Moonbounce on 1296

[VK2OMD]

Dave,

If you have looked into the G/T model, you will see that it depends on assumptions for all noise sources, including spillover noise.

I have updated the G/T model to 'calibrate' it to your preliminary Sun noise rise measurement of about 12.5dB and other assumed gain, noise and loss values. (I have reservations about the accuracy of that measure... but lets use it for the interim.)

Above is a pie chart showing the contributions to total system noise from system components. Pretty much, I have filled in everything we reasonably know (including optimal gain), and adjusted spillover noise to obtain the same G/T as indicated by your first measurements.

As you make more accurate measurements of the Sun noise rise, we get a better fix on G/T ratio. We can then adjust the model to achieve that measured G/T ratio, making some further assumptions about the contribution of components. This process drives focus on the larger components to get a better handle on them and perhaps find cost effective improvements.

Everyone looks at the same quiet sky, the same sun, they probably have a reasonable fix on the NF of their receiver system, it is the antennas and ambient noise environment that vary. Antenna gain and spillover noise become the big variables.

In Googling around, I do not see reliable reports of 'state of the art' G/T for ham stations on 1296. Perhaps I have looked in the wrong places.

I have seen reports of Sun noise rise and solar flux, but my suspicion is that they are 10.7cm solar flux figures, or they use an inaccurate estimate of 23cm solar flux from 10.7cm solar flux, or worse. Some of this is a result of the false idea that 10.7cm solar flux is just some derived index and that noise at all frequencies correlates exactly with that metric. (The error in this approach is demonstrated in Fig 3 and Fig 4 of Quiet sun radio flux interpolations.)

Reliable reports of Sun noise rise at a stated 23cm solar flux are very usable, G/T can be calculated directly from that information.

Sun noise rise reports for 1296 at known dates are of use as the measured solar flux can be extracted from archives.

Message for you in all of this, keep record of your measurements, what you measured, the setup, the date and time. You might capture information that is of value to you at a later time.

I should mention that the model does not include mismatch losses other than that the coax has been estimated at a load end VSWR of 1.5. They could be included, but they are very small, much smaller than the uncertainty associated with most things that we have estimated. I know one calculator around seems to implement the concept that the reflected wave is lost from the system. That is a simplistic and misguided notion, a myth that appeals to FLs more than technicians with an understanding of transmission lines.

Owen

[VK2OMD]

Dave,

I have updated the G/T model using estimates of spillover noise and efficiency from a study by Paul Wade of the Septum feed.

I don't know your f/D ratio, but I have guessed that it is somewhere about 0.3 from your pic.

The model is mostly estimates of things, but they are probably not far from the mark, and indicate the best achievable with the equipment configuration.

I have searched the 'net looking for reliable G/T figures of other stations. My though is that normalising G/T by dividing by the diameter^2 gives an idea of state of the art (though of course small dishes will suffer more from blocking).

The task is made difficult by a lack of reliable G/T reports.

There are plenty of Sun noise rise reports that are given in absence of solar flux, or apparently using 10.7cm solar flux (as if solar flux was contant with time and frequency), or date. Sun noise rise AND solar flux at 23cm can be used to calculate G/T.

The above shows how Sun noise rise at 23cm depends on solar flux at 23cm. Estimates of 23cm flux have error, some methods have more error than others.

Which reminds me that a single Sun noise measurement is of limited value, you need to find the largest Sun noise rise that you consistently obtain when the Sun is quiet. If you do not move the antenna during a measurement (ie point the antenna at a place where the Sun will be located soon, and allow the Sun to transit the boresight), you eliminate the statistical noise caused by changing spillover noise with antenna movement.

