Return to KLUBNL.PL main page

rsgb_lf_group
[Top] [All Lists]

Re: LF: 137.5kHz WSPR Tuesday evening - reports appreciated

To: [email protected]
Subject: Re: LF: 137.5kHz WSPR Tuesday evening - reports appreciated
From: Stefan Schäfer <[email protected]>
Date: Tue, 07 Sep 2010 22:25:27 +0200
In-reply-to: <[email protected]>
References: <[email protected]> <[email protected]> <[email protected]> <[email protected]>
Reply-to: [email protected]
Sender: [email protected]
User-agent: Mozilla/5.0 (Windows; U; Windows NT 6.1; de; rv:1.9.1.11) Gecko/20100711 Thunderbird/3.0.6
Hi Roger,

40 Vpp is just normal if the power supply has about 12V! You can assume about 3,5...4 times the supply voltage for u(DS), if well matched!
Most programs will give you just an imagination on how the values must be, since you oftendon't know the Q of the coils and Cs. You always have to tune the PA for optimal efficiency, not for maximum output power! Thus it is essential to match the antenna to 50 Ohm, if this is your designed output impedance, not to a maximum antenna current. A square wave gate voltage is essential too. Use a tiny MOSFET driver such as ICL7667. And use impulse Cs, no ceramic Cs! I am using WIMA FKP-1 2kV types (not sure if available in UK. There are surely others such as the popular "silver mica")
I do not have a ready to use design for 12V/20W (do you need one for 1 kW?), just some recommendations. Instead of a IRF510 i would recommend a IRF540N. If your antenna matching is reasonably OK you can choose e.g. a IRFZ48N. These FETs have a smaller ON resistance and will produce lower losses. Switching times are uncritical at 137 kHz. A TO220 device is totally OK for 20W. In my Class E PAs i am using no output transformer at all! There is a L between Drain and +, a C parallel to D-S and a L at Drain which is in series to a C is connected to ground. Between L and C you will find 50 Ohm when correctly dimensioned. Then you have to decouple the DC part e.g. by a 1 uF series C and pass the signal through a standard 5 pole low passfilter (12,2nF/53µH/24,7nF/53µH/12,2nF). That's all. You should get an efficiency of at least 90 % !

Best was to optimise the PA is to watch the U(DS) wave form. You will find infos how it should look. Ah, it should look as on the homepage of http://classe.monkeypuppet.com/ :-)

Wish you good luck und fun in optimising!

73, Stefan

PS: It is helpful to create an excel file where you type input voltage and input current and output power and let you calculate the efficiency to see if it has increased or not. You can start with an adjustable voltage source if the driver has its own supply! :-)

Am 07.09.2010 21:08, schrieb Roger Lapthorn:
Stefan/Mal,

Clearly my design was not working in class E properly as the PA was dissipating a fair amount of heat and it was taking about 0.8A to produce 40V p-p output from a 13.8V supply which is about half as efficient as it should be. I took a PA design start from a design program I found on the net by Tonne Software called "Class E" and the values seemed to be close to optimum when I adjusted caps and L up and down a small amount, but the dissipation seems miles off what a good class E design should take. The IRF510 gate voltage is set just below 3V with a resistive pot. The drive is not a square wave though, so I suppose this could be part of the issue?

Stefan, if you have a better class E PA circuit for 137kHz suitable for around the 10-20W out region I'd be interested in seeing it. I would prefer to use toroid inductors because of size. I have several T106-2 cores to hand as well as several 16mm diameter 3C90 cores.

I'm still learning..... and not too proud to ask for help from those with more experience.

73s
Roger G3XBM





2010/9/7 Stefan Schäfer <[email protected]>
Hello Roger,

Maybe you can send us a circuit with its values. I have built class E PA for 2200/160/80/40/30/17/10m and have some experience. If you are using a IRF510 and stay in the QRP range (< 10 W) at 137 in a class E stage you wouldn't even need a heat sink at all if well dimensioned!

See the infos at http://www.classeradio.com/  This is most useful if you are starting with that design. On 137 kHz problems are much smaller than on HF :-)

Best 73, Stefan/DK7FC

Am 07.09.2010 20:17, schrieb Roger Lapthorn:
Just hit a bit of a snag and blew up my (last) IRF510 PA because of totally inadequate heatsink. So, I'm on 137.5kHz WSPR but only with about 1W from the simpler PA at present, so ERP low uWs.  I need to get some heatsink and more FETs!

Any reports (now less likely) still appreciated.

[Mal - it started out as a Class E design attempt but hadn't been optimised, so I guess this counts as class D. ]

73s
Roger G3XBM



On 7 September 2010 17:34, Roger Lapthorn <[email protected]> wrote:
My 137.5kHz WSPR TX transverter is now working fine, so I'll be on-air with WSPR tonight (Sept 7th) from 5.30pm - 11pm UK time.

After my success on QRSS3 with G3XIZ I'm hopeful he'll now decode my WSPR signal tonight, assuming Chris can take a listen. If anyone else within, say, 100km of Burwell, Cambs JO02dg cares to take a look for the WSPR beacon (running around 35% TX time) I'd be most grateful.

Not sure if anyone else has had this problem, but I was using the small 16mm 3C90 cores (cream colour) as the PA bifilar output transformer and I had all sorts of issues with the darn thing overheating and even one cracking. I replaced this with a choke wound on a T106-2 core and made the output network work in class D/E and now everything is working very predictably. The larger diameter 3C90 cores seem fine in the step-down transformer matching the "in the air" loop.

The WSPR beacon is now going ON.....

73s
Roger G3XBM


--
http://g3xbm-qrp.blogspot.com/
http://www.g3xbm.co.uk
http://www.youtube.com/user/g3xbm
G3XBM   GQRP 1678    ISWL G11088



--
http://g3xbm-qrp.blogspot.com/
http://www.g3xbm.co.uk
http://www.youtube.com/user/g3xbm
G3XBM   GQRP 1678    ISWL G11088



--
http://g3xbm-qrp.blogspot.com/
http://www.g3xbm.co.uk
http://www.youtube.com/user/g3xbm
G3XBM   GQRP 1678    ISWL G11088
<Prev in Thread] Current Thread [Next in Thread>