Sine wave generator $2

[MasterT]

Small project to get 30-100 MHz frequency range for test purposes. My Rigol 1022 could provide 20 MHz at max, and recently I need more for evaluation new wideband op amps. So here it is, quick & easy way  to build SG, no fpga required.

[coppercone2]

is that a dip switch?

[MasterT]

KiCAD doesn't have rotary switch, so I draw a dip. Does it matter?

[themadhippy]

KiCAD doesn't have rotary switch

try sw_rotary  in the parts search you might find something usefull

[moffy]

I like the design a lot, that is pretty remarkable for an opamp with a GBW of 100MHz at gain=2, I guess the Q of the LC must be helping. What is the bandwidth of your scope?

[MasterT]

It's Rigol DS1202Z-E,  200 MHz

SN10501 is the best on distortion, 100 MHz is the maximim I could get, DS presents GBW with 150 Ohm load. 
I also  tried  another IC, in this circuits, some could get 150 MHz (1-2 turn coils!), but distortion or stability,  worse than with  SN-10501.

[coppercone2]

thought you maybe tested it with parallel inductors

[fourfathom]

Schematic questions:
Where is the oscillator positive feedback coming from?
I assume the 1N5711 / MCP-6282 / 1N916 circuit provides amplitude leveling by varying the current / impedance of the 1N916 diode acting as a variable load on the tank circuit (correct???)
The two potentiometers show no voltage connection to pins 3.  These are for tuning and level, right?

[moffy]

Schematic questions:
I assume the 1N5711 / MCP-6282 / 1N916 circuit provides amplitude leveling by varying the current / impedance of the 1N916 diode acting as a variable load on the tank circuit (correct???)
The two potentiometers show no voltage connection to pins 3.  These are for tuning and level, right?

I had the same thought, I thought that the MCP-6282 integrator output would be negative with the 1N916 absorbing energy largely on the negative cycle/peak. I assumed also that both pots would be connected to the +5v.

[MasterT]

Schematic questions:
Where is the oscillator positive feedback coming from?
I assume the 1N5711 / MCP-6282 / 1N916 circuit provides amplitude leveling by varying the current / impedance of the 1N916 diode acting as a variable load on the tank circuit (correct???)
The two potentiometers show no voltage connection to pins 3.  These are for tuning and level, right?

Positive feedback has many path, as internaly inside IC and parasitic external coupling.  Internaly I could observe that oscilator works even with gain set to 1. My understanding that "current feedback" OPA is tricky things, that has much wider GBW , probably > 1 GHZ is measured with small signal Gain=100 & F=10 MHz.  External capacitive feedbak drops at low freq. range, < 20 MHz, so I have to install R=1k between output and pin3 (+).  But doing so THD goes up.

Assumption is correct,  pots are magnitude & freq. control. They connected to ground and +6V power rail. It's obvoulsly. MCP6282 outputs 200-400 mV DC only, controling diode resistance by "current mode", small shift in frequency exists <1-2% at max. What I measured, LC tank running at 50-100 mV  when output of the OPA has 10 Vp-p, this is why CFB is tricky part IMHO. Gain is Not set by two external R. Changing gain ffrom 1 to 10 does liitle effect on overall performance.

[dobsonr741]

Wow, this circuit packs quite a few unusuals.  You could say the end justifies means and you needed a test signal. The items that felt strange me were to see you rely on parasitics to build the oscillator, the not so proper use of the dual varicap and the stripborad to build it, like it would be a guitar amp for a 4KHz tops signal vs your target of 100MHz. At the end, kudos to make it work.

[MasterT]

Some people mistakenly think, that reading text book gives them full knowledge of the world around. 
  I have a vague  idea of what extent concept of the OPA expllained in those books, but what I know for sure is a subject of misinformation, when composite amplifier misleadingly interpreted as CFB.

[fourfathom]

OK, so the required feedback comes via unspecified parasitic and stray paths.  I can easily believe this, as we've all seen amplifiers that turn out to be oscillators.  Often this is due to the presence of intended negative feedback elements that turn into positive feedback due to amplifier phase shift, but in this case it appears that this positive feedback comes from other sources, perhaps stray layout coupling capacitance or inductance.  Interesting, but difficult to reproduce.

[Alex Eisenhut]

KiCAD doesn't have rotary switch, so I draw a dip. Does it matter?

I'd say so, a DIP switch would allow no connection or all connections, which a rotary switch doesn't. Also, a rotary switch would have the break-before-make, or not, distinction.

[MasterT]

OK, so the required feedback comes via unspecified parasitic and stray paths.  I can easily believe this, as we've all seen amplifiers that turn out to be oscillators.  Often this is due to the presence of intended negative feedback elements that turn into positive feedback due to amplifier phase shift, but in this case it appears that this positive feedback comes from other sources, perhaps stray layout coupling capacitance or inductance.  Interesting, but difficult to reproduce.

There are list of about dozens OP amp's I tried on a breadboard, all of them oscilate with G=2 and LC tank at the non inverting input. The specific of fast speed amplifiers is out of expertize level for most engineers. Especialy CFB or Composite amplifiers, that intrinsicaly unstable, and precise model of parameters does not exist and could not be created.

[dobsonr741]

High speed analog is sensitive to decoupling, stray inductances and capacitances. Breadboard/veroboard above audio range is not going to bring any consistent results. Copying here two awesome app notes - AN47 with a little guidance, and AN202 on decoupling/grounding. AN47 shows how to prototype high speed analog over copper clad ground plane and how to attach probes.

https://www.analog.com/media/en/technical-documentation/application-notes/an47fa.pdf
"This 132-page app note is Jim's magnum opus, and it contains a wealth of valuable information on high-speed measurement techniques, test equipment, oscilloscope probes, and, of course, applications. It should be required reading for all electronics engineers. Read it twice. Some of the highlights include "Mr. Murphy's gallery of high speed amplifier problems" (pages 7 to 15), "About oscilloscopes" (a gallery of scope and probe responses, pages 20 to 24), and "Breadboarding techniques" (two pictorial tutorials, Figures 62 to 65 and Figures F1 to F25)"

An IC Amplifier User’s Guide to Decoupling, Grounding, and Making Things Go Right for a Change
https://www.analog.com/media/en/technical-documentation/application-notes/AN-202.pdf

[MasterT]

Wikipedist. Does it word existed or I just invented it?

[moffy]

MasterT, thanks for the fuller explanation, so many subtleties involved, definitely not one for SPICE. It probably wouldn't work on a ground plane as that would tend to kill the feedback parasitics.

[MasterT]

Add 1k resistor, I told in reply #9. Parasitics save 1 resistor !
 
Regarding magnitude control, there is LMH-6732 that has "build-in" bandwidth control. It saves a diode 1n916.

[SiliconWizard]

As far as I can see, the highest harmonic is the 3rd one at around -40dB, not bad for such a "simple" design.

[moffy]

Add 1k resistor, I told in reply #9. Parasitics save 1 resistor !
 
Regarding magnitude control, there is LMH-6732 that has "build-in" bandwidth control. It saves a diode 1n916.

But parasitics are so much more fun.

[MasterT]

As far as I can see, the highest harmonic is the 3rd one at around -40dB, not bad for such a "simple" design.

SN-10501 has sharp roll-off above 100 MHz, likely design was tailored to get max. bandwidth of the output stage.  Because of this "feature" HD-2 & HD-3 heavily attenuated, and 100 MHz IC beats up LMH-6732 (>500 MHz GBW) and even AD8008 on THD level.
 Unfortunately, this does not work at 30 MHz, and THD rise up slightly