Electronics > RF, Microwave, Ham Radio

Help understanding and modeling an amp

(1/2) > >>

I was recently gifted with a tube of MRF2628's so I figured I would take a walk on the wild side away from my usual stuff and give an RF project a shot for fun.

I am perhaps lucky that the transistor datasheet has a reference circuit.  I have been looking at that and pulling my hair out trying to figure it out.  I have spent a lot of time with chatgpt asking it questions and working through the values of the various inductors.  I now have a fairly flushed out schematic. 

I have downloaded the latest version of simnec, which is the latest "version" of symsmith and I have the input matching network laid out, but I do not see how to model the base of the transistor.  In the datasheet this is a smith chart, and they broke off the resistance and reactance at a few different frequencies.  So, how in this program do I model terminating into a complex impedance when I know both parts of it?

Also, looking at the chart, over the frequency range it covers at any rate, the resistance seems to be pretty stable, and low, as I would expect.  It is the "real" part so the fact ir changes at all is a bit surprising, and it is low as I would expect going into the base of a bipolar transistor.  I think I understand that part fairly well.  One thing, on the smith chart, it states that Zo = 10 Ω, does this mean the "normal" 50 ohm smith chart is now normalized to ten ohms?  And is that only for the output?  Also, does that have any bearing on the chart they broke out with some values at different frequencies?  This is all  new to me.  Most of my "education" is from this YouTube video:

I have built amps from kits before, the last one being based on an MRF186 and the biasing on that I totally understood.  I do not understand the biasing on this at all, unless this reference design is running class C.  I am assuming that most linear RF amps would be class B ish. 

Any insights on this would be appreciated.  I want to move more from gee, that looks about right, to having a decent model and understanding of this.  I am also not opposed to a more complex but intuitive biasing scheme.  I also, perhaps wrongly, suspect that the complexity of the input and output matched have to do with the wide spectrum the reference amp covers.   So when I get it modeled, that may change into something simpler.

The images, the amp is the reference out of the databook, the smith chart is out of the databook, and the last one is my simnec simulation so far.  If you are interested, it also containes the inductance values I came up with for the coils.

This device is intended for FM operation (see red circle in attached image) - hence, class C.

To simplify your simulation, initially try matching at one frequency first before attempting a wideband match like the datasheet's fig. 1

You are right on the "Zo = 10 Ω" text meaning the smith chart is normalized to 10 Ω. However, you can read the un-normalized impedance directly from the table below the said chart. 

The datasheet's information supports designing at 136, 150, 175, 220 MHz and the intermediate values only. I see your simulation is running at 100 MHz. To simulate outside the datasheet frequencies, you will need to characterize the transistor at your frequency.

This transistor seems working in Class-C in datasheet. Output and Input Optimum Impedances have been given for max. Delivered Power toward to Load with Z0=10 Ohm Normalized Characteristic Impedance.( there is also a table )
Any Nonlinear models is not available for the transistor so all you can do is to implement this application circuit carefully and measure. That's it.
Simulation is not possible unless the nonlinear or at least linear s-parameters are available.

First, thank you for the input. 

On the input side I now understand that the named frequencies on the smith chart 136,150,175, and 220 Mhz the resistance and reactance listed next to them are correct for those frequencies.   So the big question is how do I represent that in SimNec?  I am figuring someplace around 178 Mhz (center of 136  - 220) the impedance match will click, and it will go down hill on both sides of that.  I want to see that.  After that, I can work with 100Mhz and scale values so that is in the center.  I can interpolate what the base of the transistor is going to look like from the given numbers, or try to read the chart.  It might be worth looking at the standard deviation to get a feel for how accurate an interpellation might be.  So the question I have now is how do I represent the base of the transistor in SimNec?

On the output side, you are correct, the amp should run class C.  I was concerned about a class C amp and FM as to have low distortion it depends on the output circuit ringing, and that is resonant at one frequency.  Apparently it as a low enough Q ring with stimulation near that frequency as well.  The modulation does make a gross difference.  I still only kind of get the bias as it looks like it is being held at ground, so the positive side of the RF wave turns it on?  I guess that makes sense.  It turns on for the half cycle, and the output tank both matches the impendence to the coax/antenna and rings enough to simulate the otehr half of the wave.  Does that sound about right?  Also, the output tank should be a filter to catch harmonic spurs from the distortion from the transistor switching on and off. even though these should be very close to the zero point.  Does that sound about correct?

The table's data are in R+jX format. So, model the transistor's base using a series RC (or RL at 225 MHz) component.

The transistor output should be connected to an impedance transformation network that matches to 50 ohm. After the impedance network, then you add a low pass filter to attenuate the harmonics. 


[0] Message Index

[#] Next page

There was an error while thanking
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod