Re: DOXCO from Z3801
There are lots of newer, high performance prescalers available too, for
the much higher frequencies - especially in GaAs and other technologies.
Check this out:
https://www.analog.com/media/en/technical-documentation/data-sheets/hmc493.pdf
Note that the architecture is basically the same as in the oldies. With
the right combination of prescaler stages and baseband counting, you
could conceivably build yourself a fairly decent 18 GHz CW microwave
counter. Modern parts that are in production usually have evaluation
boards available, so putting something together isn't too hard (but
costs extra for convenient packaging).
Ed
Thanks all for the responses, given me a lot to think about.
Regards,
Dave
On 12/05/2023 18:13, ed breya via time-nuts wrote:
There are lots of newer, high performance prescalers available too, for
the much higher frequencies - especially in GaAs and other technologies.
Check this out:
https://www.analog.com/media/en/technical-documentation/data-sheets/hmc493.pdf
Note that the architecture is basically the same as in the oldies. With
the right combination of prescaler stages and baseband counting, you
could conceivably build yourself a fairly decent 18 GHz CW microwave
counter. Modern parts that are in production usually have evaluation
boards available, so putting something together isn't too hard (but
costs extra for convenient packaging).
Ed
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe send an email to time-nuts-leave@lists.febo.com
I worked for the HP Santa Clara Division frequency counter
R&D section in the mid-1980's. Although I was not personally
involved in designing microwave counters, I did pick up
a considerable amount of knowledge about them. The idea
that a simple frequency divider could replace the
traditional harmonic sampler architecture used in microwave
counters is way too simplistic. Especially if that divider
is the dynamic variety, which is the only kind available
on the open market, AFAIK. Trying to take the entire LF to
18 GHz spectrum, and somehow process it as a single
broadband signal without filtering, and then trigger a
flip flop is not realistic, as a general purpose piece of
test equipment, when there is noise and spurs that have
to be tolerated. The HP 5340, circa 1970, actually
achieved a single coaxial input that could measure
any frequency from 10 Hz to 18 GHz, automatically.
I would venture to say that this 50+ year old technology
could still handily beat any prescaler IC today. The
project manager of the 5340 happened to sit in the
cubicle next to me.
Rick N6RK
On 5/12/2023 10:13 AM, ed breya via time-nuts wrote:
There are lots of newer, high performance prescalers available too, for
the much higher frequencies - especially in GaAs and other technologies.
Check this out:
https://www.analog.com/media/en/technical-documentation/data-sheets/hmc493.pdf
Note that the architecture is basically the same as in the oldies. With
the right combination of prescaler stages and baseband counting, you
could conceivably build yourself a fairly decent 18 GHz CW microwave
counter. Modern parts that are in production usually have evaluation
boards available, so putting something together isn't too hard (but
costs extra for convenient packaging).
Ed
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe send an email to time-nuts-leave@lists.febo.com
Yes of course this simple kind of arrangement can't handle complicated
signals with modulation, lots of spurs or pulsed bursts and such, like a
commercial product. As I recall, the 5340 can discriminate between (pick
the largest of) spurs only a few dB apart. BTW I happen to be trying to
fix a 5340, and need some help figuring out the adjustment process. More
on that later.
A prescaler should be OK with a CW signal that's totally dominant, and
within the acceptable power range. It's just another divider that's very
fast. If you've got something like an oscillator or generator signal to
count, it should be good to go.
Ed
Rick, I don't see how you have the impression that any of these
prescalers are "dynamic" dividers. I remember those dynamic divider
circuits used in the old days to make 1 MHz and 100 kHz from 5 MHz, or
the more modern kind based on mixers, filters, and amplifiers, to say
make 5 from 10 Mhz, but I'm pretty sure I have never seen a
VHF/UHF/microwave prescaler IC that was not completely static - they are
chains of flip-flops and logic, pure and simple. The way they are
typically used may hide their nature a bit, but they are fundamentally
static counting machines.
Perhaps we have different definitions for dynamic vs static. My
definition of static is that there's a unique, stable machine state for
each possible count value, and you can stop it at any count value and
restart it, with the right logic activity, over any period of time. The
"right logic activity" would be to take logical control of the inputs
rather than set them up for counting RF directly in an analog
self-biased mode. In other words, you could put a gate in front of the
input to control the counting.
For instance, in the data sheet for the HMC493 it says in the General
Description:
"The HMC493LP3 & HMC493LP3E are low noise
Divide-by-4 Static Dividers utilizing InGaP GaAs HBT
technology packaged in leadless 3x3 mm QFN sur-
face mount plastic packages. This device operates
from DC (with a square wave input) to 18 GHz input
frequency from a single +5V DC supply."
The part about operating at DC "with a square wave input" means at the
normal logic levels of the device, I'm pretty sure. And usually there's
a minimum toggle edge speed for the various logic types, so there is a
little bit of dynamic-ness there, if you were to try to count at very
low rates.
