"Robert Baer" a écrit dans le message news:
...
Paul Burridge wrote:

On Sat, 10 Apr 2004 08:22:21 GMT, Robert Baer
wrote:

Not a varactor (diode), but a switching diode; fast. Snap-off is lotz
better.
And they are a *LOT* smaller!

Thanks, Robert (and all else)
I'd prefer a diode solution as there'd be no power supply requirements
for each device. The tiny size is an added bonus, of course. But
what's the drawback with superfast switching diodes as against active
inverters? I'm sure there must be (at least) one...

An active inverter is S L O W by comparison; a good, fast diode can
have turnoff times in the tens of picoseconds region.


Don't let Paul dream too much about picoseconds.
In order to achieve results that good, one have to pay very careful
attention to "wiring". There's still a lot of room for Paul to improve his
PCB design skills ;-)

Thanks,
Fred.

On Sun, 11 Apr 2004 14:14:56 +0200, "Fred Bartoli"
r_AndThisToo wrote:

Don't let Paul dream too much about picoseconds.

Well, quite. I wouldn't know what to do with all those extra
harmonics, anyway. :-)
Diodes sound great, but I'm heavily put off by that comment earlier in
the thread that they're prone to self oscillation. :-( What's a
"parametric device" anyway?

In order to achieve results that good, one have to pay very careful
attention to "wiring". There's still a lot of room for Paul to improve his
PCB design skills ;-)

Fair comment. But I *am* working on it!

"Paul Burridge" wrote in message
news
On Sun, 11 Apr 2004 14:14:56 +0200, "Fred Bartoli"
r_AndThisToo wrote:

Don't let Paul dream too much about picoseconds.

Well, quite. I wouldn't know what to do with all those extra
harmonics, anyway. :-)
Diodes sound great, but I'm heavily put off by that comment earlier in
the thread that they're prone to self oscillation. :-( What's a
"parametric device" anyway?

Parametric operation is when you have a parameter such as diode capacitance
that varies with voltage. The non-linearity of the parameter makes it act as
a multiplier, mixer or an amplifier. Parametric amplifiers (they need to be
'pumped' with an oscilllator) used to be very popular for microwave use. If
you have inadvertently make yourself a parametric amplifier with your diode
it could oscillate given the right conditions and do all sorts of strange
things. In your hands, this is a strong possibility. 8-)

Leon

Parametric operation is when you have a parameter such as diode
capacitance
that varies with voltage. The non-linearity of the parameter makes it act
as
a multiplier, mixer or an amplifier. Parametric amplifiers (they need to
be
'pumped' with an oscilllator) used to be very popular for microwave use.
If
you have inadvertently make yourself a parametric amplifier with your
diode
it could oscillate given the right conditions and do all sorts of strange
things. In your hands, this is a strong possibility. 8-)


I think you have a point Leon. Paul, use a Schottky diode, (HP 5082-2835 or
similar), use ZERO lead lengths. Make your PCBoard for the multiplier double
sided, drill a hole the diameter of the diode length, wrap a copper tape
ground from top to bottom on one edge of the hole, and drop the diode in the
hole and solder both ends. One end to the ground tape, the other to the land
to the filter for the harmonic you want. Then, drive with not more than +10
dBm of fundamental and hope a lot. It may STILL oscillate, but you've
maximized your chances. When you start getting output from the filter (on
the desired frequency which ought to be the only output you can get with the
filter in place) you can start matching the input to the diode with "L"
networks to maximize the output. Generally, the diode looks like a quite low
impedance so the capacitor used, ought to wind up on the side of the
inductor AWAY from the diode.

You can use an MMIC in the same manner to generate harmonics. Overdrive it
and under feed it with DC. More output, more chance of self oscillation if
you don't pay attention to detail.

A superior (less touchy) method of harmonic generation is just a full wave
diode rectifier. Using signal diodes and a center tapped transformer, make a
full wave doubler. Ground the DC out with an RF choke and you have 2F the
input frequency with good suppression of the fundamental and odd harmonics
built in. It's then a lot easier to filter the rest of the undesired out. Do
two of those to get 4F out rather than trying to do it all in one stage.
LOTS more efficiency.

