EMRFD Message Archive 11867
Message Date From Subject 11867 2015-11-02 22:35:27 Adrian Scripcă W7ZOI/K7TAU spectrum analyzer saga continues Hi all,Since I last wrote on this group, I acquired some Hammond 1590B and 1590BB boxes in order to shield the SA modules I have built so far. I also ordered some SMA female to SMA bulkhead SMA adapters and feed-through capacitors from Max Gain Systems(thanks Vince for the tip).What I have yet to build are the VHF and LPF filters and effectively drill the shielding cases and mount the modules inside. A couple of questions for the more knowledgeable ones:- since I have never used feed through capacitors before, how would one mount them? which side makes it inside the box? I don't feel there should be any electrical reasoning to mount them in a certain way but maybe there are mechanical reasons (the capacitors I have can be seen here http://www.mgs4u.com/RF-Microwave/doorknob-capacitors.htm by searching "feed")- how are interconnections made between related feed-through terminals? are there some kind of special connectors available?- since I intend to use bulkhead connectors to route the RF inside/outside the boxes, does it still mean I need to use shielded coax attached to a SMA male connector to drive the RF to the board (I have pin headers there)?- the Kanga files I have do not specify the VHF/LPF filter construction details, although the components are there. What version should I use? (I do have piston capacitors)Thanks,Adrian, YO6SSW 11868 2015-11-03 09:13:39 Gary Winblad Re: W7ZOI/K7TAU spectrum analyzer saga continues Good questions from Adrian...
I would also like to know the viability of using copper flashing or
blank copper clad vs. the expensive Hammond boxes (as done by VU2ESE
in his version). Are the Hammond boxes that much better?
----- Original Message -----
11869 2015-11-03 10:27:19 Adrian Scripcă Re: W7ZOI/K7TAU spectrum analyzer saga continues Hi Gary,I wouldn't think there would be any problems using copper clad to shield your modules from a performance point of view; to me the decision was merely a convenience since my mechanical skills leave a lot to desire. If I were to build the pcb boxes myself I would have to cut the pcb sheets first (I don't really have the proper tools), attempt to solder the walls so they'll be at a right angle after the solder cools, find a way to mount the lids without soldering so I can get them off whenever needed to make adjustments, make sure the shield is really closed when putting the lid on (good mechanical connection between the cooper sides), etc.73,Adrian, YO6SSW 11870 2015-11-03 10:43:19 jim Re: W7ZOI/K7TAU spectrum analyzer saga continues --------------------------------------------
11871 2015-11-03 10:53:50 Robert Fish Re: W7ZOI/K7TAU spectrum analyzer saga continues I generally use PCB for the floor of my boxes and I use tin sheet for the walls and ceiling. You can cut tin with scissors, you can solder to it and it is cheap.
I have used it on many projects and have been mostly happy with the results. I use a bending brake to make nice sharp 90 bends, but a vise will work as well.
I bought my tin sheet at the local hardware store, but here is a link to the same stuff from the same Manufacturer:
If you are a real cheapskate you can use coffee can lids too. Same stuff.
I wouldn't think there would be any problems using copper clad to shield your modules from a performance point of view; to me the decision was merely a convenience since my mechanical skills leave a lot to desire. If I were to build the pcb boxes myself I would have to cut the pcb sheets first (I don't really have the proper tools), attempt to solder the walls so they'll be at a right angle after the solder cools, find a way to mount the lids without soldering so I can get them off whenever needed to make adjustments, make sure the shield is really closed when putting the lid on (good mechanical connection between the cooper sides), etc.
11872 2015-11-03 14:56:37 David J Nushardt Re: W7ZOI/K7TAU spectrum analyzer saga continues
Adrian, I am also building a Sa ,can you contact me off list.
Sent from my Kindle fire HDX 8.9
11873 2015-11-03 18:53:37 K5ESS Re: W7ZOI/K7TAU spectrum analyzer saga continues
An additional advantage to using a solderable material for constructing module enclosures is that you can use cheaper feed through capacitors (solder in place vs threaded versions). Can be 12¢ ea vs 75 to 65¢ ea.
If aesthetics is important you have to be pretty skilled to make your own enclosures that look as good as the Hammonds.
