Hi!. I want to tie a ~40MHz VCO to a frequency stabiliser. A PIC measures initial f and then makes corrections. Resolution is 10Hz. The PIC count is shown on a LCD module.
The VCO is a classical JFET Hartley. Gate: 100k and 1N4148 to GND, 100pF to LC end. Source: to 25% tap.
After warmup, the tens of Hz show an acceptable jitter of +/-10Hz most of the times. But suddenly I witness a BAD worsening, the tens become mad and even the hundreds are affected by jitter. This can last several minutes. I go for a walk, come back, and peace reigns again.
I read about flicker (1/f, Random Telegraph) noise, but this behavior is strange. I made sure Vdd is clean, there are no cellphones or other wireless devices around, there are no creeping surface currents because I used "ugly construction" in the air. Replacing the varicap/varactor with a fixed C made no difference.
This improved somewhat by connecting the gate directly to the hot end with no R nor diode, and inserting a parallel 1kohm/1nF in series with the source, in order to provide low-frequency degeneration.
I made sure the level at the PIC input is ample. A 43MHz xtal oscillator gives always rock-steady reading.
- Is there a name for this "intermittent-1/f" noise?.
- Has anyboy had succes in reducing it?.
- Off topic (not radio): how can a PC read the values from the PIC to the LCD?. E. g. via parallel port, with some code written in ASM, and booting in DOS to keep Windows out of the way.
Many thanks!. Daniel LW1ECP

"lw1ecp" wrote in message
...
Hi!. I want to tie a ~40MHz VCO to a frequency stabiliser. A PIC measures
initial f and then makes corrections. Resolution is 10Hz. The PIC count is
shown on a LCD module.

Hi
I would suspect le reverse voltage on the varicap diode getting too close
to zero.
Try adding a small battery in serial to the control loop.

Many thanks, bilou!. But as I mentioned, the problema also happens without the varicap. The fact is, I've never closed the loop yet... the freq jump would surely fool the stabiliser into thinking I am purposely turning the knob.
Maybe this phenomenom can be explained by some friend with close knowledge of audio preamplifiers? (because this must be a low freq noise modulating the oscillator). Meanwhile I'll try with other FETs and/or increasing the source resistor. I read dr. Ulrich Rohde advised not to use the gate clamping diode.

On Tue, 30 Jul 2013 14:55:23 -0700, lw1ecp wrote:

Hi!. I want to tie a ~40MHz VCO to a frequency stabiliser. A PIC
measures initial f and then makes corrections. Resolution is 10Hz. The
PIC count is shown on a LCD module.
The VCO is a classical JFET Hartley. Gate: 100k and 1N4148 to GND, 100pF
to LC end. Source: to 25% tap.
After warmup, the tens of Hz show an acceptable jitter of +/-10Hz most
of the times. But suddenly I witness a BAD worsening, the tens become
mad and even the hundreds are affected by jitter. This can last several
minutes. I go for a walk, come back, and peace reigns again.
I read about flicker (1/f, Random Telegraph) noise, but this behavior is
strange. I made sure Vdd is clean, there are no cellphones or other
wireless devices around, there are no creeping surface currents because
I used "ugly construction" in the air. Replacing the varicap/varactor
with a fixed C made no difference.
This improved somewhat by connecting the gate directly to the hot end
with no R nor diode, and inserting a parallel 1kohm/1nF in series with
the source, in order to provide low-frequency degeneration.
I made sure the level at the PIC input is ample. A 43MHz xtal oscillator
gives always rock-steady reading.
- Is there a name for this "intermittent-1/f" noise?.
- Has anyboy had succes in reducing it?.
- Off topic (not radio): how can a PC read the values from the PIC to
the LCD?. E. g. via parallel port, with some code written in ASM, and
booting in DOS to keep Windows out of the way.
Many thanks!. Daniel LW1ECP

You mentioned cell phones and other wireless devices. How about police
cars or other radio equipped vehicles driving by? What about other
sources of RFI (TV sets, light dimmers, power tools, etc.)?

Since you used "ugly construction", I assume that your device isn't
shielded. Put it in a cardboard box wrapped in aluminum foil. Ground
the foil to the circuit and filter all the wires going in and out. If
done carefully, this may point to the source of your problem.