W4ZCB

Parametric operation is when you have a parameter such as diode
capacitance
that varies with voltage. The non-linearity of the parameter makes it act
as
a multiplier, mixer or an amplifier. Parametric amplifiers (they need to
be
'pumped' with an oscilllator) used to be very popular for microwave use.
If
you have inadvertently make yourself a parametric amplifier with your
diode
it could oscillate given the right conditions and do all sorts of strange
things. In your hands, this is a strong possibility. 8-)


I think you have a point Leon. Paul, use a Schottky diode, (HP 5082-2835 or
similar), use ZERO lead lengths. Make your PCBoard for the multiplier double
sided, drill a hole the diameter of the diode length, wrap a copper tape
ground from top to bottom on one edge of the hole, and drop the diode in the
hole and solder both ends. One end to the ground tape, the other to the land
to the filter for the harmonic you want. Then, drive with not more than +10
dBm of fundamental and hope a lot. It may STILL oscillate, but you've
maximized your chances. When you start getting output from the filter (on
the desired frequency which ought to be the only output you can get with the
filter in place) you can start matching the input to the diode with "L"
networks to maximize the output. Generally, the diode looks like a quite low
impedance so the capacitor used, ought to wind up on the side of the
inductor AWAY from the diode.

You can use an MMIC in the same manner to generate harmonics. Overdrive it
and under feed it with DC. More output, more chance of self oscillation if
you don't pay attention to detail.

A superior (less touchy) method of harmonic generation is just a full wave
diode rectifier. Using signal diodes and a center tapped transformer, make a
full wave doubler. Ground the DC out with an RF choke and you have 2F the
input frequency with good suppression of the fundamental and odd harmonics
built in. It's then a lot easier to filter the rest of the undesired out. Do
two of those to get 4F out rather than trying to do it all in one stage.
LOTS more efficiency.

W4ZCB

"Paul Burridge" wrote in message
news
On Sun, 11 Apr 2004 14:14:56 +0200, "Fred Bartoli"
r_AndThisToo wrote:

Don't let Paul dream too much about picoseconds.

Well, quite. I wouldn't know what to do with all those extra
harmonics, anyway. :-)
Diodes sound great, but I'm heavily put off by that comment earlier in
the thread that they're prone to self oscillation. :-( What's a
"parametric device" anyway?

Parametric operation is when you have a parameter such as diode capacitance
that varies with voltage. The non-linearity of the parameter makes it act as
a multiplier, mixer or an amplifier. Parametric amplifiers (they need to be
'pumped' with an oscilllator) used to be very popular for microwave use. If
you have inadvertently make yourself a parametric amplifier with your diode
it could oscillate given the right conditions and do all sorts of strange
things. In your hands, this is a strong possibility. 8-)

Leon

If I understand correctly, you're interested in harmonics up in the
70MHz region. To me it seems silly to use diodes capable of
generating a comb of harmonics out to 20GHz for that. An advantage--a
big advantage--of the tiny logic is that you get considerable power
gain in the stage, and the sot-23 package is likely no larger than the
diode you might have used anyway. The edges, as others have pointed
out, are PLENTY fast enough for what I believe you want to do.

Whether you use diodes or gates, your size problem will be the
harmonic-selection filter. As you've learned (and I trust not
forgotten already), you need either multiple poles in your filter or a
rather high Q. You _could_ get the high Q with crystals, but of
course then you're locked down to particular frequencies. For high
multiplication factors to low-VHF frequencies, it's probably hard to
beat a PLL for (potentially) small size.

In offering suggestions, it would also be helpful to us to know your
actual needs for signal purity, both close-in (phase noise) and
broadband (other harmonics, etc.)

If you do use diodes for higher-order harmonic generation, and not
just a simple full-wave-rectifier type frequency doubler, I suppose
you want something of the nature of a step recovery diode. That
implies minority carrier stored charge in the diode, and that would
preclude using a Schottky diode (which would work great in the
full-wave-rectifier type doubler). If you get into actually wanting
to generate harmonic combs out to microwave frequencies, it's probably
worthwhile looking for diodes actually characterized for step recovery
service. But I really think that's way beyond what you are trying to
accomplish right now.

