Reply by Fred Abse ●March 12, 20142014-03-12
On Wed, 05 Mar 2014 12:14:35 -0800, Fred Abse wrote:


> Run the simulation with the supplied plot file in place, and the answer
> will magically appear ;-)


Just noticed:

There's a mistake in the plot file, with respect to the output VSWR.

"(1+S22(v2))/(1-S22(v2))"

should be:

"(1+mag(S22(v2)))/(1-mag(S22(v2)))"

Sorry,my bad. I should have noticed, if I'd been awake!


Amended .plt file:

[AC Analysis]
{
   Npanes: 4
   {
      traces: 1 {3,0,"(1+mag(S22(v2)))/(1-mag(S22(v2)))"}
      X: ('G',2,10000,0,1e+009)
      Y[0]: (' ',2,0.96,0.08,1.84)
      Y[1]: ('m',0,-0.001,0.0002,0.001)
      Log: 1 0 0
      GridStyle: 1
      PltMag: 1
      Text: "" 1 (311018.192820938,1.90984126984127) ;Output VSWR 50 ohm
   },
   {
      traces: 1 {5,0,"1/RE(Yin(v2))"}
      X: ('G',2,10000,0,1e+009)
      Y[0]: ('_',1,1000,0,1e+009)
      Y[1]: ('m',1,-0.001,0.0002,0.001)
      Log: 1 1 0
      GridStyle: 1
      PltMag: 1
      Text: "" 1 (331739.531535957,2053525026.45715) ;Parallel input resistance
   },
   {
      traces: 1 {2,0,"IM((Yin(v2)))/2/pi/freq"}
      X: ('G',2,10000,0,1e+009)
      Y[0]: ('f',0,2.28e-013,3e-015,2.64e-013)
      Y[1]: ('m',1,-0.001,0.0002,0.001)
      Log: 1 1 0
      GridStyle: 1
      PltMag: 1
      Text: "" 1 (337484.183978161,2.66207094420198e-013) ;Parallel  input capacitance
   },
   {
      traces: 1 {524292,0,"V(out)"}
      X: ('G',2,10000,0,1e+009)
      Y[0]: (' ',0,0.0794328234724281,2,1)
      Y[1]: (' ',0,-240,30,90)
      Log: 1 2 0
      GridStyle: 1
      PltMag: 1
      PltPhi: 1 0
      Text: "" 1 (341818.444827778,1.14101582205948) ;Gain
   }
}

-- 
"Design is the reverse of analysis"
                   (R.D. Middlebrook)
Reply by Fred Abse ●March 5, 20142014-03-05
On Wed, 05 Mar 2014 10:49:33 -0600, amdx wrote:


> I finally got the file opened and run.
> I went to the LTspice forum for help.
> I see you ran it out to 1GHz, the design was for the 500kHz to 1700KHz,
> but author said the original was good to 10MHz. Your changes make the 
> -3db point about 25Mhz.


It's always a good idea to run beyond design limits. That's how you find
out about stability problems. Phase margin, and all that...

500kHz to 1700kHz is a bit restrictive. Most, even ancient, Q meters go to
50MHz. My Marconi goes to 300MHz. 


>    What is the info after [AC Analysis] and how do I use it.
> The file didn't run if I included it in the .asc file.


Plot definitions file .It goes in the same directory as the .asc, but with
a .plt extension.

If your simulation is "foo.asc", call the plot file "foo.plt". If it's in
the same directory, it will automatically open when you run the
simulation, and display the intended plots.



>   How do the changes alter the input impedance numbers.


