Conversion of the Maspro SCE-975 LNB for 10GHz receive use.
The unit itself
The Maspro SCE-975 LNB, was originally used for the FSS (fixed satellite service) band use, block-down-converting 10.950-11.750 GHz. It is a very small squat waveguide-terminated LNB, that uses HEMT (high electron mobility Field Effect Transistor) technology to produce an LNB with (for the time in which it was in production) a low noise figure, around the 1.3dB mark, possibly better.
Inside the unit, the modular construction of this unit is apparent, when the screening covers are removed from the top and bottom of the unit. On the top, a three stage HEMT amplifier, followed by an interdigitated filter, with a single schottky junction mixer (which is located under a smaller screening cover). On the underside, a MIC (Microwave IC) IF amplifier, followed by a discrete stage, with the power supply circuitry, and required regulation stages. The two boards are held down with around 8 screws per side, and the two boards are linked with a small piece of etched p.c.b, supplying DC, and returning the IF signal from the mixer.
Visible from the IF board side is the local oscillator capsule. The adjuster screw is the only visible part of this module, however at this stage, we need to get everything separated if we are going to convert the unit for 10 GHz operation.
SEZME! - static electricity zaps microwave electronics. So make sure you have provided all the necessary earthing precautions before you even touch this unit. HEMT transistors are particularly vulnerable, and you may not even realise you've damaged them (a poor noise figure is sometimes the only indication of damage).
The conversion is fairly simple, but due to the nature of the construction of the LNB, some care is required to get it right.
Start by using desolder braid of a good quality (chemtronics chem-wik lite is very good) to desolder the linking p.c.b. from the IF board side of the LNB. Once you are satisfied the boards are separated (wiggle the linkage p.c.b gently to make sure), turn the LNB over, and on the RF side, using the desolder braid again, desolder the probe from the waveguide aperture at the (waveguide) end of the unit. Try not to move the probe up or down. Watch out for static especially here. Once you're sure that the probe is separated, remove all the screws holding the RF board down, and gently pull the RF board up and out of the LNB casting by holding a small pair of pliers on the twist-tag that held the screening cover in place. Hopefully the board will separate without any problems. If however the waveguide probe is still attached gently heat this whilst pulling up on the board. The RF board should now be free.
Put the rest of the LNB to one side, and turn the RF board over. Underneath is the LO (Local Oscillator) capsule. This is a self-contained 10 GHz oscillator. We want to modify this to run at 9 GHz so that the 10 GHz band tunes-out on a satellite receiver at 1000-1500 MHz.
To remove the LO from the RF board, we have to first desolder the two coupling pins that link through to the RF board. These carry around 4V in to the oscillator, and the 10 GHz signal out of the oscillator. Desolder them with braid, and make sure you've got all the solder off. The RF board uses plated through holes for these connections, and we don't really want to damage them, even though the striplines on the top of the board are the only part of the p.c.b that really matters.
Using desolder braid again, desolder the sides of the tin can from the RF p.c.b. The can is only soldered at two ends, once the bulk of the solder has been removed, gently heat and pry up the can from the RF p.c.b until they separate. Now you should have the module separate, and that means we can put the RF p.c.b to one side safely in an antistatic bag whilst we work on the LO.
To get inside the LO requires considerable care. Enough heat is required to melt the solder on the can and free the p.c.b, but not so much that the surface mount devices inside will also come undone and move around! Probably the best way to start is to attempt to desolder around the underside seam where the can and p.c.b meet, and then to heat the can gently, with a small gas torch and pry each side of the p.c.b out using a pin. The pin can be easily inserted in the hole at each of the four corners of the p.c.b. With the right amount of heat, and with most of the solder removed, however, the board will probably come out first time.
Once you have the board out, look it over. It should resemble the layout as shown here. The dielectric resonator is the yellow cylinder perched on the ceramic pillar. Remove this and put it in your spares box. Look at the point labelled #1. At this point use a sharp knife (scalpel is great) to cut the source stripline off just after the source lead on the transistor. Make two cuts close together so that the stripline is no longer connected to the source, and remove the p.c.b trace inbetween the cuts.
Now look at the point labelled #2. At this point you need some copper or brass foil (0.003 to 0.005 shim stock) to make up the length of the stripline to the length of the second tuning marker (you should see two tuning makers there on the board). If you can't see the tabs on the board, make the stripline as long again as it is wide. Solder two wires onto the unserside of the p.c.b. One goes to the groundplane on the underside, and the other goes to the 4V input pin, located in the stripline you modified at point #2.
Put the board to one side for a moment, and using some desolder braid, clean all the excess solder from the inside of the can's soldered seams. This makes things easier to do the alignment with. The can should ideally be a loose-ish fit over the oscillator p.c.b now.
