class c rf amp with bfr93a

Fred Abse wrote:
On Sat, 28 Aug 2010 14:34:24 -0700, Fred Abse wrote:

Sunday's exercise might be to compare the (Philips) model's S-parameters
against Philips' data sheet values. Wanna bet they don't match?

Pretty close match, apart from S22, which doesn't show the pronounced kink
shown in the datasheet.

Shame LTSpice won't do polar plots. Since it does impedance and
S-params, it really ought to.

Found a way round that. Set LTSpice to do ASCII rawfiles, run, edit two or
three LT-specific lines out of the rawfile, and load it into Berkeley
Spice 3f4's Nutmeg, then plot polar and Smith from there.
Much easier than trying to get LTSpice netlists to run in 3f4.

I still think the facility should be in LTSpice, though.


I think Jim Thompson once mentioned a software that could render raw
data into all sorts of graphics including polar. Jim?

On Sun, 29 Aug 2010 13:51:14 -0700, Joerg <invalid_at_invalid.invalid
wrote:

Fred Abse wrote:
On Sat, 28 Aug 2010 14:34:24 -0700, Fred Abse wrote:

Sunday's exercise might be to compare the (Philips) model's S-parameters
against Philips' data sheet values. Wanna bet they don't match?
Pretty close match, apart from S22, which doesn't show the pronounced kink
shown in the datasheet.

Shame LTSpice won't do polar plots. Since it does impedance and
S-params, it really ought to.
Found a way round that. Set LTSpice to do ASCII rawfiles, run, edit two or
three LT-specific lines out of the rawfile, and load it into Berkeley
Spice 3f4's Nutmeg, then plot polar and Smith from there.
Much easier than trying to get LTSpice netlists to run in 3f4.

I still think the facility should be in LTSpice, though.

I think Jim Thompson once mentioned a software that could render raw
data into all sorts of graphics including polar. Jim?

Scratching head... I don't remember

Jim Thompson wrote:
On Sun, 29 Aug 2010 13:51:14 -0700, Joerg <invalid_at_invalid.invalid
wrote:

Fred Abse wrote:
On Sat, 28 Aug 2010 14:34:24 -0700, Fred Abse wrote:

Sunday's exercise might be to compare the (Philips) model's S-parameters
against Philips' data sheet values. Wanna bet they don't match?
Pretty close match, apart from S22, which doesn't show the pronounced kink
shown in the datasheet.

Shame LTSpice won't do polar plots. Since it does impedance and
S-params, it really ought to.
Found a way round that. Set LTSpice to do ASCII rawfiles, run, edit two or
three LT-specific lines out of the rawfile, and load it into Berkeley
Spice 3f4's Nutmeg, then plot polar and Smith from there.
Much easier than trying to get LTSpice netlists to run in 3f4.

I still think the facility should be in LTSpice, though.

I think Jim Thompson once mentioned a software that could render raw
data into all sorts of graphics including polar. Jim?

Scratching head... I don't remember


I believe it was a software that you could also throw scanned in plots.

On Sun, 29 Aug 2010 15:21:15 -0700, Joerg <invalid_at_invalid.invalid
wrote:

Jim Thompson wrote:
On Sun, 29 Aug 2010 13:51:14 -0700, Joerg <invalid_at_invalid.invalid
wrote:

Fred Abse wrote:
On Sat, 28 Aug 2010 14:34:24 -0700, Fred Abse wrote:

Sunday's exercise might be to compare the (Philips) model's S-parameters
against Philips' data sheet values. Wanna bet they don't match?
Pretty close match, apart from S22, which doesn't show the pronounced kink
shown in the datasheet.

Shame LTSpice won't do polar plots. Since it does impedance and
S-params, it really ought to.
Found a way round that. Set LTSpice to do ASCII rawfiles, run, edit two or
three LT-specific lines out of the rawfile, and load it into Berkeley
Spice 3f4's Nutmeg, then plot polar and Smith from there.
Much easier than trying to get LTSpice netlists to run in 3f4.

I still think the facility should be in LTSpice, though.

