Date   
Re: Beacon details and recording

Neil Smith G4DBN
 

That tactic was certainly in evidence on the cluster in the 1990s to put the blind callers off the trail. Ditto spotting on bands where the DX wasn't operating.  Decent way to remove (and detect) those without receivers.

Not clear why anyone would do spam spotting of VHF/UHF/SHF beacons though?  Little to be gained from that.  There is an EU station who is notorious for his "impossible" 6m spots, so anything he spots needs to be treated with a very large pinch of salt.

What would be useful is the facility to delete or edit my own spots which I have entered into the cluster with a typo.  The other day my browser auto-filled RS as the mode in a TR spot and I didn't notice. It would be good to be able to amend that in the database.  Luckily, B'Spot is bright enough to realise and marked it as TR from the report.

Neil G4DBN

On 19/10/2019 22:17, Alan G3XAQ wrote:
Perhaps I didn't express myself clearly. I'll try again. If this DX Cluster system of which you speak is the same one as is used on HF (for example, telnet to gb7djk.dxcluster.net on port 7300), then there are arguments for spotters to deliberately issue slightly inaccurate frequencies in spots in some circumstances. Examples include trying to avoid the weekend warriors who blindly point and click from landing on the transmit frequency of a DX station operating split, or more especially having loads of them show their lack of skill by constant calling on top of a "rare one" operating simplex in a contest.

It's horses for courses. One tool can't please everyone all the time.

73, Alan G3XAQ


"Some concerns" is putting it very mildly Alan, but it is a very rich
source of information without which we couldn't have anywhere near as
much detail.  Just needs a lot of human filtering to remove the obvious
dross.  As I said, you need to assess who it was spotted a particular
frequency and assign a level of trust to it.

Of course, if all beacons were locked, we wouldn't need to care about
the frequencies being reported.  I've just received another box of £4
GPS boards with PPS output.  Not exactly expensive to lock things to GPS
these days.

Neil G4DBN

On 19/10/2019 20:23, Alan G3XAQ wrote:
Neil,

Is this the same distributed DX Cluster that the HF DXing community uses? If
so, I have some concerns about reporting accurate frequencies and pileup
management.
Just asking,

Alan G3XAQ



--
Neil
<a href="http://g4dbn.uk/"><small>g4dbn.uk</small></a>

Re: Beacon details and recording

Alan G3XAQ
 

Perhaps I didn't express myself clearly. I'll try again. If this DX Cluster system of which you speak is the same one as is used on HF (for example, telnet to gb7djk.dxcluster.net on port 7300), then there are arguments for spotters to deliberately issue slightly inaccurate frequencies in spots in some circumstances. Examples include trying to avoid the weekend warriors who blindly point and click from landing on the transmit frequency of a DX station operating split, or more especially having loads of them show their lack of skill by constant calling on top of a "rare one" operating simplex in a contest.

It's horses for courses. One tool can't please everyone all the time.

73, Alan G3XAQ


"Some concerns" is putting it very mildly Alan, but it is a very rich
source of information without which we couldn't have anywhere near as
much detail.  Just needs a lot of human filtering to remove the obvious
dross.  As I said, you need to assess who it was spotted a particular
frequency and assign a level of trust to it.

Of course, if all beacons were locked, we wouldn't need to care about
the frequencies being reported.  I've just received another box of £4
GPS boards with PPS output.  Not exactly expensive to lock things to GPS
these days.

Neil G4DBN

On 19/10/2019 20:23, Alan G3XAQ wrote:
Neil,

Is this the same distributed DX Cluster that the HF DXing community uses? If
so, I have some concerns about reporting accurate frequencies and pileup
management.

Just asking,

Alan G3XAQ



--
Neil
<a href="http://g4dbn.uk/"><small>g4dbn.uk</small></a>

Re: PLL Lock Detect indication. ( Was Re: 122 GHz)

Dave_G0WBX
 

:-)

As I said:-

"I'd hope that embedded PLL lock detect schemes have improved over recent years."

73

Dave G8KBV.


From: Andy G4JNT
Date: Sat, 19 Oct 2019 08:46:57 PDT
Yes, but the ADF4153 is not an 'early chip', and has a very good digitally derived in-lock indicator

-- 
Created on and sent from a Unix like PC running and using free and open source software:

Re: Beacon details and recording

Neil Smith G4DBN
 

"Some concerns" is putting it very mildly Alan, but it is a very rich source of information without which we couldn't have anywhere near as much detail.  Just needs a lot of human filtering to remove the obvious dross.  As I said, you need to assess who it was spotted a particular frequency and assign a level of trust to it.

Of course, if all beacons were locked, we wouldn't need to care about the frequencies being reported.  I've just received another box of £4 GPS boards with PPS output.  Not exactly expensive to lock things to GPS these days.

Neil G4DBN

On 19/10/2019 20:23, Alan G3XAQ wrote:
Neil,

Is this the same distributed DX Cluster that the HF DXing community uses? If so, I have some concerns about reporting accurate frequencies and pileup management.

