calculating a link budget

Hi there - I'm pretty much RF handicapped. I'm trying to figure out some basic RF things, however. Specifically, I'm trying to figure out how to make a link budget.

So, for example, let's say I want two Nordic Semiconductor nRF24L01+s (datasheet:

formatting link
to talk to each other. On page 8 of the datasheet, it says it has a max output power of 0dBm, and -94dBm sensitivity at 250Kbps. So my understanding there is that the max output power is 1mw and the minimum received power is 251fw (10^(-96/10) mw) . Can that even be right? I mean, that seems like an incredibly small amount of power to receive.

OK, next up - so when figuring out a link budget, you first calculate out the link margin, which would be the max output power - minimum input power, or in this case 0 dBm - (-) 94dBm, giving a link margin of 94dB. Then you just sum up all the losses and gains in between the output and the input. So for example let's say we have a perfect connection from the IC to the antenna. Let's say I use the patch antennas shown here:

formatting link
for both sides of the link. So as long as they are both facing each other I'll have 9dBi gain on both sides of the link. So that means I can now lose 0 dBm - (-)94dBm + 9dBi + 9dBi = 112dB to all other factors (cable, free space/path loss). Let's just say I am also losing another 5db on each side due to connectors, antennas, etc. And let's leave 2db link margin. So that gives me 100db to free space loss. According to wikipedia, free space loss = 32.45 dB +

20*log[frequency(MHz)] + 20*log[distance(km)]. So if I solve for 100 = 32.45 + 20log(2.4E3) + 20log(distance) I find that I can operate at a range of about 1Km.

Did I do that all correctly?

Now - what about if I were to put an amp on the antenna? Specifically, how about the RF Arrays RWF111 (datasheet:

formatting link
I don't think this is possible as it looks like the nRF24L01+ doesn't have any pin to control the rx/tx switch in the amp, but bear with me anyways :) (or is there a way to do this?). So that part says it gives

19dBm for 802.11b. Since this Nordic part is even slower than 802.11b, I think I can assume that I'd get 19dBm or better. Now, I believe that means my output power would be 19dBm, instead of 0dBm before. Now, if the Nordic part had 5dBm output power, instead of 0dBm, I believe I would still have 19dBm output power as they say that the power, not power gain, is 19dBm. Is that right? It says in the transmit path the small signal gain is 28dB, so I guess that means that as long as the inputted transmitter signal had -9dBm (19dBm - 28dB) or greater power it'd output 19dBm. Otherwise it'd output the inputted signal power + 29dB. Is that all right? Lastly - the small signal gain in the receiver path is 14dB. So without an additional RX amp the Nordic part needed -94dBm. But with this extra amp the received signal could have as low as 108dBm (-94dBm - 14dB).

Sorry for writing such a book - I wanted to make an example that would illustrate a whole lot of things that I'm unsure about.

Thanks so much for anybody that can shed some light on this!

-Michael

Reply to
Michael
Loading thread data ...

Here's an example of how to do it for a wi-fi link:

Note that this makes quite a few simplifications and assumptions. It also results in the best case results, which are never the case.

formatting link

Nope. I'm not sure what you're doing, but the receive sensitivity is

-94dBm into 50 ohms. To convert that to watts, use: milliWatts = 10^(dBm/10) or use a web page based calculator:

-94 dBm = 4^-10 milliwatts or 400 pico watts. Get used to going between dBm and milliwatts, as it's a very calculation.

Nope. It's called "fade margin" or "System Operating Margin". It's (my definition) the amount of signal level you would normally operate expressed in dB over the point where your system craps out and belches unacceptable error levels. It directly correlates to the system reliability: SOM dB Reliability % Downtime per year 8 90 876 hours 18 99 88 hours 28 99.9 8.8 hours 38 99.99 53 minutes 48 99.999 5.3 minutes 58 99.9999 32 seconds

The fade margin is not really calculated. You need to determine how reliable your wireless link needs to be. From that, you can obtain a target value for fade margin. For most Wi-Fi applications, anything less than 10dB fade margin is going to barely work. For some semblance of reliability, aim for about 20dB fade margin. Note that a

10dB increase in fade margin means that your antennas will need to have 10dB more gain which is a HUGE change in antenna size.

Nope. You forgot just about everything that goes between the receiver and the transmitter. There are two antennas with gain (or loss). There are two coax cables to add some losses. There's the all important free space loss, which is a function of distance. See my example at:

or look at a simple link budget calculator:

formatting link

Nope. Let's try it my way. Start with a table of gains and losses starting at one end of the link and going to the other. Some assumptions:

  1. This is all at 2.4Ghz.
  2. There's coax cable between the devices and the antenna.
  3. The manufacturers actually meets their own specs.
  4. A receiver sensitivity of -94dBm is at 250Kbits/sec.