11874 2015-11-03 20:25:49 Gary Winblad Re: W7ZOI/K7TAU spectrum analyzer saga continues Ok, thanks guys. Confirms my feelings.
I have been using copper clad, tin and some aluminum sheet covers
in my HF projects. The up-coming SA project will push my tech into
the low VHF so wanted to learn..
Thanks and 73,
----- Original Message -----
11875 2015-11-04 06:28:57 DuWayne Schmidlko... Re: W7ZOI/K7TAU spectrum analyzer saga continues I have been working on Farhan's version of the W7ZOI/K7TAU spectrum analyzer. For shielding I have been using .020" double sided circuit board material. It is very easy to cut with a scissors or paper cutter. Much easier to solder than copper flashing. I am using SMD parts for most of the build with almost everything mounted on the top layer of the circuit board. I lay out the boards with solder pads where the shielding is going to be to make it easier to solder and keep everything straight. Instead of feed through capacitors, I place a SMD on each side of where the shield is going to go. Cover for the shielded box is either tack soldered on or held in place with adhesive copper foil. Pictures of completed boards on my blog kv4qb.blogspot.com
11876 2015-11-04 15:05:12 Kirk Kleinschmidt Re: W7ZOI/K7TAU spectrum analyzer saga continues Fantastic blog, DuWayne.Thanks for sharing all of that good stuff.--Kirk, NT0ZMy book, "Stealth Amateur Radio," is now available from www.stealthamateur.com and on the Amazon Kindle (soon) 11877 2015-11-04 15:25:34 David J Nushardt Re: W7ZOI/K7TAU spectrum analyzer saga continues
Nice,board DuWane, what did you use to layout the board?
Sent from my Kindle fire HDX 8.9
11878 2015-11-04 16:09:22 DuWayne Schmidlko... Re: W7ZOI/K7TAU spectrum analyzer saga continues I use the eagle from cadsoft, the free version that is limited in size
of board. For non SMD parts I use the through hole library part to get
the proper spacing, then use the line tool with .056" lines to create
solder pads. After I get the pads in place I then remove the original part.
12204 2016-01-19 04:18:02 Adrian Scripcă Re: W7ZOI/K7TAU spectrum analyzer saga continues Hi guys,I managed to find some time and slowly made progress on the VHF filter for my W7ZOI/K7TAU spectrum analyzer build.Here are some close up details of the filter build.Here's the test setup.Here's the measured characteristic.Test setup- wrote some code to hook up an Arduino Uno to a NT7S Si5351 board- the Si5351 drives the VHF filter- the other port of the filter drives a W7ZOI power-meter- a DVM reads the powermeter output DC voltage (the KangaUS version stipulates 60mV/dB)- an encoder allows frequency tuning and step size manipulation- made the Si5351 output a 110MHz signal and tuned the piston capacitors for maximum power meter outputSome questions- can you spot anything blatantly wrong with the build? anything that can be improved for better performance?- can you spot anything wrong with the test setup?- what is the effect of having Si535 output a square wave on the measurement?- how does one exactly tune the gimmick capacitors and what should be the expected response?- is the filter response "right" for SA use?- the inner PCB walls used for interstage shielding are soldered to the external walls but not to the floor; they are about 4mm up from the floor. is this problematic?- what other measurements make sense in order to confirm that the filter works as expected?- I attempted to measure the insertion loss of the filter by simply disconnecting the filter and feeding the raw Si5351 output to the power meter. The power meter output difference between the tuned filter signal and no-filter signal was ~0.8V which translates to approximately 13dB IL given the 60mV/dB. Is my testing approach right? Is the IL value too much? What can I do in order to improve insertion loss?You can find photos of the on-going project here.Thank you es 73 de Adrian, YO6SSWP.S: I started a git repository containing some vectorial drawings of RF system building blocks. You can find the symbols here in case you are interested. 12205 2016-01-19 04:34:14 Thomas S. Knutsen Re: W7ZOI/K7TAU spectrum analyzer saga continues Your filter is over coupled. This gives the double hump as you see.Start by adding the lid, the lid should always be 12206 2016-01-19 09:05:01 John Lawson Re: W7ZOI/K7TAU spectrum analyzer saga continues Hi Adrian. Your insertion loss on the properly completed tuned up filter should be around 5 dB. Bob Kopski and I both have built Wes' filter and those are the results we both obtained. We both used an analog sine wave signal at 110 MHz at zero dBm for tuning the filter. We also did a sweep of our filters from 80 MHz to about 120 MHz to confirm the filter's shape. Be sure to have the lid on the filter when tuning the filter as the shape of the filter will be a little different without the lid attached. Experimentation with the coupling wires to the center resonator was the key to get the tuning of the filter correct. Presently your over coupled, see Thomas's post. Bob and I are located 1300 miles apart and the filters were not built at one QTH.I hope this will be of help to you. John K5IRK
Sent from my iPad