Just asking,

Alan G3XAQ


Re: Beacon details and recording

Neil Smith G4DBN
 

For comparison, this is from DXHeat, just showing the dupes with mangled frequencies:

Something out there is still causing the "nearest 5kHz" rounding bug, and even that is buggy, when 945.0 is rounded DOWN to 940.0 in that 24GHz spot.  Worst thing is not knowing which nodes, running what software, are doing this, so we can't name and shame if/when they refuse to fix it.  Whoever it is must be aware of the bug, it has been there for years now. 

I've reported the dxcluster.co.uk issues to the author and I'm sure he'll fix the fault with rounding to 100kHz.

Doubtless there would be screams of anguish from authors of logging programs and other systems using the cluster protocol if the extra digits were to be permitted, as they would need to ensure they were compliant with frequencies to the nearest 1Hz and with frequencies above 76GHz, but we are not talking Quantum Theory here, just basic data formatting. Oh, and "Testing".

OK, rather than trying to cause a revolution and international incident, perhaps a fix for those of us who care about precision would be to allow spots on B'Spot to 1Hz from registered users, but still only send them out to the cluster rounded to the nearest 100Hz?

Neil G4DBN

On 19/10/2019 20:00, Neil Smith G4DBN wrote:

Its comes down to the level of trust you place in the person making the spot. There are some that I take with a very VERY large pinch of salt and others that I trust. In general, the wisdom of the crowd applies, and if someone's spot is an outlier, you can probably disregard that either as a typo if they are normally accurate, or mark them down further towards the "untrusted" list.

Others to ignore are where the beacon is not locked and is known to wander, but gets spotted on the published frequency, even if that it 4kHz away.

Luckily, the Beaconspot engine is bright enough to ignore the spots from the bug-ridden clusters out there which receive a spot on 10368.8450 and repeat it out on 10369.8500 or 10368.900

Some are now rounding 10GHz spots to the nearest 100kHz and not removing dupes even though they are identical. I guess nobody does testing any more.  I wouldn't trust any spot by the G4DBN bloke, he obviously doesn't know what frequency he's on.

I might decide to have a huge global rant at the cluster software writers demanding 1Hz resolution and support beyong 100GHz, I mean how hard can it possibly be to do a managed migration and set a date for the change and cut off any nodes which don't upgrade or which continue to repeat mangled frequencies?  Still scarred by my long, dark years as a cluster sysop...

Back to B'Spot. Not sure what the problem is about requiring registration. Anyone with a legitimate interest can register as a user.  If we had unlimited bandwidth and server capacity, it might not be necessary.  The wording could be slightly more friendly though, I agree.

Neil G4DBN


On 19/10/2019 17:01, Andy G4JNT wrote:
If your registration doesn't work, email admin and let them know - address at the bottom of that page.
There's no cross checking of frequencies.  If you don't believe you can measure its frequency accurately enough, then don't enter a measured value.



On Sat, 19 Oct 2019 at 16:54, David J Taylor GM8ARV via Groups.Io <gm8arv=yahoo.co.uk@groups.io> wrote:
To whom should a measured beacon frequency be reported, and what cross-check
is there?  Thanks.

By the way, when I try beaconspot I see "no guest".  See attachment.  Yes, I
have registered but not logged in to show the screen which a non-user might
see.

73,
David GM8ARV
--
SatSignal Software - Quality software for you
Web: http://www.satsignal.eu
Email: david-taylor@...
Twitter: @gm8arv



-- 
Neil
<a href="http://g4dbn.uk/"><small>g4dbn.uk</small></a>
-- 
Neil
<a href="http://g4dbn.uk/"><small>g4dbn.uk</small></a>

Re: Beacon details and recording

Alan G3XAQ
 

Neil,

Is this the same distributed DX Cluster that the HF DXing community uses? If so, I have some concerns about reporting accurate frequencies and pileup management.

Just asking,

Alan G3XAQ

Re: Beacon details and recording

Neil Smith G4DBN
 

Its comes down to the level of trust you place in the person making the spot. There are some that I take with a very VERY large pinch of salt and others that I trust. In general, the wisdom of the crowd applies, and if someone's spot is an outlier, you can probably disregard that either as a typo if they are normally accurate, or mark them down further towards the "untrusted" list.

Others to ignore are where the beacon is not locked and is known to wander, but gets spotted on the published frequency, even if that it 4kHz away.

Luckily, the Beaconspot engine is bright enough to ignore the spots from the bug-ridden clusters out there which receive a spot on 10368.8450 and repeat it out on 10369.8500 or 10368.900

Some are now rounding 10GHz spots to the nearest 100kHz and not removing dupes even though they are identical. I guess nobody does testing any more.  I wouldn't trust any spot by the G4DBN bloke, he obviously doesn't know what frequency he's on.

