Empathy List Archives

GH

Gerhard Hoffmann

Fri, May 15, 2020 12:58 AM

I have a potential project in the electron spin spectroscopy sector and
I need
one or two clean signal sources in the 10 GHz range. Phase noise at,
say, 50 Hz offset
is important, but anything below 110 dBc does not care.
That probably calls for a multiplied crystal. These Hittite PLLs from AD
seem to be
just not good enough, maybe they'd work if pushed, but no reserve left.
Are there any known proven multipliers chains from VHF to 10 GHz?

Cheers, Gerhard

I have a potential project in the electron spin spectroscopy sector and I need one or two clean signal sources in the 10 GHz range. Phase noise at, say, 50 Hz offset is important, but anything below 110 dBc does not care. That probably calls for a multiplied crystal. These Hittite PLLs from AD seem to be just not good enough, maybe they'd work if pushed, but no reserve left. Are there any known proven multipliers chains from VHF to 10 GHz? Cheers, Gerhard

MG

Mike Garvey

Fri, May 15, 2020 1:24 AM

Cesium is 9.192 GHz; use a Cs standard design as a point of departure...?
Mike

-----Original Message-----
From: time-nuts [mailto:time-nuts-bounces@lists.febo.com] On Behalf Of Gerhard Hoffmann
Sent: Thursday, May 14, 2020 20:58
To: Discussion of precise time and frequency measurement
Subject: [time-nuts] f-multipliers from VHF to 10 GHz

I have a potential project in the electron spin spectroscopy sector and
I need
one or two clean signal sources in the 10 GHz range. Phase noise at,
say, 50 Hz offset
is important, but anything below 110 dBc does not care.
That probably calls for a multiplied crystal. These Hittite PLLs from AD
seem to be
just not good enough, maybe they'd work if pushed, but no reserve left.
Are there any known proven multipliers chains from VHF to 10 GHz?

Cheers, Gerhard

time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.

Cesium is 9.192 GHz; use a Cs standard design as a point of departure...? Mike -----Original Message----- From: time-nuts [mailto:time-nuts-bounces@lists.febo.com] On Behalf Of Gerhard Hoffmann Sent: Thursday, May 14, 2020 20:58 To: Discussion of precise time and frequency measurement Subject: [time-nuts] f-multipliers from VHF to 10 GHz I have a potential project in the electron spin spectroscopy sector and I need one or two clean signal sources in the 10 GHz range. Phase noise at, say, 50 Hz offset is important, but anything below 110 dBc does not care. That probably calls for a multiplied crystal. These Hittite PLLs from AD seem to be just not good enough, maybe they'd work if pushed, but no reserve left. Are there any known proven multipliers chains from VHF to 10 GHz? Cheers, Gerhard _______________________________________________ time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there.

BB

Bill Byrom

Fri, May 15, 2020 2:04 AM

Wenzel Associates (www.wenzel.com) In Austin TX can build custom rack mounted multiplied very low phase noise crystal sources. I have used their custom microwave multiplied crystal sources at my pre-retirement job (Tektronix RF Application Engineer) with a 12.5 GHz output. A few of my customers also evaluated and purchased these sources for driving the 12.5 GHz external clock input of the Tektronix AWG70000A/B series AWG's and 70KSX series oscilloscopes. My experience with this multiplied source was about 2 to 4 years ago, but they probably can still build these. I think the price was roughly US $15,000 for a rack mounted semi-custom system source.

One customer purchased the Wenzel golden multiplied 12.5 GHz source to get very low phase noise output of a Tektronix AWG70K AWG to generate microwave chirps for use in electron spin spectroscopy. The AWG70K has an internal moderate performance PLL synthesizer but that wasn't good enough for the spectroscopy experiments. When you use the Wenzel 12.5 GHz external clock, the phase noise of the AWG is dominated by the external clock. There is some small additional phase noise due to the AWG clock edge detection and SiGe DAC wideband noise. Another customer needed to generat very low phase noise RADAR signals.

