DB
David Bengtson
Fri, Mar 24, 2023 12:45 AM
Has anyone run across any publications on best practices or examples
of 10MHz Lab wide distribution networks? I'm looking for a discussion
on how to physically locate oscillators/distribution amplifiers, cable
types and runs, RFI mitigation etc. I haven't come across any, and I'm
starting to build one at work. We've got a Cs oscillator and I'd like
to make sure we deliver that performance across our systems lab.
Thanks
Dave
Has anyone run across any publications on best practices or examples
of 10MHz Lab wide distribution networks? I'm looking for a discussion
on how to physically locate oscillators/distribution amplifiers, cable
types and runs, RFI mitigation etc. I haven't come across any, and I'm
starting to build one at work. We've got a Cs oscillator and I'd like
to make sure we deliver that performance across our systems lab.
Thanks
Dave
JM
John Miles
Fri, Mar 24, 2023 2:28 AM
Has anyone run across any publications on best practices or examples
of 10MHz Lab wide distribution networks? I'm looking for a discussion
on how to physically locate oscillators/distribution amplifiers, cable
types and runs, RFI mitigation etc. I haven't come across any, and I'm
starting to build one at work. We've got a Cs oscillator and I'd like
to make sure we deliver that performance across our systems lab.
A lot of things can be said in favor of low-noise distribution amps with
good VSWR, good channel isolation, and good PSRR (not so fast, HP 5087A.)
But over the years, I've learned that the three most important factors when
it comes to piping 10 MHz signals around are:
- Shield resistance
- Shield resistance
- Shield resistance
I've spent a lot of time recommending double-shielded coax in the TimeLab
manual and elsewhere, and I still stand by that advice, but what I've come
to realize is that this is really just a proxy for low shield resistance.
Good grades of single-shielded cable are basically as effective at HF as
double-shielded cable. To the extent your cable ground shield exhibits
resistance, it's not a shield, it's a resistor.
Avoiding ground loops is on the list too, but further down. Never lift a
ground to avoid a ground loop. Use coax-to-coax baluns only when you can
see a beneficial effect. Focus instead on providing a shared low-resistance
common ground to your entire network -- ideally not the ground all the way
back at the service entrance -- and rely on low shield resistance on the RF
side to do the rest.
Every installation is different and your mileage will most certainly vary,
but this is my take on it.
-- john
> Has anyone run across any publications on best practices or examples
> of 10MHz Lab wide distribution networks? I'm looking for a discussion
> on how to physically locate oscillators/distribution amplifiers, cable
> types and runs, RFI mitigation etc. I haven't come across any, and I'm
> starting to build one at work. We've got a Cs oscillator and I'd like
> to make sure we deliver that performance across our systems lab.
A lot of things can be said in favor of low-noise distribution amps with
good VSWR, good channel isolation, and good PSRR (not so fast, HP 5087A.)
But over the years, I've learned that the three most important factors when
it comes to piping 10 MHz signals around are:
1) Shield resistance
2) Shield resistance
3) Shield resistance
I've spent a lot of time recommending double-shielded coax in the TimeLab
manual and elsewhere, and I still stand by that advice, but what I've come
to realize is that this is really just a proxy for low shield resistance.
Good grades of single-shielded cable are basically as effective at HF as
double-shielded cable. To the extent your cable ground shield exhibits
resistance, it's not a shield, it's a resistor.
Avoiding ground loops is on the list too, but further down. Never lift a
ground to avoid a ground loop. Use coax-to-coax baluns only when you can
see a beneficial effect. Focus instead on providing a shared low-resistance
common ground to your entire network -- ideally not the ground all the way
back at the service entrance -- and rely on low shield resistance on the RF
side to do the rest.
Every installation is different and your mileage will most certainly vary,
but this is my take on it.
-- john
BC
Bob Camp
Fri, Mar 24, 2023 2:53 AM
Hi
The most basic starting point is: hlow many digits to the left of decimal are in the budget?
Somebody with major money to spend will approach this a bit differently than joe cheap skate.
One of the most basic cost drivers is the coax used. You can get some very exotic coax
if you have a big budget…..
Equally important is just how the system gets used. Un-terminated coax can be a disaster
no matter how you set things up. The longer the coax, the worse the impact.
Yes, this is already making a bunch of assumptions about what the “lab” is getting used for.
If phase noise / adev measurement isn’t part of that use, that has a budget impact.
Bob
On Mar 23, 2023, at 8:45 PM, David Bengtson via time-nuts time-nuts@lists.febo.com wrote:
Has anyone run across any publications on best practices or examples
of 10MHz Lab wide distribution networks? I'm looking for a discussion
on how to physically locate oscillators/distribution amplifiers, cable
types and runs, RFI mitigation etc. I haven't come across any, and I'm
starting to build one at work. We've got a Cs oscillator and I'd like
to make sure we deliver that performance across our systems lab.
Thanks
Dave
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe send an email to time-nuts-leave@lists.febo.com
Hi
The most basic starting point is: hlow many digits to the left of decimal are in the budget?
Somebody with major money to spend will approach this a bit differently than joe cheap skate.
One of the most basic cost drivers is the coax used. You can get some *very* exotic coax
if you have a big budget…..
Equally important is just how the system gets used. Un-terminated coax can be a disaster
no matter how you set things up. The longer the coax, the worse the impact.
Yes, this is already making a bunch of assumptions about what the “lab” is getting used for.
If phase noise / adev measurement isn’t part of that use, that has a budget impact.
Bob
> On Mar 23, 2023, at 8:45 PM, David Bengtson via time-nuts <time-nuts@lists.febo.com> wrote:
>
> Has anyone run across any publications on best practices or examples
> of 10MHz Lab wide distribution networks? I'm looking for a discussion
> on how to physically locate oscillators/distribution amplifiers, cable
> types and runs, RFI mitigation etc. I haven't come across any, and I'm
> starting to build one at work. We've got a Cs oscillator and I'd like
> to make sure we deliver that performance across our systems lab.