Run the simulation with the supplied plot file in place, and the answer
will magically appear ;-)

-- 
"Design is the reverse of analysis"
                   (R.D. Middlebrook)
Reply by Fred Abse ●March 5, 20142014-03-05
On Tue, 04 Mar 2014 16:26:06 -0600, amdx wrote:


> I truncated the file at,
>   SYMATTR SpiceLine V=10 Irms=10.541 Rser=0.004 Lser=0 mfg="KEMET"
> pn="C0603C105K8PAC" type="X5R"
> I then added this as a Spice directive. Doesn't run.
> I know I need to install the SUBCKT for the BFT92. How do I do that?
> 
> 2N3904 - wipeout! I suspect that is not good, could you define it a little
> more?
>   If needed I can try to install the BFT92, but I don't know
> if after my mod for the 2N3904 what that may involve. What do I need to do
> to prevent it from oscillating with a new install.


Just open the supplied .asc file in LTspice, and plot using the supplied
.plt file. Nothing simpler.

RTFM.

You did turn off line wrapping, before copying the files, didn't you?
Split lines screw up LTspice.

The subcircuit is included in the .asc file. It should appear on the
schematic, and automatically be included.

The oscillation you got was due to parasitic L and C. Getting rid of the
"Christmas tree" of decoupling capacitors would be a good start. Bad
practice.

Wipeout - gain drops by 40dB at 500kHz. 2N3904 is not a suitable choice.


-- 
"Design is the reverse of analysis"
                   (R.D. Middlebrook)
Reply by Fred Abse ●March 5, 20142014-03-05
On Tue, 04 Mar 2014 16:51:55 -0600, amdx wrote:


> Added the subckt from filename to ends.
> I get the error,
> Unknown subcircuit called in:
> xq1 n008 n014 n007 bft92


Are you actually using LTspice, or some other Spice?

The subcircuit is in the .asc file, should appear as a directive on the
schematic, and automatically be included in the netlist.

-- 
"Design is the reverse of analysis"
                   (R.D. Middlebrook)
Reply by Fred Abse ●March 5, 20142014-03-05
On Tue, 04 Mar 2014 16:41:20 -0600, amdx wrote:


> I'm into the bipolar .models file, but your model is not nothing like 
> the files I see, I don't know how much of the text to add.


It isn't a .model, it's a .subckt. Different thing. It includes package
parasitics.

If you've copied the file *exactly*, it should just run.

-- 
"Design is the reverse of analysis"
                   (R.D. Middlebrook)
Reply by Fred Abse ●March 5, 20142014-03-05
On Wed, 05 Mar 2014 10:49:33 -0600, amdx wrote:


>    What is the info after [AC Analysis] and how do I use it.
> The file didn't run if I included it in the .asc file.


That's the plot file.

Save it in the same directory as the .asc file, with the same name as the
.asc file and a .plt extension.

It defines the plots.


-- 
"Design is the reverse of analysis"
                   (R.D. Middlebrook)
Reply by amdx ●March 5, 20142014-03-05
On 3/4/2014 4:51 PM, amdx wrote:

> On 3/4/2014 4:41 PM, amdx wrote:


I finally got the file opened and run.
I went to the LTspice forum for help.
I see you ran it out to 1GHz, the design was for the 500kHz to 1700KHz,
but author said the original was good to 10MHz. Your changes make the 
-3db point about 25Mhz.

   What is the info after [AC Analysis] and how do I use it.
The file didn't run if I included it in the .asc file.

  How do the changes alter the input impedance numbers.

        Thanks, Mikek
Reply by amdx ●March 4, 20142014-03-04
On 3/4/2014 4:41 PM, amdx wrote:

> On 3/4/2014 4:26 PM, amdx wrote:
>> On 3/4/2014 1:26 PM, Fred Abse wrote:
>>> On Sun, 02 Mar 2014 15:19:29 -0600, amdx wrote:



>>> I ran the simulation with your 2N3904 - wipeout!
>>>
>>
>> You'll need to walk me through this.
>> I truncated the file at,
>>   SYMATTR SpiceLine V=10 Irms=10.541 Rser=0.004 Lser=0 mfg="KEMET"
>> pn="C0603C105K8PAC" type="X5R"
>> I then added this as a Spice directive.
>> Doesn't run.
>> I know I need to install the SUBCKT for the BFT92.
>> How do I do that?
>>
>> 2N3904 - wipeout! I suspect that is not good, could you define it a
>> little more?
>>   If needed I can try to install the BFT92, but I don't know
>> if after my mod for the 2N3904 what that may involve. What do I need to
>> do to prevent it from oscillating with a new install.
>>                       Thanks,  Mikek
>
> I'm into the bipolar .models file, but your model is not nothing like
> the files I see, I don't know how much of the text to add.
>                                     Mikek


