On Sat, 13 Mar 2004 10:00:52 +1000, Tony wrote:

The 5th harmonic should be only 14dB below the fundamental, although it will
drop fairly quickly as the sides of the input square wave deviate from vertical.

Does the 3.44MHz have a 50% duty cycle?

Okay, I've now tweaked the osc. to get as near to 50% as possible.
Alas, still no sign of any 5th present in the multiplier's output.
Here's a shot of the (fundamental) output from the inverters. I can't
see any real problem with why it shouldn't be good for a reasonable
comb of harmonics, but our experts may know better. BTW, settings were
2V/div. and 0.1uS/div.

http://www.burridge8333.fsbusiness.co.uk/trace.gif

This other trace was snapped at the base of the transistor stage that
does the multiplying. All there is (circuitry-wise) between this trace
and the last one is a 330 ohm series resistor and a 47pF cap. I'd have
expected to see some clamping action due to the b/e junction, but the
waveshape seems very odd - but does concur with the Spice simulation.
Is there anything amiss, here? (Setting here is 0.5V/div)

http://www.burridge8333.fsbusiness.co.uk/trace2.gif

Still not a sniff of a fifth! :-(
Currently building Reg's 17.2Mhz BPF to see if that can expose it.
--

The BBC: Licensed at public expense to spread lies.

Here's the solution. Build a 17.2 one-transistor LC oscillator. It will
easily lock to an injection of the 5th harmonic of 3.44 Mhz at any output
level you like.

Whatever you do don't go for phase locked loops.

Junk your Spice. It tells you nothing you didn't already ought to know. If
you don't know it then you shouldn't be doing the job anyway.

You must have plenty of room on the PCB. But why don't you invest in a 17.2
MHz quartz crystal. They're cheap enough. You've already spent more time
and trouble on research and investigation. Cut your losses.

Redundency is a serious crime. It degrades reliabilty. Reliability is
Quality versus Time. Every individual component must serve a recognisable,
readily understandable purpose. If it's not understood then it shouldn't be
there.

High Quality is adherence to the specified intended purpose.

Now I suppose some bright spark will say deliberate redundency can improve
reliability. But ONLY when all else has failed.
----
Reg.


"Paul Burridge" wrote in message
...
On Sat, 13 Mar 2004 10:00:52 +1000, Tony wrote:

The 5th harmonic should be only 14dB below the fundamental, although it
will
drop fairly quickly as the sides of the input square wave deviate from
vertical.

Does the 3.44MHz have a 50% duty cycle?

Okay, I've now tweaked the osc. to get as near to 50% as possible.
Alas, still no sign of any 5th present in the multiplier's output.
Here's a shot of the (fundamental) output from the inverters. I can't
see any real problem with why it shouldn't be good for a reasonable
comb of harmonics, but our experts may know better. BTW, settings were
2V/div. and 0.1uS/div.

http://www.burridge8333.fsbusiness.co.uk/trace.gif

This other trace was snapped at the base of the transistor stage that
does the multiplying. All there is (circuitry-wise) between this trace
and the last one is a 330 ohm series resistor and a 47pF cap. I'd have
expected to see some clamping action due to the b/e junction, but the
waveshape seems very odd - but does concur with the Spice simulation.
Is there anything amiss, here? (Setting here is 0.5V/div)

http://www.burridge8333.fsbusiness.co.uk/trace2.gif

Still not a sniff of a fifth! :-(
Currently building Reg's 17.2Mhz BPF to see if that can expose it.
--

The BBC: Licensed at public expense to spread lies.

On Sat, 13 Mar 2004 17:45:23 +0000 (UTC), "Reg Edwards"
wrote:

Here's the solution. Build a 17.2 one-transistor LC oscillator. It will
easily lock to an injection of the 5th harmonic of 3.44 Mhz at any output
level you like.

Whatever you do don't go for phase locked loops.

Junk your Spice. It tells you nothing you didn't already ought to know. If
you don't know it then you shouldn't be doing the job anyway.

You must have plenty of room on the PCB. But why don't you invest in a 17.2
MHz quartz crystal. They're cheap enough. You've already spent more time
and trouble on research and investigation. Cut your losses.

Hi Reg,

I'm not into cutting my losses and taking the easy way out; I wouldn't
*learn* anything by doing so. This circuit *ought* to work and I hope
to find out why it doesn't and remedy the situation.
I can't build a 17.2Mhz osc. I need to start much lower to multiply up
the 'pullability' of the fundamental to the best part of half a Meg.
Try doing that with a quartz xtal at 17Mhz! Sorry, I mean no offence I
know you're unaware of the background to this project.
BTW, I've built the filter your s/ware designed and it works pretty
much as advertised - bang on 17.2Mhz pass[1] and the third harmonic of
the fundamental at just over 10Mhz is *well* down by a factor of 80X
(sorry, can't be arsed to work out the dB equivalent). The coils I
used to build it were designed with one of your other programs, BTW!
Many thanks indeed. I urge anyone else reading this in r.r.a.hb to
grab Reg's programs whilst they can; they're *extremely* useful and
Reg deserves a vote of thanks from us all for taking the time and
trouble to write them.

