300W BLF278 PA 48V (PA0V Nanko)

[VK4DD]

Follow link to see working amplifier pictures..

viewtopic.php?f=11&t=8497&start=15

Hi Gents,

I have started this post because I had a lot of requests for an 48V solution on 2m.
As such I am posting the the design Nanko PA0V had on his web page.
Nanko has now moved to big tubes and removed this solid state amplifier from his web page.
(He used to operate 5 of these amps in a big EME amplifier).
But is still a very good project and it would be a pity if the information disappeared.
The amplifier and the boards have been named after Nanko as a credit of the work he has done.

The design is based on the BLF278 or MRF151G. Nanko even build a big EME PA with these modules.
I am led to believe that the MRF151G is identical to the BLF278.

Cheers
Ron
VK4DD

[VK4DD]

Hi Gents,

This is the follow up.
I am showing here some detailed pictures from Nanko PA0V on how to make the 4:1 balun.

5 of these boards with very good efficiency by Nanko.
See table below:

Amp-- Pin- - - - Ud-----Id-------Po------Eff-----Bias
1-----3,2 W-----50 V--7,5 A----285 W--76 %---200 mA
2-----3,2 W-----50 V--7,6 A----290 W--76 %---200 mA
3-----3,2 W-----50 V--7,6 A----295 W--77 %---200 mA
4-----3,2 W-----50 V--7,7 A----300 W--78 %---200 mA
5-----3,2 W-----50 V--7,6 A----290 W--76 %---200 mA

(note these measurement can vary per batch of transistors)

Instructions for Coax for 4:1 balun:
Semi rigid 50 Ohm. used 6mm minimum.
(normal UT141 is to thin and gets to warm, low impedance ==> high currents).

Cut the lenght to 145mm and strip 7.5 mm off at both ends.
Make one up and check how it fits on the PCB before you proceed and cut the next one.

The two shield can be soldered together along the full lenght of the balun.
The shield will also be soldered to the PCB track on the output.

Make the other connections as shown in the picture/diagram.

It might look complicated but honestly it is a very simple balun.
Have a look at the detailed pictures and you will work it out.

73's
Ron
VK4DD

[VK4DD]

Hi Gents,

Here is the part list of this simple BLF278 amp.

Firstly you need an NXP BLF278

INPUT PARTS

This is the original design from PA0V list for the input.

Part list.. part 1 (input)

Changes made by VK4DD: for C1 and C2 15 pF ATC 100B, for C3 150pF ATC 100B for C5 20pF Trimmer foil, R3 use 500 Ohm 10 turns Bourns



OUTPUT PARTS

This is the original design from PA0V for the output

output parts.JPG (11.1 KiB) Viewed 22348 times

VK4DD uses for R2 510 Ohm/5W (connect to 48V), C1 replace mica compression trimmer by 2 x 39pF ATC parallel
(received feedback from Nanko PA0V that this is probably a good choice as he has burnt out a few compression trimmers)

Coax for 4:1 balun:
Semi rigid 50 Ohm. Diameter used 6mm minimum.
Do not use UT141-50 Ohm. This is too thin and gets to hot.

Alternative solution VK4DD for coax: (If you can't buy 6mm semi rigid telfon coax)
viewtopic.php?f=11&t=8497&start=9

Source: (old home page Nanko PA0V)

73
Ron
VK4DD

[VK4DD]

Hi Gents,

Here are some more detailed pictures from the input and output circuits.
This will make it easy to copy the boards.

Cheers
Ron
VK4DD

[VK4DD]

Hi Gents,

The circuit boards currently in production.
This saves a lot of work and to make your project look good

Be very careful not to over drive these Fets as they don't like that.
I would also recommend to install a protection circuit to switch of the supply and bias if the amp sees a bad VSWR load.
This amplifier needs very little drive about 3.2 W per board.
This means if you have a 100W rig you will need to make an attenuator of some sort to reduce the drive level !!
Don't rely on the internal power control, beccause with most rigs it is know to give spikes up to 100W on the output (IC910 etc).
No need to mention that over driving the gate means a dead FET. Same for no antenna connected or a short.
VSWR 20:1 means either 20 x 50 Ohm = 1000 Ohm or 50/20= 2.5 Ohm. Of course complex impedance's with 20:1 VSWR also included.