I think Jim Thompson once mentioned a software that could render raw
data into all sorts of graphics including polar. Jim?
Scratching head... I don't remember

I believe it was a software that you could also throw scanned in plots.

Aha! I did recently buy "GetData Graph Digitizer" which can create
numerical data points from graphs in data sheets.

Handy for creating subcircuits when all you have is a data sheet.

PSpice can do about any x-y plot you'd like, but I don't know how
you'd do polar except with some innovative macros

Jim Thompson wrote:
On Sun, 29 Aug 2010 15:21:15 -0700, Joerg <invalid_at_invalid.invalid
wrote:

Jim Thompson wrote:
On Sun, 29 Aug 2010 13:51:14 -0700, Joerg <invalid_at_invalid.invalid
wrote:

Fred Abse wrote:
On Sat, 28 Aug 2010 14:34:24 -0700, Fred Abse wrote:

Sunday's exercise might be to compare the (Philips) model's S-parameters
against Philips' data sheet values. Wanna bet they don't match?
Pretty close match, apart from S22, which doesn't show the pronounced kink
shown in the datasheet.

Shame LTSpice won't do polar plots. Since it does impedance and
S-params, it really ought to.
Found a way round that. Set LTSpice to do ASCII rawfiles, run, edit two or
three LT-specific lines out of the rawfile, and load it into Berkeley
Spice 3f4's Nutmeg, then plot polar and Smith from there.
Much easier than trying to get LTSpice netlists to run in 3f4.

I still think the facility should be in LTSpice, though.

I think Jim Thompson once mentioned a software that could render raw
data into all sorts of graphics including polar. Jim?
Scratching head... I don't remember

I believe it was a software that you could also throw scanned in plots.

Aha! I did recently buy "GetData Graph Digitizer" which can create
numerical data points from graphs in data sheets.

Handy for creating subcircuits when all you have is a data sheet.


Reverse engineering ... ?

PSpice can do about any x-y plot you'd like, but I don't know how
you'd do polar except with some innovative macros


I think LTSpice can output polar data but I have never done it. I
believe the command is "meascplxfmt polar".

Jim Thompson wrote:
On Sun, 29 Aug 2010 15:21:15 -0700, Joerg <invalid_at_invalid.invalid
wrote:

Jim Thompson wrote:
On Sun, 29 Aug 2010 13:51:14 -0700, Joerg <invalid_at_invalid.invalid
wrote:

Fred Abse wrote:
On Sat, 28 Aug 2010 14:34:24 -0700, Fred Abse wrote:

Sunday's exercise might be to compare the (Philips) model's S-parameters
against Philips' data sheet values. Wanna bet they don't match?
Pretty close match, apart from S22, which doesn't show the pronounced kink
shown in the datasheet.

Shame LTSpice won't do polar plots. Since it does impedance and
S-params, it really ought to.
Found a way round that. Set LTSpice to do ASCII rawfiles, run, edit two or
three LT-specific lines out of the rawfile, and load it into Berkeley
Spice 3f4's Nutmeg, then plot polar and Smith from there.
Much easier than trying to get LTSpice netlists to run in 3f4.

I still think the facility should be in LTSpice, though.

I think Jim Thompson once mentioned a software that could render raw
data into all sorts of graphics including polar. Jim?
Scratching head... I don't remember

I believe it was a software that you could also throw scanned in plots.

Aha! I did recently buy "GetData Graph Digitizer" which can create
numerical data points from graphs in data sheets.

Handy for creating subcircuits when all you have is a data sheet.


Reverse engineering ... ?


PSpice can do about any x-y plot you'd like, but I don't know how
you'd do polar except with some innovative macros


I think LTSpice can output polar data but I have never done it. I
believe the command is "meascplxfmt polar".

