RF front end solutions for 2.4Ghz transciever.
I have been working on a small project of mine for a while now, involving wireless digital audio streaming, gathering prototype RF modules along the way. Although I'm leaning towards the NRF24L01, which have built in protocol management, I would like to experiment with my own baseband processing. Thankfully I have some old samples of ML2724's, which are very similar to the new ML2726 (in fact they are almost identical):
http://www.cdiweb.com/datasheets/mic...r/ML2726DH.pdf
This would allow me to use my own baseband protocol, and its 'limitless transmit time' comes in handy for simplex comms.
There is however one major downside: it does not have integrated LNA, matching network, or antenna switch as the NRF24L01 does. Meaning I must add these externally. I am not an RF engineer, so this stuff scares me a bit (RF analogue...) but I have done much research on matching networks and possible discrete solutions. As this is for short range experimenting, I don't think I will need a power amp for the transmit side but any integrated solution that would reduce the external part count (and therefore ease the PCB layout) woudl be handy.
I have looked at the new CC2951 from TI, and the T7024 from atmel, however, the latter doesn't contain an internal T/R switch, and the CC2951, aside from being rather expensive, uses the same pins for RF in and RF out. Great for adding a 'range extender' to the CC2500 line of chips, but I have no clue how to connect this to the ML2724/6.
The chip I'm using has a differential (or complimentary?) transmit path, but a single ended receive, on separate pins. Looking at an application note I have for it, the schematic shows a balun (to convert differential to single ended?), and a couple of 2.4Ghz band-pass filters in the transmit path, along with an LNA and bandpass filter in the receive. This adds up to quite a few passives, which normally wouldn't bother me but with RF layout being critical, I cannot trust my PCB design skills to create a semi reliable system. The antenna can be external via an SMA, as a PCB trace antenna is just too much hassle. The app note is a number of years old, so I was hoping that there have been relatively new advances in front end integration to make my life easier.
So, does anyone know of integrated solutions which can take the sting out of this? Preferably an LNA, and T/R switch on chip, as well as a matching network (tall order I know..). Power isn't an issue, I'm not after long range, it is purely from a design convenience point of view, so I don't have to worry too much about critical layout, and searching for 'baluns' that do the job.
Hope someone can help,
Buriedcode
I wasn't sure if this was the right part of the forum to post this, perhaps it should be in the 'digital communications' sub forum. If an admin can move it there, I would be most grateful :)
SKYWORKS : Products - Front-End Modules for ISM / ZigBee? / 802.15.4
Thanks for the answer! I had looked at those, and they seem VERY promising, looking for a small quantity supplier now. Now comes the bit that confuses/scares me. Impedance matching. I understand the baluns (be they integrated or external) convert a balanced output from the tx, to a single ended out, and likewise for the Rx.
The SKY65344-21's differentential port impedances are, TX: 29+j75, RX: 68+j38. My ML2724, which has a differential transmit path, (and single ended Rx) has TX: 12+j0, and RX: 2.2+j0. I am completely out of my depth with this sort of impedance matching, the only reference design I have uses a 50/50 ohm balun in the transmit path. So, would I need to use a 50/50 Balun to convert the differential Tx, to single ended, and then another balun to convert this impedance to whatever 'front end' device I use?
I realise this isn't a case of 'plugging things together', all paths will need some form of filtering, and matching, but it is starting to look like I'll need all these passives after all. I guess I should learn to walk before I cna run, so perhaps seperating it into simple building blocks (balun, PA, TR switch, LNA, BPF etc..) may prove more useful in debugging. Any pointers in tutorials or simulation software on this? It is by no means a professional setup, just tinkering on my part.
“Thinking is more interesting than knowing, but less interesting than looking”. --Goethe
I believe you think too deep when things actually are not such complicated. You can use the Sky module as they recommend in the application note (page 6) and just add a balun on the TX side of the TRX module.
