Testing the receiver of transceiver chip.

[dkumar]

Hi! Everybody!!

I have to test the receiver chin of ZigBee based transceiver.The input being LNA inputs and output the form ADC.I also have a output pin in the chip form the filter which is placed after the down conversion mixer.

Please let me know waht are the specifications that i should test for.following is the list that i found:
a)Gain testing.
b)Receiver linearity
c)Receiver sensitivity
But i still feel that there are few more things that i may test.For example I/Q mismatch etc.

I request you to please tell me what are the standard testing parameters for a receiver chain and how to go about testing that.
I have all the instruments like function generator,spectrum analyzer,logic analyzer and VNA.Also if i need anything else please do tell them.

Looking for an early response and thanking in advance..

-dk

 

[Mazz]

Hey it seem interesting: it can be a lot of work, indeed.

Let's start from the beginning:
usually the test is driven from a specification that have to be met. Do you have such a document or not?

If so, you have to measure all the parameters in the datasheet.

Mazz

 

[dkumar]

Hi! Mazz

Yes its interesting job...
Actually i just have a transceiver chip which is based on ZigBee protocol.Its a zero-IF architecture.

The chip was designed by one of the previous student here at my university and i have read the performance they designed for.So as such there is no document but my job is to make a data sheet level document by testing it.

I have already made the PCB board for the chip and also have tested the frequency synthesizer in the chip.Not i want to test the receiver chin where the specs i could find, i listed above.Problem is since i am going to pass analog data through the LNA input( may be form a function generator) and after getting some analog signal processing in side the chip i am getting the digital output form the ADC at the end of chain.

How will i ensure that, that is the expected output and then i can check on the other specifications...

I hope i am making my self clear.

 

[Mazz]

I think you can make a couple of things.

First, you surely need an RF signal generator that you haven't mentioned.
Applying an in band CW (slightly offset respect to LO) tone to the LNA, you can acquire the ADC output (a logic analyzer is OK for that) and do the math to see the output spectrum, so you can calculate the gain, linearity, passband and so on.

Or you can test the analog separately from ADC, having access to the baseband filter output.

For NF meas you will need a calibrated noise source.

Mazz

 

[dkumar]

"First, you surely need an RF signal generator that you haven't mentioned."

Yes i arranged that last day.

"Applying an in band CW (slightly offset respect to LO) tone to the LNA, "
I didnt get what you meant by CW tone. and why should it have offset form LO frequency? I guess the zero-IF architecture will have the LO frequency same as that of incoming signal (which is modulated)

"you can acquire the ADC output (a logic analyzer is OK for that)"

Yes this looks fine..

" and do the math to see the output spectrum, so you can calculate the gain, linearity, passband and so on."

Could you please elaborate on math? i have not done much of communication based course. What math might be needed to find the specs form the so obtained logic levels?


"Or you can test the analog separately from ADC, having access to the baseband filter output."

test analog separately means??

""For NF meas you will need a calibrated noise source. "
calibrate what??

 

[Mazz]

OK, one by one:

""Applying an in band CW (slightly offset respect to LO) tone to the LNA, "
I didnt get what you meant by CW tone. and why should it have offset form LO frequency? I guess the zero-IF architecture will have the LO frequency same as that of incoming signal (which is modulated)"
CW: continous wave, a simple sine wave. To measure the small signal gain usually it is used a CW, and you need to apply it in the freq interval LO+\-BW/2 to measure it in baseband circuit at 0-BW/2.

"Could you please elaborate on math? i have not done much of communication based course. What math might be needed to find the specs form the so obtained logic levels? "
The CW signal in 0-BW/2 bandwidth goes through the ADC, is aquired by logic analyzer, then it has to be converted into a spectrum, to measure its spectral characteristics. I don't have large experience in that, but you can look for many reference in ADC literature.

"test analog separately means?? "
If you pick the analog signal before the ADC (from the test point you mentioned), you measure only the analog circuits.

"calibrate what??"
NF measurement are usually based on Hot/Cold (or so called Y-factor) method.
You need a noise diode that generates a well known noise levels, so it is "calibrated". See for example cp.literature.agilent.com/litweb/pdf/5952-3706E.pdf

Mazz

 

[dkumar]

Hi mazz..

I did the most things...but 1 thing if you could tell me....

I have the output form the filter from where i can take the analog output. Say if i input a sine wave of LO+Δ then i will get some output at that point which i can wither see on scope or in spectrum analyzer.But how will find the SNR output?? can i also find the sensitivity and linearity form form this method??

Looking for your reply....

 

[Mazz]

SNR can be evaluated measuring Gain and noise Figure separately.
Sensitivity depends on min SNR requested at demodulator. Min SNR depends on demodulator itself and on the choosen modulation scheme, so you can calculate the sensitivity from min SNR requested (and sytem bandwidth, of course).

For linearity, you can measure IIP3 injecting 2 RF signals, both a the same power and at freq LO+delta1 and LO+delta2. You will find at the output only delta1 and delta2 tones and their intermodulation products, so you will be able to measure receiver IP3. Take care to use freqs as the delta1/2 and intermods will fall into you bandwidth.

I hope it can help.
Mazz

 

[dkumar]

Thanks a lot for those help.
Also,
Do i have to calculate the NF manually or is there any instrument??

for sensitivity(dBm) = -174 + B(dB) + NF + SNR has to be used??

since there is only one input, LNA, how to inject two RF signal ??And what exactly you meant when you said to take care of frequency and delta1/2 will be in my bandwidth??

looking for your reply

-dk

 

[RFCMOS]

Hi,

You can use an RF combiner to inject the two signals.
You have to take care that the third order intermodulation products will fall in band. In other words, that they will not be filter out by the filters present in your receive chain.

 

[Mazz]

yeah
agree with RFCMOS.

Mazz

 

[dkumar]

HI!

Please let me know if i have to calculate the NF manually or i can use some instrument.I need that for SNR calculation.

Looking for help

Thanks.

 

[Mazz]

You can do both.
Being a ZIF receiver, you should use a Noise source with band (and ENR) starting from low freq (usually Noise Sources low band is 10 MHz), in order to calibrate your instrument also in the sub-1MHz band.
For further details, please refer to:
**broken link removed**

I hope it can help.
Mazz