On Sun, 29 Aug 2010 16:41:26 -0700, Joerg wrote:
Jim Thompson wrote:
On Sun, 29 Aug 2010 15:21:15 -0700, Joerg <inva...@invalid.invalid
wrote:

Jim Thompson wrote:
On Sun, 29 Aug 2010 13:51:14 -0700, Joerg <inva...@invalid.invalid
wrote:

Fred Abse wrote:
On Sat, 28 Aug 2010 14:34:24 -0700, Fred Abse wrote:

Sunday's exercise might be to compare the (Philips) model's S-parameters
against Philips' data sheet values. Wanna bet they don't match?
Pretty close match, apart from S22, which doesn't show the pronounced kink
shown in the datasheet.

Shame LTSpice won't do polar plots. Since it does impedance and
S-params, it really ought to.
Found a way round that. Set LTSpice to do ASCII rawfiles, run, edit two or
three LT-specific lines out of the rawfile, and load it into Berkeley
Spice 3f4's Nutmeg, then plot polar and Smith from there.
Much easier than trying to get LTSpice netlists to run in 3f4.

I still think the facility should be in LTSpice, though.

I think Jim Thompson once mentioned a software that could render raw
data into all sorts of graphics including polar. Jim?
Scratching head... I don't remember :-(

I believe it was a software that you could also throw scanned in plots.

Aha!  I did recently buy "GetData Graph Digitizer" which can create
numerical data points from graphs in data sheets.

Handy for creating subcircuits when all you have is a data sheet.

Reverse engineering ... ?

PSpice can do about any x-y plot you'd like, but I don't know how
you'd do polar except with some innovative macros :-)

I think LTSpice can output polar data but I have never done it. I
believe the command is "meascplxfmt polar".

Can't find any reference. .meas statements usually start with an analysis
type, like ac, tran, etc.

Polar plots uploaded to a.b.s.e. Unfortunately my hand-added annotations
of frequency end points haven't survived the upload. Added in gimp,
they're present in the .pngs here.

--
"For a successful technology, reality must take precedence
over public relations, for nature cannot be fooled."
                                       (Richard Feynman)

On Sat, 28 Aug 2010 13:03:06 -0700, Joerg wrote:
So unless you I misunderstand you drive at 80mW to get 50mW out? I'd do
that with three resistors instead of a BFR93A ... <duck and run

Sorry, I wasn't clear. I didn't match the source. Input is 370 +j76 (I
overcompensated the input capacitance)  It actually takes about 11mW
drive Still goddamn poor at about 6dB power gain. It will do over 100mW
with more or less the same drive, just with a change in loading. Maybe
could squeeze it as far as 10dB, if I were interested enough. It's a
non-starter anyway, those capacitance values are silly.

I'll maybe have a play with simulating transmission line coupling,
(microstrip?). That might appeal to your parsimonious nature <grin>.

And I'd think about replacing the BFR93A since it is over 10c a pop. A
BFS17A can be had for 3c less. Ok, a bit less than half the ft but that
ought to do here. Maybe a grounded base architecture.

Sounds reasonable. The OP said BFR93A, so that's what I played with. I'd
instinctively go for grounded base and microstrip for the flea power he
wants.

Sunday's exercise might be to compare the (Philips) model's S-parameters
against Philips' data sheet values. Wanna bet they don't match?

Shame LTSpice won't do polar plots. Since it does impedance and S-params,
it really ought to.

--
"For a successful technology, reality must take precedence
over public relations, for nature cannot be fooled."
                                       (Richard Feynman)

On Sun, 29 Aug 2010 16:41:26 -0700, Joerg wrote:

Jim Thompson wrote:
On Sun, 29 Aug 2010 15:21:15 -0700, Joerg <invalid_at_invalid.invalid
wrote:

Jim Thompson wrote:
On Sun, 29 Aug 2010 13:51:14 -0700, Joerg <invalid_at_invalid.invalid
wrote:

Fred Abse wrote:
On Sat, 28 Aug 2010 14:34:24 -0700, Fred Abse wrote:

Sunday's exercise might be to compare the (Philips) model's S-parameters
against Philips' data sheet values. Wanna bet they don't match?
Pretty close match, apart from S22, which doesn't show the pronounced kink
shown in the datasheet.