The PCB has got a few extras like room for fuse holders in the 48V circuit.
A fuse is a handy thing to have. But the status has to be monitored.

Cheers
Ron
VK4DD

[VK4DD]

Hi Gents,

I have just downloaded the data sheet of the MRF151G. It shows almost the same output circuit with a 4:1 coax transformer.
They specify semi rigid Co-aX 16-18 Ohms 62-90 Mils OD.
So I need to look that up how thick that is. But I agree with who ever made this schematic that it makes sense to use the lower impedance coax
because it means a fat center conductor. The current trough this coax are high because on the Fet side the impedance is a couple of Ohms.
High current/Low voltages. As such you are better of with a bigger conductor here. Running the coax a lot cooler.
The insulation material needs to be Teflon because it can handle high temperatures so much better.

At the end of the day you don't have to worry if the cable is 50, 35, 25, 16 or 18 Ohms.
Important is theat the center conductor is thick enough to handle the current and a it has Teflon isolation.
It will still be a 4:1 impedance transformer regardless which coax impedance is used.

Update 5 june 2009 calculating mean Z

However.... Fred N1DPM emails me that...

The closer to the mean Z the broader the bandwidth and the better the actual match as the losses will be minimized (VSWR losses) as it's the best Z match you can get.
6.25 Ohm for each output of the fet, as such 12.5 ohms in total because of the push pull (symetrical) circuit putting the Z of each 1/2 of the part in series with the other half.
The thing is....I have to believe the output z is less than 6.25 ohms per side. I'm guessing it is more like 4.5 ohms or so total as the part draws 10 amps at 48 volts (easy estimations). I haven't looked at the datasheet. This is where the LC circuit comes to play as there is a 1.4 VSWR from 4.5 to 6.25 ohms (roughly) and because it is such a low Z the line loss because of current is high. Again more power less heat. This is part of the finer point of matching. Close works good and tweak from there as there are many outside factors that are tough to model / calculate.

Fred N1DPM

Calculating mean Z:
This means 10 Ohm (2 x output impedance BLF278) * 50 Ohm = 500
SQRT(500)= 22.4 Ohm ideally...

or if you assume the impedance of the fet is 2 x 6 Ohm.
Than the mean Z is 25 Ohm.
No need to be too fuzzy the mean Z is some where between 16 and 25 Ohm.

Other than that... the bandwidth does not have to maximized for this particular application...
so more than one solution is possible. The amp will work.

I will get back to the coax......

See also
Attachment MRF151G and study the coax there ...

73
Ron
VK4DD

[VK4DD]

Hi Gents,

The new BLF278 board has arrived and it came out real nice.
I also received the Mark II boards and the new board for an HF amp.

I will add this board shortly to my web shop.
http://www.vk4dd.com

Isn't this a pretty little board. SWR detector build into the PCB.

Cheers
Ron
VK4DD

[VK4DD]

Hi Gents,

This is probably the most economic way to make power. As the BLF278 can be purchased as low as US$ 100 to US$ 105.
Input and output are 50 Ohm. There is a coupler integrated in the PCB.
For the input no ATC capacitors are used, as the drive level is only a few watts.
With also very little ATC caps on the output this is a cheap solution to make 300W on 2m.
The other components are probably no more than US30 all up.

It is compact, cheap. It is not that dificult to build and output tuning is easy. Just one variable capacitor on the output.
Supply voltage is 48V DC. In some cases this is easier to obtain than 28Vdc.

There is room for a fuse holder on the board. So it is a pretty complete concept.

The cheapest way to combine two boards is with 75 Ohm coax. Just make a Wilkinson combiner out of coax.

All in all a big bang for the buck.

