Multiplier chains
 

Hello all,
I am trying to muliply a 72MHz crystal oscillator to 1296 MHz. My
first approach was to just make a series of X2 or X3 transistor
mulipliers to get the require X18 multiplication fact. However, I have
found two pieces of equipment in the junk box which just use diode
multipliers. One is an old Electrophone UHF CB radio which multiplies
the crystal oscillator by 17 (I think, from memory)then uses helical
resonators to filter the desired harmonic. The same approach is used in
an old King aircraft transponder where the 138 MHz crystal is multiplied
to 960 MHz with just a diode and uses the first two stages of an
interdigital mixer to get rid of unwanted harmonics.
So, the question is, which is the better approach? I just want a
reasonably clean signal source to test a 1296 MHz down-converter.
The diode approach seems simpler but is it likely to contain more
spurious signals than a transistor multiplier chain?
Alan
VK2ADB

Multiplier chains
 

On Wed, 21 May 2008, Alan Peake wrote:

Hello all,
I am trying to muliply a 72MHz crystal oscillator to 1296 MHz. My first
approach was to just make a series of X2 or X3 transistor mulipliers to get
the require X18 multiplication fact. However, I have found two pieces of
equipment in the junk box which just use diode multipliers. One is an old
Electrophone UHF CB radio which multiplies the crystal oscillator by 17 (I
think, from memory)then uses helical resonators to filter the desired
harmonic. The same approach is used in an old King aircraft transponder where
the 138 MHz crystal is multiplied to 960 MHz with just a diode and uses the
first two stages of an interdigital mixer to get rid of unwanted harmonics.
So, the question is, which is the better approach? I just want a reasonably
clean signal source to test a 1296 MHz down-converter.
The diode approach seems simpler but is it likely to contain more spurious
signals than a transistor multiplier chain?
Alan
VK2ADB


I don't know, and one thing to remember is that what was done years ago
may no longer be the solution because other things have come along.

A single stage of multiplication is of course simplest. But, if you
do it in one step, the signal may be so weak that you need stages of
amplification at the ultimate frequency. Once upon a time, frequency
limits may have made that unfeasible.

Also, if you have one stage that basically generates harmonics, and
then you expect to pick off the desired frequency, that filtering
may need to be much better than multiple stages. If you start
with a low enough crystal frequency, the next harmonic may be too close
and some of it will get through the filter on the ultimate frequency.
If you have a string of multipliers, each does filtering so the next
stage only has to deal with filtering out a relatively high frequency.

Note that your two examples aren't comparable. The first example you
say multiplies by 17, while the second only multiplies by 6. Even in
the old days, it wasn't uncommon to see a jump like 6, but something
like 17 was less common.

The real trick seems to be to start with as high a frequency as possible.
Then the multiplication needed is limited, and it's far easier to filter
out harmonics from a higher frequency crystal than a lower one.

Michael VE2BVW

Multiplier chains
 

On Wed, 21 May 2008 17:00:40 +1000, Alan Peake wrote:
Hello all,
I am trying to muliply a 72MHz crystal oscillator to 1296 MHz. My
first approach was to just make a series of X2 or X3 transistor
mulipliers to get the require X18 multiplication fact. However, I have
found two pieces of equipment in the junk box which just use diode
multipliers. One is an old Electrophone UHF CB radio which multiplies
the crystal oscillator by 17 (I think, from memory)then uses helical
resonators to filter the desired harmonic. The same approach is used in
an old King aircraft transponder where the 138 MHz crystal is multiplied
to 960 MHz with just a diode and uses the first two stages of an
interdigital mixer to get rid of unwanted harmonics.

So, the question is, which is the better approach?

Are you building tens of thousands like Electrophone and King did?
There was more than 'good' engineering involved with those boxes.
The bean counters were all over it, too -- I'm sure.

Jonesy
--
Marvin L Jones | jonz | W3DHJ | linux
38.24N 104.55W | @ config.com | Jonesy | OS/2
*** Killfiling google posts: http://jonz.net/ng.htm

Multiplier chains
 

On Wed, 21 May 2008 17:00:40 +1000, Alan Peake
wrote:

However, I have
found two pieces of equipment in the junk box which just use diode
multipliers. One is an old Electrophone UHF CB radio which multiplies
the crystal oscillator by 17 (I think, from memory)then uses helical
resonators to filter the desired harmonic.