As I remember the Wenzel product block diagram, they used a 10 MHz internal crystal reference oscillator as the master frequency reference with optional EFC electrical frequency control input. A higher frequency very low phase noise VHF crystal oscillator (between 100-200 MHz, I believe) was locked to the 10 MHz internal or external reference, and that VHF oscillator was multiplied and filtered in several stages to get to the microwave range (such as 10 to 12.5 GHz). They could customize their system to provide auxiliary lower frequency outputs from any of the multiplier stages.

The Tek AWG70K series arbitrary waveform generators have an external clock input with a range of 6.25-12.5 GHz. The external clock is double clocked on both rising and falling edges. So with a 12.5 GHz sine input, the internal 10-bit SiGe DAC is clocked at 25 GS/s. The AWG70001A and AWG70001B offset two of the 25 GS/s DACs by 180 degrees to get 50 GS/s AWG performance with a sin(x)/x usable bandwidth of nearly 20 GHz.

Bill Byrom N5BB
Retired in Nov 2019 from Tektronix after 32 years as an AE

On Thu, May 14, 2020, at 7:58 PM, Gerhard Hoffmann wrote:

I have a potential project in the electron spin spectroscopy sector and
I need
one or two clean signal sources in the 10 GHz range. Phase noise at,
say, 50 Hz offset
is important, but anything below 110 dBc does not care.
That probably calls for a multiplied crystal. These Hittite PLLs from AD
seem to be
just not good enough, maybe they'd work if pushed, but no reserve left.
Are there any known proven multipliers chains from VHF to 10 GHz?

Cheers, Gerhard

time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.

Wenzel Associates (www.wenzel.com) In Austin TX can build custom rack mounted multiplied very low phase noise crystal sources. I have used their custom microwave multiplied crystal sources at my pre-retirement job (Tektronix RF Application Engineer) with a 12.5 GHz output. A few of my customers also evaluated and purchased these sources for driving the 12.5 GHz external clock input of the Tektronix AWG70000A/B series AWG's and 70KSX series oscilloscopes. My experience with this multiplied source was about 2 to 4 years ago, but they probably can still build these. I think the price was roughly US $15,000 for a rack mounted semi-custom system source. One customer purchased the Wenzel golden multiplied 12.5 GHz source to get very low phase noise output of a Tektronix AWG70K AWG to generate microwave chirps for use in electron spin spectroscopy. The AWG70K has an internal moderate performance PLL synthesizer but that wasn't good enough for the spectroscopy experiments. When you use the Wenzel 12.5 GHz external clock, the phase noise of the AWG is dominated by the external clock. There is some small additional phase noise due to the AWG clock edge detection and SiGe DAC wideband noise. Another customer needed to generat very low phase noise RADAR signals. As I remember the Wenzel product block diagram, they used a 10 MHz internal crystal reference oscillator as the master frequency reference with optional EFC electrical frequency control input. A higher frequency very low phase noise VHF crystal oscillator (between 100-200 MHz, I believe) was locked to the 10 MHz internal or external reference, and that VHF oscillator was multiplied and filtered in several stages to get to the microwave range (such as 10 to 12.5 GHz). They could customize their system to provide auxiliary lower frequency outputs from any of the multiplier stages. The Tek AWG70K series arbitrary waveform generators have an external clock input with a range of 6.25-12.5 GHz. The external clock is double clocked on both rising and falling edges. So with a 12.5 GHz sine input, the internal 10-bit SiGe DAC is clocked at 25 GS/s. The AWG70001A and AWG70001B offset two of the 25 GS/s DACs by 180 degrees to get 50 GS/s AWG performance with a sin(x)/x usable bandwidth of nearly 20 GHz. -- Bill Byrom N5BB Retired in Nov 2019 from Tektronix after 32 years as an AE On Thu, May 14, 2020, at 7:58 PM, Gerhard Hoffmann wrote: > I have a potential project in the electron spin spectroscopy sector and > I need > one or two clean signal sources in the 10 GHz range. Phase noise at, > say, 50 Hz offset > is important, but anything below 110 dBc does not care. > That probably calls for a multiplied crystal. These Hittite PLLs from AD > seem to be > just not good enough, maybe they'd work if pushed, but no reserve left. > Are there any known proven multipliers chains from VHF to 10 GHz? > > Cheers, Gerhard > > _______________________________________________ > time-nuts mailing list -- time-nuts@lists.febo.com > To unsubscribe, go to > http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > and follow the instructions there. >

EB

ed breya

Fri, May 15, 2020 7:27 AM

Gerhard, you didn't mention a budget number for this. You should be able
to find a commercial synthesizer ready to go if you have enough money.
If it's a one-off, DIY thing, then it's a different story. If this is
the case, I'd recommend looking at the methods used in microwave
counters. They often use synthesized VHF sources and SRD multipliers and
YTFs to make and select sampling frequencies for down-converting
microwave inputs.

All I can say is that there are lots of ways, and the usual things apply.

Start as high as possible. An XO may be good up to 200 MHz or so. A SAW
resonator may be good to a GHz - noisier, but less multiplication
needed. You may be able to PLL a DRO well enough directly at 10 GHz, and
not need multiplication, with a high enough low-noise reference
frequency for comparison.

Going with a multiplier chain is reliable and straightforward, but can
get complicated due to the number of stages and filters that may be
needed. My favorite multiplier is the classic SRD. If you start with a
high enough drive frequency, selecting the desired harmonic is fairly
easy with a fixed or yig-tuned BPF. I always figure that if SRDs are
good enough to multiply by hundreds of times in Cs and Rb standards,
they're good enough for my simple needs.

An alternative to the SRD is to use extremely fast logic or line or
laser driver type parts running at VHF, that have edge speeds with
enough juice at 10 GHz, then BP filtered and amplified.

Ed

Gerhard, you didn't mention a budget number for this. You should be able to find a commercial synthesizer ready to go if you have enough money. If it's a one-off, DIY thing, then it's a different story. If this is the case, I'd recommend looking at the methods used in microwave counters. They often use synthesized VHF sources and SRD multipliers and YTFs to make and select sampling frequencies for down-converting microwave inputs. All I can say is that there are lots of ways, and the usual things apply. Start as high as possible. An XO may be good up to 200 MHz or so. A SAW resonator may be good to a GHz - noisier, but less multiplication needed. You may be able to PLL a DRO well enough directly at 10 GHz, and not need multiplication, with a high enough low-noise reference frequency for comparison. Going with a multiplier chain is reliable and straightforward, but can get complicated due to the number of stages and filters that may be needed. My favorite multiplier is the classic SRD. If you start with a high enough drive frequency, selecting the desired harmonic is fairly easy with a fixed or yig-tuned BPF. I always figure that if SRDs are good enough to multiply by hundreds of times in Cs and Rb standards, they're good enough for my simple needs. An alternative to the SRD is to use extremely fast logic or line or laser driver type parts running at VHF, that have edge speeds with enough juice at 10 GHz, then BP filtered and amplified. Ed

J

jimlux

Fri, May 15, 2020 3:21 PM

On 5/14/20 5:58 PM, Gerhard Hoffmann wrote:

I have a potential project in the electron spin spectroscopy sector and
I need
one or two clean signal sources in the 10 GHz range. Phase noise at,
say, 50 Hz offset
is important, but anything below 110 dBc does not care.
That probably calls for a multiplied crystal. These Hittite PLLs from AD
seem to be
just not good enough, maybe they'd work if pushed, but no reserve left.
Are there any known proven multipliers chains from VHF to 10 GHz?

-110dBc @ 50 Hz offset at 10GHz? That's -170dBc @ 50Hz with a 10MHz ref.
That's going to be a chore.. the Wenzel ULNs are -167dBc at 100Hz
offset, and they're pretty good.