73 Ron
VK4DD

[VK4DD]

Hi Gents,

I just had a look on Ebay... and it seems that the BLF278 are now selling for US$ 85.
Or if you buy 20 you pay US $ 60 each ...
And there is only 1 ATC ... of 1 NF in this project. (yes only 1).

73
Ron
VK4DD

[VK4DD]

Hi Gents,

Ok well here is my solution for the 4:1 coax on the output circuit.

(see also calculating mean Z from previous posting....)

Phillipe F1HQM about coax. Can I use RG142?
While RG142 is specified to handle 2400W on 100 Mhz it is certainly capable of handling a lot of power on 2m.
However the cable is not a semi rigid, and as such you will have trouble soldering the shield to the PCB and to the other cables.
The second problem is that you will get an air gap between the shield and the insulation.
As such the whole coax is sort of distorted.... I would not recommend to go that way.

If you don't have access to big flea markets where people sell the coax Nanko was showing, than I suggest to use my idea.

My idea is to use parallel 2 pieces of Sucoflex SM-141-35. That is 3.5mm semi rigid (hand formable) with an impedance of 35 Ohm.
This is actually quite ideal, as parallel it becomes an 17.5 Ohm coax.

On top of that layer as the second layer use a single layer of this same coax.

In the first layer the currents will be much higher as the impedance's are very low.
17.5 Ohm = 2 x 35 Ohm parallel is a perfect impedance to have here
(check the schematic of see one of the previous postings MRF151G data sheet).
The second layer the impedance are much higher.
As such the voltage will be higher and the currents lower.
Lower current means less heat. So a single layer is enough.
Other than that, I think 35 Ohm is a real good impedance to have here

If you buy 1 meter of this coax you will be able to get at least 9 pieces out.
Which means it is enough to build 2 of these PA modules.

I actually think this is quite a good solution.
I tried bending a few coaxes and it fits really nice on the PCB.
Will post some pictures later....

More to come on the coax... keep on reading...

73 Ron VK4DD

[VK4DD]

Hi All,

Nanko PA0V emailed me the other day a picture from a Nokia surplus power meter he had for sale... (about 2kg)
Shipping to VK made it too expensive.

I really liked the Meter it self... so I asked him if he could scan the scale and email it to me.
I just happened to have an identical instrument, but with a voltmeter scale. So all I need to do is print it with my DYMO label printer and I have an adhesive scale
Stick on top of the existing scale, cut of the edges with a sharp knife.
And there is my cheap power meter... with Power and SWR scale.

If you want a 600W scale, you can put a multiplier above the scale or edit the number in paint....

Alternatives...

An other option is to download software from the internet. And design your own scale.
I have downloaded this software, played hours with it, but never been successful making a good scale that fitted my meter with this software.

Cheers
Ron
VK4DD

[VK4DD]

Hi Gents,

On the input I am using a coax 316 (teflon) it is 2.6 mm thick.
Because the input is low power, this material is ideal for the balanced to unbalanced balun.

The length for coax can be calculated with this formula.

F = Freq in Mhz
L= length in Meter
V=velocity factor (for Telfon coax 0.7)

A 1/4 wave length is

F = 144 Mhz
Teflon V= 0.7 (=70% velocity factor)

L = V x 300/(4 x F)
L = 0.7 x 300/(4 x144) = 0.365 meter. or 36.5 cm.

This is not the coax for the output, this is the little coax rolled up on the input circuit.

73 Ron
Vk4DD

[VK4DD]

Hi All,

I hoped to build this amp in one day..., but I got stuck when I run out of parts....
I decided that the amp was going to be for SSB only, as such I think it should be possible to mount the BLF278 direct on the aluminum.
The BLF278 is cheaper, so I am willing to take a little more risk in that perspective...

The base of the heat sink is almost 10mm thick and that helps also to spread out the heat.

Any way.. I you want to build it in a day, you don't want to start drilling and mounting copper etc.

I also used some ATC's on the input, which is not mandatory. But I had them in stock, so I decided to use them.
They are nice and small and I like to work with them.