>
> Thanks
>
> Dave
> _______________________________________________
> time-nuts mailing list -- time-nuts@lists.febo.com
> To unsubscribe send an email to time-nuts-leave@lists.febo.com
K
Kitski
Fri, Mar 24, 2023 5:30 AM
I'll second John's recommendations, particularly double-shielded coax, but
with a twist (pardon the pun).
There are various grades of 'double shielded' coax - some with loose braid
coverage and others with really tight weaves (loosely spec'd as 'coverage').
Use 95% or forget about it.
Next up (down ?) this rabbit hole is triax which by definition is double
shielded. More to the point, the braids are insulated from each other.
Another benefit is that BNC triax connectors are not that expensive. As the
outer braid is typically at 'mains/safety earth' potential, the independent
inner braid can be used in a variety of hum-busting ways.
Then another rabbit hole to peer down is re-wiring your AC mains facility
with separate electrical earth and technical earths. Not for the
faint-hearted this one.
Mitigating leakage LF thru to UHF (and crosstalk) in and out of your
facility are part and parcel of MIL requirements in sensitive
establishments. Measuring their effectiveness (BT-DT) is an interesting
past-time.
My 10c worth (currently 67c US).
Kit
Canberra, Australia
-----Original Message-----
A lot of things can be said in favor of low-noise distribution amps with
good VSWR, good channel isolation, and good PSRR (not so fast, HP 5087A.)
But over the years, I've learned that the three most important factors when
it comes to piping 10 MHz signals around are:
- Shield resistance
- Shield resistance
- Shield resistance
I've spent a lot of time recommending double-shielded coax in the TimeLab
manual and elsewhere, and I still stand by that advice, but what I've come
to realize is that this is really just a proxy for low shield resistance.
Good grades of single-shielded cable are basically as effective at HF as
double-shielded cable. To the extent your cable ground shield exhibits
resistance, it's not a shield, it's a resistor.
Avoiding ground loops is on the list too, but further down. Never lift a
ground to avoid a ground loop. Use coax-to-coax baluns only when you can
see a beneficial effect. Focus instead on providing a shared low-resistance
common ground to your entire network -- ideally not the ground all the way
back at the service entrance -- and rely on low shield resistance on the RF
side to do the rest.
Every installation is different and your mileage will most certainly vary,
but this is my take on it.
-- john
time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe send an
email to time-nuts-leave@lists.febo.com
I'll second John's recommendations, particularly double-shielded coax, but
with a twist (pardon the pun).
There are various grades of 'double shielded' coax - some with loose braid
coverage and others with really tight weaves (loosely spec'd as 'coverage').
Use 95% or forget about it.
Next up (down ?) this rabbit hole is triax which by definition is double
shielded. More to the point, the braids are insulated from each other.
Another benefit is that BNC triax connectors are not that expensive. As the
outer braid is typically at 'mains/safety earth' potential, the independent
inner braid can be used in a variety of hum-busting ways.
Then another rabbit hole to peer down is re-wiring your AC mains facility
with separate electrical earth and technical earths. Not for the
faint-hearted this one.
Mitigating leakage LF thru to UHF (and crosstalk) in and out of your
facility are part and parcel of MIL requirements in sensitive
establishments. Measuring their effectiveness (BT-DT) is an interesting
past-time.
My 10c worth (currently 67c US).
Kit
Canberra, Australia
-----Original Message-----
A lot of things can be said in favor of low-noise distribution amps with
good VSWR, good channel isolation, and good PSRR (not so fast, HP 5087A.)
But over the years, I've learned that the three most important factors when
it comes to piping 10 MHz signals around are:
1) Shield resistance
2) Shield resistance
3) Shield resistance
I've spent a lot of time recommending double-shielded coax in the TimeLab
manual and elsewhere, and I still stand by that advice, but what I've come
to realize is that this is really just a proxy for low shield resistance.
Good grades of single-shielded cable are basically as effective at HF as
double-shielded cable. To the extent your cable ground shield exhibits
resistance, it's not a shield, it's a resistor.
Avoiding ground loops is on the list too, but further down. Never lift a
ground to avoid a ground loop. Use coax-to-coax baluns only when you can
see a beneficial effect. Focus instead on providing a shared low-resistance
common ground to your entire network -- ideally not the ground all the way
back at the service entrance -- and rely on low shield resistance on the RF
side to do the rest.
Every installation is different and your mileage will most certainly vary,
but this is my take on it.
-- john
_______________________________________________
time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe send an
email to time-nuts-leave@lists.febo.com
MD
Magnus Danielson
Fri, Mar 24, 2023 12:14 PM
Hi,
On 2023-03-24 03:28, John Miles via time-nuts wrote:
Has anyone run across any publications on best practices or examples
of 10MHz Lab wide distribution networks? I'm looking for a discussion
on how to physically locate oscillators/distribution amplifiers, cable
types and runs, RFI mitigation etc. I haven't come across any, and I'm
starting to build one at work. We've got a Cs oscillator and I'd like
to make sure we deliver that performance across our systems lab.
A lot of things can be said in favor of low-noise distribution amps with
good VSWR, good channel isolation, and good PSRR (not so fast, HP 5087A.)
But over the years, I've learned that the three most important factors when
it comes to piping 10 MHz signals around are:
- Shield resistance
- Shield resistance
- Shield resistance
I've spent a lot of time recommending double-shielded coax in the TimeLab
manual and elsewhere, and I still stand by that advice, but what I've come
to realize is that this is really just a proxy for low shield resistance.
Good grades of single-shielded cable are basically as effective at HF as
double-shielded cable. To the extent your cable ground shield exhibits
resistance, it's not a shield, it's a resistor.
You for sure want your cable tight as such, but for any low frequency
variations, you might benefit from laying extra ground-wires in parallel
with coax cables. The coax cable will tie grounds together, and the
resistance in that will for sure limit how things are damped. A lab is
just like a telecom setup. To further help, bring additional grounding
wire to shunt the resistance of the coax shield is recommended.