I might decide to have a huge global rant at the cluster software writers demanding 1Hz resolution and support beyong 100GHz, I mean how hard can it possibly be to do a managed migration and set a date for the change and cut off any nodes which don't upgrade or which continue to repeat mangled frequencies?  Still scarred by my long, dark years as a cluster sysop...

Back to B'Spot. Not sure what the problem is about requiring registration. Anyone with a legitimate interest can register as a user.  If we had unlimited bandwidth and server capacity, it might not be necessary.  The wording could be slightly more friendly though, I agree.

Neil G4DBN


On 19/10/2019 17:01, Andy G4JNT wrote:
If your registration doesn't work, email admin and let them know - address at the bottom of that page.
There's no cross checking of frequencies.  If you don't believe you can measure its frequency accurately enough, then don't enter a measured value.



On Sat, 19 Oct 2019 at 16:54, David J Taylor GM8ARV via Groups.Io <gm8arv=yahoo.co.uk@groups.io> wrote:
To whom should a measured beacon frequency be reported, and what cross-check
is there?  Thanks.

By the way, when I try beaconspot I see "no guest".  See attachment.  Yes, I
have registered but not logged in to show the screen which a non-user might
see.

73,
David GM8ARV
--
SatSignal Software - Quality software for you
Web: http://www.satsignal.eu
Email: david-taylor@...
Twitter: @gm8arv



-- 
Neil
<a href="http://g4dbn.uk/"><small>g4dbn.uk</small></a>

Re: 122 GHz

Paul Bicknell
 

Hi Neal 

I think it is more the case of what we can find but as far as I am concerned it will not be aluminium 

It is a right bacon to solder

Paul

 


From: RSGB-Workshop@groups.io [mailto:RSGB-Workshop@groups.io] On Behalf Of Neil Smith G4DBN
Sent: 19 October 2019 11:44
To: RSGB-Workshop@groups.io
Subject: Re: [RSGB-Workshop] 122 GHz

 

I guess the loss numbers above cutoff only apply when there is a perfect match? Rather like coax, where if there is a high SWR, the losses go up.

Below cutoff, the guide is going to look like a pretty serious mismatch anyway, isn't it?

There are dire warnings about surface finish causing extra loss at these frequencies. Using brass with a reamed finish is not going to be as good as a honed/polished finish, but as the longest guide section I'm making is maybe 40mm, the loss is small enough to be unimportant.  (Probably about 4dB/metre for copper, 5dB/m for aluminium and perhaps 7dB/m for brass).  Skin depth in copper at 122GHz is arout 200 nanometres.  Brass skin depth is about twice that, but resistivity is about 4 x that of copper. I can imagine that with such a tiny skin depth, any machining marks over a micrometre deep are going to look like serious corrugations, so the loss is likely to be considerably higher than simple theory suggests.  However, as 40mm of brass with a perfect surface should have a loss of 7 x 0.04 = 0.28dB, and copper would only drop that to 0.16dB, and it is harder to get a really good finish inside copper, I'm sticking to brass or aluminium.

Of course, for these tiny components, it would be perfectly possible to make them from silver bar, but then we'd have to be careful about atmospheric corrosion.

Nice bit of bling though.

With atmospheric loss even in dry air approaching 1dB/km, we need to look for every possible way to improve performance.  There is a nice general-interest paper on the challenges and possibilities at https://www.ericsson.com/en/ericsson-technology-review/archive/2017/microwave-backhaul-evolution-reaching-beyond-100ghz

It mentions adaptive modulation, which is one of things I was banging on about in my talk at the Convention.

Neil G4DBN
http://g4dbn.uk

On 19/10/2019 10:56, John Rabson wrote:

Make sure the guide is not reactively loaded (see George Craven in Electronics Letters 1969) or you may get funny answers.

 

John F5VLF



On 19 Oct 2019, at 00:57, Neil Smith G4DBN <neil@...> wrote:

 

I used a standard formula from Hemming (1992), not taking skin effect into account.  Loss(dB)=0.00181*L*f*((Fc/f)^2+1) where Fc is 17577/tube diameter in cm.  All units cm and MHz.

For a solution taking into account the skin effect, in rectangular guide, try http://www.wa1mba.org/WGCalc.htm

In this example for WR08, cutoff is 74GHz, and loss at 66GHz is about 62dB/cm

<hdlkkelpbgpdacpk.png>

 

For a good treatment that has some fairly serious maths, have a look at https://sci-hub.se/10.1049/ip-smt:20041023

For waveguides above cutoff, surface finish matters a LOT.  Charts at https://www.microwaves101.com/encyclopedias/waveguide-loss and explanations of metal loss.

A recent paper, rather heavy on maths again is https://sci-hub.se/10.1080/02726343.2017.1301198 - just watch the units, they seem keen on Nepers instead of dB, but a Neper is just the natural log equivalent of a Bel, so 1 Neper  is about 8.6dB.