Sounds like a step recovery diode (SRD) multiplier, which generates a
very broad spectrum. In extreme multiplication situations a tunable
harmonic analyser could be used to select out a single harmonic from
the broad SRD spectrum (mix down with a VFO, filter with a crystal
filter with bandwidth comparable to the fundamental and mix up the
filtered harmonic up to the original frequency using the _same_ VFO,
effectively canceling out any VFO drift).

However, since there have been prescalers available for quite a while
for satellite-TV indoor units, which are capable of at least 2 GHz,
why not use a PLL. Put a VCO at 1296 MHz, use a divide it by 64 using
the prescaler and phase lock it to a 20.25 MHz crystal. With such high
reference frequency, the loop filter can be quite broad, greatly
attenuating the VCO phase noise.

Paul OH3LWR

Multiplier chains
 

You could consider another approach -- phase locking a 1296 MHz
oscillator to your 72 MHz xtal.

Dr. G.


Alan Peake wrote in news:4833C898.2090302
@nosspam.activ8.net.au:

Hello all,
I am trying to muliply a 72MHz crystal oscillator to 1296 MHz. My
first approach was to just make a series of X2 or X3 transistor
mulipliers to get the require X18 multiplication fact. However, I have
found two pieces of equipment in the junk box which just use diode
multipliers. One is an old Electrophone UHF CB radio which multiplies
the crystal oscillator by 17 (I think, from memory)then uses helical
resonators to filter the desired harmonic. The same approach is used in
an old King aircraft transponder where the 138 MHz crystal is
multiplied
to 960 MHz with just a diode and uses the first two stages of an
interdigital mixer to get rid of unwanted harmonics.
So, the question is, which is the better approach? I just want a
reasonably clean signal source to test a 1296 MHz down-converter.
The diode approach seems simpler but is it likely to contain more
spurious signals than a transistor multiplier chain?
Alan
VK2ADB

Multiplier chains
 

Is there some magic to using the 72 MHz oscillator? Gigahertz synthesizers
are cheap and simple these days. Back in the bad old days multiplication
was the cheap and easy way to get stable high frequencies. Today, that
helical resonator would probably buy the synthesizer chip and associated
support circuitry, and you needn't fiddle around with stage after stage.

...

"Alan Peake" wrote in message
...
Hello all,
I am trying to muliply a 72MHz crystal oscillator to 1296 MHz. My
first approach was to just make a series of X2 or X3 transistor mulipliers
to get the require X18 multiplication fact. However, I have found two
pieces of equipment in the junk box which just use diode multipliers. One
is an old Electrophone UHF CB radio which multiplies the crystal
oscillator by 17 (I think, from memory)then uses helical resonators to
filter the desired harmonic. The same approach is used in an old King
aircraft transponder where the 138 MHz crystal is multiplied to 960 MHz
with just a diode and uses the first two stages of an interdigital mixer
to get rid of unwanted harmonics.
So, the question is, which is the better approach? I just want a
reasonably clean signal source to test a 1296 MHz down-converter.
The diode approach seems simpler but is it likely to contain more spurious
signals than a transistor multiplier chain?
Alan
VK2ADB

Multiplier chains
 

xpyttl wrote:
Is there some magic to using the 72 MHz oscillator? Gigahertz synthesizers
are cheap and simple these days. Back in the bad old days multiplication
was the cheap and easy way to get stable high frequencies. Today, that
helical resonator would probably buy the synthesizer chip and associated
support circuitry, and you needn't fiddle around with stage after stage.

I just wanted to do it with existing bits.
Alan

Multiplier chains
 

Michael Black wrote:

I don't know, and one thing to remember is that what was done years ago
may no longer be the solution because other things have come along.

Yes, as others have pointed out, a VCO and PLL is a better solution.

A single stage of multiplication is of course simplest. But, if you
do it in one step, the signal may be so weak that you need stages of
amplification at the ultimate frequency. Once upon a time, frequency
limits may have made that unfeasible.
I have a supply of MAR-1s that should do the trick there.


Also, if you have one stage that basically generates harmonics, and
then you expect to pick off the desired frequency, that filtering
may need to be much better than multiple stages. If you start
with a low enough crystal frequency, the next harmonic may be too close
and some of it will get through the filter on the ultimate frequency.
If you have a string of multipliers, each does filtering so the next
stage only has to deal with filtering out a relatively high frequency.

I wouldn't try a diode to go from 72 to 1296 but will double or triple
the oscillator and perhaps try the diode from there.

