Broadband impedance matching of series RL load

Am 14.03.2018 um 07:39 schrieb snipped-for-privacy@gmail.com:

That is very simple. - Branch off the load, calculate the impedance looking in the output, calculate the absolute value of the complex impedance and make the load that value.

Leo

I am afraid I do not have a visible schematic. I use HSpice at work, and Ngspice at home, both of which use the text input. It is sometimes bothersome to enter a new large netlist, but editing is very easy.

As an addendum to my previous message, I am trying to come up with a generalized scheme to tackle one-end reflections. I have drawn ideas from published literature by Randy Rhea. I have successfully tackled the parallel/series RC case, and am trying to tackle parallel/series RL loads. The basic idea is that based on energy conservation, the total energy entering the matching network must be equal(within some applicable tolerances) to the energy delivered to the load. Therefore I want to measure the RMS input/output power.

I am sorry about the typo. On the standard QWERTY keyboard, the "n" and "b" keys are adjacent to eachother. I totally agree about energy conservation and all that.

Using similar arguments as Clifford Heath, I want to have a general scheme to tackle one-end reflections for the case of parallel/series RL loads _ jabe already tackled parallel/series RC loads. Specifically, as the load and source resistances are identical in value, in order that the energy entering the matching network must equal(within same applicable tolerances) to the energy delivered to the load, bearing in mind that inductive reactance is directly proportional to the frequency, that is the load inductor's effect must be nullified. I have drawn some ideas from published literature by Randy Rhea.

I am not forcing anything on anyone. Please read my previous response to your previous message.

Both HSpice and Ngspice allow comma/tab separated (csv/tsv) numeric text output that can be read in by C programs, or plotted with gnuplot or Excel.

Just a sketch would help. I don't know what you are trying to do.

LT Spice lets you draw a schematic instantly. Typing (and reading!) netlists is really old-fashioned.

NGspice, HSpice noted. LTspice requires all kinds of extra steps to get to a numerically analyzable set of data points

Unless HSpice has been recently neutered, I'd hazard a guess that it's .PRINT statement will produce a table in the .OUT file that can pasted into Excel or a curve-fitting program (which is what I mostly use it for... for behavioral modeling ;-)

Then I missed that post... perhaps during my surgery yesterday?

I'd like an answer to "What does "wideband" mean? fL, fH, dB's?"

If your desired LP corner is > 2*pi*Rload/Lload you've got problems, if < 2*pi*Rload/Lload, piece-a-cake.

A definitive statement of the problem will produce definitive answers, otherwise blather and bloviations >:-} ...Jim Thompson

If you [think you]'ve solved it for one case, then invert R --> 1/R, C -->

Tim

So, let us start with the LP corner < 2.0*PI*RLoad*Lload,

What would be the solution ?

The setup is simple : A signal source feeds a low pass ladder matching network, e.g., of order 6. The source impedance is purely resistive (NO source reactance). The load is a resistor in series with an inductor. The load and source resistors have the same value So how do I compensate for the frequency dependent load inductor ?

I am afraid your suggested solution is very vague. Suppose the load inductor is 1 microHenry. Inverting this, the corresponding capacitor would have a value if 1 Mega Farad -- not physically realizable. Are you talking in terms of reactance in Ohms ? If so, what would be the frequency ?

Filters are designed at unity frequency. I can't write lowercase omega, so I said w=1 which looks similar. Use the same transformations as from any filter table to get Zo and Fo.

Tim

I think the answer is trivial, but if you refuse to show a sketch, I refuse to talk about it.

I'll post later today. Hope I got that right for frequency... bit foggy-brained after the surgery... it's _lower_ than the natural corner of Rload and LLoad ;-) ...Jim Thompson

There's a tradeoff between the log of the return loss and bandwidth--it's a theorem of Bode's, so of course it's called the Fano-Bode limit. See e.g.

Cheers

Phil Hobbs

Larkin's "gentleman/scholar" attitude speaks for itself. ...Jim Thompson

-- | James E.Thompson | mens | | Analog Innovations | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | STV, Queen Creek, AZ 85142 Skype: skypeanalog | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at

Understanding is a fountain of life to one who has it, But the instruction of fools is folly. Proverbs 16:22

My website is down for maintenance, so I've tried dropbox:

Hope that works!

Let me ponder... there might be an easy way to export a netlist from LTspice into NGspice.

If not, I believe I've seen original flavor PSpice Schematics available without the PSpice simulator. ...Jim Thompson