Ed
On 5/12/2023 6:00 PM, ed breya via time-nuts wrote:
Rick, I don't see how you have the impression that any of these
prescalers are "dynamic" dividers. \
That is what the project manager of the HP 5386 told me back in
the 80's. He used a static divider made by HP Santa Rosa. The
5334A used a dynamic divider made by HP Santa Clara. As the
5334B project manager, I tried whatever COTS dividers I could
find and they all had the same problems as the divider made
in Santa Clara. I designed that one out to save $100.
You previously said something like any prescaler will work if
the signal is coming from a signal generator. This notion is
contradicted by the HP 8660. For both the 5334A and 5334B,
the test procedure for the C channel is to drive the counter
with an 8660 at 1.3 GHz to verify the specified bandwidth.
However, the TP goes on to say that a 1 GHz high pass filter
must be inserted in front of the counter. And sure enough,
if you don't do this, it won't count 1.3 GHz. Admittedly,
the 8660 was not exactly what you would call HP's "finest
hour." I suspect you could dispense with the filter if you
used an 8662/3.
I was not aware of the HMC493, which seems to be static. With
that chip you might finally be able to homebrew a decent C channel.
The description of the technology used sounds like something
that would come out of the Santa Rosa IC fab. The price is a
bargain IMHO at $35. For $425 you can get an eval board.
So it looks like the technology has advanced somewhat in
35 years.
Rick
I remember those dynamic divider
circuits used in the old days to make 1 MHz and 100 kHz from 5 MHz, or
the more modern kind based on mixers, filters, and amplifiers, to say
make 5 from 10 Mhz, but I'm pretty sure I have never seen a
VHF/UHF/microwave prescaler IC that was not completely static - they are
chains of flip-flops and logic, pure and simple. The way they are
typically used may hide their nature a bit, but they are fundamentally
static counting machines.
Perhaps we have different definitions for dynamic vs static. My
definition of static is that there's a unique, stable machine state for
each possible count value, and you can stop it at any count value and
restart it, with the right logic activity, over any period of time. The
"right logic activity" would be to take logical control of the inputs
rather than set them up for counting RF directly in an analog
self-biased mode. In other words, you could put a gate in front of the
input to control the counting.
For instance, in the data sheet for the HMC493 it says in the General
Description:
"The HMC493LP3 & HMC493LP3E are low noise
Divide-by-4 Static Dividers utilizing InGaP GaAs HBT
technology packaged in leadless 3x3 mm QFN sur-
face mount plastic packages. This device operates
from DC (with a square wave input) to 18 GHz input
frequency from a single +5V DC supply."
The part about operating at DC "with a square wave input" means at the
normal logic levels of the device, I'm pretty sure. And usually there's
a minimum toggle edge speed for the various logic types, so there is a
little bit of dynamic-ness there, if you were to try to count at very
low rates.
Ed
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe send an email to time-nuts-leave@lists.febo.com
Am 2023-05-13 3:00, schrieb ed breya via time-nuts:
"The HMC493LP3 & HMC493LP3E are low noise
Divide-by-4 Static Dividers utilizing InGaP GaAs HBT
I have built a 1/8 prescaler using the HMC363 to extend the trigger
bandwidth of my 54750A scope from 2.5 to 12 GHz (or the SR620).
It was necessary to have a gain block in front of the prescaler
since the HMC363 pumped some of the divided signal back into its input
which contaminated the measured channel with some jitter.
A second gain block at the output makes sure that there is enough
signal left even after a power divider.
The home-etched board is still without the preamplifier.
Oscillation tendency is most easily cured by a small DC offset.
That costs some sensitivity, another reason for the preamp.
The design has been published in DUBUS last year. I'll put it
on my web site when the FTP access works again. Gerbers are available.
Cheers, Gerhard DK4XP
On 5/13/23 9:02 AM, Richard (Rick) Karlquist via time-nuts wrote:
On 5/12/2023 6:00 PM, ed breya via time-nuts wrote:
Rick, I don't see how you have the impression that any of these
prescalers are "dynamic" dividers. \
That is what the project manager of the HP 5386 told me back in
the 80's. He used a static divider made by HP Santa Rosa. The
5334A used a dynamic divider made by HP Santa Clara. As the
5334B project manager, I tried whatever COTS dividers I could
find and they all had the same problems as the divider made
in Santa Clara. I designed that one out to save $100.
You previously said something like any prescaler will work if
the signal is coming from a signal generator. This notion is
contradicted by the HP 8660. For both the 5334A and 5334B,
the test procedure for the C channel is to drive the counter
with an 8660 at 1.3 GHz to verify the specified bandwidth.
