Satellite signal switches are a common convenience item, allowing a single receiver injecting a steady or encoded 22khz tones on the DC bias on the coax line to remotely control a switch, in order to select from 2, 4 or even more satellite LNBs. Satellite enthusiasts with multiple LNBs or multiple dishes are sure to have encountered various forms of satellite signal switches, namely the 22khz two-way tone switch and the plain DiSEqC 1.0/2.0 four-way switch. More elaborate DiSEqC 1.1/2.1 eight and 16-way switches, as well as DiSEqC 1.2/2.2 motorized single axis motor control allows for even more possibilities.
However, the 22khz tone switch and 4-way DiSEqC switch still remain the most common, due to their low price. They can easily be had for under AU$10, sometimes under AU$5, which is very inexpensive. The convenience of having four LNBs, selectable at the end of a moderate length run of coax, is of great value to a hobbyist. However, things are never as rosy as they seem, as these switches occasionally fail by succumbing to heat and moisture, short circuits or transients on ports, or have compatibility problems and don’t reliably respond to the control signals.
As a result, many hobbyists have spent time purchasing different switches, swapping them in and out to find one that the receiver likes. I’m no different. Sometimes the fault is in the receiver too.
But I’ve always wondered what’s inside these switches, given their cost, so I took my spare switches, disassembled and photographed them for all to enjoy.
Gecen SW-03/Q 22khz Tone Switch
Gecen seems to be one of the most dominant names in the switch market, with their products priced very competitively and moderately good experiences from users. The SW-03/Q is one is one of their basic 22khz tone switches – the lack of 22khz results in LNB A being fed through, the presence of 22khz results in LNB B being fed through. This is suitable for users of regular plain LNBs, whereas universal (9750/10600) users cannot use this switch as the 22khz tone is used internally by the LNB to switch between the two local-oscillator (LO) frequencies. It is also not advisable to put this switch ahead of a DiSEqC switch as it will corrupt the encoded 22khz signalling to the switch. Placing one of these switches after a DiSEqC switch is possible, provided the remaining 22khz amplitude is enough to activate it.
The specifications, which seem slightly dubious from the standpoint of their “consistency”, claims:
- 950-2400Mhz frequency range
- 3dB insertion loss
- 20dB isolation
- 8dB return loss
- 12-20v working voltage
- 10mA power consumption
- 500mA current passing capability
- 75 ohm impedance
This unit itself has been outside, in the weather for a few years, and so the red Gecen logo has faded away. The date code on the sticker seems to generically label large batches of switches – this one is dated 3rd February 2011. It still functions, although, some receivers don’t seem to put out a strong enough 22khz tone to activate it, and was removed from service with a desire to upgrade the system to handle more satellites.
The Gecen switches seem to use a steel covering plate and some silicone to secure the rear cover. Unfortunately, the steel covering plate rusts when exposed to the rain. To open up the switch, score around the silicone edge repeatedly until all of it is cut off. Then, use a hard drive magnet to grab the metal plate, and pull it away. Simple!
Grabbing a soldering iron allows for us to clean away the solder connecting the centre pins and extract the PCB entirely from the cast-metal frame, which forms a continuous ground around the board and also forms the threaded outer parts of the F-connectors. The screws also serve the function of connecting ground to the PCB, and the large number of them is deliberate to ensure a low-impedance connection.
The PCB is date coded week 52 of 2010, and has a solder-resist clear-out mark of K002C. All components are surface mount, with no integrated circuits used throughout the design. It seems that the ports have their signals broken into two paths – one going through a printed inductor (zig-zag trace) which likely carries the DC component, and the other going through a capacitor or PIN diode, carrying the radio-frequency component.
The two output ports have their power switched by Y2 marked SS8850 PNP Transistors or similar, thus the load of both LNBs is not presented upstream as might be the case with some other switches. There are also other 1A.G marked BC846A NPN Transistors or similar. The outputs appear to be isolated by a pair of series diodes, and are just mixed together by putting their outputs together.
Surprisingly the board is a single layer design, and the underside of the board houses nothing but the fibreglass substrate. All that desoldering effort for nothing. Unusually, there are holes which suggest there may have been some vias and traces on this side, and may have been left-over from earlier designs.