12208 2016-01-19 10:26:51 w7zoi Re: W7ZOI/K7TAU spectrum analyzer saga continues Hello Adrain,The walls between filter section should definitely be soldered to the base. You are going to have currents flowing in the "grounded" base board. In some parts of each resonator these currents can be sizable. If you have a "wall" that is not attached, that just means that current in the ground foil of one resonator can now flow in the other one. When those currents flow, they induce signal in the rest of the structure. This is one form of coupling.An esoteric, but extremely interesting exercise is to build a high attenuation filter. Build a box with coax connector at each end. At each end, put a loop of wire from the coax center connector to ground near the connector. Put one shield in the box, creating a two section "filter." No LC, no holes in the wall for coupling wires. Just two boxes. Measure the thing without a lid, and then with a lid. Once you have things really tight, you should get over 100 dB of attenuation. But if you do measurements along the way, you will see some of the coupling things that can happen. This experiment will give you a feeling for what you can do.Try not to take as many photos. This is NOT a kit. It was never intended to be one. We see kit manuals that show every step, and this was intended to be a home-brewer's, experimenter's project. That said, you have good photos.Take a look at the 1991 QST piece that talks about the double tuned circuit. Experiment and get some experience building them. This will give you a feel for the way to adjust coupling.Most of all though, when you build these filters, do everything you can to have each resonator in a box that is as tight as possible, with all walls soldered to all others. Gaps can do bad things. If you want to temporarily attach a lid, you can do so with C-clamps. Experiment.73, Wesw7zoi 12209 2016-01-19 11:33:54 Adrian Scripcă Re: W7ZOI/K7TAU spectrum analyzer saga continues Hello Thomas, John, Wes and thank you all for the helpful replies!Thank you Thomas and John for the tips on over coupling and reassurance that the design has been independently reproduced with similar results; I never implied there is a glitch in the design but that I'm certainly doing something wrong; I just didn't know what that was :) But now that I received some hints I will persist until I manage to have comparable filter results.Thank you Wes -- for many things, in general, and for the shielding insights in particular; they were exactly what I needed! I now know what are some of the things I did wrong and can hope to repair. I am a programmer by trade hence my lack of knowledge on perhaps basic topics. I am also a self-taught amateur, reading as much as I can and trying to make some sense out of it all. This perhaps sheds some light on the question I had regarding the waveform of the signal used for testing. My reasoning was the following:- a square wave signal has its energy distributed between the fundamental and harmonics which means that coupling the signal directly to the powermeter we actually measure the whole available energy and not the fundamental's- when the signal enters the filter, the filter only allows the fundamental to pass while stopping the harmonics and thus loses the harmonic energy from the initial signal energy- the previous lost energy could provide an explanation for the relatively high insertion loss that I measured since it would mean I did not measure only the insertion loss of the filter but also the loss of the harmonics' energy. I wonder how much of a difference a sine wave would make.The previous points are only speculations based on my intuition.As for the photos, they were not supposed to be public, I just take shots for myself during the build so that I will have photo material for a blog post when the project is over; that's why there are so many of them: I will have what to choose from. I posted the link to the album on the list thinking that it could possibly help builders of the same KangaUS kit in ways unknown to me. Either that or/and receive well deserved criticism and advice on details that can be improved.P.S: Dear Wes, any other insight revealing experimental setups like the one you described in the previous reply are more than welcome!P.P.S: How come the holes in the shield are ok but gaps between the shield and floor are not?Thank you again,73 de Adrian, YO6SSW 12210 2016-01-19 18:21:54 w7zoi Re: W7ZOI/K7TAU spectrum analyzer saga continues Hello Adrain and gang,OK--My bad. After this morning's exchange I looked at my own web page and the info listed there. Turns out, the photo I had showing the interior of the VHF bandpass filter made it look like the shield sections between filter elements was not soldered to the baseplate. It was, but the solder was on the other side of the wall. I've added a photo and a few words to the web page dealing with this. Just go to the w7zoi dot net web page and then follow the technical links to the spectrum analyzer stuff. Hope that this didn't cause too much confusion. Thanks for bringing this to light Adrain.73, Wesw7zoi 12211 2016-01-20 15:33:14 kerrypwr Re: W7ZOI/K7TAU spectrum analyzer saga continues An interesting discussion; thanks, Adrian, for starting it.
