Posts by pe1hzg

    I do not have a SF8008 at the moment. But I asked many people to help which photos of the components to check wish possibilities we have. Nothing until now happened.

    I opened it up and voided the warranty on mine, and this is what I found:

    It seems this is an RDA5815. It is possible to find datasheets for it, and the specs says it goes as low as 230 MHz (!) but the datasheets i found do not include programming info.

    I'm still struggling to rebuild OpenATV from source (before I can attempt to make changes) but I thought this would be of interest.

    Yes you do need to modify the PLL, not because of stability but because the tuner will not work below 950 MHz. Therein lies your solution to offsets.

    Do we know for certain that the hardware of the SF8008 sat receiver does not tune below 950 MHz and this isn't a software limitation?

    The SF8008 also runs OpenATV. I have tried (but so far: failed) to build my own image from source code - once I have that working I can see if there is a software limit that can be bypassed.

    Did anyone open the tuner module? Do we know what silicon is used there, and if perhaps the silicon can be coerced to work on lower frequencies?

    Not in an amateur capacity, I have done IP-over DVB-S using a C-band transponder in Kampala, Abuja and a few other places.
    This was quite a common way to get bandwith to Africa before fiber infra existed. As Peter suggests, it has a pretty good bit/hz radio and DVB-S2 is probably better. In my case, this was for downstream only; upstream was using common BPSK if memory serves.

    The receivers were small PC's with a "DVB-S PCI card" running Linux. I'm not sure about the encapsulation used, too busy making the link work and dealing with other issues and the comfort of doing work in tropical conditions.

    Setting up an uplink on a commercial transponder is quite something different than what we are used to on QO100. You don't "just adjust the power"; you call the operator, put your signal on a special test frequency, have the signal optimized (maximal polarization isolation; optimal skew adjustments are a requirement!), have it adjusted for RF signal levels and when all is OK you would be told to move to your allocated frequency. If you would turn up your RF power by as much as 3dB afterwards you will find your transponder notched and you need to call the operator again.

    C-band is much more popular because it is less affected by the weather (if you've been in an equator rainstorm, you will understand).

    Latency, of course, is an issue. Using decent protocol stacks (FreeBSD) made quite the difference.

    I'm currently not set up for the WB transponder but learned *a lot* making this work. If you have the cycles, do try, if must be on a Wednesday..

    Kan dat bord onder de 35Mhz??? (weer een hack toepassen??)

    Voor output op 70Cmtr moet het lo sig toch 25.78...Mhz...zoals de xtalfilter...

    That's probably why Remco used a upb571 prescaler, to reduce the 4351 output to 25.78 MHz

    Other question - the 4351 doesn't work without being programmed, so you need an aatiny/pic to send it it's programming data, no?

    Also, why the capacitance multiplier? If you supply 13V, then I'd think you use the stabilizer in the LNB and I don't see the benefit if the capacitance multiplier?

    I don't see the need to feed LNBs with 12V when only one polarisation is used. It can be reduced until the drop voltage of the regulator is reached. This helps to avoid unnecessary dissipation.

    Especially on dual LNB's, which typically share one oscillator for both converters, you will find that the converter chip shuts down if the input voltage is too low. This is to mimick "no power == no output" behavior.

    You will find a resistor divider connected to the IF output to measure 13V/18V, the 22 kHz tone, but again, the same circuit causes the converter to switch off if the voltage is too low.

    You can circumvent this by adjusting the voltage dividers. Keep in mind that in many cases these resistors are 0402 devices, making them the most challenging part of a conversion.

    I have been considering prototype-ing something like this with a few changes:

    • To be fed from the IF output, not external power supply;
    • Allow it to be fed from 18V;
    • 18V DC power from the IF output passed-through to the input;
    • 144 MHz filter removed, or changed for 1335 MHz (1335 = 740 + 595 MHz)

    This way, one would have an in-line converter that I can put in-line between LNB and sat receiver, and which will convert the WB passband to a frequency range the sat receiver accepts. Because the sat receiver supplies 18V, the LNB would switch polarity.

    I'm contemplating trying this but I have some JOTA projects to finish and those have a deadline.