But, as far as your synthesis approach, maybe not the Hittite PLL, but
what about using the Hittite dividers and/or VCOs, and a different kind
of PLL. You're talking about the noise "inside the loop" so the not so
wonderful GaAs ring oscillator far out noise isn't an issue.

We have used those VCOs and dividers for deep space transponder
synthesizers, and that's a classic "gotta have really good close in
noise" application.

Compared to a DRO, which has microphonics, limited tuning range, etc.
they're a lot easier.

I notice that AD and TI both have a lot of new parts for <17GHz. Sure,
most are targeting wideband high data rate kinds of applications, but
they might fill the need. What about the LMX2595 or LMX2594 - I've not
used them, but they might be interesting.

The plot in the selection guide does show only -90dBc for the 2594 at
8GHz, but I don't know if that's the bare VCO, or locked to something,
and then you'd ask what it's locked to. The divider noise is something
like -150dBc at 10GHz, so that's probably not your dominant source.

On 5/14/20 5:58 PM, Gerhard Hoffmann wrote: > I have a potential project in the electron spin spectroscopy sector and > I need > one or two clean signal sources in the 10 GHz range. Phase noise at, > say, 50 Hz offset > is important, but anything below 110 dBc does not care. > That probably calls for a multiplied crystal. These Hittite PLLs from AD > seem to be > just not good enough, maybe they'd work if pushed, but no reserve left. > Are there any known proven multipliers chains from VHF to 10 GHz? > -110dBc @ 50 Hz offset at 10GHz? That's -170dBc @ 50Hz with a 10MHz ref. That's going to be a chore.. the Wenzel ULNs are -167dBc at 100Hz offset, and they're pretty good. But, as far as your synthesis approach, maybe not the Hittite PLL, but what about using the Hittite dividers and/or VCOs, and a different kind of PLL. You're talking about the noise "inside the loop" so the not so wonderful GaAs ring oscillator far out noise isn't an issue. We have used those VCOs and dividers for deep space transponder synthesizers, and that's a classic "gotta have really good close in noise" application. Compared to a DRO, which has microphonics, limited tuning range, etc. they're a lot easier. I notice that AD and TI both have a lot of new parts for <17GHz. Sure, most are targeting wideband high data rate kinds of applications, but they might fill the need. What about the LMX2595 or LMX2594 - I've not used them, but they might be interesting. The plot in the selection guide does show only -90dBc for the 2594 at 8GHz, but I don't know if that's the bare VCO, or locked to something, and then you'd ask what it's locked to. The divider noise is something like -150dBc at 10GHz, so that's probably not your dominant source.

GH

Gerhard Hoffmann

Fri, May 15, 2020 9:14 PM

No, no, no, it's not that bad :-) I should not post here in the middle
of the night. Sorry to cause that confusion.

Minimum is -90 dBc @ 50 Hz, or let's say @100 Hz @ 10 GHz.
that would equal -110 dBc@1 GHz, or -130 dBc @100 MHz, BTDT.

And then, the ~4 MHz difference between TX and RX frequency could be
done by a SSB mixer with a non-multiplied crystal. We would have some
common mode noise, but the RX-TX difference would be fairly constant.
It would not de-correlate over the 10 mm run length, not at low
offsets where it counts.

The -110 was only meant for "Don't care about multiplied white noise
floor", not in the sense of a spec but in the sense of "guaranteed
harmless". It's not such a great relaxation after all, it could be
20 dB looser.

The question was really only about a simple multiplier chain. The
style used in ham radio 10 GHz transverters has too many stages,
GaAS-Fets with 1/f and pipe cap filters. Too complicated.