The links won't work for me; a pity as I would like to see your work.
Your insertion-loss will be in error if, as I gather, you are measuring the square-wave input against the filter output using an AD8307 power meter; in one case you are measuring a square wave that contains the fundamental and all harmonics up to the power meter's frequency limit and, in the other case, you are measuring just the fundamental.
Bob Kopski wrote about this in his Jan/Feb 2004 QEX article; it must be available on-line as I have it in my files.
This phenomenon also affects diode detectors and can cause substantial errors in measuring non-sinusoidal waveforms.
If you use a low-pass filter to "clean-up" the square wave you should get better results; a quick fiddle with the free-download Elsie program shows that a ca.120MHz 5-el Butterworth LPF is achievable.
There is a difference between "gaps" and "holes" in shields.
Wes described circulating currents; these are re-directed by gaps and then flow in often-unwanted places but they simply flow around small holes and stay in the shield.
Holes are OK if they are small in terms of wavelength and operate as waveguide-beyond-cutoff; this is why a microwave oven has holes in the door screen. Gaps have one "long" dimension which can be large in terms of wavelength and therefore ineffective.
Skin depth is a factor in shielding; if the shield is thin in terms of skin depth, current flowing on one side will penetrate to the other side. Playing with a calculator such as this http://leleivre.com/rf_skindepth.html is educational.
Copper is not necessarily the best shield although I, like most of us, use it for the convenience of working with PCB material.
One-ounce copper is 35um thick; that is about the same as five skin depths (a "rule-of-thumb" for design of AC/RF conductors) at 100MHz.
Tinplate is better; its steel core is "lossy" and dissipates current and, because of its high resistivity, its skin depth is much less than copper so current is less likely to reach the other side. Modern tinplate is more than just tin-on-steel but the tin is usually about 0.4um thick with a "flash" of chrome over it and these thin layers don't have much effect on skin depth.
But double-sided PCB material with both sides connected to adjoining material works pretty well! :)
12212 2016-01-20 19:56:12 kerrypwr Re: W7ZOI/K7TAU spectrum analyzer saga continues "...a square wave that contains the fundamental and all harmonics ...".
Correction; a square wave ideally only contains the odd-order harmonics although, in practice, even-order harmonics at reduced levels are often present.
Thought I'd correct the mistake before someone picks it up.