Owen

[VK2JDS]

some very good info in there, looks like the expected G/T is close to whats being measured which makes me happy also i must keep a detailed log of all experiments from now on and pattern using a stationary dish pass of the of the sun, the trick is predicting where the sun is going to be so it will cross the centre perfectly , i have tried a few times. next job is to move the detector to speaker output and get a higher amplitude reading.
this morning the return pings were noticably better than on the weekend and all were audible.

the dish here has a skirt around the edge on the half closest to the ground made of 250mm of bird netting to help reduce ground noise spillover at low elevations but i doubt it makes any difference to actual dish gain as its rough however it will affect that spillover figure of 52.2.
i tried plugging in some different figures and looked at the expected G/T and can confirm it does all work fine using Open Office under Linux. very handy!

sm5le did do some experiments on a choke ring for the septum too , with mixed results as he has a 2.2 metre dish and the shading by the extra diameter of the ring assembly tended to negate some of the improvement but its something i will try on this dish.
the narrow filter is to make it a bit easier for working cw, its difficult with all the noise and flutter as it is at the moment. hb9hal was quite strong a few days ago, but difficult copy without a decent filter, would have been nice to work him but need some more power, hence some research into wilkinson combiners..
had some emails too from people asking about pointing and tracking hardware and software for their dishes.
once they are up and running they can do the G/T tests too.
thanks again, 73's

[VK2OMD]

...
sm5le did do some experiments on a choke ring for the septum too , with mixed results as he has a 2.2 metre dish and the shading by the extra diameter of the ring assembly tended to negate some of the improvement but its something i will try on this dish.

I still have no idea of the f/D ratio of your dish. So much depends on this number!

Have you read Paul Wade's paper on Septum feeds. He discusses a number of arrangements, including choke rings.

It seems intuitive to me that one or more chokes may reduce current flowing on the substantial outside surface of the Septum feed, which should help reduce side lobes / spillover with a small win on efficiency due to more of the energy directed to the reflector. However, in reality, it remains the balance between illuminating the reflector effectively whilst minimising side lobes / spillover, and the convenience of a dual mode circular polariser.

All of this has to be considered in context of your intended operating environment. Getting best G/T at high elevations looking at the Sun against adjacent cold sky is not necessarily a good indicator of performance on say a terrestrial shot where ambient noise will be much higher. Moon shots sit somewhere in between where the moon provides a noisy background (compared to cold sky) and ground noise may be higher than a high elevation shot, depending on the antenna pattern.

Measuring and optimising G/T at high elevation on Sun noise does allow you to find the location of the feedpoint for highest gain, and more importantly, it gives you an absolute measure of system performance that can be used to obtain confidence that your equipment is working as well as it should, and for comparison of configuration changes over a longer time frame.

High resolution of the Sun noise measurements is important to being able to reliably differentiate system improvements etc. Use the G/T model spreadsheet to explore sensitivity of Sun noise rise to configuration changes, eg if your preamp dropped 5dB of gain to 16dB, the expected Sun noise rise will fall by only 0.6dB. If the uncertainty of each of your noise measurements was +/- 0.5dB (eg digital multimeter), you might not see some changes as significant.

One of the solutions to making high res noise measurements is wideband noise meters. Using a bandwidth of say 1MHz, you would expect 0.05dB accuracy with 30ms integration time, but when you use 2kHz bandwidth, you need 15 seconds integration time for the same resolution.

NFM offers another solution to high res noise measurements by making true audio power measurement over long integration times.

Owen

[VK2OMD]

Dave,

...
its a single stage preamp, claimed gain 21 db (no need for all that extra gain with the 2 stage version, its just setting the front end NF)
0.51 is its optimised setting , i suggest we start at 0.56 if we can as most of us building this preamp wont have access to a lab.

Have you explored the effect on G/T of the two gain options, 21 and 33dB... do it for NF=0.51dB, there is a significant risk that the preamp meets spec!!!!

If it wins you a G/T improvement of say 0.3dB/K for instance, a S/N improvement of 0.3dB, isn't it just the easiest win you could make?

So, what is the modelled improvement from 21dB to 33dB gain in your case?

I know Mark states "Most receive setups including Satellite receivers will only usually require the single stage version. If the preamplifier is to be mounted at the antenna with lossy cable, then the dual stage version should be used." in his description of the EME103. The G/T model reveals the sensitivity of G/T (or S/N) to LNA gain.