    Speaking of receiving the WB converter: does anyone have data on the chips used in DVB-S2 sat receivers? I wonder if they can be programmed to receive 740 Mhz instead of the "minimal" 950 Mhz, or if there is a hardware restriction preventing this.

    I have tried if I could re-build the openatv image to remove the 950 MHz limitation but the bitbake recipes have a considerable build time and the build doesn't succeed in one go..

    With hacksaw...

    Hacksaw is one possibility, another approach is what PE1CKK did:, middle of page.

    Perhaps easier to make weather-proof.

    Another hint: you are using the plastic TRIAX LNB mount, but I found that POTY+enclosure+LNB (on longer arm, hence heavier) puts strength on the plastic bracket. Triax also sells an aluminium bracket, if you find that the strain causes things to sag perhaps that's a possibility.


    If you are planning to use a TRIAX dish, please be aware that they are changing their design.

    The old design had an option to use "wingnuts" for mounting/adjustment, very convenient especially for vacation setups, etc

    For the new design the wing nut option no longer exists. You may want to plan your purchase accordingly.

    I'll share Diavolo pictures in a next posting.

    Here are the same pictures for the Diavolo.


    Components to remove:

    The resistor from the voltage divider had a value of 220k (I find that each type of LNB defines their own values). Replacement would be 73k or so, I only had 63k in stock (in 0402 size) but that value was OK still.

    After installing new components (no words on my soldering please!)

    If you wonder why the 25MHz series circuit is in such a strange place, look at the inside of the lid of the LNB.

    Comparing the Diavolo and the High-Q, I like the Diavolo better because it has less gain and the receiver is not as noisy. But perhaps each should decide on their own.

    When buying SMD's it makes sense to purchase a few spares; like fleas, if they jump off your pincer, they are gone forever, but they aren't expensive. And make sure you order solder braid at the same time.

    Again, many thanks to those who did the 'before me" but I didn't find all the info I posted here in other places. Good luck!

    TWhere can I buy such a twin LNB where is one port modified to use the PE1CMO Transverter?

    Let me try to show a few things. "Modifying an LNB" has been covered on this forum and elsewhere, but I'm adding a few details that are hopefully useful. Most of these things I didn't invent, just standing on the shoulders of others, but the modifications I do I have not seen published yet.

    Modifying an LNB sometimes goes wrong, and I have a few LNB's that served their purpose as me getting experience, but that I ruined in the process. I found that if I'd order through aliexpress again, the types of the next shipment would be different, or the type I want no longer available, e.g. starcom sra-3602 ACE is now replaced by sra-3602mini, which is a very different beast (and which I find is not as sensitive).

    To avoid that, and to avoid waiting for aliexpress for replacements, I bought some LNB's at Reichelt, prices higher but worth it, certainly because I can also order the SMD's and other kit I need for the mods at the same time.
    I have found the Megasat Diavolo Twin and Megasat HD-profi Twin to be more stable and less noisy even without modification, so that's what I describe here.

    Note that especially the Diavolo has been described by others (even in funkamateur magazine) but my mods are slightly different.

    When you order, my advice is to at least order one more LNB than the number you want to modify so you can have an accident with one.

    The PE1CMO transverter does a few things slightly different. It injects a low-noise 25/27 MHz reference on the one LNB cable (so no separate cable - power supply, reference and IF out all over one cable), but it feeds the LNB with a low-noise 5V power supply. For twin LNB, this adds a few issues you should be aware of.

    A Twin LNB is a slightly different beast. The two receivers usually share one crystal, so both receivers should be powered up for one of them to function, and while a single LNB has a 5V stabilizer, twin LNB's have a 6V stabilizer (one for each port) and the output is combined with diodes (6V is to overcome the 0.7V threshold of the diodes).
    Feeding such a beast from 5V is a problem because you cannot use the diode. However, since the LNB can only function if it receives its 25MHz reference signal, I decided to unconditionally power from the 5V CMO port and not use the other port to power the LNB.

    Another issue is that a LNB uses the input voltage on its port to sense what it should do: 13/18V for polarisation, 22kHz to switch LO frequency, and if there is no voltage on the port (remember that the LNB can be fed from the other port!), switch off the IF output. This switching is arranged through a voltage divider from the IF port of the LNB. Feeding the LNB from 5V may switch off the LNB IF output because the LNB thinks the port is not powered. So you need to arrange for that.

    Opening the plastic case has been described before; it takes some experience (and one or more ruined LNB's ) to know where the plastic latches are and how to avoid breaking them. It is a matter of squeezing the half of the casing that has the latches and poking a bit with a screwdriver.

    The case is closed with Torx T8 screws and white putty. I have best results if I cut through the white putty along the seams, and along each screw. If you then stick in a Torx screwdriver the screws loosen easily. Once all screws are removed, lifting the top is easy, gently stick a knife or screwdriver in the seam and jimmy open, being careful not to disturb the inner electronics.

    The hi-Q then looks like this (sorry for the bad photos, spent one evening and two LNB's to make this posting!


    I decided to make the bottom port for the CMO-transverter. I removed the following components:

    • Lower 78L06 stabilizer + capacitor between middle and right pin (== input). Also remove capacitor at end of PCB coil (left of coil, to ground)
    • Crystal (square next to the screw in the middle of the PCB, inbetween the 2 RF chips), and both capacitors
    • Diodes combining the outputs of the stabilizers (both ports, two diodes)
    • Resistor and bypass capacitor of port sensing voltage divider (under lower RF chip)

    A few words on removing components. I used a heat gun and even then I found that these components won't let go. The reason is that the PCB is held in the metal casing and trying to heat up a component only heats the casing.

    Because of that, the procedure is:

    • De-solder the connectors of the two IF ports. Whatever you do, DO NOT PUT PRESSURE ON THE TRACES - especially when heated they loose very easily. My advice is to use desolder braid to remove the tin, then gently lift the pin of the F connector, but only to verify that the pin is no longer mechanically connected to the PCB
    • Remove the screw
    • Lift the PCB (take ESD measures!), put PCB on top of the IF pins
    • Now lifting the components with hot air (pyropen) is simple.
    • After lifting the pins, wet pads with fresh solder and clean pad with desolder braid. You will appreciate this for the next steps
    • After all components are removed and the PCB has been cleaned, re-install and put screw back so that board is mechanically stable.

    After the components have been removed, the steps are:

    1. Solder capacitor 270pF (size 1208) over the outer pins (input and output) of the stabilizer you removed.
    2. Solder inductor 150nH (size 0806) on top of capacitor of step 1. After this, you can check continuity between IF poort pad and the anode pad of the diode you removed
    3. Bridge the lower diode (the one from the CMO port). The other diode doesn't get replaced.
    4. In place of the resistor/capacitor of the voltage divider, solder new resistor (original value: 56k, new value: 16k, size 0402). I used 15k instead of 16k, it's not critical.
    5. Solder IF F connector pin back to pad, then solder series circuit of 270pf (size 0806) and 150nH (size 0806) from IF connector pad. Other end of series circuit to short wire to top right xtal pad (oscillator input).
      (apologies to Remco, I didn't use an inductor coupler for real estate issues)

    The resulting mod then looks like this:

    After this, I put a F->BNC connector on the F connector of the CMO port, mark the CMO-port with a marker and secure the BNC connector with schrink tube. This is to avoid connecting that port to a regular receiver (13/18V) which would be the instant death of the LNB.

    After this, test (with casing closed and screwed tight, LNB won't work w/o top lid on!), and if both ports work, re-install plastic outer casing.

    Should you decide to drill a hole in the front of the LNB (see PE1CKK's website), make sure you measure and mark the hole you will be drilling, I found that it is quite easy to get the hole off-center.

    After this, the 2nd port (F-connector) "just works" as a LNB port, but the 1st port (CMO port) MUST be powered and have reference signal.

    I'll share Diavolo pictures in a next posting.

    still have to figure out how to keep it steady.

    I discovered that standard M4 nuts are 3.0mm thick. When soldering the top of the patch plate, I put some between it and the bottom plate to keep the distance OK. After soldering you can simply remove the (loose) nuts.

    I don't know, if I can feed such a low tone from PC's soundcard to the microfone input, a higher subtone would be better, eg 151.4 Hz is possible.

    Adding CTCSS to older transceivers is a topic that has been on the terresteral repeater usergroups for years. It isn't difficult: a typical FM transmitter has a mic pre-amplifier, clipper (limiting the deviation), low-pass filter (to remove the harmonics from the clipper and then a potmeter going to the FM-modulated oscillator. Since the clipper limits the maximal audio signal, the setting of the pot defined the max deviation the transmitter can make.

    Best place to inject CTCSS is on the runner of this potmeter, typically with a high resistor (100k) in series.

    How to generate a tone - you should use a sine wave! - these days one uses a small PIC or ATMEL processor. I use a high frequency, 100kHz or so, and pulsewith modulation. The output looks nasty, but a single RC filter then delivers a nice sine wave.

    Many years ago I made a design that would be used for fundraising for a hamradio meuseum but for various reasons this never materialized. If you can program a 12HV615 yourself it is really easy, I may have a spare chip left. All you need is a crystal, 4 capacitors, 3 resistors and a piece of breadboard.

    There are also various kits out there, for instance on ebay, many of these do have more components than my design (grin).

    I couldn't use the original 5V LNB from PE1CMO with my Transverter.

    I have modified a number of "twin LNB's" - using one port with PE1CMO modification (25 MHz reference via coax, low-noise 5V power supply), and keeping the other LNB port original.

    This 2nd port works with a regular TV receiver, probably with a touner - and if you put 18V on that port, that port will switch polarisation for the wideband transponder and the CMO port remains the same.

    The 2nd port obviously will only work if the CMO transverter is switched on since the 25 MHz reference comes from there.

    I put an adapter F->BNC on the CMO-modified port to prevent me plugging in a normal tuner on that port - the 13/18V on the modified port would instantly kill the LNB.

    And the 2nd port is rock stable as well since it uses the same reference - connecting a SDR receiver only measures the drift of the SDR dongle, and connecting a TV receiver I was able to receive Palestine TV which is on one of the transponders on BADR4 which is is on the same position.

    It switches polarisation with 13/18V and switches LO, for that port only, if 22kHz is sent, as is common with "universal LNB's".

    I can even use this modified twin LNB with a POTY.

    My personal favorite with this is the Megasat Diavolo Twin. Very good reception this way.

    pe1hzg R counter is 5 (10/5 = 2), channel spacing is 2 MHz and then you don't have a FRAC.


    Although I am not sure what loopfilter is on the standard board, this should give the lowest phase noise for this REF and RF combination. Also try to put the ADF into INT mode (LDF is now in FRAC mode) and experiment with the charge pump current.

    Thanks for reviewing my programming, it is what I was hoping to get.

    The setting of R was what I could not find (it was staring at me all the time!) and now I think it would look like this:

    I found the schematic of the board (at least, all the parts on the board I can recognize do match:


    How would I optimize the charge pump current? I have an aged Hameg analyzer but it can't look that deep.

    Please keep me updated on the proceedings of your project. Doing something like that is exactly what I was thinking about

    Status update, as requested. To recap, I wanted to add a 432-28 MHz transverter to my setup so I can drive the PE1CMO transverter kit QO100 station over a looooong coax cable, on 28 MHz, using non-critical RG58 coax.
    (I'm using a separate cat5 cable for remote control and monitoring).

    Unfortunately, the Ukrainian transverter's LO drifts, expecially when it is warmed up using the groundplane of the transverterboard. My solution was to replace the 404MHz LO with a LO from an ADF4351 board (Ebay), whose reference is the same 10MHz GPS lock as used by the PE1CMO kit.

    This posting is not a "how-to" but perhaps gives inspiration to others. Many of the info is copied from others, the least I can do is make my stuff available too.

    The ADF4351 board came with a 25 MHz TCXO, but it is possible to use an external (GPS-locked) reference by lifting a few SMD's, see the 2nd picture.

    I completely lifted the 404MHz TXCO off the transverterboard to have solderpads to solder a sort coax to the output of the ADF4351. The transverterboard provides termination; the ADF4351 board has two outputs, as recommended I terminated the other output with 50E.

    The ADF4351 board requires 5V so I needed to add a 7805. This stabilizer is noisy but it is only used to power the 3V3 stabilizer on board.