The paper I used for the origfinal calculation was:

Hemming, L. H. (1992). Applying The Waveguide Below Cut-off Principle To Shielded Enclosure Design. International Symposium on Electromagnetic Compatibility. doi:10.1109/isemc.1992.626096 

Extract here:

<nfjllphnhpjmcjmk.png>

Just be VERY careful with the units (cm and MHz).  There are serious short-cuts involved in those constants.  Avoid the area close to a few percent of cutoff as you will get silly numbers.

I bet there is on online Javascript calculatoir for round w/g below cutoff, but my google-fu is weak tonight.

Neil G4DBN

On 18/10/2019 23:28, Paul Bicknell wrote:

Hi Neal 

 

I have been looking for the attenuation figure below cut of relative to length for ages where did you get 

"the more than about 8mm long, the loss is >60dB per cm"

Regards Paul 

 

-----Original Message-----
From: RSGB-Workshop@groups.io [mailto:RSGB-Workshop@groups.io] On Behalf Of Neil Smith G4DBN
Sent: 18 October 2019 23:07
To: RSGB-Workshop@groups.io
Subject: Re: [RSGB-Workshop] 122 GHz

 

Where there is a multiplier chain starting from below 50GHz, I use my 

ancient HP specan, below 20GHz, my janky old Marconi counter.  The final 

frequency then has to traverse a length of waveguide with a cutoff 

usually around 20% lower than the frequency of interest.  Say I am

multiplying from 40GHz to 240GHz, and the waveguide tube is 0.9mm 

diameter.  So long as the guide is more than about 8mm long, the loss is 

 >60dB per cm at the 5th harmonic 200GHz, so the chances of any leakage 

are pretty much zero.

 

You can find the frequency using a diffraction grating, measuring the 

angle between detected peaks using a precision rotary table. Using 

d*sin(theta)=n*lambda, where theta is the measured angle between peaks,

lambda is wavelength to be measured, d is the grating spacing and n is 

0,1,2,3,...

 

At 240.0 GHz, with a 6.00 mm grating, the peaks are at 11.57 degrees.

 

Neil G4DBN

 

On 18/10/2019 22:19, Gareth G4SDW (ne G8DXY) GQRP #3339 via Groups.Io wrote:

> On Fri, Oct 18, 2019 at 08:55 PM, Paul Bicknell wrote:

>> Hi

>> There is a divide by 64 so we can measure that and x  by 64

> So not possible to verify the output frequency should there be a

> failure within the chip.

>> Alternatively use a HP 8566B spectrum analyser

>> 

>  From googling, that would seem to solve the 22, but what of the other 100GHz?  

-- 

Neil

<a href="http://g4dbn.uk/"><small>g4dbn.uk</small></a>

 

 

 

 

 

-----

No virus found in this message.

Checked by AVG - www.avg.com

Version: 2016.0.8048 / Virus Database: 4793/15886 - Release Date: 08/14/18

Internal Virus Database is out of date.

 

-- 
Neil
<a href="http://g4dbn.uk/"><small>g4dbn.uk</small></a>

 

-- 
Neil
<a href="http://g4dbn.uk/"><small>g4dbn.uk</small></a>

No virus found in this message.
Checked by AVG - www.avg.com
Version: 2016.0.8048 / Virus Database: 4793/15886 - Release Date: 08/14/18
Internal Virus Database is out of date.

Re: Beacon details and recording

Andy G4JNT
 

If your registration doesn't work, email admin and let them know - address at the bottom of that page.
There's no cross checking of frequencies.  If you don't believe you can measure its frequency accurately enough, then don't enter a measured value.



On Sat, 19 Oct 2019 at 16:54, David J Taylor GM8ARV via Groups.Io <gm8arv=yahoo.co.uk@groups.io> wrote:
To whom should a measured beacon frequency be reported, and what cross-check
is there?  Thanks.

By the way, when I try beaconspot I see "no guest".  See attachment.  Yes, I
have registered but not logged in to show the screen which a non-user might
see.

73,
David GM8ARV
--
SatSignal Software - Quality software for you
Web: http://www.satsignal.eu
Email: david-taylor@...
Twitter: @gm8arv



Beacon details and recording

David J Taylor GM8ARV
 

To whom should a measured beacon frequency be reported, and what cross-check is there? Thanks.

By the way, when I try beaconspot I see "no guest". See attachment. Yes, I have registered but not logged in to show the screen which a non-user might see.

73,
David GM8ARV
--
SatSignal Software - Quality software for you
Web: http://www.satsignal.eu
Email: david-taylor@...
Twitter: @gm8arv

Re: PLL Lock Detect indication. ( Was Re: 122 GHz)

Andy G4JNT
 

Yes, but the ADF4153 is not an 'early chip', and has a very good digitally derived in-lock indicator



On Sat, 19 Oct 2019 at 16:42, Dave_G0WBX via Groups.Io <g8kbvdave=googlemail.com@groups.io> wrote:

Hi.

That depends.  I have in the past seen other PLL IC's indicate "in lock", where in fact the VCO had failed entirely.

