1. 2PAD Analogue System Tim Ikin SKADS - Limelette
Good morning. My name is Tim Ikin, and today I will be presenting the design, and measurements of the 2PAD analogue back end system, as well as considering the future through PrepSKA. This is collective work from Universities of Manchester, Oxford and Cambridge.
6th November 2009
SKADS - Limelette

2. Gain chain & cable & splitter
Freq
Balun
LNA
Filter & Gain
Gain chain & cable & splitter
Signal Conditioning
500
-0.15dB
20.3dB
-42.7dB
45.1dB
4.7dB
875
-0.42dB
18.8dB
+17.2dB
39.5dB
2.76dB
In this slide I am showing a simplified schematic of the analogue front and back end of the 2PAD system. I will be going through this step by step in some detail.
Here we have the balun connecting between antenna and LNA, then we have a filter which also has gain. Next we have the gain chain module, CAT-7 cable, splitter module and signal conditioning module which takes us all the way into the data acquisition card and the ADC.
I have drawn lines on here to show the barrier of our shielded bunker and also the transition to the digital domain here.
Every cable and every item of the system has been exhaustively tested in the lab at Oxford University by Sascha Schediwi and his colleagues. They have made the results publically available on the web. The purpose of this testing was two fold. It was to give a good measure of performance item by item, but also to provide the DSP engineers with the phase performance of each item.
I have used their measured data from one example chain just to bring out a couple of frequency points to show the insertion effects of each item. None of these components is ever flat in response. Here we see a snapshot of the filter performance.
6th November 2009
SKADS - Limelette

3. Front end system SKADS - Limelette
Here we look in detail at electronics in the front end.
First I show the balun, which takes us from 150 ohm balanced signal to 50 ohm single ended. This is simply a placeholder for the LNA under development by Danielle George and her team. It does have a small insertion loss, which defines the noise performance of our system. The copper tracks of the balun are joined to the aluminium antenna by means of 2mm screws with a beryllium washer between them.
Next we have a low noise broad-band amplifier from Mini-Circuits. This has a gain of ~20dB and operates at 3V. This device has good headroom to cope with any transmitted signals in the region.
Next we have the filter module. This is a narrow band filter specified to exploit a 40MHz region of relatively low noise in our band. It has two amplifiers either side so has a net gain of 17dB in the pass band.
The gain chain is a development between Oxford and Manchester, and Sascha has developed this to this present form. It simply takes in four SMA lanes from the LNA and applies 57dB of gain to each one. The four signals are converted to differential mode before being sent down the 20m of twisted pair cable. There is no frequency conversion going on and no multoplexing – this is a very simple system.
At the output side of the gain chain module we see the CAT-7 connector, and this is what I will speak about next.
6th November 2009
SKADS - Limelette

4. Analogue Transport SKADS - Limelette
The analogue signals are transported down 20m of CAT-7 cable to a remote RFI shielded bunker. The CAT-7 cable was selected because of low cross-talk, low cost and high signal density. Each of the four twisted pairs are foil wrapped and there is an overall braid. The target market for this cable is for low cost 10Gb links. We are not pretending this is the only choice for SKA AA, and I will discuss this later.
At the end of the CAT-7 cable we have a separator card which brings the differential signals out as single ended again in SMA connections. The shielding interface is at the SMA connectors at this point. We have tested the shielding performance of this interface and it is extremely effective. Much of this design is influenced by work of Paul van der Merwe and Professor. Howard Reader.
6th November 2009
SKADS - Limelette

5. SCM to DAQ SKADS - Limelette
Signals from the field are treated in the Signal Conditioning Module.
This is an important part of the system because it applies a little gain about 5dB, but it applies the equalisation filter required to correct for the steep roll-off of the CAT-7 cable, and it also contains the anti-aliasing filters required by the ADC on the next card. Additionally here we have attenuators built in which are digitally controlled. Every signal path is flexiblly switched so the signal may be routed to by-pass all filters, or isolated from the ADC. Finally a monitoring amplifier will send the signal to the front panel which has been crucial for diagnostic purposes.
The control of this card is by means of an I squared C bus which comes from the instrument management subsystem which is on the next card.
6th November 2009
SKADS - Limelette