Shame LTSpice won't do polar plots. Since it does impedance and
S-params, it really ought to.
Found a way round that. Set LTSpice to do ASCII rawfiles, run, edit two or
three LT-specific lines out of the rawfile, and load it into Berkeley
Spice 3f4's Nutmeg, then plot polar and Smith from there.
Much easier than trying to get LTSpice netlists to run in 3f4.

I still think the facility should be in LTSpice, though.

I think Jim Thompson once mentioned a software that could render raw
data into all sorts of graphics including polar. Jim?
Scratching head... I don't remember

I believe it was a software that you could also throw scanned in plots.
Aha! I did recently buy "GetData Graph Digitizer" which can create
numerical data points from graphs in data sheets.

Handy for creating subcircuits when all you have is a data sheet.

Reverse engineering ... ?


PSpice can do about any x-y plot you'd like, but I don't know how
you'd do polar except with some innovative macros

I think LTSpice can output polar data but I have never done it. I
believe the command is "meascplxfmt polar".

Can't find any reference. .meas statements usually start with an analysis
type, like ac, tran, etc.

Polar plots uploaded to a.b.s.e. Unfortunately my hand-added annotations
of frequency end points haven't survived the upload. Added in gimp,
they're present in the .pngs here.

Hi there.
Where did you get the input impadance of 300+j76 ohms from.

Can you get it from LTSpice , or do you have a datasheet that shows
that.
I can't find it on the datasheet that I have. They do have a S11
plot , but it does
show anything like that. Maybe I'm just being a dumbo. A distinct
possability :0)

Same for the collector impedance of 700 ohms.How do you get
it?

On Mon, 30 Aug 2010 06:29:37 -0700, neddie wrote:

Hi there.
Where did you get the input impadance of 300+j76 ohms from.

From an LTSpice simulation of the circuit, using an .ac analysis with a
.net directive.

Can you get it from LTSpice , or do you have a datasheet that shows
that.
I can't find it on the datasheet that I have. They do have a S11
plot , but it does
show anything like that. Maybe I'm just being a dumbo. A distinct
possability :0)

The S-param plots are for the device alone. The model gets near(ish) the
measured values in the datasheet (assuming they *are* measured!), but I'd
say no cigar.

Same for the collector impedance of 700 ohms.How do you get
it?

It's not the looking-backwards collector impedance, that's more or less
irrelevant, despite what some people would have you believe. Forget
maximum power transfer theorem (Joerg has anecdotes about that, and his
onetime Prof).

700 ohms is the load that happens to give about 50mW in this case, using a
Vcc of 6V.

The usual rule of thumb is:

Rl=Vcc^2/(2*Pout), for a class C amplifier.

This can be out by quite a lot, depending on other factors, but it's a
start. In fact, it gives 360 ohms, which gives about 100mW in this case,
so I throttled it back.

The design procedure I use is this:

Select the collector feed inductor to resonate with the device output
capacitance.

The only reason I've gor a load of 200 ohms is...
The anternna is going to be a piece of track on the pcb.
Analysing it in Sonnet shows it has an impedance of about 200 ohms at
433Mhz

On Mon, 30 Aug 2010 06:29:37 -0700, neddie wrote:
Hi there.
Where did you get the input impadance of 300+j76 ohms from.

From an LTSpice simulation of the circuit, using an .ac analysis with a
.net directive.

Can you get it from LTSpice , or do you have a datasheet that shows
that.
I can't find it on the datasheet that I have. They do have a S11
plot , but it does
show anything like that. Maybe I'm just being a dumbo. A distinct
possability :0)

The S-param plots are for the device alone. The model gets near(ish) the
measured values in the datasheet (assuming they *are* measured!), but I'd
say no cigar.

Same for the collector impedance of 700 ohms.How do you get

it?

It's not the looking-backwards collector impedance, that's more or less
irrelevant, despite what some people would have you believe. Forget
maximum power transfer theorem (Joerg has anecdotes about that, and his
onetime Prof).

700 ohms is the load that happens to give about 50mW in this case, using a
Vcc of 6V.

The usual rule of thumb is:

Rl=Vcc^2/(2*Pout), for a class C amplifier.

This can be out by quite a lot, depending on other factors, but it's a
start. In fact, it gives 360 ohms, which gives about 100mW in this case,
so I throttled it back.