Avoiding ground loops is on the list too, but further down. Never lift a
ground to avoid a ground loop. Use coax-to-coax baluns only when you can
see a beneficial effect.
Figure out if the isolation needs to be at low or high frequency. A
transformer achieves galvanic isolation and is still achieving some
isolation all the way up at 50 or 60 Hz, but for RF it's just a cap that
shorts quickly. If it is RF-power, then a common-mode choke that is the
right solution, and some care should go into figuring out which
frequency it will operate on.
Focus instead on providing a shared low-resistance
common ground to your entire network -- ideally not the ground all the way
back at the service entrance -- and rely on low shield resistance on the RF
side to do the rest.
Agree.
Look at ITU-T K.27 and read the Mesh-BN parts.
Cheers,
Magnus
Hi,
On 2023-03-24 03:28, John Miles via time-nuts wrote:
>> Has anyone run across any publications on best practices or examples
>> of 10MHz Lab wide distribution networks? I'm looking for a discussion
>> on how to physically locate oscillators/distribution amplifiers, cable
>> types and runs, RFI mitigation etc. I haven't come across any, and I'm
>> starting to build one at work. We've got a Cs oscillator and I'd like
>> to make sure we deliver that performance across our systems lab.
> A lot of things can be said in favor of low-noise distribution amps with
> good VSWR, good channel isolation, and good PSRR (not so fast, HP 5087A.)
> But over the years, I've learned that the three most important factors when
> it comes to piping 10 MHz signals around are:
>
> 1) Shield resistance
> 2) Shield resistance
> 3) Shield resistance
>
> I've spent a lot of time recommending double-shielded coax in the TimeLab
> manual and elsewhere, and I still stand by that advice, but what I've come
> to realize is that this is really just a proxy for low shield resistance.
> Good grades of single-shielded cable are basically as effective at HF as
> double-shielded cable. To the extent your cable ground shield exhibits
> resistance, it's not a shield, it's a resistor.
You for sure want your cable tight as such, but for any low frequency
variations, you might benefit from laying extra ground-wires in parallel
with coax cables. The coax cable will tie grounds together, and the
resistance in that will for sure limit how things are damped. A lab is
just like a telecom setup. To further help, bring additional grounding
wire to shunt the resistance of the coax shield is recommended.
> Avoiding ground loops is on the list too, but further down. Never lift a
> ground to avoid a ground loop. Use coax-to-coax baluns only when you can
> see a beneficial effect.
Figure out if the isolation needs to be at low or high frequency. A
transformer achieves galvanic isolation and is still achieving some
isolation all the way up at 50 or 60 Hz, but for RF it's just a cap that
shorts quickly. If it is RF-power, then a common-mode choke that is the
right solution, and some care should go into figuring out which
frequency it will operate on.
> Focus instead on providing a shared low-resistance
> common ground to your entire network -- ideally not the ground all the way
> back at the service entrance -- and rely on low shield resistance on the RF
> side to do the rest.
Agree.
Look at ITU-T K.27 and read the Mesh-BN parts.
Cheers,
Magnus
>
> Every installation is different and your mileage will most certainly vary,
> but this is my take on it.
>
> -- john
>
>
> _______________________________________________
> time-nuts mailing list -- time-nuts@lists.febo.com
> To unsubscribe send an email to time-nuts-leave@lists.febo.com
BC
Bob Camp
Fri, Mar 24, 2023 12:26 PM
Hi
There are indeed multi year sort of courses (actually sets of courses)
that various folks put on about the EMI side of this. Unfortunately it does
quickly get pretty deep in theory and thus the long slog to get to this or
that level of the knowledge base.
How big is “work” and how big is “lab”? I’ve seen places that measured
the answers in kilometers. I’ve also seen places that had everything within
20 meters of everything else. What you do in one is profoundly different
than another.
How many “destinations” do you have that need this or that? Again, some
will have an answer in dozens, for others it will be in the thousands or even
tens of thousands.
What connects to those end points? If the answer (as noted earlier) is
a bunch of ADEV / phase noise gear, that’s very different than more
conventional test gear. Keeping the 10 MHz away from the DUT can be
a very big deal.
Like it or not, much of this has been decided for you. That piece of test
gear has a very normal BNC connector on the back of it. They are a
terrible thing for this, but that’s what you get. The cable you use is
going to mate with them. If you use off the shelf gear for the distribution,
it also has BNC’s on it.
Cable needs to be run from here to there. Loss is rarely an issue. Skin
depth vs the shield is very much an issue at 10 MHz. Foil as a shield is
useless in this case. Big thick multiple layers of braid is needed. That
isn’t just a “get RG-400” sort of thing. It’s also who you get it from. The
expensive source may be the one ….
There is a tendency to overdo this kind of thing. Think through how
many endpoints you need. Cable up the reasonable minimum and
see how it goes. Plan on a re-visit in 6 months and maybe again in
12 months. Having a lot of cable to nowhere is a problem in a number
of ways ….
Fun
Bob
On Mar 23, 2023, at 8:45 PM, David Bengtson via time-nuts time-nuts@lists.febo.com wrote:
Has anyone run across any publications on best practices or examples
of 10MHz Lab wide distribution networks? I'm looking for a discussion
on how to physically locate oscillators/distribution amplifiers, cable
types and runs, RFI mitigation etc. I haven't come across any, and I'm
starting to build one at work. We've got a Cs oscillator and I'd like
to make sure we deliver that performance across our systems lab.
Thanks
Dave
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe send an email to time-nuts-leave@lists.febo.com
Hi
There are indeed multi year sort of courses (actually sets of courses)
that various folks put on about the EMI side of this. Unfortunately it does
quickly get pretty deep in theory and thus the long slog to get to this or
that level of the knowledge base.
How big is “work” and how big is “lab”? I’ve seen places that measured
the answers in kilometers. I’ve also seen places that had everything within
20 meters of everything else. What you do in one is profoundly different
than another.
How many “destinations” do you have that need this or that? Again, some
will have an answer in dozens, for others it will be in the thousands or even
tens of thousands.