Or, maybe not. I was once called on to find the cause of low frequency
noise in a large instrument that was being developed. The noise would
come and go. I spent several days without much luck until I decided to
put the preamp in a box to shield it from the room lights (some
transistors, ICs, and diodes are light sensitive). The noise went away.
But removing the box and turning off the lights didn't make it go away.
Then I noticed that the noise appeared only when the air conditioning fan
came on. Since putting it in a cardboard box made the noise disappear,
it couldn't have been an electrical problem. I visually examined the
board and found that the designer had used Z5U capacitors in a filter
circuit. Replacing these with (much larger, requiring a new board
layout) polyester capacitors solved the problem.

The explanation: Z5U capacitors are sensitive to just about everything.
They change value with temperature, voltage, and frequency. They are
also piezoelectric. In this case, it was the piezoelectric effect. The
capacitors were picking up the air pressure changes caused by the air
conditioning and injecting stray voltages into the circuit.

Not all ceramic capacitors are like this. The C0G types are among the
best capacitors that are readily available. X7R types are intermediate.

Z5U capacitors are excellent for power supply bypassing but not much else.

--
Jim Mueller

To get my real email address, replace wrongname with dadoheadman.
Then replace nospam with fastmail. Lastly, replace com with us.

Many thanks, bilou!. But as I mentioned, the problema also happens without the varicap.
The fact is, I've never closed the loop yet... the freq jump would surely fool the
stabiliser into thinking I am purposely turning the knob.
Maybe this phenomenom can be explained by some friend with close knowledge of audio
preamplifiers? (because this must be a low freq noise modulating the oscillator).
Meanwhile I'll try with other FETs and/or increasing the source resistor. I read dr.
Ulrich Rohde advised not to use the gate clamping diode.

Do you have any diodes in or around the oscillator circuit which have
glass cases? It's not unusual for such diodes to pick up 50/60 Hz
noise from lights in the room - they act (inadvertently) as
photodiodes. Fluorescent lights in particular have a strong enough
flicker to cause a significant amount of photo-optical noise pickup.

Reportedly, some transistors can suffer a similar
problem... metal-case transistors with ceramic seals can "leak" light
in as far as the dice, and I think I've heard of cases in which even
"opaque" plastic encapsulation could allow enough IR transmission to
disturb sensitive circuits.

It might be worth your while to build yourself a custom shielding case
for this oscillator circuit... make it out of double-sided FR-4
copperclad PC board material, solder the sides and base together, use
capacitive feed-throughs and ferrite beads on the power and
VCO-voltage line, bring out the signal using SMA or BNC
connectors, and fasten on a lid using some sort of light-tight and
well-grounded arrangement. This would exclude both light, and other
forms of EMI. Even though you say that there are no wireless devices
nearby, it's entirely possible that RF pickup from a transmitter you
don't know about, blocks away, might be leaking into the power or
control lines and causing problems.

Do you have a radio receiver you can tune to the 40 MHz output of the
VCO? Might be interesting to listen to its output, or look at the
spectrum on a spectrum analyzer... there might be audible sidebands
(e.g. hum or buzz or audio pickup) which could give you a clue as to
where the problem originates.

Jim, David: thanks for the hints. Yes, it will be a good idea be to shield the stage. I didn't suspect of external sources because the xtal osc. gave perfect stability, but who knows. I was pointing to weird causes such as surface leak currents, e. g. the FET case had some years' dirt on the bottom.
Light?. Yes, I once used a power RF transistor for a VCO because they recommend operating the active device at low current density to get lower phase noise, but mains frequency sidebands were horrible... till I realized the fluorescent light was penetrating the transistor's ceramic body!.
Listening to the demodulated oscillator output, or amplifying the drain current variations seems also a great idea, maybe it's not popcorn what I hear..
Fortunately I won't need Z5Us at 40MHz, I've also had headaches from their microphonics in the past.
I'll let you know!

I have been far from this project for the last months, I have no definitive conclussions yet, but just for keeping muy promise I'll tell you that the culprit seemed to be that particular transistor. After repeated desoldering/resoldering the gate short circuited with the channel, so I replaced the FET and frequency jitter became quite lower. More on this later!