Mobile Linear

[n1ps]

Well...finally finished the linear and got it on the air this week.  Got good reports.  Very clean amp, no garbage out (except what the 706 puts in).  This is for anyone who is interested in bulding one of these puppies...I'm sure it can be improved.
 
At the moment I have the amp set at 100W (carrier) out with 8W in.  The only issue I had turned out to be the 706 (shocker eh?).  It has pretty bad asymetrical modulation in the wrong direction and sends the drain current downward with voice (and thus the RF output).  I have not looked at this yet in the 706 manual but more suspect the carrier balance.  So in the meantime it is installed and working.  It is one band only right now....80M.
 
Because this is not your usual linear...using switchmode fets at 3.8MHz never mind using as a linear  , there was some Xperimenting.  Mostly for me it was the bifilar choke T2.  I thought that 3 turns would do it (the calculator told me it was OK), but needed 10 turns on a 43 core to clean up any noise.  I filtered and bypassed the hell out of all ins and outs.  The case happened to be one I had on hand as well as the PC board which I etched from double sided stock.  In fact all the components I had on hand except the DC on-off relay which came from Roger N1XP.  I had all the cores and had some IRFP260N FETs (the same as the Class E modulator).  Total cost was what I paid for the parts over time at festers (I know, priceless). 

Not seen in the trunk photo on the other side of the amp case is the heatsink,fan and LP filter.  The sink was a little small, so I added a PC fan (onto the sink where it belongs).  I was particularly concerned with heat on these FETs.  I cut some 3/8" thick solid copper as a heat spreader for each FET. The FETs mount directly to the copper and they are insulated from the sink using silpads (and lots of heat grease). I'll worry about overheating the FETs somewhat less on one of those hot summer days....
 
Sorry for the hand sketched (and erased) print.  All I have time for right now.  Well, first I have to thank Mark KA2QFX.  When I heard that Bob K1KBW and Ken ???UDZ were making amps using this method, Mark kindly informed me what he believed was used and then sent over an LTSpice diagram of what he recommended so I could take it for a virtual spin.  Over time I modified it to my liking; if you would like a copy email me.  Mark and I conversed many times via email and he provided much advice.  TNX Mark!

Circuit decription (for those that care): It is a 2 FET unit as you can see and draws about 10A at 100W carrier and 13.8V DC input.  The in and out xformers are binocular types per Helge of Motorola fame.  The output xformer I completely over-did with 6 cores  .  But in my defence I did not really have a good handle on figuring out what the saturation points were for the cores.  The conductors are enameled types and/or teflon. 

I expected the bias setting to be touchy as the transconductance of these FETs is very high.  I ended up setting  the bias at about 1.5V.  At 3.5V some instability began to occur....I was pretty surprised that the FETs did not begin to show idle current until about 3V.  I initially set the bias at 3V to set the temp compensation. Then turned it back to 1.5V.

As mentioned I also added a simple thermal compensation circuit using a small xistor to keep the FETs from running away and getting hot about it.  The xistor is mounted to one of the copper spreaders and provides neg temp. compensation.

A last minute change was to add series caps in the drain circuit.  I used 715 series orange drops.  They have some inductance but I was more concerned about RF current.

The cores are the FB-43-1020 as used in the E rig.  I used these for T2 and T3.  I used smaller cores for T1. T1 and T3 are binocular types as noted. T2 has a neg feedback winding to reduce the gain of the amp and add some stability.

I think the most work was all the control circuits and relays. Also the remote control head and the control cable.  I decided that I wanted remote control, so I made a simple control head with a drain current meter and on-off switches for the fan and the linear. Also I added a low pass filter, a 5 pole unit on the outpoot so my PW little machine will not swamp out those nice 28 meter foreign  broadcasters .  I think the 3rd harmonic was about 40 dB down as measured on my spectrum analyzer.  Enough for a mobile.
 
I may get ambitious and build a 4 FET unit next year (like the BIG boys).  We'll see how this one does (or doesn't). BTW I saw Bob's 4 FET unit and Ken's 8 FET (!!!) unit...beautiful stuff...mine is pretty ugly with single sided PCB construction.
 
The trunk photo shows the rats nest.  The linear is under the 706...the device to the right is an audio amp as the 706 has PW speaker audio. 
 
