[Amps] Stripline design

Thu Jan 5 16:06:27 PST 2012

I've had a casual interest in how striplines work for a while, but not to the point that I have started a detailed study.  You've provied a great explanation.  From what I  can gather, the "stripline" is just being used in the same way as for example a coaxial transformer.  The difference is "we" are creating the coax out of the equipment case and the bar connected to the tube anode.  Based on the bar size and the case size, the impedance is determined.   This seems very similar in idea to calculating impedance of a coax cable by using the 138log(D/d) equation.  If this is roughly true, then increasing the strip size, thereby reducing the distance between the strip and the case will make the impedance lower, this would be the case of extending the strip on one end.

I also think you have hit the nail on the head.  the 2x3cx800 k1fo amp uses the same enclosuer so the stripline sides are closer to the case sides, I presume this would make less inductance and make it shorter to maintain the right inductance.  While the single 3cx800 is narrow so it must compensate for being longer.

This might seem like an off the wall idea, but perhaps the simpleton approach is to calculate the original area of 3.5" x 8.625" = 30.1875.  Then take the new length of 3.5 x 9.0 = 31.5, an increase of 1.3125".  To maintain the same area, reduce the width by 0.07292"  on each side so the new width is 3.35416".   We should have the same impedance because the same insulative area is maintained.

Your first email mentioned that 

"The capaitance reactance of the tube and strays is given by

Xc=1/(2 Pi f C)"

Where does the "C" paramater come from?  Is this one of the paramaters on the tube data sheet such as Cin, Cout, Cpk, Chtr.  I would guess Cout.   
3cx400A7/3cx800A7
Cin   = 20.5/26.0
Cout = 6.0/6.1
Cpk = 0.03/0.05

Cout between the tubes is only 0.1 pF difference, so this seems very small.  

My thought would be to leave the stripline at the same height as original, this makes the tube anode connector stand up higher then the original, but there is enough room on the top for that.

I do have a grid dip meter I could use, I'm not sure how it would help since the stripline likely has to be enclosed with the lid on to give the proper results.  How would I couple it to the enclosed stripline?  I would guess I would leave the tube out of circuit to use the dip meter.  Would it not be good enough to somehow connect a probe and simply measure the impedance perhaps with an antenna analyzer or with a VNA (i dont have a vna)?  

I have not found any construction articles on this RF deck so far, I bought it off of Bill Olson K1DY half a year ago.  He said it had been sitting collecting dust for 20 years and he built it 35+ years ago.  I did some clean up and traced out the circuit, threw in some control, metering, and a power supply and away it goes.

I will take it apart this weekend and get some of the dimensions as best I can, it can never hurt to have those laying around.  

Thanks again,
Paul 

----- Original Message -----
From: Dr. David Kirkby <david.kirkby at onetel.net>
To: Paul Decker <kg7hf at comcast.net>
Cc: amps at contesting.com
Sent: Thu, 05 Jan 2012 22:31:50 -0000 (UTC)
Subject: Re: [Amps] Stripline design

On 01/ 5/12 09:38 PM, Paul Decker wrote:
> Hi David,
> Thanks for the reply and the great information.  I can put together some specifics and more pictuers of the inside of the deck, and using a copper foil wouldbe an easy way to test it.  The 3cx800 tube height is taller than the 3cx400 so this might play a role.  Perhaps the stripline size difference in the k1fo's amp is due to the cabinet size, he does use a larger rf chamber than what I have.

The case has quite a major effect if it's distance is not much larger than the 
spacing between the chassis and the stripline. I'd say if the distance between 
teh stripline and the side walls of the case are less than 2-3 times the height 
of the stripline, then the walls will have an appreciable effect on the 
impedance of the line.

We need to know where the DC voltage is fed, and where the RF 
coupling/decoupling capacitors are. Scan the original publications of the projects.

TXLine is a free program to compute the impedance of transmission lines. It 
works with microstrip line, which is your stripline, with Er=1.0. But if the 
chassis walls are close, then TXLine will be inaccurate. ATLC could be your 
friend, as that takes the presence of the chassis, top panel and the two side 
panels, but it's a Unix program, which you may or may not be happy with.

ATLC can't take the end panels into account, as its a 2D simulator, not a 3D 
one. A true 3D simulator is difficult to use and expensive.

Another modeling option might be FEKO lite.

If you can send some detailed diagrams, I could be tempted to try to model this 
in HFSS

http://www.ansoft.com/products/hf/hfss/

using it's Eignemode solver. I've never used that feature, but I reckon that 
might actually be useful.

If I were in your shoes, I'd not worry about the extra length. It will effect 
the system, but in a way which you can easily compensate for by changing the 
width of the line, so from a practical point of view, it's not a major headache.

The theory is a bit more complex, but I've given you some of it. Depending on 
your background, you might prefer to ignore that or research the topic in more 
depth.

> Just a quick question, what is a "GDO" you mention near the end of your message?

Grid dip oscillator.

http://en.wikipedia.org/wiki/Grid_dip_oscillator

> Get some rest, 73,
> Paul

-- 
A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?