Gecen GD-41C 4-way DiSEqC 2.0 Switch
Gecen makes a large variety of switches, although they mainly seem to differ on the port configuration and exterior styling. Older switches had boxy outsides (like the one above) with the newer switches having a curved design. The GD-41C is a four-way DiSEqC 2.0 switch which is backwards compatible to DiSEqC 1.0 which is supported by a vast majority of receivers. The specifications are identical to the one above so I won’t copy it again.
This switch had succumbed to weather, and was attacked by heat and moisture over several years. It didn’t help that I used some cheap compression F-connectors from China that weren’t waterproof despite their claims. This unit’s crazed label has had the red Gecen logo faded away, with a date code of 23rd December 2008. The rear steel plate has pretty much rusted to death to the point the Gecen holographic label has come off entirely.
Removing the rear plate had confirmed that the rust had gotten through to the other side of the plate, meaning moisture had entered the circuitry which explains the inconsistent performance. After some solder-sucking, we had the PCB removed.
This PCB is marked with a date code of week 2 of 2009, and its silk-screen print says 106H6. The PCB seems to have a very unusual layout, with all output ports having two connection finger areas, with them appearing to be isolated, but they’re not! They’re only separated by solder resist and are otherwise continuous – likely to accommodate different chassis with different port spacings.
The more sophisticated DiSEqC protocol is handled by an unmarked SOIC-8. The power to each pair of ports is controlled by a BC807DS Dual PNP transistor marked N2. There is also a PMST3904 NPN transistor marked W1A. A similar diode combination/isolation of signals seems to be taking place, although with two hierarchies – two series to isolate and select which port from each pair of ports, and then one more to select which pair. I’m not sure how good that arrangement is.
The rear of the PCB has a continuous grounding plane with some vias and traces connected underneath. No solder resist was applied and it seems some traces had started to corrode slightly.
Chieta WSD-2041 Heavy Duty 4-way DiSEqC Switch
This switch seems to be the recommended switch by satellite forum users when it comes to eliminating problems. It’s apparently claimed to have much better quality, although supply is limited as they (Zhuhai Chieta Electrical Co. Ltd.) seems to have stopped manufacturing switches altogether.
The unit itself is light, but bulky compared to the Gecen switch above. It features specific grounding terminals, if you need to use them. It claims a frequency range of 900-2400Mhz (as they all do), but a lower insertion loss of 1.5dB and a higher isolation of >30dB. It uses slightly more power at 25mA, and operates off the same 12-20V DC range. Some sources claim the passing power capacity is 450mA, which is a little short of what most receivers can power, but most LNBs don’t consume more than 200mA anyway.
The rear plate on this unit is aluminium, which doesn’t rust. That’s good. But the tolerances between the plate and the shell show a large gap, and the silicone beading wasn’t particularly well done. Its companion succumbed to water a long time ago, this unit remains functional but taken out of service. I decided we should probably see what quality looks like.
More desoldering … and viola.
Quality is a very empty PCB marked with EMC4104-2009-4. It’s also one where each centre pin is only connected to one side. It features an unmarked SOIC-8 package to control this switch, and discrete Y2 marked SS8850 PNP Transistors for each port’s power. There is a 1AM marked MMBT3904 NPN Transistor as well.
The star difference seems to be the lack of strange diode-arrays like the Gecen units have – instead these are mixed together with A8 marked BAP50-03 General Purpose PIN Diodes. The diode array above is likely to ensure better isolation between each port, but is an arrangement of just single diodes better? It seems unlikely, but I’m not sure – it would have to depend on what the specification of the diodes are (and the Gecen ones are unknown).
While the construction quality seems better, and the layout is nicely spaced and symmetrical, and the casing doesn’t rust, any functional advantage isn’t obvious.
The underside of the PCB doesn’t hold any particular surprises for us. It is solder-resist covered, for protection, and features a large ground plane with vias interconnecting it to the top ground planes. The case is used to provide ground shielding all around the board with screws ensuring pressure between the exposed board hole pads and the casing.
For one reason or another, satellite hobbyists always will have some bucket of spare DiSEqC switches. I’ve managed to take them apart, and take a look at what their design looks like, how they (approximately) function, and what components are used. On the whole, the switches are cheap because their designs are simple. It seems that even the quality Chieta switch doesn’t offer many surprises in its design either.
I’ve moved on to slightly more sophisticated 8-way DiSEqC 1.1/2.1 switches for now, but I’m not taking that one apart just yet!
No switches were harmed in the making of this post. All of them were re-soldered together in a similar condition to when they were taken apart.