The antenna many have seen before.  Mercedes enthusiasts look at it in horror.  Its a single band homebrew center coil type.

That's my story...the experiment will continue....
~ps

[KD6VXI]

I feed my FT857 into a 350 watt amp, and then use that to drive an old 5.1 sys sub and sat's.  Feed the 857 into one channel and another radio into the other (scanner, cb, etc, depending on where / what I'm going / doing).  I am actually surprised at how much low end is actually out there...  My sub gets more of a workout than I thought it would.

My 16X2879s fall apart below 40, so this is VERY intriguing...  And a LOT cheaper!  for a 4X2879 amp, you'd be looking at over 100 dollars in transistors:  I sourced the FETs you used at about 4 dollars each at mouser!  NICE!!!

Any chance for some inside pix? 

The downward mod on AM is normal on those rigs.  It's the ALC action.  Take a look at the thread by KE7TRP and others about the alc jack input...  Throw a couple volts at the radio, and all of a sudden, GREAT AM reports.  I get a little controlled carrier action at high levels of mod, but have no problem getting 90++ percent mod out of my 857, nor do I have a problem getting good audio reports from it.. 

--Shane
KD6VXI

[WA1GFZ]

I don't see how you can get 400 watts PEP from a 12 volt final with a 1:4 transformer ratio. Look inside any 100 watt 12 volt rig. The final output transformer is 1:4 or 1:5. I can push my TR7 to 200 watts with a 1:5 ratio but avoid doing it. This doesn't add up?? I wonder if you ever performed any two tone testing?

[WD5JKO]

At the moment I have the amp set at 100W (carrier) out with 8W in.  
Circuit decription (for those that care): It is a 2 FET unit as you can see and draws about 10A at 100W carrier and 13.8V DC input.  

    Nice write up! I'm still a little puzzled though as to the power point of this brick. You say 100W at 10 amps with 13.8 volts. That is over 70% efficient, so I presume that is flat out, and for AM linear use this is a 25 watt carrier out amplifier?  Unless I'm mistaken, doing AM linear with headroom for 100% upward modulation still means the carrier efficiency is ~ 1/2 what we get at full output. The CB crowd violate this rule by compromising upward modulation for more carrier output where they might settle for 100w carrier, 20% upward modulation, and 90% downward modulation.

    Sure would like to learn more about this amplifier, thanks again for the thought stimulating work!

Edit, I bet an amplifier like this is ideal for high level modulation using one of those big car stereo amplifiers in a bridge configuration, and a suitable modulation transformer. On a smaller scale, the IRF530 in the Retro-75 has a perfect trapezoid pattern up to 200% modulation. Getting 100W plus on AM mobile while modulating the tar out of it (>130% upward peaks) sure would be an effective setup.

Jim
WD5JKO

[WA1GFZ]

In a linear amplifier a push pull stage each FET will will swing between 0 and 24 volts assuming no losses. 48 volts times 4 (turns ratio) is 192 P-P volts. Convert to RMS divide by 2.8 or 68 volts. 68 *68 / 50 ohms = 94 watts. Usually the voltage is closer to 14 volts so you hit 100 watts. Now drive it harder into square wave operation (no longer a linear) and the power goes way up but so does distortion. Square wave operation I can see it hit 400 watts.
So to increase power you need to increase the turns ratio or shift impedance in the low pass filter.
When you drive into saturation the transformer stores energy so the off flyback voltage swings higher than 24 volts.

[n1ps]

TNX all on the comments.  This is a continuing experiment.

Shane on the ALC idea:  yes I agree and I may add a simple ALC voltage and try it.

Frank - yes you are correct.  When I do get the audio right side up, it will be a 40-50W linear at best and that will be driving it 5:1.  The limiting item here is the 14V power supply (the car battery) and the 24V drain voltage.  I'm going to see how far I can push it and see where it goes.  The only way to get 400WPEP is to either double the DC supply (switch in a series battery) or double up on the FETs.  That will be a future project. The idea was to start with something manageable and better than the 15W I currently run on the 706.

Next time I get into it (3-4 weeks) I will send along some board pics if anyone is interested. 

Pete

[WD5JKO]

In a linear amplifier a push pull stage each FET will will swing between 0 and 24 volts assuming no losses.