I have used 35 Ohm coax for the output circuit.
The first layer has 2 x 35 Ohm coax parallel resulting in quite a bit of copper for the center conductor.. (that is the main thing).

I did some calculations and compared the center conductor of a 50 Ohm, 6mm teflon coax with the 35 Ohm center conductor.
And believe it or not, there is only 0.2 mm difference in thickness between these 2 cables.
However 2 x 35 Ohm center conductors has 75% more copper compared with a 6mm 50 Ohm center conductor.
So I believe that 35 Ohm solution will run cooler than the 6mm 50 Ohm coax.

The second layer is only 1 x 35 Ohm. That is because the impedance is higher and as such lower currents trough this piece of coax.

First I bended the coax in shape and stripped and bend the leads in position. Than I soldered the first two 35 Ohm coaxes (which are parallel) together from below.
I bended them again into shape. I tinned the PCB. And than soldered them to the board using two solder irons. I think you need at least one 80W Iron and a 40W iron.
Any thing less does not have enough heat. Soldering bit by bit, making sure it stays clear from the gnd.
I would not say that soldering these coaxes is an easy job.... there is not much room, but if you don't make a mistake, it will fit just fine.
The third coax on top was easy to solder. Connect the leads as shown in the picture.
All the shields are interconnected and soldered to the big "U" on the circuit board.

Any way have a look at the pictures...how it all worked out.

A few words about the coax...
Soldering the 2 x 35 Ohm cables is possible, but it is a bit of a challenge. If you can get the thick 6mm cable, go for it.
That would be easy to solder.

After soldering the coax, clean up the solder flux with alcohol or other special flux cleaner
(I used nail remover, very agressive and not optimal... but make sure it does not solve the green coating of the PCB).

But at the end of the day, I think it is going very well
It all seems to fit, and need to order some bits on Monday.
I need to chase these parts up in VK. No idea were I can get a big compression trimmer...

73 Ron
VK4DD

[VK4DD]

Hi Gents,

I received some questions about which heat sink I am using...

I buy my heat sink mostly from Ebay.
Seller: barrredboss

5 3/8 WIDE
.375" BASE THICKNESS
1" FIN HEIGHT
21 TOTAL FINS [VERY DENSE]

I ask barrredboss to cut this heat sink for me ==> 10 inch long.

This is how it looks on his Ebay site...

You will also need a 120mm FAN to produce air across the heat sink.
This will give you this nice low heat resistance you are after.

73 Ron
VK4DD

[VK4DD]

Hi Gents,

Pino IK0SMG emailed me an interesting schematic he uses for the protection. It uses a thyristor for the protection.
I would like to use this idea to change the cicuit from W6PQL because I prefer to have a reset when a fault occurs.

The circuit I got from Pino uses a PNP transistor Q1 to interrupt the 12V to the coax relays.
Q2 activates the PTT and energizes the coax relays.
Different jumper settings for JP1 and JP2 are possible depending on how you activate the PTT circuit. (to GND or not)

The rectified voltage from the coupler should be just high enough in a fault situation to trip the Thyristor (SCR). One the Thyristor is tripped (actived) it will start to conduct and the LED (DL2 ALARM) will come up.
Coax relays and Transistor Q2 will be cut off from the +12V supply because Q1 has stopped conducting.
The coax relays will no longer be energized. The amp is tripped.

RV2 sets the level at which the Thyristor (SCR) trips. Some damping capacitors are on the gate of SCR to reduce false trips.
(note that some transmitters have very bad function of ALC and that they reduce very large power spikes when the power is reduced to several watts, the capacitor should not be increased
for this sort situations, because it makes the circuit too insensitive).

I think that Q1 should be an NPN, but that is probably because the drawing needs to be updated.

The standby/reset switch will have to be opened so that the Thyristor will no longer carry current.
It will stop conducting and the circuit is reset.