Note that your two examples aren't comparable. The first example you
say multiplies by 17, while the second only multiplies by 6. Even in
the old days, it wasn't uncommon to see a jump like 6, but something
like 17 was less common.
Sorry, it was a factor of 9, not 17. I just found the circuit:) The xtal
was 17 MHz.


The real trick seems to be to start with as high a frequency as possible.
Then the multiplication needed is limited, and it's far easier to filter
out harmonics from a higher frequency crystal than a lower one.



Michael VE2BVW
Thanks for your thoughts.
Alan

Multiplier chains
 

Allodoxaphobia wrote:

Are you building tens of thousands like Electrophone and King did?
There was more than 'good' engineering involved with those boxes.
The bean counters were all over it, too -- I'm sure.

Jonesy

No, this is a "one-off" from my junk box.
Alan

Multiplier chains
 

Paul Keinanen wrote:

Sounds like a step recovery diode (SRD) multiplier, which generates a
very broad spectrum.
......
I have a couple of SRDs but they seem to be optimised for 500MHz input
and I don't know if they would work at lower frequencies. I would assume
that the diodes I mentioned are some sort of SRD or "snap" diode so I
might pull one out and try it.

Paul OH3LWR

Multiplier chains
 

"Alan Peake" wrote in message
...
Hello all,
I am trying to muliply a 72MHz crystal oscillator to 1296 MHz. My
first approach was to just make a series of X2 or X3 transistor mulipliers
to get the require X18 multiplication fact. However, I have found two
pieces of equipment in the junk box which just use diode multipliers. One
is an old Electrophone UHF CB radio which multiplies the crystal
oscillator by 17 (I think, from memory)then uses helical resonators to
filter the desired harmonic. The same approach is used in an old King
aircraft transponder where the 138 MHz crystal is multiplied to 960 MHz
with just a diode and uses the first two stages of an interdigital mixer
to get rid of unwanted harmonics.
So, the question is, which is the better approach? I just want a
reasonably clean signal source to test a 1296 MHz down-converter.
The diode approach seems simpler but is it likely to contain more spurious
signals than a transistor multiplier chain?
Alan
VK2ADB

Here's an interesting simple circuit to generate odd harmonics.
"New Topology Multiplier Generates Odd Harmonics"
http://www.wenzel.com/pdffiles1/pdfs/RFDesign2.pdf
Mike

Multiplier chains
 

amdx wrote:

Here's an interesting simple circuit to generate odd harmonics.
"New Topology Multiplier Generates Odd Harmonics"
http://www.wenzel.com/pdffiles1/pdfs/RFDesign2.pdf
Mike
Yeah, saw that. Trouble is that I need EVEN harmonics :)
Anyway, there was some interesting info on snap diodes etc. Makes me
think that the diode used in the gear I have, aren't SRDs as these seem
to be mainly for microwave use. The other info I gleaned was that it's
easier to get low phase noise with straight multiplication than it is
with PLLs. Not that I have an urgent need for low phase noise - but it's
interesting nonetheless.
Alan

Multiplier chains
 

In message , Alan Peake
writes


amdx wrote:

Here's an interesting simple circuit to generate odd harmonics.
"New Topology Multiplier Generates Odd Harmonics"
http://www.wenzel.com/pdffiles1/pdfs/RFDesign2.pdf
Mike
Yeah, saw that. Trouble is that I need EVEN harmonics :)

If you want EVEN harmonics, then use the 'fullwave rectifier' circuit.
This is essentially identical to what you use to rectify the mains, but
it works at RF. One version is the simple bi-phase circuit, consisting
of a ferrite input transformer (ratio 1 : 1+1) and two rectifying
diodes. An alternative is the fullwave bridge version.

I have used the former with great success to multiply by 2, 4 and even 6
but, of course, you get a lot less with the higher order harmonics.
Also, the relative suppression of the even harmonics is less.

Anyway, there was some interesting info on snap diodes etc. Makes me
think that the diode used in the gear I have, aren't SRDs as these seem
to be mainly for microwave use.

I did get involved with a comb generator which used a single step
recovery diode, driven at 6, 7 or 8MHz, to generate fairly equal
amplitude harmonics (both odds and evens) to around 600MHz.