So, this is what we know so far. TX power 0 dBm TX coax loss -2 dB TX ant gain 9 dB Distance unknown RX ant gain 9 dB RX coax loss -2 dB RX sens -94 dBm Fade Margin 20 dB

Plug that into a link budget calculator and try various ranges until you get the desired 20 dB fade margin.

I get 0.15 miles or 240 meters. Again, note that this is the best case and makes quite a few assumptions, such as no Fresnel Zone issues, perfectly aligned antennas, no interference, no atmospheric issues (rain or fog), no folliage attenuation, and no other losses.

Fair choice. The amp is made to be driven at below about -5 dBm, so you should have enough drive power. If you go for full output power at +22dBm, it will probably go non-linear, belch errors, and never pass FCC Part 19 tests. My guess(tm) is that you can get about +19dBm out of it and meet specs.

Wrong. Pins 1 and 15 are TX enable and RX enable. These form the parts of the TX/RX switch.

Damn, I'm good today at guessing.

Adobe reader just crashed on me so I can't look at the original data sheet. There's a difference in modulation density between the various

802.11 speeds, and 802.11g OFDM modes. The result is that the average power output for OFDM is about -2dB less than for 802.11b modes. You'll see this on the various FCC type certification reports, where the power output for 802.11b and 802.11g are different.

Nowever, none of this applies to the Nordic chip, which runs pure FM with no amplitude components. No need to be particularly linear in the power amplfier (except to avoid harmonic generation). No issues with modulation density. My guess(tm) is that with the Nordic chip running FM only, you could get the full +22dBm out of the power amp.

You're describing the effects of overdriving a power amplifier. For a Class C amplifier, that's the way it's done. However, this is a Class A amplifier, that is intended to operate in the linear region of the device, which results in somewhere around -19dBm output. You can overdrive it by 3dB and get the full +22dBm output without any apparent violations of the printed specifications. No clue what it will do if you go over that. As a rule, use the minimum amount of power necessary to get the full output power. Any more is wasted.

Nope. That's not how it works. You can't just add more and more and more gain in front of a receiver to improve the sensitivity. Sensitivity is primarily determined by the NF (noise factor) of the input stage. In this case, it's the NF of the 14dB RF RX amplifier in the power amplifier. In general (a really bad assumption), the NF of the devices inside the amp will be the same as the NF of the devices inside the receiver. If that's true, then the receiver sensitivity will remain at -94dBm for 250Kbits/sec. The main thing the RX amp does is eliminate the losses inherent in the T/R switching and any interconnecting spagetti between the amp and the receiver.

Perahps you should first read a book on the topic instead of writing one?

Rule of thumb:

- To double the range, you need 4 times the xmit power (+6dB).

- Double the size of the antenna, and you get 3dB more gain.

- 6dB of system gain is good for twice the range. 12dB for 4 times the range.

- Double the data rate is good for a -6dB decrease in RX sensitivity.

Also, everyone lies on their data sheets. For 802.11b/g see:

--
Jeff Liebermann     jeffl@cruzio.com
150 Felker St #D    http://www.LearnByDestroying.com
 Click to see the full signature
Reply to
Jeff Liebermann

formatting link

(-94dBm is actually 398fW...)

The only thing that limits the ability of a receiver to detect a signal is the internal noise floor of the receiver which -- in a perfect world -- is k*T*b, where k is Boltzmann's constant, T is the temperature in Kelvin, and b is the bandwidth. The ratio between your incoming signal and the internal noise floor gets you the signal to noise ratio, so then the question is what SNR do you need to get a decent signal out? Nordic is telling you that with a -94dBm input signal, they still have enough SNR to produce some "reasonable" bit error rate (presumably they define what the BER is somewhere -- possibly

10^-3...) Note that if you average signals the noise tends to cancel out so you actually can build quite workable systems where the SNR is negative -- this is the basis of direct-sequence spread-spectrum systems, such as GPS, where the incoming signal power is something ridiculously low: Some really good receivers can keep a GPS signal locked with only -150dBm (1aW!) input power.

-94dBm is OK -- but not great -- for a 250Kbps system. Once the price of the chip is taken into consideration, though, it's entirely acceptable. (For comparison... something like SSB voice operation that amateur radio guys use is typically ~3kHz wide -- so on a good day you *might* get 3kbps over such a link -- and good receivers will produce decent audio down to perhaps -120dBm. If you want to spend $5 or less building a receiver, it's still no problem to do better than -100dBm. At the other extreme, for WiFi connections --

12Mbps -- you often need a signal in the -60dBm ballpark to get the full speed.)

Unfortunately I don't have the time right now to keep responding here, but I'll try to get back to it again if no one else responds.

