The BUF602 is a very well controlled part, 1 GHz bw,
low output impedance, closed-loop gain = 1.00. So I expect
the real oscillator to match the sim very well.

This is a sort-of-Colpitts.

Version 4
SHEET 1 1812 708
WIRE 1008 32 912 32
WIRE 912 64 912 32
WIRE 48 80 -32 80
WIRE 176 80 128 80
WIRE 304 80 240 80
WIRE 368 80 304 80
WIRE 432 80 368 80
WIRE 576 80 432 80
WIRE 704 80 576 80
WIRE 864 80 704 80
WIRE 704 112 704 80
WIRE -32 128 -32 80
WIRE 864 128 832 128
WIRE 304 144 304 80
WIRE 432 144 432 80
WIRE 576 144 576 80
WIRE 832 160 832 128
WIRE 912 160 912 144
WIRE 912 160 832 160
WIRE 912 176 912 160
WIRE 704 224 704 176
WIRE 736 224 704 224
WIRE 784 224 736 224
WIRE 896 224 864 224
WIRE 1008 224 1008 32
WIRE 1008 224 976 224
WIRE 704 240 704 224
WIRE -32 272 -32 208
WIRE 304 272 304 224
WIRE 432 272 432 208
WIRE 432 272 304 272
WIRE 576 272 576 224
WIRE 576 272 432 272
WIRE 576 320 576 272
WIRE 704 320 704 304
FLAG 576 320 0
FLAG 368 80 LC
FLAG 704 320 0
FLAG 912 176 0
FLAG 736 224 R
FLAG -32 272 0
SYMBOL ind 288 128 R0
WINDOW 0 -63 28 Left 2
WINDOW 3 -76 61 Left 2
SYMATTR InstName L1
SYMATTR Value 150n
SYMATTR SpiceLine Rser=0.9
SYMBOL cap 416 144 R0
WINDOW 0 -43 17 Left 2
WINDOW 3 -47 45 Left 2
SYMATTR InstName C1
SYMATTR Value 42p
SYMBOL res 560 128 R0
WINDOW 0 -46 27 Left 2
WINDOW 3 -44 57 Left 2
SYMATTR InstName Rq
SYMATTR Value 3K
SYMBOL cap 720 176 R180
WINDOW 0 61 21 Left 2
WINDOW 3 52 -8 Left 2
SYMATTR InstName C2
SYMATTR Value 22p
SYMBOL cap 688 240 R0
WINDOW 0 -54 20 Left 2
WINDOW 3 -60 48 Left 2
SYMATTR InstName C3
SYMATTR Value 75p
SYMBOL e 912 48 R0
WINDOW 0 45 44 Left 2
WINDOW 3 50 70 Left 2
SYMATTR InstName E2
SYMATTR Value 1
SYMBOL ind 768 240 R270
WINDOW 0 -29 54 VTop 2
WINDOW 3 -35 56 VBottom 2
SYMATTR InstName L2
SYMATTR Value 100n
SYMBOL res 992 208 R90
WINDOW 0 64 57 VBottom 2
WINDOW 3 70 58 VTop 2
SYMATTR InstName R1
SYMATTR Value 50
SYMBOL schottky 240 64 R90
WINDOW 0 -44 31 VBottom 2
WINDOW 3 -35 35 VTop 2
SYMATTR InstName D1
SYMATTR Value BAT54
SYMATTR Description Diode
SYMATTR Type diode
SYMBOL voltage -32 112 R0
WINDOW 0 57 37 Left 2
WINDOW 3 62 67 Left 2
SYMATTR InstName V1
SYMATTR Value 1
SYMBOL res 32 96 R270
WINDOW 0 77 64 VTop 2
WINDOW 3 62 63 VBottom 2
SYMATTR InstName R2
SYMATTR Value 50
TEXT 288 -64 Left 2 !.tran 0 500n 0 500f
TEXT 0 -64 Left 2 ;JL Feb 24 2024
TEXT -32 -104 Left 2 ;BUF602 50 MHz Oscillator
TEXT 200 224 Left 2 ;0.9R dc
TEXT 792 40 Left 2 ;BUF602
TEXT 576 -104 Left 2 ;L2 100n F = 50.012
TEXT 576 -72 Left 2 ;L2 120n F = 49.405
TEXT 576 -8 Left 2 ;F ratio 1.012
TEXT 576 -40 Left 2 ;L ratio 1.2
TEXT 600 -136 Left 2 ;L2 Sensitivity
TEXT 160 256 Left 2 !.ic I(L1)=1m

On 25/02/2024 4:12 am, John Larkin wrote:
> The BUF602 is a very well controlled part, 1 GHz bw,
> low output impedance, closed-loop gain = 1.00. So I expect
> the real oscillator to match the sim very well.

Perhaps. Neither inductor has any parallel capacitance, and L2 hasn't
got any series resistance either.

Replacing L1 with a 150n Wurth ferrite bead, which does come with full
set of data - the part I picked was 742794 WE-CBF 1806 - stopped it from
working.

A YAGEO company - Pulse sells a 150nH part you can buy from Mouser

PE-0603CD151GTT 150 @ 250MHz 28 @ 150MHz 990 0.92 2

The 990 number is the series resonant frequency of 990MHz which implies
a parallel capacitance of 17pF. The Q was 28 at 150MHz

Plugging that in, and dropping C1 to 25pF did give a working circuit
with a oscillation frequncy of 50.483012MHz.

The second harmonic was 22dB below the fundamental, and the third was
32db down.

A sim is only as good as the component data you plug in.