Macom still make a SRD diode, but probably it is easiest to phaselock
a ceramic puck or an on-chip VCO to a 100+ MHz crystal. The offset-
mixing removes the need for a low reference frequency or fractional
voodoo.

cheers, Gerhard

No, no, no, it's not that bad :-) I should not post here in the middle of the night. Sorry to cause that confusion. Minimum is -90 dBc @ 50 Hz, or let's say @100 Hz @ 10 GHz. that would equal -110 dBc@1 GHz, or -130 dBc @100 MHz, BTDT. And then, the ~4 MHz difference between TX and RX frequency could be done by a SSB mixer with a non-multiplied crystal. We would have some common mode noise, but the RX-TX difference would be fairly constant. It would not de-correlate over the 10 mm run length, not at low offsets where it counts. The -110 was only meant for "Don't care about multiplied white noise floor", not in the sense of a spec but in the sense of "guaranteed harmless". It's not such a great relaxation after all, it could be 20 dB looser. The question was really only about a _simple_ multiplier chain. The style used in ham radio 10 GHz transverters has too many stages, GaAS-Fets with 1/f and pipe cap filters. Too complicated. Macom still make a SRD diode, but probably it is easiest to phaselock a ceramic puck or an on-chip VCO to a 100+ MHz crystal. The offset- mixing removes the need for a low reference frequency or fractional voodoo. cheers, Gerhard

BG

Bruce Griffiths

Fri, May 15, 2020 9:45 PM

Macom also do NLTL comb generators which are much quieter than SRDs:
https://www.macom.com/products/product-detail/MLPNC-7100-SMA850

Bruce

On 16 May 2020 at 09:14 Gerhard Hoffmann ghf@hoffmann-hochfrequenz.de wrote:

No, no, no, it's not that bad :-) I should not post here in the middle
of the night. Sorry to cause that confusion.

Minimum is -90 dBc @ 50 Hz, or let's say @100 Hz @ 10 GHz.
that would equal -110 dBc@1 GHz, or -130 dBc @100 MHz, BTDT.

And then, the ~4 MHz difference between TX and RX frequency could be
done by a SSB mixer with a non-multiplied crystal. We would have some
common mode noise, but the RX-TX difference would be fairly constant.
It would not de-correlate over the 10 mm run length, not at low
offsets where it counts.

The -110 was only meant for "Don't care about multiplied white noise
floor", not in the sense of a spec but in the sense of "guaranteed
harmless". It's not such a great relaxation after all, it could be
20 dB looser.

The question was really only about a simple multiplier chain. The
style used in ham radio 10 GHz transverters has too many stages,
GaAS-Fets with 1/f and pipe cap filters. Too complicated.

Macom still make a SRD diode, but probably it is easiest to phaselock
a ceramic puck or an on-chip VCO to a 100+ MHz crystal. The offset-
mixing removes the need for a low reference frequency or fractional
voodoo.

cheers, Gerhard

time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.

Macom also do NLTL comb generators which are much quieter than SRDs: https://www.macom.com/products/product-detail/MLPNC-7100-SMA850 Bruce > On 16 May 2020 at 09:14 Gerhard Hoffmann <ghf@hoffmann-hochfrequenz.de> wrote: > > > No, no, no, it's not that bad :-) I should not post here in the middle > of the night. Sorry to cause that confusion. > > Minimum is -90 dBc @ 50 Hz, or let's say @100 Hz @ 10 GHz. > that would equal -110 dBc@1 GHz, or -130 dBc @100 MHz, BTDT. > > And then, the ~4 MHz difference between TX and RX frequency could be > done by a SSB mixer with a non-multiplied crystal. We would have some > common mode noise, but the RX-TX difference would be fairly constant. > It would not de-correlate over the 10 mm run length, not at low > offsets where it counts. > > The -110 was only meant for "Don't care about multiplied white noise > floor", not in the sense of a spec but in the sense of "guaranteed > harmless". It's not such a great relaxation after all, it could be > 20 dB looser. > > The question was really only about a _simple_ multiplier chain. The > style used in ham radio 10 GHz transverters has too many stages, > GaAS-Fets with 1/f and pipe cap filters. Too complicated. > > Macom still make a SRD diode, but probably it is easiest to phaselock > a ceramic puck or an on-chip VCO to a 100+ MHz crystal. The offset- > mixing removes the need for a low reference frequency or fractional > voodoo. > > cheers, Gerhard > > > _______________________________________________ > time-nuts mailing list -- time-nuts@lists.febo.com > To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > and follow the instructions there.