Ideas for further improvements by VK4DD

I think this is a good circuit, but I am going to make a few changes to it....
I would like to use a sold state relay or big FET to replace the function of RLY1.
This would cut of the current to the amplifier much quicker, which will allow the fet to live longer.
I would also like to double the trip circuit so that an input over drive circuit can be added as well.

Thank you very much Pino for emailing me this very good and simple Amplifier services schematic.
I am sure we are going to see more of Pino's work.

73's
Ron
VK4DD

Continue to next page

viewtopic.php?f=11&t=8497&start=15

.

[VK4DD]

Hi Gents,

I got my BLF278 going 300W on 144 Mhz with 4.2W drive, but I made a few changes in the design, which I have documented in previous postings.
This is because I don't want to double up postings and create confusion. (I will write on a posting, which date it was updated...)

I drove the amp with 4.2W because that was full power from my little Icom hand held radio.
Half power gave from my HT is 1.8W. That was good for 240W out. The input VSWR is excellent, but I made some changes to get it down further.
The trimmer PA0V specified was to big. I found that the optimum was with the plates fully open (minimum capacity).
As such I changed the trimmer for a smaller one from only 20pF. This trimmer is now nearly at minimum capcity.
My guesstimate is that this capacity of is now in the order of 5 to 7pF. As such the 20pF trimmer is a better choice.

I also noticed that PA0V had a small capacitor in series with the coax. I think that was done to improve the VSWR, but there was no need for it in my case.
Perhaps because I used different and ATC caps on the input? perhaps because this is a different PCB?
When I drove the amp with 4.2W I could only just see the meter moving when I measured the reverse power.
As such reverse power is 30mw ? 50mW? too small to measure with a 10W bird slug.
This means that the return loss is about 20dB I did not even bother to further improve this.

Output circuit...

I had only a 4-50pF compression trimmer, so I decided to use it and have a 47pF ATC parallel.
The power was less at minimum capacity so I changed the 47pF ATC for a 68pF ATC and started tuning the compression trimmer again.
I could not find a sharp dip of peak... so I increased it to 100pF parallel on my compression trimmer.
That gave slightly less output, so I went back to just one ATC of 82pF, also tested 100pF, and even 200pF.

Best results: with 2 x 39pF ATC on the output.
One ATC can only just handle the 300W. But the 2 x 39pF is a better choice as it will also be capable of driving a VSWR of 2:1 load.

Efficiency....
I got 9.8 Amp x 48V = 470W input, output was 300W on the bird.
300/470= 64% efficiency (bias was set to 0.2Amp each fet).
I think that is spot of what can be expected for a BLF278 on 144 Mhz.
My bird meter should be spot on, and my amp meter is a digital meter from Fluke.

Bias current

Firstly make sure the trimmer resistor is not set on maximum voltage to the BLF278...... that will give you 10Amp idle curerent or more...
I accidentally did this It is suggested to avoid this. I recommend to do first bias alignment with 3Amp fast fuse for this reason....

Class AB is recommended for SSB use:

For 144 Mhz you need to measure 0.9A. That means each fet gets 450mA (2 fets in one transistor). I suggest to setup your bias on a warm heat sink.

If you going to use it at WSJT/FM just set it at 0.4 Amp total (2 x 0.2A). that is enough.

It is clear that the lower the bias current, the higher the efficiency.
It is also clear that the higher the bias, the more linear the amp....
In SSB the amp needs to be linear so that is why the idle current is higher.

On the input I made a few minor changes, which I have noted under the material list in bold letters.
You can try both, and choose what is best for you. Lets us know what you did and what the results are.
(I am very happy with this input VSWR).

Stability...
The amp appeared to be nice and stable, no funny things. Very good indeed.

Bias resistor...
I used a 510 Ohm/5W resistor and connected it to the 48V.
If you use 24V or 12V to switch your bias, just change the value.

But at the end of the day, I am a very happy chappy.
The BLF278 works like a rocket

300W out, for very little money, that is excellent.

More goodies to come... I will make a posting on how to make an Harmonic filter, how to combine two amps.
I will also be happy to answer specific questions. Please feel free to email me these.