What connects to those end points? If the answer (as noted earlier) is
a bunch of ADEV / phase noise gear, that’s very different than more
conventional test gear. Keeping the 10 MHz away from the DUT can be
a *very* big deal.
Like it or not, much of this has been decided for you. That piece of test
gear has a very normal BNC connector on the back of it. They are a
terrible thing for this, but that’s what you get. The cable you use is
going to mate with them. If you use off the shelf gear for the distribution,
it also has BNC’s on it.
Cable needs to be run from here to there. Loss is rarely an issue. Skin
depth vs the shield is very much an issue at 10 MHz. Foil as a shield is
useless in this case. Big thick multiple layers of braid is needed. That
isn’t just a “get RG-400” sort of thing. It’s also who you get it from. The
expensive source may be the one ….
There is a tendency to overdo this kind of thing. Think through how
many endpoints you need. Cable up the reasonable minimum and
see how it goes. Plan on a re-visit in 6 months and maybe again in
12 months. Having a lot of cable to nowhere is a problem in a number
of ways ….
Fun
Bob
> On Mar 23, 2023, at 8:45 PM, David Bengtson via time-nuts <time-nuts@lists.febo.com> wrote:
>
> Has anyone run across any publications on best practices or examples
> of 10MHz Lab wide distribution networks? I'm looking for a discussion
> on how to physically locate oscillators/distribution amplifiers, cable
> types and runs, RFI mitigation etc. I haven't come across any, and I'm
> starting to build one at work. We've got a Cs oscillator and I'd like
> to make sure we deliver that performance across our systems lab.
>
> Thanks
>
> Dave
> _______________________________________________
> time-nuts mailing list -- time-nuts@lists.febo.com
> To unsubscribe send an email to time-nuts-leave@lists.febo.com
JV
John Vendely
Fri, Mar 24, 2023 2:15 PM
Bob makes a good point about the devices terminating the ends of
reference distribution cables. Many frequency counters have little
shielding, and can be a significant source of 10 MHz radiation,
regardless of the shielding effectiveness of the reference cable runs
feeding them. Counters such as the HP 53132 and 5335 are good examples
of this. Even some signal generators and spectrum analyzers can exhibit
this problem. Despite good shielding in their RF sections, sometimes
inadequate attention was paid to shielding of the external reference inputs.
I encountered this problem when building a reference distribution system
for my low phase noise house standard. The distribution amp was
meticulously shielded and double shielded cable was used throughout,
with runs up to 60'. When individually terminated, the cables showed
very low leakage. However, significant 10 MHz radiation resulted when
connected to numerous test instruments...
73,
John K9WT
On 3/24/2023 8:26 AM, Bob Camp via time-nuts wrote:
Hi
There are indeed multi year sort of courses (actually sets of courses)
that various folks put on about the EMI side of this. Unfortunately it does
quickly get pretty deep in theory and thus the long slog to get to this or
that level of the knowledge base.
How big is “work” and how big is “lab”? I’ve seen places that measured
the answers in kilometers. I’ve also seen places that had everything within
20 meters of everything else. What you do in one is profoundly different
than another.
How many “destinations” do you have that need this or that? Again, some
will have an answer in dozens, for others it will be in the thousands or even
tens of thousands.
What connects to those end points? If the answer (as noted earlier) is
a bunch of ADEV / phase noise gear, that’s very different than more
conventional test gear. Keeping the 10 MHz away from the DUT can be
a very big deal.
Like it or not, much of this has been decided for you. That piece of test
gear has a very normal BNC connector on the back of it. They are a
terrible thing for this, but that’s what you get. The cable you use is
going to mate with them. If you use off the shelf gear for the distribution,
it also has BNC’s on it.
Cable needs to be run from here to there. Loss is rarely an issue. Skin
depth vs the shield is very much an issue at 10 MHz. Foil as a shield is
useless in this case. Big thick multiple layers of braid is needed. That
isn’t just a “get RG-400” sort of thing. It’s also who you get it from. The
expensive source may be the one ….
There is a tendency to overdo this kind of thing. Think through how
many endpoints you need. Cable up the reasonable minimum and
see how it goes. Plan on a re-visit in 6 months and maybe again in
12 months. Having a lot of cable to nowhere is a problem in a number
of ways ….
Fun
Bob
On Mar 23, 2023, at 8:45 PM, David Bengtson via time-nuts time-nuts@lists.febo.com wrote:
Has anyone run across any publications on best practices or examples
of 10MHz Lab wide distribution networks? I'm looking for a discussion
on how to physically locate oscillators/distribution amplifiers, cable
types and runs, RFI mitigation etc. I haven't come across any, and I'm
starting to build one at work. We've got a Cs oscillator and I'd like
to make sure we deliver that performance across our systems lab.
Thanks
Dave
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe send an email to time-nuts-leave@lists.febo.com
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe send an email to time-nuts-leave@lists.febo.com
Bob makes a good point about the devices terminating the ends of
reference distribution cables. Many frequency counters have little
shielding, and can be a significant source of 10 MHz radiation,
regardless of the shielding effectiveness of the reference cable runs
feeding them. Counters such as the HP 53132 and 5335 are good examples
of this. Even some signal generators and spectrum analyzers can exhibit
this problem. Despite good shielding in their RF sections, sometimes
inadequate attention was paid to shielding of the external reference inputs.
I encountered this problem when building a reference distribution system
for my low phase noise house standard. The distribution amp was
meticulously shielded and double shielded cable was used throughout,
with runs up to 60'. When individually terminated, the cables showed
very low leakage. However, significant 10 MHz radiation resulted when
connected to numerous test instruments...
73,
John K9WT
On 3/24/2023 8:26 AM, Bob Camp via time-nuts wrote:
> Hi
>
> There are indeed multi year sort of courses (actually sets of courses)
> that various folks put on about the EMI side of this. Unfortunately it does
> quickly get pretty deep in theory and thus the long slog to get to this or
> that level of the knowledge base.