Cheers,
Tom


Paul Burridge wrote in message . ..
On Sun, 11 Apr 2004 14:14:56 +0200, "Fred Bartoli"
r_AndThisToo wrote:

Don't let Paul dream too much about picoseconds.

Well, quite. I wouldn't know what to do with all those extra
harmonics, anyway. :-)
Diodes sound great, but I'm heavily put off by that comment earlier in
the thread that they're prone to self oscillation. :-( What's a
"parametric device" anyway?

In order to achieve results that good, one have to pay very careful
attention to "wiring". There's still a lot of room for Paul to improve his
PCB design skills ;-)

Fair comment. But I *am* working on it!

If you do use diodes for higher-order harmonic generation, and not
just a simple full-wave-rectifier type frequency doubler, I suppose
you want something of the nature of a step recovery diode. That
implies minority carrier stored charge in the diode, and that would
preclude using a Schottky diode (which would work great in the
full-wave-rectifier type doubler). If you get into actually wanting
to generate harmonic combs out to microwave frequencies, it's probably
worthwhile looking for diodes actually characterized for step recovery
service. But I really think that's way beyond what you are trying to
accomplish right now.

My turn to learn something here. Tom, would you elaborate a bit on the above
please? I know SRD's are comb generators out to visible light, but they're
also 50 percent hard to find and 50 percent magic. I've been using
Schottky's for X16 multipliers to 2 GHz, am I doing something wrong? (I keep
promising myself that I'm gonna substitute an MMIC for that one day, I DID
find the "Filter Gain" in the line length from generator to filter), THAT
was both impressive AND helpful. If I go with the MMIC, any preference of
Silicon over GaAs?

Regards

W4ZCB

On Mon, 12 Apr 2004 19:09:51 GMT, "Harold E. Johnson"
wrote:


If you do use diodes for higher-order harmonic generation, and not
just a simple full-wave-rectifier type frequency doubler, I suppose
you want something of the nature of a step recovery diode. That
implies minority carrier stored charge in the diode, and that would
preclude using a Schottky diode (which would work great in the
full-wave-rectifier type doubler). If you get into actually wanting
to generate harmonic combs out to microwave frequencies, it's probably
worthwhile looking for diodes actually characterized for step recovery
service. But I really think that's way beyond what you are trying to
accomplish right now.

My turn to learn something here. Tom, would you elaborate a bit on the above
please? I know SRD's are comb generators out to visible light, but they're
also 50 percent hard to find and 50 percent magic. I've been using
Schottky's for X16 multipliers to 2 GHz, am I doing something wrong? (I keep
promising myself that I'm gonna substitute an MMIC for that one day, I DID
find the "Filter Gain" in the line length from generator to filter), THAT
was both impressive AND helpful. If I go with the MMIC, any preference of
Silicon over GaAs?

Regards

W4ZCB


The only distributor-stock SRDs I know of are the M/Acom MA44767,
MA44768, MA44769 parts, all SOT-23 and dirt cheap. I think Penstock
carries them. The '68 or '69 should be good for multiplication to 2
GHz. For high ratios, an SRD will beat a plain diode by a huge amount.
There are lots of appnotes around about using them as multipliers.

I have a bunch in stock and can send a few to anybody who wants to
play.

John

The only distributor-stock SRDs I know of are the M/Acom MA44767,
MA44768, MA44769 parts, all SOT-23 and dirt cheap. I think Penstock
carries them. The '68 or '69 should be good for multiplication to 2
GHz. For high ratios, an SRD will beat a plain diode by a huge amount.
There are lots of appnotes around about using them as multipliers.

I have a bunch in stock and can send a few to anybody who wants to
play.

I'd like to talk you out of a couple, and thanks for the supplier info.
Wonder what MaCom's min order is? Anything you need in trade or a few green
stamps?

W4ZCB
Call Book Address.