AN762 Bias test & The beginnings of a 5W driver amp for the SDR-Cube...

This weekend I applied power to the AN-762 Amplifier for the first time prior to installing the final transistors (MRF454's) to verify that the bias regulator (the LM723H and bias diode circuit).  CCI stated that a 15 to 100 Ohm load needed to be applied B-E on one of the two MRF454 junctions for some load to the linear regulator/pass transistor circuit.  Then the adjustment range should be 0.5 to 1.0V across the resistor.

I did two 100 Ohm 1/4W resistors in parallel (50 Ohm load) and saw that the range was from 0.41V to 1.11V or so.  So it looks good.  That range is supposedly safe for the PA transistors.  As CCI suggested, the bias was preset to 0.50V.  Nominal operating voltage to the base should be 0.68V or so for class AB but starting off low is a safety margin.

Eventually, the correct way to bias the amplifier with the MRF454's installed is to adjust the bias voltage (really current on a RF Bipolar Junction transistor like the MRF454) so that the supply current drawn with no RF drive is, in the case of the MRF454's, 100mA (minumum) per device, or 200 mA total in Class AB push-pull service.  This can be adjusted up a small amount for IMD improvements, but will end up being in this general ballpark.

I still need to build the K5OOR switched filter module.  I really need access to a L/C meter when winding so many toroids (uuuughhhhh.. necessary here for many reasons though) , so I'm thinking of bringing the board in to work next weekend Sunday or after work several nights to do this.  I need a decent LC meter (or at least "L") some day... it is on the list.  The one by AADE seems to be the best hobbyist priced unit, and likely that is what I'll eventually get.

In the meantime, I decided to start experimenting with the Mitsubishi RD16HHF1 RF MOSFET that is readily available from RF Parts and the "Toroid King W8DIZ" (Kits and Parts.Com)

It's a single device, Class-A design.  I suspect that the part is intended for use in CB radios, so a singe device at 4-6W.. seems about right for that service.. :O)

Mike Collins KF4BQ has spent a lot of time playing around with this device in the requisite power range (5W ish) needed to drive the CCI / AN762 Amplifier to full power (on the "Mobo" project) and it seems it is capable of this with low-IMD output in a single device configuration if one is willing to trade off efficiency and run the device in Class-A.  What do I mean by this?  KF4BQ recommends 400-600 mA of bias to run these in Class A... the 100W AN-762 in Class-AB only needs 200 mA quiescent!  So there is a definite efficiency trade off.  This means for fixed station use, it's probably fine.  For battery, AB with worse IMD on the 5W driver might be a better design.  (Or of course Class-C for CW folks and maybe someday class-E and the more esoteric efficient amps...)

Class A is a good way to start experimenting with a device, obviously, for linear service.   I decided to take one of my 10 RD16HHF1's and built a prototype testbed for this.  If it is sufficiently clean to drive the AN762 with 400 mA of Bias, I can live with that for fixed station or even 12V mobile use for ICAS amateur service.  The circuit is simple enough.

Here is a drawing of the first version prototype.  It's meant to be driven initally on 20m with the TXPA (PAF-3) at about 750mW (I can drive it down now with the 1.01 SW).

The circuit is more or less stolen from Mike who in turn stole it from the "box.de" group.  However there are not many different ways of doing a N-Channel MOSFET class A amp, anyway.  The only clever thing about these amps are the "balun" designs which...effectively are RF Autotransformers instead of the normal transformers with two unconnected windings for broadband transformation.

The input impedance is set by R3/R4 to 50 Ohms (which will likely change if another driver stage is used).  There is roughly 8 dB of attenuation shown here so I can run it with the 750 mW or so that I'll put out of the SDR-Cube.  I'm guessing about 17dB of gain once I get the feedback tweaked.. maybe I can lower that to 15 dB or so for better IMD.  We will see in several weeks.  It took me quite a while to draw this up and build it because it took me a long time to re-learn Eagle (light) but the advantage to this is that once I get the amp working.. I'll have a few PCB's made up at Batch PCB (or elsewhere)...it's easy to do if you start in the PCB layout software from the beginning.

I intend on putting a small driver in front of this amp.  I am guessing that the drive at the gate needs to be right about +20 dBm (100 mW) so I really want to make a broadband amp (probably another class-A device -- smaller).  I am thinking that the 2N5109 at kitsandparts.com is ideal, but I have a lot of surplus 2N2219A's in the drawer of junque.

So likely there will be a 2N2219A amp built in front of this for the 100 mW ish drive for this driver in the next version of the amp.

Anyway, the one above looks like this when hacked together of old parts:

The bias circuitry on a MOSFET is voltage controlled.  I assumed that 5V would be enough.  It would in Class-AB push-pull.  To get this comfortably into Class-A range, apparently 400 mA of quiescent drain current is needed.  This was at about 4.82V on my device.  Many of KF4BQ's data seems to make me think that up to 600 mA might bee needed for good IMD.  That's at about 5.11V with my device. Not going to get there with 5.00V.  

I solved this by putting a "cheater" diode in between the 78L05 ground connection and real ground.  This raises the maximum bias voltage to 5.71V or so which is probably still safe for the device and allows up to 675 mA (on my device) to be "dialed" in with the bias pot.  Also I added a 8.6K resistor in series (on the "low side") so the overall adjustment rate of the 5K pot (13.6K total.. okay since the gate has such a high impedance and low bias current draw.. it's a MOSFET remember).. is more in the "actively biased" range.. rather than having all of the action at the last 15% of the pot travel it's now more like 65% of travel.... it's good.

I have my amp adjusted to 400 mA.  I will next feed in RF from a sig gen without the 8 dB atten, or the SDR-Cube with.. see first what RF power output we get and verify nothing starts smoking.  If it looks right then a directional coupler, attenuator and the Spectrum Analyzer will come out and I'll independently verify Mike Collins' results.

Then the 0 dBM to +20 dBM BJT amp would be next.. and then.. well the 100W amp.  I should finally at that point be able to go from 1 mW to 100W with the solid state and then 800W with tube amp.. all with home built equipment.

It'll still be awhile but I'm headed there.

One immediate change to the amp is that some sort of PNP or P-channel device, in a "high-side switch" configuration,  will be needed to turn on/off the regulator for bias.  With the bias off the device draws almost no current, so switching the entire amp off with a relay is not needed.

I've done this all the time with P-channel Enhancement mode MOSFET's... but I'd like to use something smaller and cheaper.  I'm wondering if the J176's I have would work.. that or 2N3906's.  The 3906's would but an additional NPN stage is needed for the safety of the 3.3V or 5V microcontroller.

I still need to think that one through.