[1] Significant spurious pass response at 13.2Mhz for some reason
(probably my use of inappropriate capacitors for the sake of
expediency - Spice didn't predict it) fortunately it just misses the
fourth harmonic!
--

The BBC: Licensed at public expense to spread lies.

On Sat, 13 Mar 2004 17:45:23 +0000 (UTC), "Reg Edwards"
wrote:

Here's the solution. Build a 17.2 one-transistor LC oscillator. It will
easily lock to an injection of the 5th harmonic of 3.44 Mhz at any output
level you like.

Whatever you do don't go for phase locked loops.

Junk your Spice. It tells you nothing you didn't already ought to know. If
you don't know it then you shouldn't be doing the job anyway.

You must have plenty of room on the PCB. But why don't you invest in a 17.2
MHz quartz crystal. They're cheap enough. You've already spent more time
and trouble on research and investigation. Cut your losses.

Hi Reg,

I'm not into cutting my losses and taking the easy way out; I wouldn't
*learn* anything by doing so. This circuit *ought* to work and I hope
to find out why it doesn't and remedy the situation.
I can't build a 17.2Mhz osc. I need to start much lower to multiply up
the 'pullability' of the fundamental to the best part of half a Meg.
Try doing that with a quartz xtal at 17Mhz! Sorry, I mean no offence I
know you're unaware of the background to this project.
BTW, I've built the filter your s/ware designed and it works pretty
much as advertised - bang on 17.2Mhz pass[1] and the third harmonic of
the fundamental at just over 10Mhz is *well* down by a factor of 80X
(sorry, can't be arsed to work out the dB equivalent). The coils I
used to build it were designed with one of your other programs, BTW!
Many thanks indeed. I urge anyone else reading this in r.r.a.hb to
grab Reg's programs whilst they can; they're *extremely* useful and
Reg deserves a vote of thanks from us all for taking the time and
trouble to write them.

[1] Significant spurious pass response at 13.2Mhz for some reason
(probably my use of inappropriate capacitors for the sake of
expediency - Spice didn't predict it) fortunately it just misses the
fourth harmonic!
--

The BBC: Licensed at public expense to spread lies.

On Sat, 13 Mar 2004 16:26:10 +0000, Paul Burridge
wrote:


Okay, I've now tweaked the osc. to get as near to 50% as possible.
Alas, still no sign of any 5th present in the multiplier's output.
Here's a shot of the (fundamental) output from the inverters. I can't
see any real problem with why it shouldn't be good for a reasonable
comb of harmonics, but our experts may know better. BTW, settings were
2V/div. and 0.1uS/div.

http://www.burridge8333.fsbusiness.co.uk/trace.gif



That waveform *has* bunches of 5th harmonic. All you need is a
properly functioning bandpass filter to pluck it out.

John

On Sat, 13 Mar 2004 10:37:13 -0800, John Larkin
wrote:

On Sat, 13 Mar 2004 16:26:10 +0000, Paul Burridge
wrote:


Okay, I've now tweaked the osc. to get as near to 50% as possible.
Alas, still no sign of any 5th present in the multiplier's output.
Here's a shot of the (fundamental) output from the inverters. I can't
see any real problem with why it shouldn't be good for a reasonable
comb of harmonics, but our experts may know better. BTW, settings were
2V/div. and 0.1uS/div.

http://www.burridge8333.fsbusiness.co.uk/trace.gif



That waveform *has* bunches of 5th harmonic. All you need is a
properly functioning bandpass filter to pluck it out.

You must have bloody good eyesight, John! :-)
BTW, can you recommend a sub nS Schmitt inverter that's easily
obtainable?
--

The BBC: Licensed at public expense to spread lies.

On Sat, 13 Mar 2004 19:38:47 +0000, Paul Burridge
wrote:

On Sat, 13 Mar 2004 10:37:13 -0800, John Larkin
wrote:

On Sat, 13 Mar 2004 16:26:10 +0000, Paul Burridge
wrote:


Okay, I've now tweaked the osc. to get as near to 50% as possible.
Alas, still no sign of any 5th present in the multiplier's output.
Here's a shot of the (fundamental) output from the inverters. I can't
see any real problem with why it shouldn't be good for a reasonable
comb of harmonics, but our experts may know better. BTW, settings were
2V/div. and 0.1uS/div.

http://www.burridge8333.fsbusiness.co.uk/trace.gif



That waveform *has* bunches of 5th harmonic. All you need is a
properly functioning bandpass filter to pluck it out.