In one case (an early synthesized PMR radio, I forget the model) the oscillator was running at a multiple of where it should have been due to a component failure (a shorted turn in the VCO inductor.)  So technically, though it was "in lock" (only just though) it produced a signal that the downstream system mostly filtered out, so (thankfully) little to no RF got to the PA.

I'd hope that embedded PLL lock detect schemes have improved over recent years.

~ ~ ~

(Related..)  Why on earth, did Yaesu design the FT736 with all those PLL's, and not monitor any of the "lock detect" status lines?   Those rig's will (I have first hand experience of this!) TX out of band if some of them fail, as the VCO free runs outside of the loop lock range.

I wonder how many other commercial rig's, old or new, potentially could suffer that too.  I note that many ex PMR rigs DO monitor the PLL status, and inhibit TX if things are bad.  Some even indicate an error on the front panel, if just a status light.

73.

Dave G8KBV.   (Hearing some interesting aircraft Doppler sounds, with the beam way off axis for GB3UHF)

>><<


Re: 122 GHz
From: Andy G4JNT
Date: Fri, 18 Oct 2019 14:24:04 PDT

If the chip were to fail, the PLL would be unlocked so you'd know the output frequency was wrong.

Andywww.g4jnt.com
-- 
Created on and sent from a Unix like PC running and using free and open source software:

PLL Lock Detect indication. ( Was Re: 122 GHz)

Dave_G0WBX
 

Hi.

That depends.  I have in the past seen other PLL IC's indicate "in lock", where in fact the VCO had failed entirely.

In one case (an early synthesized PMR radio, I forget the model) the oscillator was running at a multiple of where it should have been due to a component failure (a shorted turn in the VCO inductor.)  So technically, though it was "in lock" (only just though) it produced a signal that the downstream system mostly filtered out, so (thankfully) little to no RF got to the PA.

I'd hope that embedded PLL lock detect schemes have improved over recent years.

~ ~ ~

(Related..)  Why on earth, did Yaesu design the FT736 with all those PLL's, and not monitor any of the "lock detect" status lines?   Those rig's will (I have first hand experience of this!) TX out of band if some of them fail, as the VCO free runs outside of the loop lock range.

I wonder how many other commercial rig's, old or new, potentially could suffer that too.  I note that many ex PMR rigs DO monitor the PLL status, and inhibit TX if things are bad.  Some even indicate an error on the front panel, if just a status light.

73.

Dave G8KBV.   (Hearing some interesting aircraft Doppler sounds, with the beam way off axis for GB3UHF)

>><<


Re: 122 GHz
From: Andy G4JNT
Date: Fri, 18 Oct 2019 14:24:04 PDT

If the chip were to fail, the PLL would be unlocked so you'd know the output frequency was wrong.

Andywww.g4jnt.com
-- 
Created on and sent from a Unix like PC running and using free and open source software:

Re: Latest Radcom

Andy G4JNT
 

There seems to have been some serious thread creep here, spinning off into three directions:
   RadCom Proof Reading (the original thread)
   Beacon details and recording  
   PI4 and optimum data modes for VHF beacons

All of which are interesting and no doubt will continue for a while
Can I suggest starting a new thread for the two later breakaways. 




Re: Latest Radcom

Dave_G0WBX
 

Thanks John.

A good view indeed.

Sadly as a Linux user, I find that recent versions of MSHV have some functionality issues with PI4, often just freezing, or making tuning "difficult" in the presence of QRM.  The other modes that software supports, don't seem to suffer the same issues.

Unfortunately the PI-RX program (that works well on Windows) needs not just wine, but 'wine mono' on Linux Mint, and for whatever reason that fails to install here, on either of two Mint 19.2 machines.

73.

    Dave G8KBV.



From: John G4SWX
Date: Fri, 18 Oct 2019 11:35:32 PDT

On Fri, Oct 18, 2019 at 11:27 AM, Dave_G0WBX wrote:

but PI4 does seem to be gaining ground elsewhere,
all-be-it slowly.
Unfortunately none of Joe Taylor's stable are up to the job! I spent a couple of years telling him about the issues of terrestrial propagation at VHF but he chose to listen more to US 'experts'. The excellent presentation from last year's RSGB convention by Bo Hansen OZ2M is here. It is well worth several views.
https://www.youtube.com/watch?v=jFcCncceSN4

John G4SWX


-- 
Created on and sent from a Unix like PC running and using free and open source software:

Re: Hairui Soldering irons

DAVE LEE
 

Hi Brian, have a look at the Circuit Specialist web site.   They have a wide range of solder stations, prices range from reasonable to high.  I have had the CSI450+ (£47.99) for over 6 years and I am very pleased with it. 
There is also a good selection of bits and spare parts available

https://www.circuitspecialists.eu/soldering/soldering-stations/

Regard
Dave Lee
G8ZZK

Re: 122 GHz

Neil Smith G4DBN
 

Hi Peter, the tricky bit is the size, the round guides are 2mm diameter for the large version and a little over 1.8mm for the small ones.  Certainly for rectangular guide, broaching works at larger sizes, but at this sort of dimension, I think the only really practical way to do rectangular guide in a home workshop is to make the part in two halves with pins and then machine each half of the rectangular guide separately.