EB

ed breya

Fri, May 15, 2020 10:54 PM

Bruce wrote

Macom also do NLTL comb generators which are much quieter than SRDs:
https://www.macom.com/products/product-detail/MLPNC-7100-SMA850

Wow, those are nice. I've studied and searched for NLTLs over the years,
and found all sorts of research papers about making monolithic ones, but
not such a finished piece that you can order up and go - even though
these may have been around for some time. I've even considered trying to
build some discretely and experiment, but got dissuaded when I found out
how many compression stages are needed for meaningful results. I think
that's why the research was all geared toward making them in integrated
form, where you can get lots of similar items repeated in tiny form. I'd
guess that these may have 20-100 stages, or maybe nowadays it can be one
continuous line with spatially variable characteristics..

It's cool to see, but probably quite spendy - I'd guess 2 grand or more.
I tried to find out the price looking at a distributor, but it's a
special order, call for quote type deal. I'm sure it's quite a few dB$
above what I can justify for any of my projects.

Ed

Bruce wrote Macom also do NLTL comb generators which are much quieter than SRDs: https://www.macom.com/products/product-detail/MLPNC-7100-SMA850 Wow, those are nice. I've studied and searched for NLTLs over the years, and found all sorts of research papers about making monolithic ones, but not such a finished piece that you can order up and go - even though these may have been around for some time. I've even considered trying to build some discretely and experiment, but got dissuaded when I found out how many compression stages are needed for meaningful results. I think that's why the research was all geared toward making them in integrated form, where you can get lots of similar items repeated in tiny form. I'd guess that these may have 20-100 stages, or maybe nowadays it can be one continuous line with spatially variable characteristics.. It's cool to see, but probably quite spendy - I'd guess 2 grand or more. I tried to find out the price looking at a distributor, but it's a special order, call for quote type deal. I'm sure it's quite a few dB$ above what I can justify for any of my projects. Ed

J

jimlux

Fri, May 15, 2020 11:15 PM

On 5/15/20 2:14 PM, Gerhard Hoffmann wrote:

No, no, no, it's not that bad :-) I should not post here in the middle
of the night. Sorry to cause that confusion.

Minimum is -90 dBc @ 50 Hz, or let's say @100 Hz @ 10 GHz.
that would equal -110 dBc@1 GHz, or -130 dBc @100 MHz, BTDT.

That's much easier.

Use one of the single chip synthesizers - they can probably do just what
you want, with adequate performance.

The question was really only about a simple multiplier chain. The
style used in ham radio 10 GHz transverters has too many stages,
GaAS-Fets with 1/f and pipe cap filters. Too complicated.

Yes, that would be an ordeal, and is completely unnecessary today.

Macom still make a SRD diode, but probably it is easiest to phaselock
a ceramic puck or an on-chip VCO to a 100+ MHz crystal. The offset-
mixing removes the need for a low reference frequency or fractional
voodoo.
o

Trivially easy to lock the onchip VCO to a 10 or 100 MHz oscillator.
These days, dividing down from 10GHz isn't a chore, like it was 15 years
ago.

SRDs are a pain, they need huge drive power (100mW?) to work well, as do
the Sampling Phase Detector/ harmonic mixer equivalents. It's hard to
get a real low noise +20dBm signal . You'll spend as much time on that
as the other stuff.

Sampling phase detector and DRO from 2004
https://tmo.jpl.nasa.gov/progress_report/42-156/156C.pdf

GaAs and PLL from 2006
https://tmo.jpl.nasa.gov/progress_report/42-166/166A.pdf

You can probably do it with an off the shelf eval board.