>
> How big is “work” and how big is “lab”? I’ve seen places that measured
> the answers in kilometers. I’ve also seen places that had everything within
> 20 meters of everything else. What you do in one is profoundly different
> than another.
>
> How many “destinations” do you have that need this or that? Again, some
> will have an answer in dozens, for others it will be in the thousands or even
> tens of thousands.
>
> What connects to those end points? If the answer (as noted earlier) is
> a bunch of ADEV / phase noise gear, that’s very different than more
> conventional test gear. Keeping the 10 MHz away from the DUT can be
> a *very* big deal.
>
> Like it or not, much of this has been decided for you. That piece of test
> gear has a very normal BNC connector on the back of it. They are a
> terrible thing for this, but that’s what you get. The cable you use is
> going to mate with them. If you use off the shelf gear for the distribution,
> it also has BNC’s on it.
>
> Cable needs to be run from here to there. Loss is rarely an issue. Skin
> depth vs the shield is very much an issue at 10 MHz. Foil as a shield is
> useless in this case. Big thick multiple layers of braid is needed. That
> isn’t just a “get RG-400” sort of thing. It’s also who you get it from. The
> expensive source may be the one ….
>
> There is a tendency to overdo this kind of thing. Think through how
> many endpoints you need. Cable up the reasonable minimum and
> see how it goes. Plan on a re-visit in 6 months and maybe again in
> 12 months. Having a lot of cable to nowhere is a problem in a number
> of ways ….
>
> Fun
>
> Bob
>
>
>
>> On Mar 23, 2023, at 8:45 PM, David Bengtson via time-nuts <time-nuts@lists.febo.com> wrote:
>>
>> Has anyone run across any publications on best practices or examples
>> of 10MHz Lab wide distribution networks? I'm looking for a discussion
>> on how to physically locate oscillators/distribution amplifiers, cable
>> types and runs, RFI mitigation etc. I haven't come across any, and I'm
>> starting to build one at work. We've got a Cs oscillator and I'd like
>> to make sure we deliver that performance across our systems lab.
>>
>> Thanks
>>
>> Dave
>> _______________________________________________
>> time-nuts mailing list -- time-nuts@lists.febo.com
>> To unsubscribe send an email to time-nuts-leave@lists.febo.com
> _______________________________________________
> time-nuts mailing list -- time-nuts@lists.febo.com
> To unsubscribe send an email to time-nuts-leave@lists.febo.com
MD
Magnus Danielson
Fri, Mar 24, 2023 3:13 PM
Hi,
On that note I noticed that the HP5370A has a severe 5 MHz pulse
polution coming from a 10 MHz external signal detector which
implementation in ECL spews out a lot of noise. All to provide a
diagnostic LED only vissible if you lift the lid. A strategic short
killed the 5 MHz loop and removed the noise, without loosing any
significan function other than this signal present.
I did not make any attempt to measure to which degree this affected
actual measurements on the instruments, which I naturally should have done.
So yes, at times one find that counters are not as clean as you wish
them to be.
Cheers,
Magnus
On 2023-03-24 15:15, John Vendely via time-nuts wrote:
Bob makes a good point about the devices terminating the ends of
reference distribution cables. Many frequency counters have little
shielding, and can be a significant source of 10 MHz radiation,
regardless of the shielding effectiveness of the reference cable runs
feeding them. Counters such as the HP 53132 and 5335 are good
examples of this. Even some signal generators and spectrum analyzers
can exhibit this problem. Despite good shielding in their RF
sections, sometimes inadequate attention was paid to shielding of the
external reference inputs.
I encountered this problem when building a reference distribution
system for my low phase noise house standard. The distribution amp
was meticulously shielded and double shielded cable was used
throughout, with runs up to 60'. When individually terminated, the
cables showed very low leakage. However, significant 10 MHz radiation
resulted when connected to numerous test instruments...
73,
John K9WT
On 3/24/2023 8:26 AM, Bob Camp via time-nuts wrote:
Hi
There are indeed multi year sort of courses (actually sets of courses)
that various folks put on about the EMI side of this. Unfortunately
it does
quickly get pretty deep in theory and thus the long slog to get to
this or
that level of the knowledge base.
How big is “work” and how big is “lab”? I’ve seen places that measured
the answers in kilometers. I’ve also seen places that had everything
within
20 meters of everything else. What you do in one is profoundly different
than another.
How many “destinations” do you have that need this or that? Again, some
will have an answer in dozens, for others it will be in the thousands
or even
tens of thousands.
What connects to those end points? If the answer (as noted earlier) is
a bunch of ADEV / phase noise gear, that’s very different than more
conventional test gear. Keeping the 10 MHz away from the DUT can be
a very big deal.
Like it or not, much of this has been decided for you. That piece of
test
gear has a very normal BNC connector on the back of it. They are a
terrible thing for this, but that’s what you get. The cable you use is
going to mate with them. If you use off the shelf gear for the
distribution,
it also has BNC’s on it.
Cable needs to be run from here to there. Loss is rarely an issue. Skin
depth vs the shield is very much an issue at 10 MHz. Foil as a shield is
useless in this case. Big thick multiple layers of braid is needed. That
isn’t just a “get RG-400” sort of thing. It’s also who you get it
from. The
expensive source may be the one ….
There is a tendency to overdo this kind of thing. Think through how
many endpoints you need. Cable up the reasonable minimum and
see how it goes. Plan on a re-visit in 6 months and maybe again in
12 months. Having a lot of cable to nowhere is a problem in a number
of ways ….
Fun
Bob
On Mar 23, 2023, at 8:45 PM, David Bengtson via time-nuts
time-nuts@lists.febo.com wrote:
Has anyone run across any publications on best practices or examples
of 10MHz Lab wide distribution networks? I'm looking for a discussion
on how to physically locate oscillators/distribution amplifiers, cable
types and runs, RFI mitigation etc. I haven't come across any, and I'm
starting to build one at work. We've got a Cs oscillator and I'd like
to make sure we deliver that performance across our systems lab.