One possibility that comes to mind is using a polished hard steel mandrel and a compression die in a press (or vice?) to forge a slightly annealed copper tube around the mandrel. That woudl wor-harden the copper and the bore could then be honed to a fine polish using diamond paste and a soft bronze hone.


A sharp  2.00mm reamer does appear to produce an optically-reflective finish which has only sub-micrometre defects though, so perhaps all is good.


Every dB I can save on antenna/feed losses and every dB I win on gain gets me up to another 2km distance at these frequencies, and 4km if the improvements are done at each end.


Many folks will probably wonder why I'm going to these lengths for fractions of a dB.  I blame a character defect which makes me ludicrously competitive against the Laws of Physics.  Also the journey is just as interesting as the destination when you have a hacker mindset.


Neil G4DBN


On 19/10/2019 12:44, Peter Chadwick via Groups.Io wrote:

Neil,


Possibly a broach would give the necessary finish.



Re: 122 GHz

Peter Chadwick
 

Neil,


Possibly a broach would give the necessary finish.


73


Peter G3RZP



------ Original Message ------
From: "Neil Smith G4DBN" <neil@...>
To: RSGB-Workshop@groups.io
Sent: Saturday, 19 Oct, 19 At 11:44
Subject: Re: [RSGB-Workshop] 122 GHz

I guess the loss numbers above cutoff only apply when there is a perfect match? Rather like coax, where if there is a high SWR, the losses go up.

Below cutoff, the guide is going to look like a pretty serious mismatch anyway, isn't it?

There are dire warnings about surface finish causing extra loss at these frequencies. Using brass with a reamed finish is not going to be as good as a honed/polished finish, but as the longest guide section I'm making is maybe 40mm, the loss is small enough to be unimportant. (Probably about 4dB/metre for copper, 5dB/m for aluminium and perhaps 7dB/m for brass). Skin depth in copper at 122GHz is arout 200 nanometres. Brass skin depth is about twice that, but resistivity is about 4 x that of copper. I can imagine that with such a tiny skin depth, any machining marks over a micrometre deep are going to look like serious corrugations, so the loss is likely to be considerably higher than simple theory suggests. However, as 40mm of brass with a perfect surface should have a loss of 7 x 0.04 = 0.28dB, and copper would only drop that to 0.16dB, and it is harder to get a really good finish inside copper, I'm sticking to brass or aluminium.

Of course, for these tiny components, it would be perfectly possible to make them from silver bar, but then we'd have to be careful about atmospheric corrosion.

Nice bit of bling though.

With atmospheric loss even in dry air approaching 1dB/km, we need to look for every possible way to improve performance. There is a nice general-interest paper on the challenges and possibilities at https://www.ericsson.com/en/ericsson-technology-review/archive/2017/microwave-backhaul-evolution-reaching-beyond-100ghz

It mentions adaptive modulation, which is one of things I was banging on about in my talk at the Convention.

Neil G4DBN
http://g4dbn.uk

On 19/10/2019 10:56, John Rabson wrote:
Make sure the guide is not reactively loaded (see George Craven in Electronics Letters 1969) or you may get funny answers.

John F5VLF

On 19 Oct 2019, at 00:57, Neil Smith G4DBN <neil@...> wrote:

I used a standard formula from Hemming (1992), not taking skin effect into account. Loss(dB)=0.00181*L*f*((Fc/f)^2+1) where Fc is 17577/tube diameter in cm. All units cm and MHz.

For a solution taking into account the skin effect, in rectangular guide, try http://www.wa1mba.org/WGCalc.htm

In this example for WR08, cutoff is 74GHz, and loss at 66GHz is about 62dB/cm

<hdlkkelpbgpdacpk.png>


For a good treatment that has some fairly serious maths, have a look at https://sci-hub.se/10.1049/ip-smt:20041023

For waveguides above cutoff, surface finish matters a LOT. Charts at https://www.microwaves101.com/encyclopedias/waveguide-loss and explanations of metal loss.

A recent paper, rather heavy on maths again is https://sci-hub.se/10.1080/02726343.2017.1301198 - just watch the units, they seem keen on Nepers instead of dB, but a Neper is just the natural log equivalent of a Bel, so 1 Neper is about 8.6dB.

The paper I used for the origfinal calculation was:

Hemming, L. H. (1992). Applying The Waveguide Below Cut-off Principle To Shielded Enclosure Design. International Symposium on Electromagnetic Compatibility. doi:10.1109/isemc.1992.626096
Extract here:

<nfjllphnhpjmcjmk.png>

Just be VERY careful with the units (cm and MHz). There are serious short-cuts involved in those constants. Avoid the area close to a few percent of cutoff as you will get silly numbers.

I bet there is on online Javascript calculatoir for round w/g below cutoff, but my google-fu is weak tonight.