73 Ron
VK4DD

See the action pictures...

[VK4DD]

Hi Gents,

I have updated the parts list. See posting 3.
I have also updated other pages, to make them more accurate..please read all postings again
I will attach a picture showing you how the compression trimmer is replaced by 2 x 39pF ATC.
This is not only a cheaper solution, it is a better solution also.
I have blown up many compression trimmers over the years, besides that, they can start to arc etc.
I just don't like these things in a final and will get rid of them if possible.

Attached is a picture showing you how I used these 3 islands (or dots on the PCB).
The ATC have a narrow and a wide side.
I soldered the ATC next to each other with the narrow side up.
This gives them more clearance from the gnd.

Here is more data on the amp.
400mW in gives 80W out.
1.8W in gives 240W out.
4.2W in (well into saturation level) gives 300W out.

HEAT SINK
No copper heat spreader was used. That seems to be fine for SSB work.
Modes like FM/WSJT will require a copper heat spreader. A fan will need to be used in combination with the heat sink to keep the amp cool.

35 Ohm SM141-35 coax.
After running it in the test, it was still only finger warm. Very good.

Gerd DJ5BV emailed me also that he is working on getting 35 Ohm real cheap.
But so far I am still waiting on him. I think he is planning to buy this stuff in bulk as other hams in Germany are also interested in it
Gerd told me he is using this 35 Ohm coax to match his DK7ZB 28 Ohm yagi. Any way it seems that the applications for the 35 Ohm coax have grown
steep after the BLF248 PA, the BLF278 PA. I hope to see one day 6mm 35 Ohm for sale by the meter as well ...

I noticed that http://www.rfmicrowave.it is also selling 25 Ohm coax. UT141-25.
This really close to the mean Z of 22.3 Ohm we calculated earlier.
The only disadvantage is price and the fact it is rigid. Semi rigid is ideal. But I am sure you can bend it...
It is 17 Euro per meter, but 1 meter is enough to build 3 amps... so that is less than 6 euro coax per amp.
This would also be the guy where you can purchase ATC's etc.
(apart from me off course... I got about 1500 ATC's size 100B, but not the complete range...., that is probably enough for my own experimental needs )

That is probably the easy way to build this amp. Only one coax of 25 Ohm on the bottom, and one coax of 25 Ohm on the top.
1 meter should be enough to build 3 of these amps.
That would make soldering the 4:1 balun into a piece of cake

73
Ron
VK4DD

[VK4DD]

Hi Gents,

I have experimented a little bit more...
with a 39pF and a 27pf parallel I get 320W out at 10.5 Amps/48V.
That means the efficiency is 63.5 %
a little bit more, power efficiency only 0.5% less...

I don't recommend to get more power out of your BLF278, I think the 300W is already enough.

Next step would be to build an harmonic filter.

Well here is one..

http://www.dj9kw.de/dj9kw/projekte/afu/ ... owpass.htm

if you find etching difficult, or you have no access to pcb material than here is an other method to make the capacitors...
UT-141 coax has a capacity of approx 1pF per cm.
So 34pF is 34 cm.
It is best is you cut this into 2 pieces of 17 cm. Than connect the begin with the end so the coax is a loop.
Don't cut it to short try first with a longer piece and make sure you end up with the correct capacity.

You can solder the coax direct to blank (not etched) piece of FR4 PCB.

It is that easy.

Cheers
Ron
VK4DD





73 Ron
VK4DD

[VK5AYD]

Good evening.....
I am looking at buying one of the devices from simpli_city212 on ebay.....am wondering if anyone has had problems with this supplier before I make a move......
Thanks for any comment......

[VK4DD]

Shipping with that guy is US$ 35... so your BLF278 will cost US$110 landed. No experience with this seller.
Suggest to contact VK3KAI, he bought 2 x BLF278 for this project via Ebay for US$85 each.
2 x BLF78 is a very conservative way to make 400W (legal limit VK) that would not stress the devices very much at all.

73 Ron
VK4DD