13061 2016-08-09 04:01:36 Adrian Scripcă Re: W7ZOI/K7TAU spectrum analyzer saga continues Hi guys, it's been a while since my last post. I managed to put together all modules and quickly ran a test to confirm the system as a whole works. I used a W8DIZ frequency reference kit ss signal source which is described to output a 3V PP in 50 Ohm translating to somewhere around +13.5dBm. The frequency reference is set up to output a 10MHz square wave signal.Due to lack of potentiometers I decided to skip both IF GAIN and LOG CAL controls and here is what I got:With no IF gain:With max IF gain (+15V on gain control):Both scope channels are set to 0.5V/div and clearly the output is too low even when the IF gain is at maximum. I tried to peak both the 110MHz as well as the BW res filter before but that was the most I got.Since I do not have a professional signal generator I canibalized a G3UUR oscillator I used for crystal parameter measurements, added a voltage regulator, installed a 10MHz xtal and followed the oscillator with a 30 meter 7-element Chebyshev LPF acquired as a kit from Diz long ago (http://www.partsandkits.com/univlpfilter.php). I coupled it to a 50 ohm load and measured an output voltage of 256 mV peak to peak across the load; this equates to something like -7.8 dBm. Not exactly the -10dBm calibration source specified in article but good enough for the tests I hope.Here's how the signal source looks in the SA:With no IF gain:With max IF gain (+15V on gain control):I attempted to look for the culprit starting from the scope side of the system, testing the logamp and bw res together. I injected the aforementioned -7.8 dBm signal into the IF AMP + LOG detector module (the MC3356 Kanga version) and got the following measurements:* 1.35V with no input* 6.25V with -7.8 dBm at input and no IF gain* 7.12V with -7.8 dBm at input and max IF gainThis tells me that both the bw res filter as well as the if amp + log detector modules work as specified and leave me clueless as to how to proceed from here onwards.I have a restricted set of tools that I can use in debugging:- DVM- 30 MHz AD9851 DDS generator- 40 MHz oscilloscope- 50 ohm load- uncalibrated W7ZOI powermeter- the -7.8 dBm source presented earlier- W8DIZ signal reference (http://www.kitsandparts.com/fref.php)I would appreciate any tips and/or hints on how to continue the search for the culprit.Thank you,Adrian, YO6SSW 13069 2016-08-09 23:04:21 Adrian Scripcă Re: W7ZOI/K7TAU spectrum analyzer saga continues Here's one more update to my SA debugging saga, hoping I don't spam anyone.Since I do not possess the needed test gear I have to get creative when trying to test things out. I have the feeling that part of my SA build problem is the 110MHz band pass filter but I need to somehow prove it.I performed the following experiment: I powered up the 1st LO + 1st Mixer board, hooked a frequency meter to the VCO sample output and used a variable power source to determine the control voltage for 110 MHz output.I coupled the VCO output to a W7ZOI AD8307 power meter and read the output.Afterwards, I inserted the 110 MHz BPF between the VCO and power meter, read the output again and calculated the insertion loss. I assumed that all ports are 50 ohm compliant and I don't need impedance matching.The voltage difference between the readings is 800mV which, at a slope of 60mV/dB translates to somewhere around 13.3 dB. Instinct tells me an IL of 13.3 dB is way too high and I need to rework that filter but would this be all that's missing from the picture?The zero spur also looks quite weird in this shot: http://hq.scene.ro/sa/-7.8dbm-in-sa-max-gain.jpg Virtually all pictures I've seen of others' builds had the zero spur sky high.I'm still trying to figure out a way to test that the outputs of my build's 1st LO + 1stMixer and the 2nd LO + 2nd Mixer boards are as designed.Could someone please let me know what the output on their 1st LO + Mixer board is when feeding an input of -7.8 dBm?Thank you,Adrian, YO6SSW 13073 2016-08-11 09:07:24 David J Nushardt Re: W7ZOI/K7TAU spectrum analyzer saga continues Adrian, I strongly agree about the 110mhz band pass filter being the culprit. It is by far the most difficult and critical part of the SA.
That is why aI left it for last, I wonder if you have any pictures of it, that may shed some light as to weather that is the problem or not.
I also wonder if you applied all the updates on Wes's site, I know you said your following the original schematic first then doing the updates,which makes me wonder if you applied all updates, If notthat could be your problem. The updates are necessary for proper operation. If you have pictures of the filters and an overall picture how everything is hooked up , I would love to see them.
Actually that is the last I heard from you , It looks like your doing a good job with everything up to the filter, the brass has to be flat aginst the boards on both sides top and bottom and soldered continuously all around ,not easy to do, as long as the entire outside edge is continuously soldered I wouldnt think the inside would matter, but that is just my thinking , I have no proof.
I would imagine you would need about a 100 watt soldering iron to do the job, again just my thinking. Although I do have some experience with this. Different application though. Flux or past would definitely help.
The heat conduction is so high you need to be able to supply heat fast enough , the only way is high wattage you have to go fairly slow but not so slow that the tip sticks.
As for the the pos vco you could rig up a pot and regulated 12 volt supply to have a 0-12 volt supply to the varactor and manually sweep the filter or use the ramp generator of the sa itself to drive the vco varactor. POS150 building the bob kopski sweeper would take too long.
I never did hear how you sweeped the filter, the filters are the real challenge of the project and where the learning comes in the updates are where some knowledge and magic happens.