You must have bloody good eyesight, John! :-)

You can count the graticule lines fairly well; it's close to 50%, and
the edges are fairly fast. And no, in fact I have terrible eyesight.

BTW, can you recommend a sub nS Schmitt inverter that's easily
obtainable?

I don't know of any really fast Schmitts. An HC14 followed by an AC04
should have fast edges. My favorite thing like this is an OnSemi
NL37WZ16 with all three sections in parallel. Powered from +6 or so,
it puts 5 volts into 50 ohms in something like 750 ps.

The old original RCA AC-series parts were sub-ns - crude and rude,
they were - but some ACs are now a little slower to reduce ground
bounce.

Most of the LVDS-to-TTL LVDS line receivers make damned fine
comparators with sub-ns output edges.

For screaming edges, there's always the step-recovery diode, or a
medium-power gaasfet like the CLY2.

John

On Sat, 13 Mar 2004 13:31:26 -0800, John Larkin
wrote:

I don't know of any really fast Schmitts. An HC14 followed by an AC04
should have fast edges. My favorite thing like this is an OnSemi
NL37WZ16 with all three sections in parallel. Powered from +6 or so,
it puts 5 volts into 50 ohms in something like 750 ps.

The old original RCA AC-series parts were sub-ns - crude and rude,
they were - but some ACs are now a little slower to reduce ground
bounce.

Most of the LVDS-to-TTL LVDS line receivers make damned fine
comparators with sub-ns output edges.

For screaming edges, there's always the step-recovery diode, or a
medium-power gaasfet like the CLY2.

Thanks, John.
I don't see the need for anything super-fast in this instance (I mean
- 3 to4 Mhz for God's sake) but was just curious as to what they use
in UHF and beyond...
2morrow I'm going to stick Reg's 17.2Mhz BPF in line and see if that
kills the 3rd enough to allow the 5th to thrive.
--

The BBC: Licensed at public expense to spread lies.

On Sat, 13 Mar 2004 13:31:26 -0800, John Larkin
wrote:

I don't know of any really fast Schmitts. An HC14 followed by an AC04
should have fast edges. My favorite thing like this is an OnSemi
NL37WZ16 with all three sections in parallel. Powered from +6 or so,
it puts 5 volts into 50 ohms in something like 750 ps.

The old original RCA AC-series parts were sub-ns - crude and rude,
they were - but some ACs are now a little slower to reduce ground
bounce.

Most of the LVDS-to-TTL LVDS line receivers make damned fine
comparators with sub-ns output edges.

For screaming edges, there's always the step-recovery diode, or a
medium-power gaasfet like the CLY2.

Thanks, John.
I don't see the need for anything super-fast in this instance (I mean
- 3 to4 Mhz for God's sake) but was just curious as to what they use
in UHF and beyond...
2morrow I'm going to stick Reg's 17.2Mhz BPF in line and see if that
kills the 3rd enough to allow the 5th to thrive.
--

The BBC: Licensed at public expense to spread lies.

On Sat, 13 Mar 2004 19:38:47 +0000, Paul Burridge
wrote:

On Sat, 13 Mar 2004 10:37:13 -0800, John Larkin
wrote:

On Sat, 13 Mar 2004 16:26:10 +0000, Paul Burridge
wrote:


Okay, I've now tweaked the osc. to get as near to 50% as possible.
Alas, still no sign of any 5th present in the multiplier's output.
Here's a shot of the (fundamental) output from the inverters. I can't
see any real problem with why it shouldn't be good for a reasonable
comb of harmonics, but our experts may know better. BTW, settings were
2V/div. and 0.1uS/div.

http://www.burridge8333.fsbusiness.co.uk/trace.gif



That waveform *has* bunches of 5th harmonic. All you need is a
properly functioning bandpass filter to pluck it out.

You must have bloody good eyesight, John! :-)

You can count the graticule lines fairly well; it's close to 50%, and
the edges are fairly fast. And no, in fact I have terrible eyesight.

BTW, can you recommend a sub nS Schmitt inverter that's easily
obtainable?

I don't know of any really fast Schmitts. An HC14 followed by an AC04
should have fast edges. My favorite thing like this is an OnSemi
NL37WZ16 with all three sections in parallel. Powered from +6 or so,
it puts 5 volts into 50 ohms in something like 750 ps.

The old original RCA AC-series parts were sub-ns - crude and rude,
they were - but some ACs are now a little slower to reduce ground
bounce.

Most of the LVDS-to-TTL LVDS line receivers make damned fine
comparators with sub-ns output edges.

For screaming edges, there's always the step-recovery diode, or a
medium-power gaasfet like the CLY2.

John