---Joel

Reply to
Joel Koltner

formatting link

In all that one should not forget that a cheap vacuum cleaner or uncle Leroy's angle grinder can generate noise. Previously I didn't think that was possible at 2.45GHz (assuming that's what the OP wants, can't see his post). Once I had a receiver here I found that it was possible :-(

--
Regards, Joerg

http://www.analogconsultants.com/
 Click to see the full signature
Reply to
Joerg

Joel Koltner wrote: "

Not entirely true: In practice, the thermal noise floor is somewhat worse than KTB by some number resulting from losses and thermal noise in active devices in the receiver. Plus, the ambient noise from other emitters in the band, microwave ovens, WiFi cards etc will establish the external noise floor which may exceed the receiver thermal noise floor.

If this is a critical design, the designer should factor in some reserve transmitter power in the link to achieve the distance required. If that reserve is not practical, (regulatory restrictions) the link path will be less than expected.

--
Joe Leikhim K4SAT
"The RFI-EMI-GUY"©
 Click to see the full signature
Reply to
RFI-EMI-GUY

Hi Jeff - So that seems to be exactly how I was calculating it, except that I was using the wrong terms and didn't know that I needed a fade margin. By the way, I'm getting a slightly different number for that calculation:

tx power: +15DbM TX/RX coax loss: 4dB distance: 2 miles -> 110.225dB free space loss (36.6 + 20log(2.4E3) +

20log(2)) fade margin: 20dB RX sensitivity: -84dBm

RX sensitivity So, for example, let's say I want two Nordic Semiconductor nRF24L01+s

I accidentally thought it was 96dBm, not 94, hence the wrong number :)

Does this table only apply for 802.11 or can it be used for anything?

Not forgetting - just saving for later :)

here:

formatting link

I get 249.625 meters. Our numbers are close enough for government work, methinks :)

Sure - but what about on the Nordic part? What would drive the rx/tx enable lines on the RF amp? Doesn't it need to be told whether the Nordic part is sending or receiving

So when a power amplifier says that it is a 22dBm amplifier, it means that it's actually adding +22dBm to the strength of the signal, not bringing the signal to 22dBm?

So if the NF of the amp was 5dB worse than the NF of the receiver, the receiver sensitivity would go from -94dBm to -89dBm? Also, are the terms small signal gain and noise factor interchangeable in this context? Also, so the RX part of the amplifier does nothing to boost the signal? That doesn't seem right...

If there is a good book on the topic, I would be interested! Bear in mind I have a degree in EE, so I'm not a *complete* dolt. (just 80% of one)

Thanks so much for your help,

-Michael

Reply to
Michael

I'm not sure what he's trying to accomplish. Probably a class project or assignment.

Would you believe that an Ace Hardware wireless doorbell does a nice job of trashing my wi-fi? Push the button and my access point sometimes drops the connection.

List of other sources of interference:

I forgot to add the new 2.4GHz model airplane radio controls.

Also, vehicle trackers, game pads, PDA's, wireless TIVO, wireless video, and wireless PowerPoint(yech) projectors. I'll add these and more to my list later (when I find my password).

Noise floor? I never see the floor through the ever increasing fuzz growing out of the carpet.

The surest sign of success is pollution. I guess 2.4Ghz unlicenced services are successful.

--
Jeff Liebermann     jeffl@cruzio.com
150 Felker St #D    http://www.LearnByDestroying.com
 Click to see the full signature
Reply to
Jeff Liebermann

Nope - not in school. RF has just always been black magic to me so I'm slowly trying to fill in the gaps.

-Michael

Reply to
Michael

I'll decode your calculations later tonite. Gotta run and rescue a customer from the mistakes I made yesterday. I hate it when that happens.

Any RF link. Going from reliability to downtime is easy enough. Just multiply the % reliability by the number of minutes in a year. I forgot how going from fade margin to reliabily was calculated. I gotta dig that out of a book somewhere. Sorry.

Read a little on signifigant figures. Generating a 6 digit accurate number from two digit data, is rediculous. You can maintain such level of accuracy for intermediate calculations, but your final result will only be accurate to two digits. Try 250 meters.

Dunno. Acrobat refuses to resurrect itself until I reboot. I'm in a rush and don't to wait forever for the reboot. I'll look at it later. If there's no tx/rx output from the Nordic chip, you get to build an RF detector, that will sense input RF power, and use that to key the power amplifier.

The power amp does NOT add the input power to the output. Amplification converts the input power to output power. The output power is +22dBm. The input power is whatever drive level the data sheet says is required to get +22dBm output.

No. The equation includes the gain of the amplifier. The input device amplifies both the signal AND the noise. See calculations for cascading RF amplifier stages:

RF is magic. A sheepskin doesn't help with RF.