Thanks
Dave
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe send an email to time-nuts-leave@lists.febo.com
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe send an email to time-nuts-leave@lists.febo.com
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe send an email to time-nuts-leave@lists.febo.com
Hi,
On that note I noticed that the HP5370A has a severe 5 MHz pulse
polution coming from a 10 MHz external signal detector which
implementation in ECL spews out a lot of noise. All to provide a
diagnostic LED only vissible if you lift the lid. A strategic short
killed the 5 MHz loop and removed the noise, without loosing any
significan function other than this signal present.
I did not make any attempt to measure to which degree this affected
actual measurements on the instruments, which I naturally should have done.
So yes, at times one find that counters are not as clean as you wish
them to be.
Cheers,
Magnus
On 2023-03-24 15:15, John Vendely via time-nuts wrote:
> Bob makes a good point about the devices terminating the ends of
> reference distribution cables. Many frequency counters have little
> shielding, and can be a significant source of 10 MHz radiation,
> regardless of the shielding effectiveness of the reference cable runs
> feeding them. Counters such as the HP 53132 and 5335 are good
> examples of this. Even some signal generators and spectrum analyzers
> can exhibit this problem. Despite good shielding in their RF
> sections, sometimes inadequate attention was paid to shielding of the
> external reference inputs.
>
> I encountered this problem when building a reference distribution
> system for my low phase noise house standard. The distribution amp
> was meticulously shielded and double shielded cable was used
> throughout, with runs up to 60'. When individually terminated, the
> cables showed very low leakage. However, significant 10 MHz radiation
> resulted when connected to numerous test instruments...
>
> 73,
>
> John K9WT
>
>
> On 3/24/2023 8:26 AM, Bob Camp via time-nuts wrote:
>> Hi
>>
>> There are indeed multi year sort of courses (actually sets of courses)
>> that various folks put on about the EMI side of this. Unfortunately
>> it does
>> quickly get pretty deep in theory and thus the long slog to get to
>> this or
>> that level of the knowledge base.
>>
>> How big is “work” and how big is “lab”? I’ve seen places that measured
>> the answers in kilometers. I’ve also seen places that had everything
>> within
>> 20 meters of everything else. What you do in one is profoundly different
>> than another.
>>
>> How many “destinations” do you have that need this or that? Again, some
>> will have an answer in dozens, for others it will be in the thousands
>> or even
>> tens of thousands.
>>
>> What connects to those end points? If the answer (as noted earlier) is
>> a bunch of ADEV / phase noise gear, that’s very different than more
>> conventional test gear. Keeping the 10 MHz away from the DUT can be
>> a *very* big deal.
>>
>> Like it or not, much of this has been decided for you. That piece of
>> test
>> gear has a very normal BNC connector on the back of it. They are a
>> terrible thing for this, but that’s what you get. The cable you use is
>> going to mate with them. If you use off the shelf gear for the
>> distribution,
>> it also has BNC’s on it.
>>
>> Cable needs to be run from here to there. Loss is rarely an issue. Skin
>> depth vs the shield is very much an issue at 10 MHz. Foil as a shield is
>> useless in this case. Big thick multiple layers of braid is needed. That
>> isn’t just a “get RG-400” sort of thing. It’s also who you get it
>> from. The
>> expensive source may be the one ….
>>
>> There is a tendency to overdo this kind of thing. Think through how
>> many endpoints you need. Cable up the reasonable minimum and
>> see how it goes. Plan on a re-visit in 6 months and maybe again in
>> 12 months. Having a lot of cable to nowhere is a problem in a number
>> of ways ….
>>
>> Fun
>>
>> Bob
>>
>>
>>
>>> On Mar 23, 2023, at 8:45 PM, David Bengtson via time-nuts
>>> <time-nuts@lists.febo.com> wrote:
>>>
>>> Has anyone run across any publications on best practices or examples
>>> of 10MHz Lab wide distribution networks? I'm looking for a discussion
>>> on how to physically locate oscillators/distribution amplifiers, cable
>>> types and runs, RFI mitigation etc. I haven't come across any, and I'm
>>> starting to build one at work. We've got a Cs oscillator and I'd like
>>> to make sure we deliver that performance across our systems lab.
>>>
>>> Thanks
>>>
>>> Dave
>>> _______________________________________________
>>> time-nuts mailing list -- time-nuts@lists.febo.com
>>> To unsubscribe send an email to time-nuts-leave@lists.febo.com
>> _______________________________________________
>> time-nuts mailing list -- time-nuts@lists.febo.com
>> To unsubscribe send an email to time-nuts-leave@lists.febo.com
> _______________________________________________
> time-nuts mailing list -- time-nuts@lists.febo.com
> To unsubscribe send an email to time-nuts-leave@lists.febo.com
BI
Burt I. Weiner
Fri, Mar 24, 2023 3:59 PM
Gang,
I know this probably goes against good
engineering... I have two DATUM 9390 GPS Time
and Frequency "standards" that I feed to several
devices. For distribution I use a Winegard TV
splitter, the blue type in a metal case. The
Winegard splitter is mounted to an unpainted rack
panel in my rack o' stuffs. The output level of
the DATUM standards is high enough so that the
loss in the splitter is not a problem for the
equipment I'm driving; at least nothing
complains. All of my cables are Quad RG-6U cable
with F-Connectors. The cables are on the average
about 4 feet long, with one exception of a cable
that is about 25 feet long. At each receiving
end I have F female to BNC male adaptors - I said
this was not good engineering! I use to have a
Heathkit "Worlds most accurate clock" in my
shoppe that often parked and worked on WWV-10
without any problems. I also have a few
receivers, with an outside dipole antenna, that I
can tune to WWV-10 and barely hear the 10 MHz
from my standards and unorthodox distribution
system. I didn't expect the standard's 10 MHz to
be non-existent in the receivers, but it's a lot lower than expected.
