FET power amp rating

[VE3ELQ]

I have seen many power output claims of single BLF188 FET amps of 1150, 1250 and now another by PA0NVD of 1300 watts in another thread. I dont understand how these claims are possible so I must be missing something important.
For anyone using one of these devices or equivalent could you please provide the following:
Operating voltage (Vdd)
Measured output power after low pass filtering into sign wave
Input and output impedance of the low pass filter,and
Impedance or turns ratio of output transformer.
Your data and discussions would be very helpful.
73s  Nigel

[pa0ast]

This one may be   http://pa0fri.home.xs4all.nl/Lineairs/BLF188%20HF%20amplifier/BLF188%20HF%20versterker.htm
Anton

[VE3ELQ]

This one may be   http://pa0fri.home.xs4all.nl/Lineairs/BLF188%20HF%20amplifier/BLF188%20HF%20versterker.htm
Anton

Thanks for the link.  Reveiwing the schematics and power claims just confirms my confusion.

Here is the dilema. All of the designs I have seen including the one above, are powered with 50 Volts, employ a 9 to 1 impedance ratio output transformer and use 50 ohm input to output (matched impedance) low pass filters yet claim impossible output power. Here is the problem.  These are class AB1 linear amps so are not hard switched as in the class D,E boost converter designs used for AM. The FETs can be driven only just to the point of saturation, any higher results in flat topping with no power gain just a bunch of harmonics. This means that the drain of the driven FET in a push pull pair will drop from 50V linearly down to zero and the drain of the opposing FET will rise to 100v then it will reverse for the other half cycle. This places 100V peak RF across the primary of the output transformer.  It cannot be greater than that and in reality is slightly less due to the DC resistance of the drain choke and Rds on of the FET.  A 9 to 1 impedance ratio is a 3 to 1 turns ratio (root of 9 is 3) so the secondary will see 300V peak RF.  Converting this to RMS gives 212.1 VRMS. Power is E squared divided by R load resistance. Doing the math gives 900 Watts.  Not 1050 or 1150 or 1300 as is so often claimed. The power could be increased by increasing Vdd to more than 50V, changing the impedance ratio of the output transformer to 16 to 1 (4:1 turns ratio) or employing an output lowpass filter with a lower input impedance than 50 ohms.  None of that is being done in the many designs I have seen and besides the FETs would not survive.

Measurements of my own BLF188 linear amp agree very closely to this analysis.  It will make about 875 Watts but its nasty there so I run it about 700W to keep things clean.  It will not make more than than that period, nor should it.  So what am I missing here folks or is there a bit of horse huey involved.

73s  Nigel

[DMOD]

I have seen many power output claims...

or is there a bit of horse huey involved.

73s  Nigel

YEP!  

Phil - AC0OB

[pa0ast]

Right,  if You use a 50 Ohm dummy. But  what if i use a,  lets say a 40 Ohm or even a 35 Ohm antenna ?
Or i load the pa with a tuner , symetric line , dipole and load it to max pwr output. Stil 212 Volts and in 35 Ohms its 1300 watts. May-be that's the explication.
If that's good for the failure rate of the fets is another issue. probably a little overstressed.
What do You think ?  By the way i don't own a rifined silicium power amp my self. i use stil tubes 2x QB5/1750 at 5 kV. Glows so nice..
Anton

[WD5JKO]

   It sure seems to me that some of the data is taken under overload conditions where the RMS voltage on the FET Drains approaches the peak voltage. This would create a huge odd (mostly third harmonic) output which gets recorded anyway when using a resistive 50 ohm dummy load. Notice that the output drops like a rock when they add a low pass filter.

Jim
Wd5JKO

[KD6VXI]

I looked at those fet boards long and hard.

Comparing operation,  etc...   I came to the conclusion they where pretty much like CB amps 15 years ago.

They have the maximum smoke versions,  and  somewhat clean versions.   The max smoke,  1200 watts output rating.   The clean version,  about. 8 to 1kw.

Then,  after low pass filtering,  clean input,  etc.   You end up with about half to a third of max smoke output.

Not bad,  if run conservative.   Friend in San Diego is experimenting with one.   I went class E ARF448 instead.   We both get 500 watts pep output.   Mine is am only,  his is all mode.

Mine cost less than 50 bucks to build.   I believe he has close to that in shipping.

--Shane
KD6VXI

[PA0NVD]

Hello Nigel
One point is forgotten in the calculations, the coupling factor in the transformers is NOT 100 % and has quite a spread inductance. You HAVE to tune the prim. and the secundary for good efficiency and power. For 27.12 MHz I tune the prim. with 470 pF between the drains and the sec with 22 pF. The 22 pF is dependent upon the wire diameter of the secundary. So it will be very complex to calculate what is happening, you have a structure of two coupled resonance circuits, not a simple transformer. Without tuning, the efficiency and the power go down quite dramatically. For that reason, wideband amps in general do not reach 1300 Watts and 91%.
My amps are for 27.12 MHz generators for plastic welding. The lowpass filter is fine tuned by compressing the coils a littel with a load of 50 Ohms. This indeed will change the loading impedance of the amp. More important is the phase of the third harmonics in the load. This affects the efficiency and the power quite a lot. If the phase is not correct (wich can be adjusted by the length of the cable between the amp and the LPF), the power may be a few hundred watts less and the efficiency may drop to 80% or so.
I reach 91% of efficiency at 50 Volts and 1310 Watts measured at my digital Bird (Wich may be off a few percent as all Birds may do). The max temp of the ferites doesn't reach 65 degrees C and the heatspreader is approx 56 degr. after 1 hour op full power operation.

[VE3ELQ]

PA0NVD
Thank you for your post.  That's an interesting amplifier design that you describe for RF plastic welding.  At 91 % efficiency it is certainly well into class D.  Assuming a 50 ohm load and working the math backwards implies about 120V peak RF at the FET drains which is just inside the 130V breakdown spec so technically possible but with little headroom.. With suitable output harmonic filters this amp could be a candidate for Pulse Width Modulated AM service at about 300W carrier level on any band from 6M down. An interesting project.

But my concern is not for RF welding but for the many inflated power claims for class AB1 linear service for AM and SSB which is what most of us use them for. My experiments and measurements show about 700 to 750 watts PEP with low IMD is about all you can expect with a single BLF188XR device in linear service.  I have not been able to better this and cannot understand how others claim they can.  I would like to see some hard data. 

I do have a new linear amp project nearly finished using two BLF188XRs in parallel push pull. The math says it will do about 1500W of nice clean power.  Will know for sure soon.

73s  Nigel

[PA0NVD]

I am highly interested in the results Nigel. Will it be a wideband or a single band amp?
Indeed a lot af claims are inflated if you have wide band amps, without tuning it will be difficult to reach 1 kW and with good linearity less. With a tuned amp for 1 frequency, 1300 Watts was actually quite simple. I only had to retune the cap between the drains from 390 to 470 pF and to adjust the cable length between the lowpass filter and the amp. That is almost zero now.
Please mind the problem of oscillating with the FET's in phase. When in phase, both the input and output transformers do not transform and change in completely unloaded inductances. That gives rise to a possible low frequency in-phase oscillation wich may be destructive. Not one example circuit adresses this possible oscillation mode. It seems unknown. I noticed Freescale about it, NXP doesn't even give answer to mails and questions. Here in many industries we do not use NXP anymore due to the extremely bad service. Freescale was always excelent in this aspect.
This in-phase oscillation is very simply solved by placing two 5 Ohm resistors across the half secundaries of the input transformer. That gives sufficient load to avoid this oscillation.