For simplicity I would recommend to use from the beginning just the TRX as it is (adding only a balun on the TX), not even the T/R switch. Use one small antenna for RX, and one for TX. Build the prototypes, test your digital, the protocol, and so on. And after that, you can start thinking and learn about RF side.
Thanks for the advice :D I do tend to over complicate matters, especially in a field I'm unfamiliar with. Also, the reference design I have is probably designed to be 'perfect' for showing what the transceiver can do, as manufacturers want people to see their products at their best.
A separate antenna's actually sound like a good idea, although the RF switch (from the data sheets I've seen) seem to be quite simple. Its all the numbers and specifics that are weighing heavy on my mind... that and one possibly crucial point: I am making my own PCB's :D. I've made many PCB's in the past couple of years, and can now do double sided, with 10/10 mil spacing, so I thought I might be able to get away with doing a quick and dirty prototype for these. My paranoia is trying to make sure everything else is 'ok, before I design, etch, solder on all the parts, and have two 'dead' bits of hardware to debug.
I have made a parts list of things, and thankfully it has only come to ￡4 (thats S6 US), so I may add these to my next big order, and, as suggested, start off basic. Operating range can wait, I'm happy with a few meters for now whilst I code my protocol.
Thanks for your help! The reason I came here rather than just 'collect schematics' and try to work things out myself is, the ML2724 is an old chip (2002?) and has fallen out of favour to Nordic and chipcon devices. But I have some, and I'm sure they're still useful. As always, if anyone has reference designs... :D
Well a little further research into part availability, I believe I can find all the components necessary for the transmit path. I have found the T/R switch, and band pass filter, which I may or may not use for my prototype. At the moment I am just concentrating on getting a half decent (ie: working) transmit and recieve path.
The reference design I have is quite old and is for short range game controllers. I believe there is a newer reference design for higher power purposes, but I can only imagine the difference being an added power amplifier on the transmit side. From reading datasheets almost all RF 2.4Ghz power amps have 50ohm in and out impedances, meaning it would be easy to add those to this design.
Here are snipets from the schematic, one of the TX path, and one for the RX path. I have all the components necessary for the TX path, but the LNA used (RF2472) isn't available to me. This apparently has a perfectly matched 50 ohm input, with open collector output (as many seem to have). Finding an LNA with a matched input seems to be rather difficult, meaning I would have to add my own network - would the stock reference designs given in datasheets be ok? The reference design for the RF2742 seems wildly different from the one used in this schematic. (much lower value DC blocking caps).
So, the TX side, I'm confident I the schematic provides a decent match from the output of the transciever (12+j0 ohm, balanced) to the switch (50 ohm unbalanced). I am new to this but it seems the inductors on the Tx lines are the DC current path, and the two caps with a parallal inductor (just before the balun) are the matching network.
As each output from the Tx is 12+j0 ohm, would that mean that the network must match 12+j0 to 25+j0 (25 ohm, because the balanced input to the balun is 50ohm)?
The bit I require real help with is the LNA. Given the RFin of the transciever I'm using has an impedance of 2.2+j0, I am clueless as to how to match this (roughly, not after perfect here) to the output of an LNA? Most datasheets provide a reference design for a 50ohm output, and some are internally matched, which is fine for a 50ohm input but my IC doesn't have that. Also, since the LNA used has an internall latched 50 ohm input, I imagine I would have to add external matching for an LNA that doesn't have this.
I realise I'm a newbie with RF/microwave, and its NOT an easy subject, but I would be very grateful if someone could explain the basics of these two schematics - mainly the RX path.
Apologies, but I get a bit obsessed with things when I don't understand them. This isn't a professional project, or urgent (as one can buy premade RF modules very cheaply) it is mainly academic curiosity - my University course a few years ago skimmed over RF >.<
BuriedCode.
Edit: Just noticed that the datasheet, when describing the pins of the IC, quotes:
So, how would I determine the circuit/values for connecting a commercial LNA IC to this input?