However, the TP goes on to say that a 1 GHz high pass filter
must be inserted in front of the counter. And sure enough,
if you don't do this, it won't count 1.3 GHz. Admittedly,
the 8660 was not exactly what you would call HP's "finest
hour." I suspect you could dispense with the filter if you
used an 8662/3.
I was not aware of the HMC493, which seems to be static. With
that chip you might finally be able to homebrew a decent C channel.
The description of the technology used sounds like something
that would come out of the Santa Rosa IC fab. The price is a
bargain IMHO at $35. For $425 you can get an eval board.
So it looks like the technology has advanced somewhat in
35 years.
Hittite (HMC) made a bunch of GaAs dividers (and some are still being
made) that worked up at microwave frequencies. Some of them were
originally developed under SBIR funding from NASA. We use them in
synthesizers for deep space radios. We were trying to get wideband
(50-100 MHz) tuning coverage at X-band, and our ideas with DROs with two
varactors didn't work out. So we went to MMIC VCO and dividers in a
conventional PLL sort of arrangement (as opposed to DROs and SPDs)
https://ipnpr.jpl.nasa.gov/progress_report/42-166/166A.pdf
describes the approach and measured performance in the application
(using eval boards). I can't remember what parts we used for the divide
by 8 and the divide by N, and the picture is too low res to read it off
the part. I can probably find it. (Oddly, I think I saw these
breadboards a couple years ago in a closet).
The Iris cubesat transponder uses a divide by 2 like the HMC361 from 8.5
GHz or 7.1 GHz
https://digitalcommons.usu.edu/cgi/viewcontent.cgi?article=3603&context=smallsat
It's true that in these applications, the divider is seeing only one
signal with very good SNR - it's not like a general purpose counter.
As far as I know, they're just a chain of flipflops, sometimes with some
logic (there was a variable divide ratio one we used). They aren't
exactly low power.
I think the very first prescaler IC I ever encountered, long ago, is the
11C90. I believe it's heyday was in the 1980s - it appears it was near
EOL by 1992. I discovered it in some junk 6 GHz synthesizer modules that
used a PLLed VHF VCO and a power amplifier, that was then SRD/BPF
multiplied up the rest of the way. Later, I found one used as a
prescaler in a low cost frequency counter from the 80s or 90s.
A data sheet is here - it provides some more perspective on how these
work, and their evolution.
https://pdf.datasheetcatalog.com/datasheet/nationalsemiconductor/11C90.pdf
You can see that it is basically a high speed ECL 1/10 divider (or 1/11
- dual modulus even) capable of DC to around 650 MHz. It also includes
provision to connect it for self-biased clock input at the logic
threshold for AC coupling without any other parts (except an input
capacitor depending on lowest frequency needed). This made it possible
to toggle with input levels a fair amount less than the normal ECL
range. I think the later, improved technologies used more gain stages to
get the higher sensitivities we see in the modern (1990s-present) types.
There is of course a trade-off between speed and sensitivity, since more
gain slows it down.
Anyway, if you thoroughly read the Functional Description on page 6,
you'll see the story from the horse's mouth, so to speak. BTW I believe
there is a typo in the second paragraph, where it says
"This feature ensures that the circuit will operate with clock waveforms
having very slow rise and fall times, and thus, there is no maximum
frequency restriction."
I think it was meant to say no "minimum" frequency restriction, which
makes more sense, in context.
I think around here is where prescalers were becoming a distinct
functional component class, more tailored to the huge emerging
telecommunications markets - optimized for RF, and separate from the
counters category, although used together. At heart, they are the same.
Enjoy the story.
Ed
Hi
The “junk back on the input” is not at all uncommon. There are a number of PLL
IC’s that have the same “feature”. Buffering the input signal is often a good idea.
Bob
On May 13, 2023, at 1:21 PM, Gerhard Hoffmann via time-nuts time-nuts@lists.febo.com wrote:
Am 2023-05-13 3:00, schrieb ed breya via time-nuts:
"The HMC493LP3 & HMC493LP3E are low noise
Divide-by-4 Static Dividers utilizing InGaP GaAs HBT
I have built a 1/8 prescaler using the HMC363 to extend the trigger
bandwidth of my 54750A scope from 2.5 to 12 GHz (or the SR620).
It was necessary to have a gain block in front of the prescaler
since the HMC363 pumped some of the divided signal back into its input
which contaminated the measured channel with some jitter.
A second gain block at the output makes sure that there is enough
signal left even after a power divider.
The home-etched board is still without the preamplifier.
Oscillation tendency is most easily cured by a small DC offset.
That costs some sensitivity, another reason for the preamp.
The design has been published in DUBUS last year. I'll put it
on my web site when the FTP access works again. Gerbers are available.
Cheers, Gerhard DK4XP<Auswahl_2023051318:57:49.png><Auswahl_2023051319:12:15.png>_______________________________________________
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe send an email to time-nuts-leave@lists.febo.com