On Fri, 19 Mar 2004 17:46:43 GMT, "Harold E. Johnson"
wrote:


I don't think you can get away
with short cuts at RF, especially if you're a non-expert.

3 MHz is practically DC. Why don't you ask one of your friends at the BBC to
build it for you.

I was told 40Mhz is "practically DC" too. I guess it depends on where
you're coming from.
Actually I've dumped the factory inductor as suggested by Tom and
wound-up a large, air core job on 15mm plastic water pipe. It's made
a big difference. I'm happy to report I've now got the 5th! Could be a
little cleaner but who cares? Tom made a big deal out of the
importance of high-Q so it was the obvious thing to try. Fortunately,
it's worked. God knows how I'm going to squeeze this monster coil onto
the board, though! :-|
Can I infer from this experience that SMD inductors of over a few uH
are a waste of time?
--

The BBC: Licensed at public expense to spread lies.

Harold E. Johnson wrote,


I don't think you can get away
with short cuts at RF, especially if you're a non-expert.

3 MHz is practically DC. Why don't you ask one of your friends at the BBC to
build it for you.

W4ZCB


The BBC couldn't do it unless all the leads were bent to the right.
73,
Tom Donaly, KA6RUH

Harold E. Johnson wrote,


I don't think you can get away
with short cuts at RF, especially if you're a non-expert.

3 MHz is practically DC. Why don't you ask one of your friends at the BBC to
build it for you.

W4ZCB


The BBC couldn't do it unless all the leads were bent to the right.
73,
Tom Donaly, KA6RUH

I think you can get adequate Q if you use a powdered iron toroid. I'm
not going to take the time to look; there are some Q curves on the
Amidon website. But I think that a T-37-2 core might work OK. I
could also have a look at what's possible with a lower Q coil, but
don't have the time right now. You could do that with Spice or with
RFSim99, too. If you lower the "output" capacitor, you will raise the
voltage delivered to the 18MHz gate input, given that the LC is
resonant at the desired output freq.

A note on the effects of inadequate filtering... Of course, the
result is harmonics other than just the fifth in the output. But what
does that mean after you've squared things up by running the signal
through another gate? How does a waveform which spends all its time
at either 5V or 0V have multiple frequencies? It does it by having
different high and low times on each cycle of the five output cycles
that happen for each input cycle. And if you want a pretty pure 15th
harmonic (3rd harmonic of the 5th harmonic), you are probably going to
want the cycles to be uniform, and close to 50% duty cycle. (They
WILL have plenty of third harmonic...the rise and fall times I
observed yesterday on 'HC04s were about 4 nanoseconds, and that will
get you up above 100MHz...) That will take more filtering. If your
inductor Q is only moderate, you still have a chance if you build a
higher order filter, instead of trying to do it with a single pole
(really single pole pair) filter such as I suggested. You can do it
with a single pole but it has to be quite high loaded Q, and the coil
unloaded Q must be higher still. But you can make the multi-pole
filter wider, maybe two or three MHz wide, and get by with lower Q
inductors and still have reasonably low loss and adequate rejection of
the other harmonics.

Yeah, the little moulded SMT inductors aren't very high Q. They are
intended more for power supply filtering and choking, but you can use
them in relatively wideband filters too.

Now that you've seen that it can be done, I hope you'll be inspired to
see what you can do to make it better.

Cheers,
Tom

Paul Burridge wrote in message . ..
On Fri, 19 Mar 2004 17:46:43 GMT, "Harold E. Johnson"
wrote:


I don't think you can get away
with short cuts at RF, especially if you're a non-expert.

3 MHz is practically DC. Why don't you ask one of your friends at the BBC to
build it for you.

I was told 40Mhz is "practically DC" too. I guess it depends on where
you're coming from.
Actually I've dumped the factory inductor as suggested by Tom and
wound-up a large, air core job on 15mm plastic water pipe. It's made
a big difference. I'm happy to report I've now got the 5th! Could be a
little cleaner but who cares? Tom made a big deal out of the
importance of high-Q so it was the obvious thing to try. Fortunately,
it's worked. God knows how I'm going to squeeze this monster coil onto
the board, though! :-|
Can I infer from this experience that SMD inductors of over a few uH
are a waste of time?

I think you can get adequate Q if you use a powdered iron toroid. I'm
not going to take the time to look; there are some Q curves on the
Amidon website. But I think that a T-37-2 core might work OK. I
could also have a look at what's possible with a lower Q coil, but
don't have the time right now. You could do that with Spice or with
RFSim99, too. If you lower the "output" capacitor, you will raise the
voltage delivered to the 18MHz gate input, given that the LC is
resonant at the desired output freq.