--
Bill Sloman, Sydney

On Sun, 25 Feb 2024 14:58:13 +1100, Bill Sloman <bill.sloman@ieee.org>
wrote:

>On 25/02/2024 4:12 am, John Larkin wrote:
>> The BUF602 is a very well controlled part, 1 GHz bw,
>> low output impedance, closed-loop gain = 1.00. So I expect
>> the real oscillator to match the sim very well.
>
>Perhaps. Neither inductor has any parallel capacitance, and L2 hasn't
>got any series resistance either.

Stray c will change the frequency slightly, less than component
tolerances.

There's already 40 ohms in series with L2 so a bit more doesn't
matter. About 8 ohms of the 40 is the output impedance of the BUF502.

>
>Replacing L1 with a 150n Wurth ferrite bead, which does come with full
>set of data - the part I picked was 742794 WE-CBF 1806 - stopped it from
>working.
>

Sure, oscillators are easy to break if you really want to.

>A YAGEO company - Pulse sells a 150nH part you can buy from Mouser
>
>PE-0603CD151GTT 150 @ 250MHz 28 @ 150MHz 990 0.92 2
>
>The 990 number is the series resonant frequency of 990MHz which implies
>a parallel capacitance of 17pF. The Q was 28 at 150MHz

Series resonant frequency? 17 pF resonates with 150n at 100 MHz. You
slipped a couple of decimal points. You could use my LC7.exe program.

https://www.dropbox.com/scl/fi/h7l4bkx07fjo78rj83na4/LC7.EXE?rlkey=9cpwpna4begf1u8lfdawibwrl&dl=0

https://www.dropbox.com/scl/fi/yio6yl26ag6wqliafeahc/LC7.txt?rlkey=ncfyzyiav8majni507dgc8qzg&dl=0

I'm using a Coilcraft part, 1812SMS-R15GLB. Q is about 100 at 50 MHz.
SRF 750 MHz, so c=0.3 pF, less than the PCB pads and traces. Tempco is
only 40 PPM.

>
>Plugging that in, and dropping C1 to 25pF did give a working circuit
>with a oscillation frequncy of 50.483012MHz.
>
>The second harmonic was 22dB below the fundamental, and the third was
>32db down.
>
>A sim is only as good as the component data you plug in.

One thing that expedites engineering is knowing which possible effects
don't matter, so can be ignored.

And, I suppose, which people don't matter and can be ignored.

On 26/02/2024 5:11 am, John Larkin wrote:
> On Sun, 25 Feb 2024 14:58:13 +1100, Bill Sloman <bill.sloman@ieee.org>
> wrote:
>
>> On 25/02/2024 4:12 am, John Larkin wrote:
>>> The BUF602 is a very well controlled part, 1 GHz bw,
>>> low output impedance, closed-loop gain = 1.00. So I expect
>>> the real oscillator to match the sim very well.
>>
>> Perhaps. Neither inductor has any parallel capacitance, and L2 hasn't
>> got any series resistance either.
>
> Stray c will change the frequency slightly, less than component
> tolerances.
>
> There's already 40 ohms in series with L2 so a bit more doesn't
> matter. About 8 ohms of the 40 is the output impedance of the BUF502.
>
>> Replacing L1 with a 150n Wurth ferrite bead, which does come with full
>> set of data - the part I picked was 742794 WE-CBF 1806 - stopped it from
>> working.
>
> Sure, oscillators are easy to break if you really want to.

That wasn't the point. The Wurth ferrite bead is on offer in the LTSpice
program with all the parameters you need to model it more or less
accurately. There were two other 150nH parts, which I didn't bother
trying - ferrite beads are actually intended to kill oscillations
(though they don't always succeed ).>
>> A YAGEO company - Pulse sells a 150nH part you can buy from Mouser
>>
>> PE-0603CD151GTT 150 @ 250MHz 28 @ 150MHz 990 0.92 2
>>
>> The 990 number is the series resonant frequency of 990MHz which implies
>> a parallel capacitance of 17pF. The Q was 28 at 150MHz
>
> Series resonant frequency? 17 pF resonates with 150n at 100 MHz. You
> slipped a couple of decimal points. You could use my LC7.exe program.

Just one - 99.7MHz. The correct answer is 0.17pF which is remarkably
low. It's about the parallel capacitance of an axial resistor.>
> https://www.dropbox.com/scl/fi/h7l4bkx07fjo78rj83na4/LC7.EXE?rlkey=9cpwpna4begf1u8lfdawibwrl&dl=0

Posting an executable program on the web is anti-social. Posting the
code which can be compiled into an executable program is more
respectable, if the code is properly and accurately commented.

> https://www.dropbox.com/scl/fi/yio6yl26ag6wqliafeahc/LC7.txt?rlkey=ncfyzyiav8majni507dgc8qzg&dl=0

It's a trivial program. I should have checked my calculation before I
posted it, but getting it right wouldn't have changed anything.

> I'm using a Coilcraft part, 1812SMS-R15GLB. Q is about 100 at 50 MHz.
> SRF 750 MHz, so c=0.3 pF, less than the PCB pads and traces. Tempco is
> only 40 PPM.

So why didn't you say so, and plug in the parameters you did have for
that part.

>> Plugging that in, and dropping C1 to 25pF did give a working circuit
>> with a oscillation frequency of 50.483012MHz.
>>
>> The second harmonic was 22dB below the fundamental, and the third was
>> 32db down.

I reran the simulation for 35usec - it took ages - and got a much
sharper looking Fourier transform, but with the same harmonics.>>
>> A sim is only as good as the component data you plug in.
>
> One thing that expedites engineering is knowing which possible effects
> don't matter, so can be ignored.

And one thing that screws up engineering really badly is thinking you
know what they are, and getting it wrong.

> And, I suppose, which people don't matter and can be ignored.

Which can also get you into trouble if you get it wrong.

--
Bill Sloman, Sydney