Neil G4DBN

On 18/10/2019 23:28, Paul Bicknell wrote:
Hi Neal
I have been looking for the attenuation figure below cut of relative to length for ages where did you get
"the more than about 8mm long, the loss is >60dB per cm"
Regards Paul
-----Original Message-----
From: RSGB-Workshop@groups.io [mailto:RSGB-Workshop@groups.io] On Behalf Of Neil Smith G4DBN
Sent: 18 October 2019 23:07
To: RSGB-Workshop@groups.io
Subject: Re: [RSGB-Workshop] 122 GHz
Where there is a multiplier chain starting from below 50GHz, I use my
ancient HP specan, below 20GHz, my janky old Marconi counter. The final
frequency then has to traverse a length of waveguide with a cutoff
usually around 20% lower than the frequency of interest. Say I am
multiplying from 40GHz to 240GHz, and the waveguide tube is 0.9mm
diameter. So long as the guide is more than about 8mm long, the loss is
>60dB per cm at the 5th harmonic 200GHz, so the chances of any leakage
are pretty much zero.
You can find the frequency using a diffraction grating, measuring the
angle between detected peaks using a precision rotary table. Using
d*sin(theta)=n*lambda, where theta is the measured angle between peaks,
lambda is wavelength to be measured, d is the grating spacing and n is
0,1,2,3,...
At 240.0 GHz, with a 6.00 mm grating, the peaks are at 11.57 degrees.
Neil G4DBN
On 18/10/2019 22:19, Gareth G4SDW (ne G8DXY) GQRP #3339 via Groups.Io wrote:
> On Fri, Oct 18, 2019 at 08:55 PM, Paul Bicknell wrote:
>
>> Hi
>> There is a divide by 64 so we can measure that and x by 64
> So not possible to verify the output frequency should there be a
> failure within the chip.
>
>> Alternatively use a HP 8566B spectrum analyser
>>
> From googling, that would seem to solve the 22, but what of the other 100GHz?
--
Neil
<a href="http://g4dbn.uk/"><small>g4dbn.uk</small></a>
-----
No virus found in this message.
Checked by AVG - www.avg.com
Version: 2016.0.8048 / Virus Database: 4793/15886 - Release Date: 08/14/18
Internal Virus Database is out of date.
-- 
Neil
<a href="http://g4dbn.uk/"><small>g4dbn.uk</small></a>

-- 
Neil
<a href="http://g4dbn.uk/"><small>g4dbn.uk</small></a>

Re: The Knack

Peter Chadwick
 

I read Dilbert everyday......it is surprising how accurate some of it is regarding management attitudes.


73


Peter G3RZP



------ Original Message ------
From: "Andy G4JNT" <andy.g4jnt@...>
To: RSGB-Workshop@groups.io
Sent: Saturday, 19 Oct, 19 At 11:23
Subject: [RSGB-Workshop] The Knack

It's been around quite some time, but I was reminded again just now


or here (with adverts)



Re: Latest Radcom

Peter Chadwick
 

Incidentally, Andy, going back 30+ years, RadCom had a staff of six and so there was higher degree of capability of proof reading. Additionally, articles accepted by the then T & P Committee had the page proofs sent back to the author for checking, but it did add to the turn round time, especially if the author wasn't able to proof read and return immediately. Once you have the space allocated in the magazine, you just can't be awaiting the author to return the proofs at his or her leisure.


I have an article due to appear in the ARRL National Contest Journal January/February issue next year. They will send me the proofs by email, but I doubt it will be before the middle to end of November and they will need an answer within seven days at the longest. (I have to accept that there are problems in that being a US magazine, they don't accept my spelling of certain words, which complicates proof reading!)


Things are just nowhere near as leisurely in publishing as was once the case.


vy 73


Peter G3RZP


Re: 122 GHz

Neil Smith G4DBN
 

I guess the loss numbers above cutoff only apply when there is a perfect match? Rather like coax, where if there is a high SWR, the losses go up.

Below cutoff, the guide is going to look like a pretty serious mismatch anyway, isn't it?

There are dire warnings about surface finish causing extra loss at these frequencies. Using brass with a reamed finish is not going to be as good as a honed/polished finish, but as the longest guide section I'm making is maybe 40mm, the loss is small enough to be unimportant.  (Probably about 4dB/metre for copper, 5dB/m for aluminium and perhaps 7dB/m for brass).  Skin depth in copper at 122GHz is arout 200 nanometres.  Brass skin depth is about twice that, but resistivity is about 4 x that of copper. I can imagine that with such a tiny skin depth, any machining marks over a micrometre deep are going to look like serious corrugations, so the loss is likely to be considerably higher than simple theory suggests.  However, as 40mm of brass with a perfect surface should have a loss of 7 x 0.04 = 0.28dB, and copper would only drop that to 0.16dB, and it is harder to get a really good finish inside copper, I'm sticking to brass or aluminium.