I think you need something basic: "Complete Wireless Design" by Cotter W. Sayre. A bit old (2001), but that's what fell off the bookshelf.

--
Jeff Liebermann     jeffl@cruzio.com
150 Felker St #D    http://www.LearnByDestroying.com
 Click to see the full signature
Reply to
Jeff Liebermann

Mine's the other way around. Maybe because it's from HD :-)

When we had a military exercise in the areas sometimes when the fighter jets were in the visible range our doorbell would go off. Probably their radars. Other than that the receiver in that thing is remarkably selective and the first set of batteries is almost in its 10th year. Got to check for leakage soon ...

Same here :-(

Pretty soon I'll have to make another decision for a client, or rather suggestion, on where to go. I am now leaning towards 433MHz, seems the only other band with some range that's at least somewhat internationally allowed. Luckily we need less than 300bps on this one.

--
Regards, Joerg

http://www.analogconsultants.com/
 Click to see the full signature
Reply to
Joerg

Let me guess... no "whiz bang" do-it-all chipset, but rather a discrete design for the receiver? :-)

We might be doing some analog ACSB transceivers in the near future... should be fun... I of course have your number for when it doesn't work. :-)

Reply to
Joel Koltner

I'd rather do it with chips this time since I need transceivers and this client likes stuff simple, with the least amount of parts.

--
Regards, Joerg

http://www.analogconsultants.com/
 Click to see the full signature
Reply to
Joerg

My idea of fun is driving through the local restaurant parking lot while yacking on my VHF/UHF ham radio. 50 watts does wonders to automobile burglar alarms. I can usually set off at least one. My record is getting 5 alarms going at the same time. (Hint: I get bored easily).

I deal with quite a few old computahs. The ones with the batteries soldered into the board frequently leak. I didn't know the NaOH will rot copper traces, but apparently it does. I've had a few fairly nice but old boards, that were ruined that way. Check the battery, or do a pre-emptive replacement.

That's the problem. Everyone is using 418/433MHz. All the fancy automobile key fobs work on those frequencies. Same with wireless burglar alarms remotes, air conditioner remote controls, etc. It's really crowed, but is fairly free of high power noise sources. The only way I've gotten reliable communications is to use frequency diversity, where I send the same junk on two frequencies, and use CRC's or error correcting codes to put the bits back together. If you rely on a single frequency, chances are good that some RF source is going to trash it for a while. The odds are better with 2 frequencies.

--
# Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060
# 831-336-2558            jeffl@comix.santa-cruz.ca.us
 Click to see the full signature
Reply to
Jeff Liebermann

Do yourself a huge favour and install foxit reader 2.0. Fast, free, does not crash. Opens in a split second versus several seconds for acrobat.

Mark

Reply to
TheM

Someone in Europe stalled out the engine of the car next to him. But IMHO automotive folks have often come up with, ahem, not so RF-proof designs.

That's what I plan to do and the main reason why this one won't be discrete but will have to be a chip. It'll be frequency hopping.

--
Regards, Joerg

http://www.analogconsultants.com/
 Click to see the full signature
Reply to
Joerg

[snip]

In the '60's Ford was testing electronic ignition systems on police cars in Detroit. The bad guys quickly figured out that keying a CB radio near them would stall them out ;-)

Add metal case.

...Jim Thompson

--
| James E.Thompson, P.E.                           |    mens     |
| Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson

So they didn't test their design? Ouch!

Doesn't always help. I have seen some rather horrible layouts in automotive. Spaghetti-style. Almost a guarantee that any RF riding on a connected wire would cause trouble inside.

--
Regards, Joerg

http://www.analogconsultants.com/
 Click to see the full signature
Reply to
Joerg

On this one it did... the RF was directly interfering with the chip, in spite of adding-on decoupling.

Yep. My beloved Q45's GPS sometimes loses north when near some grocery store automatic door openers ;-)

...Jim Thompson

--
| James E.Thompson, P.E.                           |    mens     |
| Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson

formatting link

formatting link

Rather a bit of reading, but very good. I now know who to call if need help in your demenses.

Reply to
JosephKK

Well.... Adobe Acrobat 8.1.2 was doing just fine until about a month ago, when I installed something that broke Acrobat. I just reinstalled Acrobat and things seem to have stabilized. I think my home computah will live.

However, thanks for the reminder about Foxit Reader. I've used it in the distant past and found problems. Time to try again. I just installed Foxit 2.3. Seem to work MUCH faster than Acrobat 8. It also opened most of my problem documents, but balked when I fed it a PDF with foreign languages and obscure fonts. Time to download some add-ons:

--
Jeff Liebermann     jeffl@cruzio.com
150 Felker St #D    http://www.LearnByDestroying.com
 Click to see the full signature
Reply to
Jeff Liebermann

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.