Owen

[VK2JDS]

thats a very interesting observation!, do you think the increased noise would affect the mixer? overdriving the front end and inducing agc action that will inadvertantly upset the receiver perhaps? i will have to experiment with this.
my E&C tafe cert microwaves and rx courses many yonks ago taught me to set the front end with lna at feed then build gain afterwards... well i guess it is building gain isnt it.

todays pings were clear and mostly audible but the moon is very difficult to see as its just a thin wedge , so i was navigating by instruments using a analogue elevation meter, then swinging the dish left and right a tad from a compass bearing till i was seeing return pings. that works.

righto , off to do some modeling with the dual stage amp under linux.

back again , well i set NF of the preamp to 0.56 as a guess until the lab in orange finishes with the preamp i supplied them and receiver noise figure at antenna connector is 0.87dB with 21dB of lna gain equating to line 31 total G/T 13.0, now plugging in 33 db of lna gain produces 0.77dB noise figure at the antenna conector and a overall G/T is now 13.3 dB and thats as far as it can be pushed by adding in more and more gain.
13.4 dB total G/T with 0.51dB lna. interesting results indeed, off to install the second stage !
73's

[VK2JDS]

after that last post I am now crashing back to earth... no luck with pings using dual stage, all noise and no signal. working on that.

now to the measurements : due to the small signal level from line output of the radio its skewed the readings by being in the nonlinear region of the diode.
setting up with oa91 diode and a 1uf to smooth it out I plugged it into the speaker output socket on the back of the set.
it was jumping about a lot, at a guess sky noise 62mV, sun 208mV = 10.5db. this ratio seemed to vary a bit with settings of the volume knob..
so i added a 47uF cap across the output to the dmm and that really settled it down. each measurement i waited 30 seconds before averaging the small changes to get a final figure. db=20*log(sunvolt/skyvolt)

today at 10:30am i adjusted the volume to set sky noise to 58 millivolts approximately and then swung up on the sun. this resulted in 184 mV of sun noise = 10.03db and 142mV when switched into a 50 ohm microwave rated dummy load.
10:50 am 180mV and 55mV = 10.3db
12:30 pm 167mV and 46.5mV = 11.1db
2 pm 167mV and 47mV = 11.01db with 123mV when switched into the dummy load.
2:30 pm 162mV and 47mV = 10.75db.

also i measured the f/d ratio finally , its 0.38 using a long bit of wood spanning the dish and measuring back to the centre
f=diameter squared/16*depth
what really surprised me was the sag in this dish. i measured 4.6 metres across the dish but 4.55 metres top to bottom.. thats bad. not to mention how much its sagged backwards with age too
73's

[VK2OMD]

after that last post I am now crashing back to earth... no luck with pings using dual stage, all noise and no signal. working on that.

now to the measurements : due to the small signal level from line output of the radio its skewed the readings by being in the nonlinear region of the diode.
setting up with oa91 diode and a 1uf to smooth it out I plugged it into the speaker output socket on the back of the set.
it was jumping about a lot, at a guess sky noise 62mV, sun 208mV = 10.5db. this ratio seemed to vary a bit with settings of the volume knob..
so i added a 47uF cap across the output to the dmm and that really settled it down. each measurement i waited 30 seconds before averaging the small changes to get a final figure. db=20*log(sunvolt/skyvolt)

today at 10:30am i adjusted the volume to set sky noise to 58 millivolts approximately and then swung up on the sun. this resulted in 184 mV of sun noise = 10.03db and 142mV when switched into a 50 ohm microwave rated dummy load.
10:50 am 180mV and 55mV = 10.3db
12:30 pm 167mV and 46.5mV = 11.1db
2 pm 167mV and 47mV = 11.01db with 123mV when switched into the dummy load.
2:30 pm 162mV and 47mV = 10.75db.

also i measured the f/d ratio finally , its 0.38 using a long bit of wood spanning the dish and measuring back to the centre
f=diameter squared/16*depth
what really surprised me was the sag in this dish. i measured 4.6 metres across the dish but 4.55 metres top to bottom.. thats bad. not to mention how much its sagged backwards with age too
73's

Dave,

Some thoughts...

Starting with the f/D ratio of 0.38.

Above is Paul Wade's NEC model results of a standard Septum feed.

With an f/D of 0.38, your dish is quite suited to the feed or vice versa without modification (eg a horn flare).

The chart suggests that your antenna should have an efficiency up to 61% (gain=33.8dBi), and spillover of about 16% (16% of 290 = 46K).

Those models are for a solid dish, so in practice, you will also encounter higher spillover noise due to the mesh dish, and gain may be lower due to structure distortion etc.

Plugging all those into the G/T model (including the 33dB gain LNA), gives a Sun noise rise of 13.3dB at solar flux 56 (as it was today). The model indicates the ideal system Teq at the antenna connector would be 41K, NF=0.57.

If you replace the feed unit with a termination, total noise power is 290+41=331K. Lets say spillover is 60K, and you look at cold sky (lets use 8K), the total noise power is 60+8+41=109K. The ratio of rx noise output for Term/ColdSky would be 331/109 or 4.8dB. Conversely, you can measure the noise output ratio and determine the external noise (knowing Teq of the receiver).

Taking your 2pm measurement of ColdSky=47mV and Term=123mV, and assuming the Rx Teq is 41K:
The noise power for the Term is proportional to 290+41=331K. The noise temp for ColdSky+spillover+RxTeq = (47/123)^2*331=48.3K. When we deduct the RxTeq of 41, it gives total external noise of 7.3K, which is much lower than expected 8K of cold sky and 60K of spillover, it is unrealistic.

What should we expect for Term / ColdSky? See above, 4.8dB, or about 173% voltage.

Making reliable accurate measurements of the noise underpins the whole method. You need to look for non linearity, eg clipping, AGC, instrument error. If you can't get the same ratios at different audio output levels, something is non-linear... and you just have to work through it.

Hope this helps.

Owen

PS: implicit in the above is that T1/T2 is a power ratio, and (V1/V2)^2 is a power ratio, in dB, 10*log(T1/T2) or 20*log(V1/V2).

[VK2OMD]

...
Taking your 2pm measurement of ColdSky=47mV and Term=123mV, and assuming the Rx Teq is 41K:
The noise power for the Term is proportional to 290+41=331K. The noise temp for ColdSky+spillover+RxTeq = (47/123)^2*331=48.3K. When we deduct the RxTeq of 41, it gives total external noise of 7.3K, which is much lower than expected 8K of cold sky and 60K of spillover, it is unrealistic.
...

You can use the Ambient noise calculator to perform this calc, here is how to use it:

The key result is Ta, which represents all external noise (ColdSky, spillover, feedthru etc). (Hint: specificying a very large attenuator in one of the steps in the calculator is equivalent to using a termination.

Owen

PS: I have updated the calculator so that you can enter rx noise temperature directly... been meaning to do it for ages. You may need to use the refresh button to pickup the latest version.

[VK2JDS]

yes , a linear detection circuit with variable rates of integration is what we need to make accurate measurements.
as an example i have measured the system here using 3 different volume levels using the diode and capacitor ac>dc converter:

local time 30-8-2008

time : sun noise : cold sky : terminator :
7:55 am 185 mV 57mV 132mV
7:57 am 305mv 100mV 215mV
8:00am 571mV 220mv 420mV

the differences between sky and sun noise vary with the volume control and get worse as the level increases
7:55am > 10.22 dB, 7:57 am > 9.68 dB, 8:00am > 8.28 dB

its proportional to sky to load ratio too
7:55am > 7.29 dB, 7:57am > 6.65 dB,8:00am > 5.61 dB

clearly this is an unscientific method of measurement of noise. so , I will go hunting around for some sort of linear detector to add to the IC910...

[VK2OMD]

yes , a linear detection circuit with variable rates of integration is what we need to make accurate measurements.
as an example i have measured the system here using 3 different volume levels using the diode and capacitor ac>dc converter:

local time 30-8-2008

time : sun noise : cold sky : terminator :
7:55 am 185 mV 57mV 132mV
7:57 am 305mv 100mV 215mV
8:00am 571mV 220mv 420mV

the differences between sky and sun noise vary with the volume control and get worse as the level increases
7:55am > 10.22 dB, 7:57 am > 9.68 dB, 8:00am > 8.28 dB

its proportional to sky to load ratio too
7:55am > 7.29 dB, 7:57am > 6.65 dB,8:00am > 5.61 dB

clearly this is an unscientific method of measurement of noise. so , I will go hunting around for some sort of linear detector to add to the IC910...

Dave,

OK, I can see the effect you are observing, and it is frustrating, but it is part of proving the measurement system.

The IC910 probably uses a transformerless audio output stage, and is specified to produce 2W (10% THD) into 8 ohms.

Assuming that the first onset of clipping is the audio output stage:

If the radio is running from 12V, and you allow 0.5V for Vces, you have 5.5Vp available, over 16 ohms you have 1.9W at the onset of clipping. Icom probably make 2W with 14V supply.

Ok, you have 5.5Vp max, lets give ourselves a little margin for other voltage drops and call it 5Vp.

The peak value of noise voltage is quite a deal higher than the RMS. The article Headroom in white noise measurements discusses the issue. Lets allow 8dB of headroom (11dB would be better, but lets use 8dB as a bare minimum)... so you don't want the audio output to be greater than 5V less 8dB or 2V RMS.

Your contrived audio voltmeter should not respond to peak value, it should respond to the average. That means the diode should not feed a capacitor input filter, just a moving coil DC voltmeter which will average the input current. Such a half wave voltmeter will produce an average voltage of close to 1V from 2V RMS noise voltage.

A capacitor input filter is less predictable.

The voltmeter will be a little non-linear due to the diode voltage drop.

Without changing the level, make 20 measurements of noise and write them down. Now cross off the very largest and the very smallest readings so the middle 90% are left. Now find the largest and smallest remaining values and calculate 10*log(largest/smallest), that is the 90% uncertainty of your measurement on 'constant' noise. This figure needs to be of the order of a tenth of a dB for what you are trying to do.

I regularly make good measurements of receiver output voltage. I use two main methods:

• - an AC millivoltmeter permananently across the receiver audio output. The AC millivoltmeter gives calibrated readings from millivolt level up to the receiver output saturation voltage; and
  - NFM for higher precision measurements.

The millivoltmeter swings around +/- 0.5dB on white noise, I would guess that the 90% uncertainly interval is probably around 0.3dB. NFM can make measurements with uncertainty interval well below 0.1dB. Consider this uncertainty to be an error in measurement, so if I measured Sun and ColdSky with the millivoltmeter, I should consider that the ratio is xdB +/- 0.6dB.

Your requirement depends on what you are measuring, and when you are taking ratios, how close they are to one. As you improve your system performance, you may want to increase the resolution or precision of your noise measurements.

Owen

[VK2OMD]

Dave,

I understand you are going to try NFM.

(Click on the pic to enlarge.)

Here is a dry run of NFM for Term/ColdSky.

Enter Thot as 290, and Tcold as say 60 (a guess from the G/T model). Make a pair of measurements of Term noise and ColdSky noise, I have entered 5dB for example. Such a measurement would calculate Teq(2) (your rx eq noise temperature at the antenna) as 46K (eq to NF=0.64dB). Since we think Teq is probably 41K, you could play with values of Tcold until it calculates Teq(2)=41K. There is a smarter way to achieve this, read the help on "Calibrating the noise source". If I follow that procedure, double click on Tcold and tell NFM that I know Teq is 41K, it calculates that Tcold must be 63.7 as in the above screen dump.

If we think that the cold sky is about 8K, then your Tcold figure of 63.7k must comprise 8K of cold sky and 55.7K of spillover / feedthru noise.

You can have a lot of fun just exploring Term/ColdSky and refining your measurement system without the hassle of following the Sun, you can even do it in the dark. When you have that sorted, go chase the Sun.

Owen

[VK2JDS]

today hasnt been a good day. i went on a mission to try and fix the vswr problem on both ports.
i can hit lowest vswr at 1275 so i then adjusted things to try and get it up to 1296.
the septum is made as-per the instructions so the dimensions of the inside of the casings are spot on.
so, what do you adjust?
the length of the probes and the spacing from the probe to the backplane.
first i tried trimming the monopoles back bit by bit , not much difference when you use the backplane adjusters for tuning.
considering a quarterwave at 1296 is 57.8 mm when i looked at what they were doing on some of the setups : 38mm, 42mm monopoles. wow, huge difference and way off resonance.
so , re-reading the notes one fellow said the spacing off the back plane made a big difference, so i moved the feed probe out (varying it)by up to 15mm and that made it worse. that was a fellow who wasnt using adjustable tuning plates and the measurement came off the 'net from a computer model.
so i went back to 57mm probes on both receive and transmit ports and adjusted as best as possible which was 2.4:1, awefull i know
what i measure on the vswr meter as i adjust the tuning screw is a steady 2.5:1 then as i screw it inwards further the vswr rises to a peak nearly infinity then reduces back down again. on the outwards side there is a small dip which is where i have set the adjusters at the moment.

its late and dark , very tired, i have given up . what a wasted day
i tried pinging the moon and its a very weak trace on spectran compared to what i used to have
any suggestions?

ps i forgot to mention a switching transistor failed and i shoved 10 watts up the back end of the preamp, so killed a gasfet again...

[VK2OMD]

today hasnt been a good day. i went on a mission to try and fix the vswr problem on both ports.
i can hit lowest vswr at 1275 so i then adjusted things to try and get it up to 1296.
the septum is made as-per the instructions so the dimensions of the inside of the casings are spot on.
so, what do you adjust?

I was a bit confused by your post... but it does appear that you built the feed with the tuning screws.

I assume that your best VSWR with the connectors screwed down soundly as per the drawings, and adjusting the tuning screws was your reported 2.4. That is a RL of just 8dB.

You should expect to get at least 20dB RL and up to 30dB would be good, that corresponds to VSWR 1.2 to 1.07 respectively.

You say you have double checked all the dimensions.

Some thoughts:

• Does the VSWR meter read properly on a known good dummy load? If you are using any adapters, they should be proven in the same way.
  
  Are you sure the coax cable connectors are clean and adequately tight. When you get down to high RL measurements, N type connectors need to be absolutely clean and quite tight.
  
  Have you dealt adequately with the risk of corrosion where the coax connector attaches to the aluminium side of the Septum feed.
  
  Have you tried the feed out of the dish and pointing at the zenith?
  
  Length and diameter of the probe affect behavior.

You know, this should be fun... so when it stops being fun, put it down for another day when you will enjoy it. Remember that you learn more from things that don't work than things that work straight off. It doesn't make sense to waste the learning opportunity on a time when you are not in the frame of mind to enjoy it... and you will solve the problems more directly when you are enjoying it.

Owen

[VK2JDS]

yes Owen 30db would be nice but i can get close to 20db return loss at 1276, not 1296...
i remeasured the septum piece, checked for updates on the net, remeasured the internals, unscrewed 40 odd bolts and checked a side piece. reassembled, trimmed the feeds, still 2.4:1 awefull. 8dB return loss is going to hurt the amp. its the same measured on both ports.