For the fortunate few, there are spectrum analysers, these will instantly tell its operator when something is working and when something is not, what frequency it is on, and whether the signal is pure or not. For the dabblers however, there is a less elegant, but viable alternative
I use a satellite receiver with an FSS LNB to look for signals around 9 GHz. This makes a cheap 9 GHz monitor for close-up work. In an ordinary (FSS) LNB the wanted sideband to be converted is in the 10.950-11.750 GHz band. This is the upper sideband that is block-converted. The lower sideband however is in the 9.050-8.250 GHz range. Although filtering inside the LNB normally removes this unwanted LSB artefact to improve both the noise figure and rejection, when there is a few mW of 9GHz around, it still manages to get though the tuned amplifiers and filtering to appear as a signal on the satellite receiver. This lower sideband, however, tunes in the opposite direction to normal, and so the tuning of the satellite receiver is reversed, which means the higher in frequency we tune up on the satellite receiver, the lower the received signal is. However, at 9GHz it's simple 10 GHz - 9 GHz = 1 GHz, so expect to see the 9 GHz signal around the 1000 MHz mark on the satellite receiver.
That's the theory. Typically you could be looking around +/- 80 MHz of the 1 GHz figure, so you have to scan around with the satellite receiver when you're looking for oscillation. Also the LNB is pretty deaf at 9 GHz, so more than a foot away, and you've lost the signal altogether. However these methods, although time-consuming are my only practical means at present, and they work for me - this is how this unit and a few more were done.
Using a 9 GHz DR, or a 9.75 GHz one especially sandwiched* for the purpose, place the DR slightly offset from its original position, without a spacer, more over the stripline on the gate of the oscillator transistor. Apply 4V d.c to the oscillator p.c.b, and bring a metal object over and above the DR. (I use the handle of a scalpel knife). If you don't see any sign of oscillation (not untypical) then experiment with the positioning of the DR by moving it around with a pin ( I use the sharp edge of the scalpel). Should the slightest whiff of oscillation be seen whilst this is done, stop, and bring the metal object back over the top of the DR, moving it up and down over the DR until oscillation is seen. If you see this, then you have the DR in almost the right place. It is hard to describe this practical work in an easy way, as seeing is the only way of learning such techniques. However I did learn them out of my own instincts. If you don't see oscillation, repeat the whole procedure, but instead bring the can of the unit above the DR, and move it up and down to look for oscillation.
Once oscillation is established, make a mental note of the position of the DR, and sparingly apply some slow setting adhesive to the underside of it (cheap nail varnish is ideal, I prefer the clear type, as coloured varieties may have different dielectric properties). Look for oscillation again, and when satisfied, try to get the unit to oscillate nearly on frequency with the can nearly on. This is a certain sign that you've nearly done it.
If you can get oscillation when the can is on, at or around the desired frequency then leave the DR to set, otherwise just keep moving the DR around by small amounts and keep checking until you can get the oscillator working. You have plenty of time as the varnish takes a while to set.
Once the oscillator is working, check it will work repeatedly, by turning off and on the 4V d.c supply to make sure that the oscillator will start up every time. This is very important, as we don't want the LNB to become tempremental. If all is OK then leave the whole thing to set for a day.
When the adhesive has had time to set, power up the oscillator and check that it still works like when you left it. If it doesn't then you will have to pry off the DR by softening the adhesive (nail varnish remover), and then repeat the whole tuning process again. If it does, however, you can proceed to put the whole LNB back together again.
Solder the four corners of the p.c.b to the can using the minimum amount of solder. Re-check that the oscillator is still working. Bear in mind that it will change frequency a little after the can is soldered. If all is still OK then solder the p.c.b back to the can using the minimum amount of solder and heat to make a good soldered seam. You have now successfully modified the local oscillator to 9GHz and deserve to pick up some good DX on 10 GHz for your efforts!
Put the oscillator block back onto the RF p.c.b, and solder the LO module to the RF p.c.b first, at each end, like before. Make sure the module sits right down on the RF p.c.b as there is not much clearance inside the LNB. Then carefully re-solder the two feedthrough pins on the top side of the RF p.c.b, with just enough solder. Carefully clean up the waveguide probe hole, and the probe itself with desolder braid, then again carefully put the RF p.c.b back into the LNB enclosure, making sure that the linkage p.c.b goes through to the IF p.c.b correctly, and that the waveguide probe pin goes through the hole on the RF p.c.b correctly also. Fix all the screws you removed from the RF p.c.b and solder first the IF p.c.b linkage side, and lastly the waveguide probe back - look out for that static again!
Now put the LNB on your favourite satellite receiver, and tune in a workshop generated 10 GHz signal. You should find it just where it should be. If not, then track your steps back and check through everything as before. If all is OK then you've got yourself a good sensitive and very small 10 GHz LNB.
At this moment in time, I haven't re-tuned the amplifier stages or re-aligned the interdigitated filter for 10 GHz, but they seem to work pretty well as they are, certainly better than other similar types of LNB in this class (I.e Grundig / Uniden types).
Although this conversion is more involved mechanically than most other units, it seems worthwhile to do. My motive was to get a physically small waveguide type LNB for use in a confined space., as these LNBs are easily and cheaply acquired at present, it seems a shame to waste a useful unit. With newer LNBs having only a feedhorn input, and the increasing number of digital services appearing in the UK you have to make hay whilst the sun shines, and get hold of these waveguide ones before they're all gone!
A.M October 1998
Back to the waveguide LNB page