How to calculate the "insertion loss" at different frequency

[Phillip L.]

Hi, I am trying to select PXI switch with the correct bandwidth rating.

I know that a switch with a bandwidth rating of 10 MHz implies that the switch has an insertion loss of -3 dB at 10 MHz. If the input signal is a 10 MHz, 1V peak-to-peak sine wave, the output signal will be a 10 MHz, 0.7 V peak-to-peak sine wave.

But how do I calculate the insertion loss at other frequencies?

e.g. if my input is 1MHz, 1V peak-to-peak sine wave, what is the output signal going to be?

Thanks

Phillip

Message Edited by Phillip L. on 06-05-2008 04:07 PM

[Brooks_C]

Hello Phillip,

A number of our switches will have a graph of insertion loss vs. frequency for the entire bandwidth available in their specifications document.  In many cases you'll find that our bandwidth specification is rather conservative with respect to the 3dB mark (in many cases our bandwidth is more like a 1.5-2 dB loss).  If you would like to post back with the specific switch model you have I can help you locate this graph.

I hope this helps, and have a great Friday!

Cheers,

[Phillip L.]

Hi Brooks,

2530 is what I am looking at.

Thanks

Phillip

[Brooks_C]

Hello Phillip,

I believe one reason that we don't provide a graph of insertion loss vs. frequency for this switch is because this is a matrix switch.  Since it is a matrix switch that means that there are a number of different possible paths between rows and columns so the insertion loss may vary between path lengths.  Thus we specify the bandwidth which means that any path in the switch will have an insertion loss of less than 3dB at frequencies up to the specified bandwidth (as you said earlier).

If your particular application requires more information than this then I would recommend measuring the insertion loss vs. frequency for a particular path in your switch.  Here is how I would measure the insertion loss:

1.  You'll need a signal source and a measurement device.  This can be a spectrum analyzer or any function generation + oscilloscope/data acquisition card should be able to make this measurement.
2.  Once you know which connection path you want in your switch you need to measure the RMS value of your signal source and then send the signal through the switch and take another RMS measurement.
3.  Once you have these two measurements you can calculate the insertion loss = 20*log10(RMS voltage through switch/RMS voltage without switch).

There are several ways to do this, if you have a network analyzer this is likely a measurement you can make directly.  More generally with any signal source and measurement device you can either split the signal source and route it through both the switch and just a cable to the measurement device or make each connection separately.  You should also keep in mind that the cables and connections you make will introduce loss of their own so you should measure this loss as well to take it into account.

Cheers,

[Phillip L.]

Hi Brooks

I am also looking at 2503 which is a multiplexer but yet the insertion loss vs. frequency graph is still missing in the datasheet and specifications.

I am also confused about why the specifications for 2593 is included in the same document.

Thanks.

Phillip

[Brooks_C]

Hello Phillip,

I believe that the same reasoning applies to the PXI-2503.  If you look at page three of the PXI-2503 Specifications you can see that the switch module is actually composed of 4 relay banks.  This means that the path length of each pole on the multiplexer will not be the same; in turn the insertion loss will also not be the same.  If you're concerned about this then you probably need to consider looking at a switch that is marked for RF use (RF switches) since most of these do specify insertion loss vs. frequency.

Also, I'm not sure what you're referring to with the PXI-2593 specifications being included.  However, it is likely a mistake and if you can post a link I'd be happy to get it fixed.

Thanks,