BATC Forum

David M0YDH presented me with yet another Portsdown 8-Band RF Switch for fault-finding at CAT 21. I am aware that a number of these switches have recently failed in-service with the expensive 8-way RF switching IC needing replacement.

I think that the failures may be caused by the use of these switches with the LimeSDR, as the PCB was originally designed for use with the Filter-Modulator Board.

The maximum output of the filter-modulator board was about +7 dBm, and the frequency dependent attenuator on the input of the 8-way RF Switch PCB (before the RF amplifier) was designed to give a maximum output level through the RF switch of about +12 dBm. The problem with the LimeSDR is that it can easily deliver +17 dBm during its calibration pulse, thus feeding +22 dBm into the RF switch, which exceeds its absolute maximum rating at 146 and 437 MHz.

The solution is to increase the attenuation on the input to the 8-way RF switch, or to use a lower gain RF amplifier.

My apologies that I did not spot this system design error earlier. More information to follow as I am about to put an 8-way RF switch into my LimeSDR based portable system.

Dave, G8GKQ

Interesting - as you know I had at least 1 and maybe 2 fail.

A solution could be to power the amplifier stage from a rail switched by the Tx line so it is only powered after the Lime cal routine.

Noel

Yes, but switching the supply is only a partial solution as it does not prevent a Lime that is set Lime Gain 100 delivering too much power on 437 MHz and blowing up the switch IC.

Dave

Better to limit the power perhaps? E.g. with PIN diodes. Either with individual diodes or a chip like the SKY16602-632LF which will limit at about 6 dBm.

You can even get them ready made https://www.ebay.co.uk/itm/312623587269 - just a couple of diodes and dc blocking capacitors.

Mike

I have designed and tested a new input circuit for the 8-way RF switch which should protect the RF switch IC. I have put a 6 dB attenuator after the MMIC so that the maximum input to the RF switch IC is +15 dBm, not +21 dBm. To compensate for the loss of gain I have reduced the input attenuator values. The overall gain of the modified switch PCB is about 3 dB, but rises to 5 dB at 2.4 GHz (where the Lime output is much lower).

The revised circuit is below. The surface mount components for the new attenuator were soldered together on the top of the PCB. I scratched away the insulation where I needed a ground connection, and put an insulation sheet inder the other connections. I have also attached a table showing the maximum levels and gain at various frequncies. While I was at it, I measured the Pluto output level to check that it was within a safe range.

The table shows the 8-way switch output level with Lime Gain at 100, the 8-way switch output level with Lime Gain at 90, the barefoot Lime output level with Lime Gain at 90, the gain through the switch PCB, and the Barefoot Pluto output level with Pluto level set to 0 (max)

If there is demand, we might be able to get a new PCB made for this design. Please register your interest here.

Dave, G8GKQ

Hi Dave,

many thanks for publishing this very quick solution to solve the issue.

The modification should be quite easy to to but I would like to order and buy two modified PCB's if they become available.

Thanks for the your excellent work.

73 Wolfgang