    To make the ADF4351 work you need to program it using the values from the ADF4351 designer program. While there are solutions with arduino I thought this was way too big, I rather like the approach of OK1CDJ, see…d-with-attiny13-kit-i351/

    With apologies to the seller I made the same thing on breadboard with an ATtiny85 (leftover from another project). See the link on the website to the github data. I did find that for the ATtiny85 I needed a 4k7 pullup to the reset signal (pin1 - pin8) to keep the AVR programmer happy.

    I made some effort to minimize the PLL noise of the ADF4351 and used the parameters below (suggestions appreciated)!

    To program, I used the AVR programmer and programmed the ATtiny85 to use the lowest clock speed to minimize noise - all it needs to do is to clock 5 32-bits words and then halts until power is removed. I modified OK1CDJ's code to make it work on Atmel Studio 7, my code is below.

    I also used the CPU board to add a Lock Detect LED using the LD pin of the eval kit and a 1K resistor.

    What else is there to say - when you case the Ukrainian transverters, make sure the case is big enough to have this addition.

    Looking at the BATC WebSDR as well as the one from IS0GRB, the CW beacon is about 15-16dB over the noise. An S-point is 6 dB so the "best" we can do (before getting a QSL card from LEILA) would be 2.5 S-point above the noise.

    The question then, is where you lie "zero". With all the mixing and amplification making an absolute measurement is hard. And I see that the 50-60dB gain of the LNB drives many receivers hard: if the noise floor is S8, are the signals well received and nice to listen to even if they are 59+10dB?

    I tend to attenuate so that the noise is S1-S2, this makes the receiver a lot better behaving to but it would mean that the best report I can give is S5 or so.

    And that doesn't even bring the poor calibration of an S-meter into the equation, I'm inclined not to use "five-nine" at all.

    Perhaps using the dB's of one of the WebSDR's is more accurate than an S-meter report?

    Geert Jan

    I have asked ARRL for advice, since the requirement also exists in the USA, and received a detailed analysis with lots, lots of data. I have suggested that ARRL make an article of the info provided by this specialist (I think this will be interesting to many).

    I can't make the analysis itself available right now - need to wait for the publication.

    I was interested in actual risk because of my plans to be active with JOTA, when the setup is seen and used by many scouts (children). Due diligence dictates that if an exposure question would come, I would need to have an answer ready.

    My question was on 20W output (PE1CMO kit), 80cm dish, 100% efficiency (50-60% is more likely in practice) and POTY patch antenna. A few comments:

    • One needs to distinguish between near-field and far-field effects. The near-fields are not as serious, because while the dish acts like a spot beam on distance, that isn't true on close distance, the field is lower, so safer.
    • FCC rules make a distinction between amateur exposure and public exposure. Public exposure, 10W/m^2 would be safe for a distance bigger than 7.8 meters.
      Amateur exposure (50W/m^2) would be safe at 3.5m.
      This would be true if the whole body is exposed, for partial exposure (arm in dish) the limits are 40W/m^2 (public) and 200W/m^2 (amateur). The field calculated never goes above 160W/m^2 so it would be entirely safe.
    • Outside the beam the field is 20dB lower, making that field a non-issue
    • The main concern would be someone from the public placing their hand in front of the patch antenna itself. The exposure level is lower than the amateur limit, but above the limit for the public.

    There are much, much more details in the analysis of Ric K5UJU and my comments above really don't do it justice (and I hope my summary is correct!), but I would like ARRL to have the privilege of publication.

    Something to ponder is that geometry dictates that even if the dish is on the ground, the beam would be 2m up at a distance of 4m, at least in the Netherlands (our inclination is roughly 30 degrees). And, given that humans are shorter than 2m, the 4m clearance area doesn't need to be bigger than that.

    So, for my JOTA work, if I clear 4m in front of the dish then there really is no risk of exposure.

    I apologise for being unable to make the detailed analysis available at this point.

    As to using "camping dish" antennas, a few years back LIDL was selling a sat kit consisting of a small dish, with all kids of mounting hardware. I've used that for tests and made QSO's with it, obviously 80cm is much better but SSB QSO's would work with the LIDL antenna.