The FET amplifiers run at 13.8V have always had limited efficiency over the "ON-Voltage" issue, where this voltage is more than an equivalent bipolar transistor. The IRFP260N however has an RDS on of .04 ohms, which makes me think we've solved the issue. Then looking at the charts on the data sheet, I see a best "on voltage" saturated of 1v with Tj @ 175 Celsius. See figures 1, 2 at:

http://www.irf.com/product-info/datasheets/data/irfp260n.pdf

I still feel that this amplifier will stand up and kick serious butt if high level modulated with the readily available off the shelf car stereo audio power amplifiers. Here the RF brick thermal issues are largely solved since the brick runs at maximum efficiency at full drive.

Jim
WD5JKO

[WA1GFZ]

You are talking peanuts maybe 1 volt (if you are lucky) of extra swing in the linear mode. The only ways to increase power is multiple amps into a combiner,  a higher turns ratio in the transformer or an inverter to increase the drain voltage. There is a reason 12 volt rigs ran 100 to 200 watts for the past 35 years. Simple algebra.

[WD5JKO]

You are talking peanuts maybe 1 volt (if you are lucky) of extra swing in the linear mode. The only ways to increase power is multiple amps into a combiner,  a higher turns ratio in the transformer or an inverter to increase the drain voltage. There is a reason 12 volt rigs ran 100 to 200 watts for the past 35 years. Simple algebra.

Ok, 1 volt out of 12 is peanuts, and high level modulation has no advantage over linear mode AM. I stand corrected.

I do wonder though about in RF service how a FET compares to a Bipolar as we approach the clip point of saturation. With a bipolar the slope changes as the gain goes into compression as we get to the clipping point. With a FET, maybe things are different.


Jim
WD5JKO

[WA1GFZ]

I've read bipolars have better IMD3 but FETs are better further out IMD5, IMD7, IMD9 so the angle into saturation is different. A bipolar is harder to drive. You don't see many low voltage RF FETs they usually like to operate at above 20 volts. Switching FETs may be different due to very low on resistance but I bet they won't go to 10 meters.

[WU2D]

This might give more rail. http://www.current-logic.com/shop/index.php?main_page=product_info&cPath=4&products_id=187

[WD5JKO]

This might give more rail. http://www.current-logic.com/shop/index.php?main_page=product_info&cPath=4&products_id=187

  I find it absolutely amazing how easy and efficient power conversion has become. So making 24V out of 12V is an attractive idea in many ways. I remember how hard this was to do 30 years ago in any reasonable size, cost, and weight. The IRF100's had just come out from IR, and Siliconix had the 'V' Fet...

  I still wonder though why not take the two FET brick as is, optimize for maximum out at maximum efficiency, and then get a garage sale audio power amplifier for under fifty bucks, and make a high level modulator out of the thing. Look for one with a DC-DC converter within, and bridge the stereo amps into one big amplifier. To me there is an attraction here in simplicity, and efficiency where we can make 100 watts of AM (carrier) at > 70% efficiency, and then modulate the tar out of it to > 500w PEP. The tricky part will be finding a suitable modulation transformer scheme. Maybe me being on night shift since Christmas has made me lose my better judgement. Or maybe you folks in the north like an extra heater in the car! 

Time to go to sleep now..

Jim
WD5JKO

[WA1GFZ]

Far better to build a PDM modulator with a step up transformer so the modulator runs close to 90% efficiency. Then you can run the final at a higher voltage.

[KD6VXI]

Far better to build a PDM modulator with a step up transformer so the modulator runs close to 90% efficiency. Then you can run the final at a higher voltage.

Frank,

What about the car stereo amp, but kill the FETs on the output...  IIRC, most of the stereos use a switcher in them to go +/- 55-130 volts...  Could one modify the Power Supply to modulate the signal?  I'd done this using an old Lambda and a bipolar class C amp a few years ago....  Could you use this method to get the higher voltages necessary?

Would be an easy way to go I think...  The power supply arrangement usually uses a single IC to generate the signal for the FETs in the power supply, then they run that into a toroidal step-up transformer.

--Shane
KD6VXI

[WA1GFZ]

It depends on the pwm control. Some can be modulated better than others.
It is very easy to do.