A note on the effects of inadequate filtering... Of course, the
result is harmonics other than just the fifth in the output. But what
does that mean after you've squared things up by running the signal
through another gate? How does a waveform which spends all its time
at either 5V or 0V have multiple frequencies? It does it by having
different high and low times on each cycle of the five output cycles
that happen for each input cycle. And if you want a pretty pure 15th
harmonic (3rd harmonic of the 5th harmonic), you are probably going to
want the cycles to be uniform, and close to 50% duty cycle. (They
WILL have plenty of third harmonic...the rise and fall times I
observed yesterday on 'HC04s were about 4 nanoseconds, and that will
get you up above 100MHz...) That will take more filtering. If your
inductor Q is only moderate, you still have a chance if you build a
higher order filter, instead of trying to do it with a single pole
(really single pole pair) filter such as I suggested. You can do it
with a single pole but it has to be quite high loaded Q, and the coil
unloaded Q must be higher still. But you can make the multi-pole
filter wider, maybe two or three MHz wide, and get by with lower Q
inductors and still have reasonably low loss and adequate rejection of
the other harmonics.

Yeah, the little moulded SMT inductors aren't very high Q. They are
intended more for power supply filtering and choking, but you can use
them in relatively wideband filters too.

Now that you've seen that it can be done, I hope you'll be inspired to
see what you can do to make it better.

Cheers,
Tom

Paul Burridge wrote in message . ..
On Fri, 19 Mar 2004 17:46:43 GMT, "Harold E. Johnson"
wrote:


I don't think you can get away
with short cuts at RF, especially if you're a non-expert.

3 MHz is practically DC. Why don't you ask one of your friends at the BBC to
build it for you.

I was told 40Mhz is "practically DC" too. I guess it depends on where
you're coming from.
Actually I've dumped the factory inductor as suggested by Tom and
wound-up a large, air core job on 15mm plastic water pipe. It's made
a big difference. I'm happy to report I've now got the 5th! Could be a
little cleaner but who cares? Tom made a big deal out of the
importance of high-Q so it was the obvious thing to try. Fortunately,
it's worked. God knows how I'm going to squeeze this monster coil onto
the board, though! :-|
Can I infer from this experience that SMD inductors of over a few uH
are a waste of time?

On 19 Mar 2004 23:01:19 -0800, (Tom Bruhns) wrote:

I think you can get adequate Q if you use a powdered iron toroid. I'm
not going to take the time to look; there are some Q curves on the
Amidon website. But I think that a T-37-2 core might work OK. I
could also have a look at what's possible with a lower Q coil, but
don't have the time right now. You could do that with Spice or with
RFSim99, too. If you lower the "output" capacitor, you will raise the
voltage delivered to the 18MHz gate input, given that the LC is
resonant at the desired output freq.

A note on the effects of inadequate filtering... Of course, the
result is harmonics other than just the fifth in the output. But what
does that mean after you've squared things up by running the signal
through another gate? How does a waveform which spends all its time
at either 5V or 0V have multiple frequencies? It does it by having
different high and low times on each cycle of the five output cycles
that happen for each input cycle. And if you want a pretty pure 15th
harmonic (3rd harmonic of the 5th harmonic), you are probably going to
want the cycles to be uniform, and close to 50% duty cycle. (They
WILL have plenty of third harmonic...the rise and fall times I
observed yesterday on 'HC04s were about 4 nanoseconds, and that will
get you up above 100MHz...) That will take more filtering. If your
inductor Q is only moderate, you still have a chance if you build a
higher order filter, instead of trying to do it with a single pole
(really single pole pair) filter such as I suggested. You can do it
with a single pole but it has to be quite high loaded Q, and the coil
unloaded Q must be higher still. But you can make the multi-pole
filter wider, maybe two or three MHz wide, and get by with lower Q
inductors and still have reasonably low loss and adequate rejection of
the other harmonics.

Yeah, the little moulded SMT inductors aren't very high Q. They are
intended more for power supply filtering and choking, but you can use
them in relatively wideband filters too.

Now that you've seen that it can be done, I hope you'll be inspired to
see what you can do to make it better.

Indeed, and thanks, Tom. I shall be following up your concept in the
manner you describe; it's a departure from what I thought would be
suggested when I commenced this thread but it's the outright winner in
my book. There's a lot I don't know about coils - I never expected to
have to say that. But there's a lot more to them than meets the eye,
too so I don't feel too ashamed to admit it. Yes, it's a great
solution and the only drawback is the physical size of the inductors
which can be remedied with a little further development as you say.
That is what I plan to do next.
Thanks again!
--

The BBC: Licensed at public expense to spread lies.