Burt, K6OQK
At 07:15 AM 3/24/2023, you wrote:
Bob makes a good point about the devices
terminating the ends of reference distribution
cables. Many frequency counters have little
shielding, and can be a significant source of 10
MHz radiation, regardless of the shielding
effectiveness of the reference cable runs
feeding them. Counters such as the HP 53132
and 5335 are good examples of this. Even some
signal generators and spectrum analyzers can
exhibit this problem. Despite good shielding
in their RF sections, sometimes inadequate
attention was paid to shielding of the external reference inputs.
I encountered this problem when building a
reference distribution system for my low phase
noise house standard. The distribution amp was
meticulously shielded and double shielded cable
was used throughout, with runs up to 60'. When
individually terminated, the cables showed very
low leakage. However, significant 10 MHz
radiation resulted when connected to numerous test instruments...
73,
John K9WT
On 3/24/2023 8:26 AM, Bob Camp via time-nuts wrote:
Hi
There are indeed multi year sort of courses (actually sets of courses)
that various folks put on about the EMI side of this. Unfortunately it does
quickly get pretty deep in theory and thus the long slog to get to this or
that level of the knowledge base.
How big is âworkâ and how big is âlabâ?
Iâve seen places that measured
the answers in kilometers. Iâve also seen places that had everything within
20 meters of everything else. What you do in one is profoundly different
than another.
How many âdestinationsâ do you have that need this or that? Again, some
will have an answer in dozens, for others it will be in the thousands or even
tens of thousands.
What connects to those end points? If the answer (as noted earlier) is
a bunch of ADEV / phase noise gear, thatâs very different than more
conventional test gear. Keeping the 10 MHz away from the DUT can be
a very big deal.
Like it or not, much of this has been decided for you. That piece of test
gear has a very normal BNC connector on the back of it. They are a
terrible thing for this, but thatâs what you get. The cable you use is
going to mate with them. If you use off the shelf gear for the distribution,
it also has BNCâs on it.
Cable needs to be run from here to there. Loss is rarely an issue. Skin
depth vs the shield is very much an issue at 10 MHz. Foil as a shield is
useless in this case. Big thick multiple layers of braid is needed. That
isnât just a âget RG-400â sort of thing.
Itâs also who you get it from. The
expensive source may be the one
.
There is a tendency to overdo this kind of thing. Think through how
many endpoints you need. Cable up the reasonable minimum and
see how it goes. Plan on a re-visit in 6 months and maybe again in
12 months. Having a lot of cable to nowhere is a problem in a number
of ways
.
Fun
Bob
On Mar 23, 2023, at 8:45 PM, David Bengtson
via time-nuts time-nuts@lists.febo.com wrote:
Has anyone run across any publications on best practices or examples
of 10MHz Lab wide distribution networks? I'm looking for a discussion
on how to physically locate oscillators/distribution amplifiers, cable
types and runs, RFI mitigation etc. I haven't come across any, and I'm
starting to build one at work. We've got a Cs oscillator and I'd like
to make sure we deliver that performance across our systems lab.
Thanks
Dave
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe send an email to time-nuts-leave@lists.febo.com
Burt I. Weiner Associates
Broadcast Technical Services
Glendale, California U.S.A.
biwa@att.net
K6OQK
Gang,
I know this probably goes against good
engineering... I have two DATUM 9390 GPS Time
and Frequency "standards" that I feed to several
devices. For distribution I use a Winegard TV
splitter, the blue type in a metal case. The
Winegard splitter is mounted to an unpainted rack
panel in my rack o' stuffs. The output level of
the DATUM standards is high enough so that the
loss in the splitter is not a problem for the
equipment I'm driving; at least nothing
complains. All of my cables are Quad RG-6U cable
with F-Connectors. The cables are on the average
about 4 feet long, with one exception of a cable
that is about 25 feet long. At each receiving
end I have F female to BNC male adaptors - I said
this was not good engineering! I use to have a
Heathkit "Worlds most accurate clock" in my
shoppe that often parked and worked on WWV-10
without any problems. I also have a few
receivers, with an outside dipole antenna, that I
can tune to WWV-10 and barely hear the 10 MHz
from my standards and unorthodox distribution
system. I didn't expect the standard's 10 MHz to
be non-existent in the receivers, but it's a lot lower than expected.
Burt, K6OQK
At 07:15 AM 3/24/2023, you wrote:
>Bob makes a good point about the devices
>terminating the ends of reference distribution
>cables. Many frequency counters have little
>shielding, and can be a significant source of 10
>MHz radiation, regardless of the shielding
>effectiveness of the reference cable runs
>feeding them. Counters such as the HP 53132
>and 5335 are good examples of this. Even some
>signal generators and spectrum analyzers can
>exhibit this problem. Despite good shielding
>in their RF sections, sometimes inadequate
>attention was paid to shielding of the external reference inputs.
>
>I encountered this problem when building a
>reference distribution system for my low phase
>noise house standard. The distribution amp was
>meticulously shielded and double shielded cable
>was used throughout, with runs up to 60'. When
>individually terminated, the cables showed very
>low leakage. However, significant 10 MHz
>radiation resulted when connected to numerous test instruments...
>
>73,
>
>John K9WT
>
>
>On 3/24/2023 8:26 AM, Bob Camp via time-nuts wrote:
>>Hi
>>
>>There are indeed multi year sort of courses (actually sets of courses)
>>that various folks put on about the EMI side of this. Unfortunately it does
>>quickly get pretty deep in theory and thus the long slog to get to this or
>>that level of the knowledge base.
>>
>>How big is âworkâ and how big is âlabâ?
>>Iâve seen places that measured
>>the answers in kilometers. Iâve also seen places that had everything within
>>20 meters of everything else. What you do in one is profoundly different
>>than another.
>>
>>How many âdestinationsâ do you have that need this or that? Again, some
>>will have an answer in dozens, for others it will be in the thousands or even
>>tens of thousands.
>>
>>What connects to those end points? If the answer (as noted earlier) is
>>a bunch of ADEV / phase noise gear, thatâs very different than more
>>conventional test gear. Keeping the 10 MHz away from the DUT can be
>>a *very* big deal.