Your way ahead of me I still have the timebase and another board to complete, the filters and the layout of the box to put it all in will be the hardest part for me to do one handed!!!
I had a hell of a time laying out and cutting and filing all the holes on the dual power supply with one hand.
Best of luck
13074 2016-08-11 09:30:41 Ashhar Farhan Re: W7ZOI/K7TAU spectrum analyzer saga continues There is a simple way to align the 110 MHz filter.1. Build the VCO, keep the sweep to zero and tune it the center frequency down to 108 MHz as monitored on a nearby FM radio.2. Feed the VCO output through the 6db pad to the 110 MHz filter, attach another 6db pad on the other side of the filter and attach it to the log amp.3. Now, fiddle around until you get a proper peak. If you peak it properly at 108 MHz, it will be useable down to 110 MHz as well.4. Connect up the specan properly and tune to a signal and lightly retouch the band pass filter.An easier band pass filter that I played around with is described on http://w7zoi.net/bpf-ese-nhi.pdf . I can confirm that it works well as I have built it several times over.- farhan 13076 2016-08-11 14:50:53 w7zoi Re: W7ZOI/K7TAU spectrum analyzer saga continues Hello all,The 110 MHz bandpass filter that we now recommend for the 1998 QST spectrum analyzer is that on my web site. Go to http://w7zoi.net/sa-stuff.html and then click on the picture of the filter. This will get you to info about the design. This filter is a simple one, but has reasonable stop-band attenuation performance at UHF. It has a narrow enough bandwidth at 110 MHz that the 90 MHz image for the second conversion is adequately suppressed.I prefer filters with variable trimmer capacitors. It is certainly possible to build working filters that do a good job that do not use the expensive trimmers, but I find them more difficult to adjust. They are, however, inexpensive, so they are worthy of experimenting.No, a 100 W soldering iron is not needed. The filters shown on my site were built with a Weller TC201T with a large tip. I also have a Weller SP80L that came from a local hardware store. It’s only a 30 W iron, but it has a massive tip and is useful for gross metal work. It was not expensive.The original 1998 filter design will still work. It needs a low pass filter to be cascaded with it to achieve the desired UHF stop band attenuation. I prefer the newer design with well isolated resonators.Farhan commented that the VCO in the analyzer itself can be used to align the 110 MHz filter. The same recommendation was made in the original paper, or in the related web info. A frequency counter will be needed. This, with a buffer amplifier such as the MAV-11 used in the SA will complete the needed RF source. Filter output can be measured with an AD-8307 based power meter.This 110 MHz bandpass filter will require experimentation. You start with construction. Then throw RF at the input and put a power meter on the output. Use RF at the filter center of 110.0 MHz. Tweak the tuning capacitors for max output as noted in a 50 Ohm detector. Now tune the RF source above and below, measuring the bandwidth and looking for ripple. Temporarily replace the filter with a "thru" connection so you can calibrate the RF source. This will then allow you to determine insertion loss. This filter IL should be a few dB, but not a lot more. It will probably be necessary to experiment with the resonator-to-resonator coupling. As you adjust this, you will have to re-adjust the resonator tuning capacitors. We live in an interactive world.Once you have built one of these filters, it will then be easy to build a similar one for a different frequency. A 144 MHz design is an easy extension, as is one for 222 MHz.If you have never built a VHF bandpass filter, a simpler design may be a good starting experiment. I'm not talking about a simpler filter for the SA, but rather an experiment that will give you experience in building and adjusting such things. A double tuned circuit is a good one to do. See the DTC tutorial piece in QST, Dec 1991, p29. This paper is on the EMRFD CD.Generally, I do NOT respond to questions that are posted on this Yahoo Group site that deal with the spectrum analyzer. If you have a SA question intended for me, just write directly to me.73, Wesw7zoi 13077 2016-08-11 16:19:30 David J Nushardt Re: W7ZOI/K7TAU spectrum analyzer saga continues
Wes, Many Thanks for the clarifications , nice work!
I,am trying to follow your advice , I been having problems with my Leukemia ,so behind 6months, I have a half dozen projects to finish , before I get back to the SA and filter, but rest assured I will finish it , if it's the last thing I do!