Of course, for these tiny components, it would be perfectly possible to make them from silver bar, but then we'd have to be careful about atmospheric corrosion.

Nice bit of bling though.

With atmospheric loss even in dry air approaching 1dB/km, we need to look for every possible way to improve performance.  There is a nice general-interest paper on the challenges and possibilities at https://www.ericsson.com/en/ericsson-technology-review/archive/2017/microwave-backhaul-evolution-reaching-beyond-100ghz

It mentions adaptive modulation, which is one of things I was banging on about in my talk at the Convention.

Neil G4DBN
http://g4dbn.uk

On 19/10/2019 10:56, John Rabson wrote:
Make sure the guide is not reactively loaded (see George Craven in Electronics Letters 1969) or you may get funny answers.

John F5VLF

On 19 Oct 2019, at 00:57, Neil Smith G4DBN <neil@...> wrote:

I used a standard formula from Hemming (1992), not taking skin effect into account.  Loss(dB)=0.00181*L*f*((Fc/f)^2+1) where Fc is 17577/tube diameter in cm.  All units cm and MHz.

For a solution taking into account the skin effect, in rectangular guide, try http://www.wa1mba.org/WGCalc.htm

In this example for WR08, cutoff is 74GHz, and loss at 66GHz is about 62dB/cm

<hdlkkelpbgpdacpk.png>


For a good treatment that has some fairly serious maths, have a look at https://sci-hub.se/10.1049/ip-smt:20041023

For waveguides above cutoff, surface finish matters a LOT.  Charts at https://www.microwaves101.com/encyclopedias/waveguide-loss and explanations of metal loss.

A recent paper, rather heavy on maths again is https://sci-hub.se/10.1080/02726343.2017.1301198 - just watch the units, they seem keen on Nepers instead of dB, but a Neper is just the natural log equivalent of a Bel, so 1 Neper  is about 8.6dB.

The paper I used for the origfinal calculation was:

Hemming, L. H. (1992). Applying The Waveguide Below Cut-off Principle To Shielded Enclosure Design. International Symposium on Electromagnetic Compatibility. doi:10.1109/isemc.1992.626096 
Extract here:

<nfjllphnhpjmcjmk.png>

Just be VERY careful with the units (cm and MHz).  There are serious short-cuts involved in those constants.  Avoid the area close to a few percent of cutoff as you will get silly numbers.

I bet there is on online Javascript calculatoir for round w/g below cutoff, but my google-fu is weak tonight.

Neil G4DBN

On 18/10/2019 23:28, Paul Bicknell wrote:
Hi Neal 
 
I have been looking for the attenuation figure below cut of relative to length for ages where did you get 
"the more than about 8mm long, the loss is >60dB per cm"
Regards Paul 
 
-----Original Message-----
From: RSGB-Workshop@groups.io [mailto:RSGB-Workshop@groups.io] On Behalf Of Neil Smith G4DBN
Sent: 18 October 2019 23:07
To: RSGB-Workshop@groups.io
Subject: Re: [RSGB-Workshop] 122 GHz
 
Where there is a multiplier chain starting from below 50GHz, I use my 
ancient HP specan, below 20GHz, my janky old Marconi counter.  The final 
frequency then has to traverse a length of waveguide with a cutoff 
usually around 20% lower than the frequency of interest.  Say I am
multiplying from 40GHz to 240GHz, and the waveguide tube is 0.9mm 
diameter.  So long as the guide is more than about 8mm long, the loss is 
 >60dB per cm at the 5th harmonic 200GHz, so the chances of any leakage 
are pretty much zero.
 
You can find the frequency using a diffraction grating, measuring the 
angle between detected peaks using a precision rotary table. Using 
d*sin(theta)=n*lambda, where theta is the measured angle between peaks,
lambda is wavelength to be measured, d is the grating spacing and n is 
0,1,2,3,...
 
At 240.0 GHz, with a 6.00 mm grating, the peaks are at 11.57 degrees.
 
Neil G4DBN
 
On 18/10/2019 22:19, Gareth G4SDW (ne G8DXY) GQRP #3339 via Groups.Io wrote:
> On Fri, Oct 18, 2019 at 08:55 PM, Paul Bicknell wrote:
>> Hi
>> There is a divide by 64 so we can measure that and x  by 64
> So not possible to verify the output frequency should there be a
> failure within the chip.
>> Alternatively use a HP 8566B spectrum analyser
>> 
>  From googling, that would seem to solve the 22, but what of the other 100GHz?  
-- 
Neil
<a href="http://g4dbn.uk/"><small>g4dbn.uk</small></a>
 
 
 
 
 
-----
No virus found in this message.
Checked by AVG - www.avg.com
Version: 2016.0.8048 / Virus Database: 4793/15886 - Release Date: 08/14/18
Internal Virus Database is out of date.
 
-- 
Neil
<a href="http://g4dbn.uk/"><small>g4dbn.uk</small></a>

-- 
Neil
<a href="http://g4dbn.uk/"><small>g4dbn.uk</small></a>