On 19 Mar 2004 23:01:19 -0800, (Tom Bruhns) wrote:

I think you can get adequate Q if you use a powdered iron toroid. I'm
not going to take the time to look; there are some Q curves on the
Amidon website. But I think that a T-37-2 core might work OK. I
could also have a look at what's possible with a lower Q coil, but
don't have the time right now. You could do that with Spice or with
RFSim99, too. If you lower the "output" capacitor, you will raise the
voltage delivered to the 18MHz gate input, given that the LC is
resonant at the desired output freq.

A note on the effects of inadequate filtering... Of course, the
result is harmonics other than just the fifth in the output. But what
does that mean after you've squared things up by running the signal
through another gate? How does a waveform which spends all its time
at either 5V or 0V have multiple frequencies? It does it by having
different high and low times on each cycle of the five output cycles
that happen for each input cycle. And if you want a pretty pure 15th
harmonic (3rd harmonic of the 5th harmonic), you are probably going to
want the cycles to be uniform, and close to 50% duty cycle. (They
WILL have plenty of third harmonic...the rise and fall times I
observed yesterday on 'HC04s were about 4 nanoseconds, and that will
get you up above 100MHz...) That will take more filtering. If your
inductor Q is only moderate, you still have a chance if you build a
higher order filter, instead of trying to do it with a single pole
(really single pole pair) filter such as I suggested. You can do it
with a single pole but it has to be quite high loaded Q, and the coil
unloaded Q must be higher still. But you can make the multi-pole
filter wider, maybe two or three MHz wide, and get by with lower Q
inductors and still have reasonably low loss and adequate rejection of
the other harmonics.

Yeah, the little moulded SMT inductors aren't very high Q. They are
intended more for power supply filtering and choking, but you can use
them in relatively wideband filters too.

Now that you've seen that it can be done, I hope you'll be inspired to
see what you can do to make it better.

Indeed, and thanks, Tom. I shall be following up your concept in the
manner you describe; it's a departure from what I thought would be
suggested when I commenced this thread but it's the outright winner in
my book. There's a lot I don't know about coils - I never expected to
have to say that. But there's a lot more to them than meets the eye,
too so I don't feel too ashamed to admit it. Yes, it's a great
solution and the only drawback is the physical size of the inductors
which can be remedied with a little further development as you say.
That is what I plan to do next.
Thanks again!
--

The BBC: Licensed at public expense to spread lies.

On 19 Mar 2004 23:01:19 -0800, (Tom Bruhns) wrote:


Now that you've seen that it can be done, I hope you'll be inspired to
see what you can do to make it better.

Just a quick thought: how about changing the ratio of the C/L pair to
increase the C so a small coil can be made; keeping the resonant
frequency the same of course. Is there any inherent problem in trying
that approach (I'm sure there is or you would have already suggested
it).

p.
--

The BBC: Licensed at public expense to spread lies.

On 19 Mar 2004 23:01:19 -0800, (Tom Bruhns) wrote:


Now that you've seen that it can be done, I hope you'll be inspired to
see what you can do to make it better.

Just a quick thought: how about changing the ratio of the C/L pair to
increase the C so a small coil can be made; keeping the resonant
frequency the same of course. Is there any inherent problem in trying
that approach (I'm sure there is or you would have already suggested
it).

p.
--

The BBC: Licensed at public expense to spread lies.

Paul,
If you are trying to keep things small, have you considered one of the high
speed versions of the 4046 PLL/OSC? Some manufacturers spec these up to 20
MHz. No coils. If you go with the LC, and you have any fixed capacitors in
parallel with the inductor, use decent capacitors, like mica or RF approved
ceramic. I have seen cheap ceramic caps meant for bypassing just not work in
applications like yours.

Tam
"Paul Burridge" wrote in message
...
On Fri, 19 Mar 2004 17:46:43 GMT, "Harold E. Johnson"
wrote:


I don't think you can get away
with short cuts at RF, especially if you're a non-expert.

3 MHz is practically DC. Why don't you ask one of your friends at the BBC
to
build it for you.

I was told 40Mhz is "practically DC" too. I guess it depends on where
you're coming from.
Actually I've dumped the factory inductor as suggested by Tom and
wound-up a large, air core job on 15mm plastic water pipe. It's made
a big difference. I'm happy to report I've now got the 5th! Could be a
little cleaner but who cares? Tom made a big deal out of the
importance of high-Q so it was the obvious thing to try. Fortunately,
it's worked. God knows how I'm going to squeeze this monster coil onto
the board, though! :-|
Can I infer from this experience that SMD inductors of over a few uH
are a waste of time?
--

The BBC: Licensed at public expense to spread lies.