fig 112 and OM6MU's version at >>

I built one for 14 MHz with an output of 15 watts. It uses an IRF540. The output is pretty clean. I've used it successfully for many DX contacts using only a ground plane. The gate input follows the OM6MU design and the output follows the EMRFD design using a shunt capacitor, a 4:1 transmission line transformer, and a pi section low pass filter.

Then I tried to replicate it for 21 MHz with an IRF510 at the same power level. But, despite considerable fiddling with input and output networks, I cannot get a clean signal--lots of blur on the scope at the 50 ohm load following the low pass filter. The output is clean if I run the 4:1 transformer directly into the 50 ohm resistive load. But it blurs badly with addition of the low pass filter. This is the same problem as noted at EMRFD page 2.42 as a "disturbing characteristic". Addition of shunt capacitor on the IRF510 drain reduces somewhat but does not appreciably change the scope result.

Any suggestions?

9772 2014-02-19 03:55:49 jorschei Re: EMRFD 2.11 power amplifier

Hi Bob,

May be the rise and fall time of the IRF540 is too high 60 and 50 ns  for this 21 MHz frequency.If you put thad in the Class-E calculator of Tonne soft it tell you to lower the frequency.  I go to experiment with the FDT86256  FET  aprox 2.7 ns  it will go to 50 MHz in the Class-E calculator bud is only 2-3 watt power.

Try;   http://www.tonnesoftware.com/  Class E-calculator




Turn-on Delay Time

Rise Time

VDD = 50 V ID = 12 A

RG = 4.7 W VGS = 10 V

(Resistive Load, Figure 3)

60ns   45ns


Turn-off Delay Time

Fall Time

50ns   20ns



 73'  Joris PE1KTH

9777 2014-02-22 18:11:16 kb1gmx Re: EMRFD 2.11 power amplifier
I've built a number of amps using the IRF510 and the results are...

*Wa2eby/K5oor  push pull pair at 26V 37W at 10M  (as much as 55 at 20M).
  A commercial kit of this is known as the HFpacker 160-10m.
*My own experimental design amp for 6M [50.150mhz] at 24V yielded 21W 
  from a single ended variant similar OM6MUs.
*Duplicated the OZ1PIF amp (one module) using 8 IRF510s and got a nice 
265W at 40V and 190W at 28V at 6M.

The blur is a likely side effect of both measurement method and insufficient loading 
of the drain.  

Things I've found to watch for with these parts.
* the DC and low (audio) frequency input impedance of power MOSFETs is high at RF
it drops significantly due to high input capacitance and miller capacitance (Drain to Gate capacitance)
so treating the input like a high impedance is fine for Bias only.  The RF path must be
matched or loaded for stability and prior stage matching.  
* lead inductance, if it has leads there are likely too long make the sort as possible.
* Delete the drain lead and use the tab as the the connecting point, Gets the Drain to Gate
  coupling a bit lower.
*  flying heatsinks, The heatsink should be at electrical ground for RF.
* Thermal resistance of the device is poor, at more than 40W input power watch for heating.
*Do not add inductance in series with the gate without loading it well as the FET has RF gain.
extending to VHF or higher.  FETs in general do not exhibit FT, parasitics limit upper frequency.
* if there are parasitic oscillations, series resistance 1-15 ohms right at the gate with short leads 
or better yet from the pad to the gate using a SMT resistor in the 1206 or 804 size.
* loading the gate to some resistance in the 10 to 100 ohms region is sometimes helpful
in assuring input matching and loading as well as assuring low Q and reduced unwanted
*Negative feedback is useful with care as its most effective at lower frequencies.  Resistors used need to be low values . 
* At frequencies above a few Mhz and a few watts use care in grounding as enough RF current 
can flow to create sneak paths or worse feedback.
* my best success with RF power amps always includes a two sided board, edges wrapped from 
top to bottom with cooper foil, and through wires to selectively insure places like next to the Source, 
Gate inputs, Drain loading components are well coupled to ground by the lowest RF impedance 

The gain of the device is to be respected.  However feeding it is a matter of trial and error.

9779 2014-02-23 03:58:13 g3xaq Re: EMRFD 2.11 power amplifier
Allison, please can you expand a little on the comment

*  flying heatsinks, The heatsink should be at electrical ground for RF.

The lure of internally mounted forced-air cooled heatsinks floated at drain voltage is tempting from a thermal point of view.

73, Alan G3XAQ
9780 2014-02-23 08:41:07 kb1gmx Re: EMRFD 2.11 power amplifier
Yes, Flying heatsinks, heatsinks that are electically at the same potential as the tab or 
case of the device.   the problem is if the device is both high voltage and high power
its both a RF burn risk and decreases stability.  Last thing needed is finals with extended 
surfaces that are RF hot with drivers and maybe even predrivers or VFO in close proximity.
Stability is often an issue with designs built like that.

Tempting as you do not need an insulator but, the thermal cost of that insulator is small.
Keep in mind TO220 and similar have poor thermal transfer compared to a good mica 
insulator.  There are TO220 and similar devices cases that do not need insulators.

I do use computer heatsinks with fans but at RF ground.  Works very well.
Also inside closed cases the thermal rise is much higher as there restricted airflow
or if the case is tight no airflow. 

9781 2014-02-23 09:31:03 g3xaq Re: EMRFD 2.11 power amplifier
OK Allison, thanks for expanding. We have somewhat different applications in mind. I am thinking of a self contained stand alone amplifier so I see a "flying" heatsink as being closely analogous to an external anode PA valve (4CX250, 3CX1500, 4CX10,000D, etc). The airflow management issues are not too dissimilar either when using fan plus heat exchanger composites such as these


I'm looking at using high voltage FETs with drain loads in the hundreds of ohms so the additional output capacitance from mica insulation kits is also unattractive, although there are Alumina insulators available that can side-step that issue. I think it's horses for courses. Everyone's milage may vary.

73, Alan G3XAQ

73, Alan
9782 2014-02-23 10:13:23 kb1gmx Re: EMRFD 2.11 power amplifier
The situation with external anode Valves is very different as they are usually used inside a shielded
compartment where the grid or cathode side is isolated from the anode side.  Something rarely done these days with semiconductors.

I'd think to develop power using FETS at high voltage you will still need load impedances 
lower than hundreds of ohms. Even at 50V I find the load for my 6 and 2M amps are down in 
the less than 5 ohm range.  So unless your using something in the 300V range hundreds of 
ohms sounds high to me.

Also calculate the capacitance of a to220 package to ground through a .003" [.08mm] thick
mica, Its not many picofarads (very few).