this evenings pings are poor as to be expected.
this all started to unravel because i had bought a nice new thick piece of aluminium to use as a backplate to replace the thin bit i had been using. I installed the tuning capacitor screws exactly like the old bit, and sealed up any gaps that could leak rf out the back end. but it just wont tune.
Nick vk2aoh wants to come out here tomorrow and have a look at the system and its just not performing. so , i got the old backplate and installed tuning screws on it earlier. maybe tomorrow if i refit the old parts it might come good.
will get started early and see how it measures up.
interestingly i did tune by using the weak signal source and keeping away from the peak (which removes the received signal) and it seemed to be best.
as for the measurements: the mechanicals of the feed and connectors are sound.
i have done all the usual stuff, run the radio>vswr meter>septum with different cables, ( heliaxes ) terminated into 3GHz rated load and vswr is flat. once that was tested the septum was then inserted in place of the load and then tested. i hope the old backplane will get it working for tomorrow , but the fundamental problem hasnt been solved.
googling for a circular septum shows that the guy who got one going hasnt released the dimensions as he is selling them commercially. so i am on the hunt for a working set of dimensions as what i have works at 1275 not 1296
btw: sun noise rise in current configuration is 8.5dB over cold sky
73's

[VK4CDI]

Hi all,
Just in the process of tuning my Septum polariser/feed assy. Started with 58mm probes 6mm dia, and VSWR was over 2 to 1, tuning screws made it worse!
Figured the probe lengths should be shorter (a la RA1AQ's 44mm) so started trimming, I am down to 49mm and VSWR is 1.17. with a trough at 1272 of 1.bugger all. The tuning screws are doing what I would expect now. Will trim some more today. Hopefully it won't be too far out when I mount it in the dish!
I am using an Agilent VNA, makes life a bit easier!

Cheers
Phil VK4CDI

[VK2JDS]

Thats interesting Phil. I will keep chopping mine then.
What did you use as a backplane for yours , and how good is the rf seal around it?

This morning I put back the old thin backplate i had been using before. Its got gaps of about 1mm maximum all around it and you can see daylight when you look up into it on the feed arm. Its got 3 screws on each side holding it on but magically the vswr has returned to 1.6:1 on both ports , with the receive dipping nicely, while the transmit side didnt but was good without adding capacitance. Running my fingers around the back of the feed while feeding half a watt into it and looking at the vswr meter it varied from 1.6:1 up to over 4:1.

The new piece was a good seal all around but it just wouldnt tune.... hmmm
interesting you see a trough at 1272

73's

[VK4CDI]

I use a 10mm thick piece of Alum, 4bolts each side. I have a 1.5mm slot 3mm deep down the middle and I made the septum 3mm longer so it fits in the groove. Dont know if that made any difference. Haven't had time to get to the VNA today, maybe tomorrow....

Cheers
Phil
VK4CDI

[VK4CDI]

OK, just got back from the VNA.....
cut a bit more of the probe, ending up at 45mm, VSWR is 1:1.0157, a RL of around 40db. Replaced test probe with machined 6mm brass probe cut to 45mm and retuned tuning screws to 1:1.04.
Moving my hand or screwdriver around the back had no effect.
Now to dig the hole for the dish mount.......

Cheers
Phil VK4CDI

[VK2JDS]

Yesterday i tuned my mark2 septum, its 10mm backplate is made of 2 pieces screwed together, a small slot for the septum to sit back into like yours Phil and 43mm copper 6mm diameter tube probes with the ends sealed by a bit of brass soldered in them.
it returned 1.4 : 1 with the tuning screws all the way out. moving them in made the vswr worse, so i left it as it was and mounted it on the dish last night.
Up at 5am this morning the cw moon pings were really solid and clearly audible with 90watts up the pipe. Barefoot 10 watts from the IC910 was just visible.
I worked a DL4 on cw but i dont think he got my callsign right, then moved to 1296.065 and worked Gerard PA0BAT at -17. Yuri RD3DA was -12 as well. Moon was low when trying for Latvia and Greece, hopefully get them tomorrow. It went down into the trees before i got back to them. 8am and it was all over.
Sun noise rise was the next thing to try and as the sun rose it produced 13.09 dB increase using the old diode and capacitor setup with cold sky set at 41millivolts which means its way better than the old feed was.
I then fitted a better regulator to the solar charged battery bank and shut the hut for the rest of the day.
Up again early tomorrow!
so far this week : california, canada and the netherlands. Phil get some quick-set concrete and get that dish up! you will love this.