Added the subckt from filename to ends.
I get the error,
Unknown subcircuit called in:
xq1 n008 n014 n007 bft92
                       Mikek
Reply by amdx ●March 4, 20142014-03-04
On 3/4/2014 4:26 PM, amdx wrote:

> On 3/4/2014 1:26 PM, Fred Abse wrote:
>> On Sun, 02 Mar 2014 15:19:29 -0600, amdx wrote:
>>
>>> Want vary high input
>>> impedance,
>>
>> You'll get that all right :-)
>>
>> On a more serious note, it's the use of a source follower that screws
>> things. You'll note that I redesigned with a unity-gain "drain follower".
>> Source, or emitter followers can have unforeseen consequences at high
>> frequencies.
>>
>> Just rearrange the first stage, with an extra resistor, taking the output
>> from the drain. You don't have to make the second stage PNP, so long as
>> you don't mind the opposite output polarity.
>>
>> The rest is just a matter of trimming values for unity overall gain, and
>> half-rail bias.
>>
>> The LTspice schematic shows you how. Values may have to be adjusted in a
>> real-life circuit.
>>
>> I ran the simulation with your 2N3904 - wipeout!
>>
>
> You'll need to walk me through this.
> I truncated the file at,
>   SYMATTR SpiceLine V=10 Irms=10.541 Rser=0.004 Lser=0 mfg="KEMET"
> pn="C0603C105K8PAC" type="X5R"
> I then added this as a Spice directive.
> Doesn't run.
> I know I need to install the SUBCKT for the BFT92.
> How do I do that?
>
> 2N3904 - wipeout! I suspect that is not good, could you define it a
> little more?
>   If needed I can try to install the BFT92, but I don't know
> if after my mod for the 2N3904 what that may involve. What do I need to
> do to prevent it from oscillating with a new install.
>                       Thanks,  Mikek


I'm into the bipolar .models file, but your model is not nothing like 
the files I see, I don't know how much of the text to add.
                                    Mikek
Reply by amdx ●March 4, 20142014-03-04
On 3/4/2014 1:26 PM, Fred Abse wrote:

> On Sun, 02 Mar 2014 15:19:29 -0600, amdx wrote:
>
>> Want vary high input
>> impedance,
>
> You'll get that all right :-)
>
> On a more serious note, it's the use of a source follower that screws
> things. You'll note that I redesigned with a unity-gain "drain follower".
> Source, or emitter followers can have unforeseen consequences at high
> frequencies.
>
> Just rearrange the first stage, with an extra resistor, taking the output
> from the drain. You don't have to make the second stage PNP, so long as
> you don't mind the opposite output polarity.
>
> The rest is just a matter of trimming values for unity overall gain, and
> half-rail bias.
>
> The LTspice schematic shows you how. Values may have to be adjusted in a
> real-life circuit.
>
> I ran the simulation with your 2N3904 - wipeout!
>


You'll need to walk me through this.
I truncated the file at,
  SYMATTR SpiceLine V=10 Irms=10.541 Rser=0.004 Lser=0 mfg="KEMET" 
pn="C0603C105K8PAC" type="X5R"
I then added this as a Spice directive.
Doesn't run.
I know I need to install the SUBCKT for the BFT92.
How do I do that?

2N3904 - wipeout! I suspect that is not good, could you define it a 
little more?
  If needed I can try to install the BFT92, but I don't know
if after my mod for the 2N3904 what that may involve. What do I need to
do to prevent it from oscillating with a new install.
                      Thanks,  Mikek