The design procedure I use is this:

Select the collector feed inductor to resonate with the device output
capacitance.

Determine what load resistance will result in the desired power.

Design a network to match that resistance to your desired load, with your
desired Q, usually about 10, any more and tuning gets difficult. Much
less, and harmonics become a problem.

Unfortunately, stepping up 200 ohms to 700 ohms at Q=10, using a simple RL
network results in practically unrealizable capacitance values, so my
circuit isn't much use outside of a theoretical simulation.

Why 200 ohms, BTW?

Here's the circuit, for what it's worth (not much).

Version 4
SHEET 1 1300 2476
WIRE 48 144 -256 144
WIRE -256 192 -256 144
WIRE 48 192 48 144
WIRE -256 304 -256 272
WIRE -224 304 -256 304
WIRE -96 304 -160 304
WIRE -48 304 -96 304
WIRE 64 304 32 304
WIRE 192 304 128 304
WIRE -256 320 -256 304
WIRE 192 320 192 304
WIRE -96 352 -96 304
WIRE -464 368 -672 368
WIRE -352 368 -400 368
WIRE -320 368 -352 368
WIRE -352 384 -352 368
WIRE -672 480 -672 448
WIRE -368 480 -672 480
WIRE -352 480 -352 464
WIRE -352 480 -368 480
WIRE -256 480 -256 416
WIRE -256 480 -352 480
WIRE -96 480 -96 432
WIRE -96 480 -256 480
WIRE 48 480 48 272
WIRE 48 480 -96 480
WIRE 192 480 192 400
WIRE 192 480 48 480
WIRE -368 496 -368 480
FLAG -368 496 0
FLAG 48 480 0
SYMBOL voltage -672 352 R0
WINDOW 3 32 84 Left 0
WINDOW 123 -97 53 Left 0
WINDOW 39 24 44 Left 0
SYMATTR Value SINE(0 3 433meg)
SYMATTR Value2 AC 1
SYMATTR SpiceLine Rser=50
SYMATTR InstName V1
SYMBOL ind -272 176 R0
SYMATTR InstName L1
SYMATTR Value 190n
SYMBOL voltage 48 176 R0
SYMATTR InstName V2
SYMATTR Value 6
SYMBOL npn2 -320 320 R0
SYMATTR InstName Q1
SYMATTR Value BFR93A
SYMATTR Prefix X
SYMBOL ind -368 368 R0
SYMATTR InstName L2
SYMATTR Value 47n
SYMBOL res 176 304 R0
SYMATTR InstName R1
SYMATTR Value 200
SYMBOL ind -64 288 M90
WINDOW 0 5 56 VBottom 0
WINDOW 3 32 56 VTop 0
SYMATTR InstName L3
SYMATTR Value 597.5n
SYMBOL cap 128 288 R90
WINDOW 0 0 32 VBottom 0
WINDOW 3 32 32 VTop 0
SYMATTR InstName C1
SYMATTR Value 0.18p
SYMBOL ind -112 336 R0
SYMATTR InstName L4
SYMATTR Value 137.5n
SYMBOL cap -224 288 M90
WINDOW 0 0 32 VBottom 0
WINDOW 3 32 32 VTop 0
SYMATTR InstName C2
SYMATTR Value .8p
SYMBOL cap -400 352 R90
WINDOW 0 0 32 VBottom 0
WINDOW 3 32 32 VTop 0
SYMATTR InstName C3
SYMATTR Value 1n
TEXT -552 280 Left 0 !.tran 0 60n 10n 100p
TEXT -160 536 Left 0 !.lib BFR93A.sub
TEXT -408 536 Left 0 !;.net I(R1) V1
TEXT -872 280 Left 0 !;ac dec 1000 350e6 550e6

--
"For a successful technology, reality must take precedence
over public relations, for nature cannot be fooled."
                                       (Richard Feynman)

Just a word of caution here: Don't do this for a design that has to go
into mass production. The capacitive device parameters are lot dependent
and vary with temperature.

using a simple RL
network