The other info I gleaned was that it's easier to get low phase noise
with straight multiplication than it is with PLLs. Not that I have an
urgent need for low phase noise - but it's interesting nonetheless.
Alan

I didn't know that the "New Topology Multiplier Generates Odd Harmonics"
circuit existed. It's every bit as simple as the fullwave rectifier
circuit for generating even harmonic!
--
Ian

Multiplier chains
 

On Thu, 22 May 2008 22:49:01 +1000, Alan Peake
wrote:

Yeah, saw that. Trouble is that I need EVEN harmonics :)

A push-pull multiplier will suppress most odd harmonics.

Paul OH3LWR

Multiplier chains
 

On May 22, 5:49 am, Alan Peake wrote:
amdx wrote:
Here's an interesting simple circuit to generate odd harmonics.
"New Topology Multiplier Generates Odd Harmonics"
http://www.wenzel.com/pdffiles1/pdfs/RFDesign2.pdf
Mike

Yeah, saw that. Trouble is that I need EVEN harmonics :)
Anyway, there was some interesting info on snap diodes etc. Makes me
think that the diode used in the gear I have, aren't SRDs as these seem
to be mainly for microwave use. The other info I gleaned was that it's
easier to get low phase noise with straight multiplication than it is
with PLLs. Not that I have an urgent need for low phase noise - but it's
interesting nonetheless.
Alan

Huh? I thought you wanted to multiply by 18. 18 = 3 * 3 * 2.

If you were specific about what you want to do with the output, I
missed it. If it's just low-level stuff (e.g., a signal to feed into
the receiver input), you don't need much power and the multiplier can
be very inefficient and still do what you want. As you note, phase
noise will be better (possibly much better) than other ways of doing
it, and it could be that the signal level will be a lot closer to what
you want. You'll have to shield things pretty carefully to get down
to microvolt levels from a PLL or other full-blown oscillator on
1.2GHz. There might already be enough 18th harmonic in the oscillator
output to do what you need. I have some little very fast CMOS single
gate chips that have square enough edges (under half a nanosecond rise
time, unloaded) to generate quite a bit of harmonic content up that
high. Though a PLL would work fine, you then need a way to set up the
PLL chip, assuming you use one of the readily available programmable
ones. I know for me, I could hack a decent analog multiplier with
filters a lot quicker than I could a decent synthesizer. I had to do
a x9 not long ago, and used a multiplication scheme. Also, I've seen
small crystal oscillator modules from two different vendors that used
a PLL to lock a 100MHz crystal to a lower frequency reference, and
then used analog multiplication up to several times 100MHz--they are
doing it for low phase noise.

Cheers,
Tom

Multiplier chains
 

K7ITM wrote:

Yeah, saw that. Trouble is that I need EVEN harmonics :)


Huh? I thought you wanted to multiply by 18. 18 = 3 * 3 * 2.
Quite right. 18 is an even harmonic of course but I forgot about the
3*3*2 bit :)


If you were specific about what you want to do with the output, I
missed it.

It's to test a 1296 down-converter basically but will try a CW
transmitter later so a bit of grunt wouldn't go astray.

..... There might already be enough 18th harmonic in the oscillator
output to do what you need.

Could be. AT the moment I have a 2N5485 as the osc and a 2N918 as a
doubler but the 2N918 runs out of puff well before 1296MHz. I have some
BFR91s so I might put them into use.

Cheers,
Tom
73 de Alan

Multiplier chains
 

On Fri, 23 May 2008, Alan Peake wrote:



K7ITM wrote:

Yeah, saw that. Trouble is that I need EVEN harmonics :)


Huh? I thought you wanted to multiply by 18. 18 = 3 * 3 * 2.
Quite right. 18 is an even harmonic of course but I forgot about the 3*3*2
bit :)


If you were specific about what you want to do with the output, I
missed it.

It's to test a 1296 down-converter basically but will try a CW transmitter
later so a bit of grunt wouldn't go astray.

But unless you need a signal to tune the converter to the right frequency,
this is quite a bit simpler. So long as the converter is tuned to the
right general frequency, merely something with good harmonics will do.
The fact that they will be attenuated at the desired frequency is not
a bad thing, since you won't want a strong signal for tuning the converter
anyway.

Your original post didn't give an end use, but it was easy to assume
you needed a clean signal for a transmitter of local oscillator injection
in a converter. You don't want unwanted signals there.

But for a signal generator, having extra harmonics isn't really a problem
(unless you can't be sure that your converter is tuned to the right
frequency). The extra harmonics will just be ignored by the converter's
tuned circuits.

There was a time when a lot of cheap signal generators actually used
harmonics for their upper ranges.

Michael VE2BVW