6. Winter Hill Pre-DSO Winter Hill Post-DSO SKADS - Limelette
Analogue 500kW
Digital 10kW
Trans.Power (log)
Winter Hill Pre-DSO
Roughly 43km to 2PAD
Local repeat 250W
1MW
100kW
10kW
1kW 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68
626MHz
634MHz
642MHz
650MHz
658MHz
666MHz
674MHz
682MHz
690MHz
698MHz
706MHz
714MHz
722MHz
730MHz
738MHz
746MHz
754MHz
762MHz
770MHz
778MHz
786MHz
794MHz
802MHz
810MHz
818MHz
826MHz
834MHz
842MHz
850MHz
Trans.Power (log)
Winter Hill Post-DSO
Roughly 43km to 2PAD
Digital 100kW
Here I am showing in very simple terms the TV broadcast signals which affect us. Right now we are in the transition from this scenario to this one. We have very powerful analogue transmission which is being turned off by 2nd December. Then the two digital broadcasts which are intruding will be moved away.
The power of the digital broadcasts will be increased by 10dB.
1MW
Local repeat 40W
100kW
10kW
1kW 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68
626MHz
634MHz
642MHz
650MHz
658MHz
666MHz
674MHz
682MHz
690MHz
698MHz
706MHz
714MHz
722MHz
730MHz
738MHz
746MHz
754MHz
762MHz
770MHz
778MHz
786MHz
794MHz
802MHz
810MHz
818MHz
826MHz
834MHz
842MHz
850MHz
6th November 2009
SKADS - Limelette

7. EMC
It is bad engineering to attempt to bolt-on a shielding can if the need arises. EMC must be designed in from the outset. Neil Roddis and others have raised this last week in the WP2 meeting, and it is not yet being treated seriously in every case.
What we have found with some of the items in my talk today is that there has been poor design decisions relating to shielding policy. As an example, we have measured RF currents flowing on outsides of cables which should be quiet. We have found sloppy attempts at building shielding enclosures without understanding the mechanisms of leakage.
The basic rules are to think of conducted leakage and radiated leakage from your electronics. There are software packages which can aid design, but these are only tools and misuse of them can be a mistake. Get onto some of the good courses on EMC which are being run around the country. Industry is obliged to comply with a very limited performance target, the SKA requirements are much tougher.
Get away from copper connections wherever possible.
While enclosures are often a major cost considerations good shielding need not dominate the system cost. We have been meeting with various industries to discuss large volume manufacturing of shielding enclosures, and the ideas are out there.
6th November 2009
SKADS - Limelette

8. Meeting power budget
SKA AA requirement 200mW max. in the analogue front end, including LNA.
We are an order of magnitude away from meeting this target with present COTS technology. We need to ask some serious questions about the basic topology, and with recent suggestions to raise the upper frequency limit on aperture arrays to 1.4GHZ, now is probably a good time to take a critical review of the analogue system we have developed here.
Should we be using a twisted pair technology with such a strong roll-off? Possibly not, and there are many and growing arguments against it. Should we be using copper at all as analogue signal transport? Not necessarily. There are many good people in SKA who are involved in RFoF technology, and this still has not been properly addressed in this application. It is very likely that fibre links are too expensive but we need to keep an open mind. We are working with Roshene’s contacts, which is a very friendly group, and we will investigate and monitor the state of the market.
Should we be transferring analogue RF in the first place? Isn’t it time to move the ADC up to the front end? If it’s done properly, there is no reason why not. See Chris Shenton’s talk.
6th November 2009
SKADS - Limelette