>>
>>Like it or not, much of this has been decided for you. That piece of test
>>gear has a very normal BNC connector on the back of it. They are a
>>terrible thing for this, but thatâs what you get. The cable you use is
>>going to mate with them. If you use off the shelf gear for the distribution,
>>it also has BNCâs on it.
>>
>>Cable needs to be run from here to there. Loss is rarely an issue. Skin
>>depth vs the shield is very much an issue at 10 MHz. Foil as a shield is
>>useless in this case. Big thick multiple layers of braid is needed. That
>>isnât just a âget RG-400â sort of thing.
>>Itâs also who you get it from. The
>>expensive source may be the one
.
>>
>>There is a tendency to overdo this kind of thing. Think through how
>>many endpoints you need. Cable up the reasonable minimum and
>>see how it goes. Plan on a re-visit in 6 months and maybe again in
>>12 months. Having a lot of cable to nowhere is a problem in a number
>>of ways
.
>>
>> Fun
>>
>>Bob
>>
>>
>>
>>>On Mar 23, 2023, at 8:45 PM, David Bengtson
>>>via time-nuts <time-nuts@lists.febo.com> wrote:
>>>
>>>Has anyone run across any publications on best practices or examples
>>>of 10MHz Lab wide distribution networks? I'm looking for a discussion
>>>on how to physically locate oscillators/distribution amplifiers, cable
>>>types and runs, RFI mitigation etc. I haven't come across any, and I'm
>>>starting to build one at work. We've got a Cs oscillator and I'd like
>>>to make sure we deliver that performance across our systems lab.
>>>
>>>Thanks
>>>
>>>Dave
>>>_______________________________________________
>>>time-nuts mailing list -- time-nuts@lists.febo.com
>>>To unsubscribe send an email to time-nuts-leave@lists.febo.com
>>_______________________________________________
>>time-nuts mailing list -- time-nuts@lists.febo.com
>>To unsubscribe send an email to
>>time-nuts-leave@lists.febo.com
>>_______________________________________________
>>time-nuts mailing list --
>>time-nuts@lists.febo.com To unsubscribe send an
>>email to time-nuts-leave@lists.febo.com
>
>Burt I. Weiner Associates
>Broadcast Technical Services
>Glendale, California U.S.A.
>biwa@att.net
>K6OQK
DW
Dana Whitlow
Fri, Mar 24, 2023 8:04 PM
What's wrong with fiber optical distribution?
Dana
On Thu, Mar 23, 2023 at 9:54 PM John Miles via time-nuts <
time-nuts@lists.febo.com> wrote:
Has anyone run across any publications on best practices or examples
of 10MHz Lab wide distribution networks? I'm looking for a discussion
on how to physically locate oscillators/distribution amplifiers, cable
types and runs, RFI mitigation etc. I haven't come across any, and I'm
starting to build one at work. We've got a Cs oscillator and I'd like
to make sure we deliver that performance across our systems lab.
A lot of things can be said in favor of low-noise distribution amps with
good VSWR, good channel isolation, and good PSRR (not so fast, HP 5087A.)
But over the years, I've learned that the three most important factors when
it comes to piping 10 MHz signals around are:
- Shield resistance
- Shield resistance
- Shield resistance
I've spent a lot of time recommending double-shielded coax in the TimeLab
manual and elsewhere, and I still stand by that advice, but what I've come
to realize is that this is really just a proxy for low shield resistance.
Good grades of single-shielded cable are basically as effective at HF as
double-shielded cable. To the extent your cable ground shield exhibits
resistance, it's not a shield, it's a resistor.
Avoiding ground loops is on the list too, but further down. Never lift a
ground to avoid a ground loop. Use coax-to-coax baluns only when you can
see a beneficial effect. Focus instead on providing a shared
low-resistance
common ground to your entire network -- ideally not the ground all the way
back at the service entrance -- and rely on low shield resistance on the RF
side to do the rest.
Every installation is different and your mileage will most certainly vary,
but this is my take on it.
-- john
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe send an email to time-nuts-leave@lists.febo.com
What's wrong with fiber optical distribution?
Dana
On Thu, Mar 23, 2023 at 9:54 PM John Miles via time-nuts <
time-nuts@lists.febo.com> wrote:
> > Has anyone run across any publications on best practices or examples
> > of 10MHz Lab wide distribution networks? I'm looking for a discussion
> > on how to physically locate oscillators/distribution amplifiers, cable
> > types and runs, RFI mitigation etc. I haven't come across any, and I'm
> > starting to build one at work. We've got a Cs oscillator and I'd like
> > to make sure we deliver that performance across our systems lab.
>
> A lot of things can be said in favor of low-noise distribution amps with
> good VSWR, good channel isolation, and good PSRR (not so fast, HP 5087A.)
> But over the years, I've learned that the three most important factors when
> it comes to piping 10 MHz signals around are:
>
> 1) Shield resistance
> 2) Shield resistance
> 3) Shield resistance
>
> I've spent a lot of time recommending double-shielded coax in the TimeLab
> manual and elsewhere, and I still stand by that advice, but what I've come
> to realize is that this is really just a proxy for low shield resistance.
> Good grades of single-shielded cable are basically as effective at HF as
> double-shielded cable. To the extent your cable ground shield exhibits
> resistance, it's not a shield, it's a resistor.
>
> Avoiding ground loops is on the list too, but further down. Never lift a
> ground to avoid a ground loop. Use coax-to-coax baluns only when you can
> see a beneficial effect. Focus instead on providing a shared
> low-resistance
> common ground to your entire network -- ideally not the ground all the way
> back at the service entrance -- and rely on low shield resistance on the RF
> side to do the rest.
>
> Every installation is different and your mileage will most certainly vary,
> but this is my take on it.
>
> -- john
>
>
> _______________________________________________
> time-nuts mailing list -- time-nuts@lists.febo.com
> To unsubscribe send an email to time-nuts-leave@lists.febo.com
>