On 5/15/20 2:14 PM, Gerhard Hoffmann wrote: > No, no, no, it's not that bad :-) I should not post here in the middle > of the night. Sorry to cause that confusion. > > Minimum is -90 dBc @ 50 Hz, or let's say @100 Hz @ 10 GHz. > that would equal -110 dBc@1 GHz, or -130 dBc @100 MHz, BTDT. That's much easier. Use one of the single chip synthesizers - they can probably do just what you want, with adequate performance. > > The question was really only about a _simple_ multiplier chain. The > style used in ham radio 10 GHz transverters has too many stages, > GaAS-Fets with 1/f and pipe cap filters. Too complicated. Yes, that would be an ordeal, and is completely unnecessary today. > > Macom still make a SRD diode, but probably it is easiest to phaselock > a ceramic puck or an on-chip VCO to a 100+ MHz crystal. The offset- > mixing removes the need for a low reference frequency or fractional > voodoo. > o Trivially easy to lock the onchip VCO to a 10 or 100 MHz oscillator. These days, dividing down from 10GHz isn't a chore, like it was 15 years ago. SRDs are a pain, they need huge drive power (100mW?) to work well, as do the Sampling Phase Detector/ harmonic mixer equivalents. It's hard to get a real low noise +20dBm signal . You'll spend as much time on that as the other stuff. Sampling phase detector and DRO from 2004 https://tmo.jpl.nasa.gov/progress_report/42-156/156C.pdf GaAs and PLL from 2006 https://tmo.jpl.nasa.gov/progress_report/42-166/166A.pdf You can probably do it with an off the shelf eval board.

BG

Bruce Griffiths

Fri, May 15, 2020 11:32 PM

Indicative price:
https://www.richardsonrfpd.com/Products/Product/MLPNC-7103-SMT6

Bruce

On 16 May 2020 at 10:54 ed breya eb@telight.com wrote:

Bruce wrote

Macom also do NLTL comb generators which are much quieter than SRDs:
https://www.macom.com/products/product-detail/MLPNC-7100-SMA850

Wow, those are nice. I've studied and searched for NLTLs over the years,
and found all sorts of research papers about making monolithic ones, but
not such a finished piece that you can order up and go - even though
these may have been around for some time. I've even considered trying to
build some discretely and experiment, but got dissuaded when I found out
how many compression stages are needed for meaningful results. I think
that's why the research was all geared toward making them in integrated
form, where you can get lots of similar items repeated in tiny form. I'd
guess that these may have 20-100 stages, or maybe nowadays it can be one
continuous line with spatially variable characteristics..

It's cool to see, but probably quite spendy - I'd guess 2 grand or more.
I tried to find out the price looking at a distributor, but it's a
special order, call for quote type deal. I'm sure it's quite a few dB$
above what I can justify for any of my projects.

Ed

time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.

Indicative price: https://www.richardsonrfpd.com/Products/Product/MLPNC-7103-SMT6 Bruce > On 16 May 2020 at 10:54 ed breya <eb@telight.com> wrote: > > > Bruce wrote > > Macom also do NLTL comb generators which are much quieter than SRDs: > https://www.macom.com/products/product-detail/MLPNC-7100-SMA850 > > Wow, those are nice. I've studied and searched for NLTLs over the years, > and found all sorts of research papers about making monolithic ones, but > not such a finished piece that you can order up and go - even though > these may have been around for some time. I've even considered trying to > build some discretely and experiment, but got dissuaded when I found out > how many compression stages are needed for meaningful results. I think > that's why the research was all geared toward making them in integrated > form, where you can get lots of similar items repeated in tiny form. I'd > guess that these may have 20-100 stages, or maybe nowadays it can be one > continuous line with spatially variable characteristics.. > > It's cool to see, but probably quite spendy - I'd guess 2 grand or more. > I tried to find out the price looking at a distributor, but it's a > special order, call for quote type deal. I'm sure it's quite a few dB$ > above what I can justify for any of my projects. > > Ed > > > _______________________________________________ > time-nuts mailing list -- time-nuts@lists.febo.com > To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > and follow the instructions there.