what does insertion/return loss mean when talking about baluns?

Insertion loss is expressed in dB and pertains to the efficiency of the balun when passing a signal. A balun with a 1.2 dB insertion loss is somewhat less efficient than one with 0.2 dB IL. For example if a device has a 3 dB insertion loss and you generate 100 watts (+50 dBm) at its input you will lose 1/2 the power and will have only 50 watts (+47 dBm) at the output. Additionally the device has to dissipate the lost energy as heat and this will also affect its longevity, unless it is power rated accordingly.

The insertion loss specification assumes that the device is properly terminated, bringing us to the next question:

Return loss pertains to the power reflected back from a device at its input (and likewise output) port. Return loss is measured with a return loss bridge or other device. This is a figure of merit of the proper matching of the source impedance to the load impedance.

Imagine a spot light focused on a mirror, nearly all of the power will be reflected back to the source. Alternately, if that light is focused on a block of charcoal, a great deal of the light will be absorbed into the charcoal and little will be reflected back. The light absorbed by the charcoal will perform "work" in that the charcoal will become heated. This may be a clumsy example, but keep in mind the concept of doing "work". A similar concept is the selection of proper gear ratio when riding a bicycle up a hill. The proper ratio (match) yields the best "work".

Ideally if a device such as a balun is properly constructed and terminated properly, the amount of power reflected will be zero. Unfortunately this is seldom the case and some amount of power will be reflected back. The ratio of the power into the device and the power reflected back is the return loss which is often expressed in dB's. Another figure of merit is voltage standing wave ratio (VSWR). The following table depicts the relationship between VSWR and return loss. A device with a return loss of 20 dB is better than a device with a return loss of 14 dB.

As far as your question about the passive devices between the chip and the antenna, the easy answer is "yes" however, proper matching of the device requires some passive components, so it would be safe to assume if the manufacturer recommends a certain matching network to yield a specific receiver sensitivity threshold, those components used in an evaluation circuit would already be included in the terminal sensitivity figure. Usually manufacturers will recommend a specific network be used to measure parameters of an RF device in their application notes.

If you have a lossless transmission line you can express all of it's behavior as power travel. Power that's going from a source to a sink is 'forward' power, and power that's going back to the source is 'reflected' power.

If you connect a network (like a balun) to a transmission line and you properly terminate the other end of the network, and then inject some RF into the transmission line, power will show up at the network terminals. Some of this power will go all the way through the network to the load

-- that's not lost. Some of it will get burnt up in the network -- that's insertion loss. And some of it will bounce out of the network and back down the transmission line -- that's return loss.

They consume power, and so reduce the amount of power available at the antenna in transmit, and the signal strength at the IC in receive.

Get a copy of the ARRL handbook. It claims to be a hobbyist publication, but it's really a book for people doing serious, practical RF work who don't have PhDs (or who find that the esoteric theory that they learned getting their PhD doesn't help them make radios work). I would guess that at least half, if not 90%, of all RF engineers own at least one copy (it gets updated annually). When you want to understand the minutia of transmission line theory you dive for an E&M book. When you want a practical system with a transmission line in it to work, you dive for the Handbook.

This makes perfect sense. Just to be sure, however: the insertion loss goes both ways, right? In other words - both received and transmitted signals suffer from insertion loss - so a 1dB insertion loss really means a 2dB link loss as it'll happen on both the receiving and transmitting ends of the link.

So on page 152 of the AT86RF212 datasheet

Thanks!

-Michael

My bet is that sensitivity it spec'd for 50 Ohm signal at pin 4, the receiver input, so it would not account for the balun.

See my answers embedded below:

Michael wrote:

Yes this is typical for passive devices. There is at least one exception- RF Isolators (AKA Circulators) that have an insertion loss in the forward direction and an isolation loss in the reverse direction. Of course a passive has to be in both the transmit and receive path to affect both equally. For example a filter used to clean up the transmitter can be put between the antenna and the antenna switch to both clean up the transmitter and provide frequency selectivity protection to the receiver, but doing so, will reduce the path performance by 2X its insertion loss (once at each end of the link). If this component does not truly provide a benefit to the receiver, it should be placed between the transmitter and the antenna switch.

I would expect so. If you build the circuit and you complain to the manufacturer that you don't measure the RX sensitivity or power output, the manufacturer is going to ask if you built it to the application note specs using the same parts. I would however caution that filter 1 (F1) is optional and the vendor may be "forgetting" to include that as a worst case. Also beware of words like "typical", may mean that they are not guaranteeing the performance. I am sure you are aware that the data rate, modulation scheme, transmitter masks and receiver IF filters will all affect the link margin.

You should build a link, put each transceiver in a well shielded can and use a variable attenuator to simulate the link performance. Beware that step attenuators may introduce glitches that will cause the link to drop out. If you can obtain a truly variable attenuator as part of your test set up, you can approach the threshold more smoothly.

Since we are on the discussion about link budgets, I suggest you read up about log normal fading, Rayleigh fading etc, so that you don't rely entirely on free space loss and become disappointed in the real world performance. There are more documents than I could even list here that cover the subject of RF propagation. Fortunately there is a lot of study in that area.

See my answers embedded above: