EVM and ACPR issue of Power amplifier

[lyxiao653]

Hi Guys,

I designed a power amplifier for both GSM and EDGE application using GaAs HBT. The saturated output power is around 31dBm, which is good. When I tested the linearity performance under linear mode at 26.5dBm, which is 4dB backoff from the saturated power under GSM mode, I found the ACPR 200KHz is Ok and the ACPR 400KHz is just so so, but the EVM is ridiculously poor, greater than 20%. I understood that ACPR and EVM are two standards that describe the linearity of the PA, but I donot get how these two things have such big difference in one PA. Is there a way to find out what is happening?

thanks

 

[:Dark_Force]

Hi

How do you measure these parameters? By means of a communication tester that offers such one-click-measurements?
Generally said, the ACPR is not necessarily connected to EVM since ACPR represents the spectral purity of the modulated signal and EVM is a figure for in-band signal-quality or -linearity. Your system may very well have a good EVM but a very poor ACPR if your modulation signal (not carrier) is too broadband. Also, a good ACPR but bad EVM is an indication for a band-limited modulation signal. Therefore, I wouldn't blame it all on the PA but also look for the preceding stages in your system, whatever they look like!

greets...

 

[lyxiao653]

Hi

How do you measure these parameters? By means of a communication tester that offers such one-click-measurements?
Generally said, the ACPR is not necessarily connected to EVM since ACPR represents the spectral purity of the modulated signal and EVM is a figure for in-band signal-quality or -linearity. Your system may very well have a good EVM but a very poor ACPR if your modulation signal (not carrier) is too broadband. Also, a good ACPR but bad EVM is an indication for a band-limited modulation signal. Therefore, I wouldn't blame it all on the PA but also look for the preceding stages in your system, whatever they look like!

greets...

Hi Dark_force,

Thank you very much for your reply.

Right, I measured the EVM and ACPR of the PA on the Phone board with a communication tester. Besides, the system is tested under EGPRS mode, which uses 8PSK modulation and has 200KHz modulation bandwidth.
Yes, I understood that the ACPR is not quite related to the EVM. The fact that confused me was that the PA designed by the other team performs good on the same phone board. I mean their PA has the same ACPR performance as mine, but the EVM of their PA is much better than mine.
I tried to tune the OMN and increase the bias current of my PA, then I found the ACPR would be improved largely, but EVM stays the same.
Do you have further suggestions?

thanks,

 

[:Dark_Force]

Can you post a schematic of your PA?
Which Frequency does it work on? 800/900MHz or also in the higher bands? If so, is the transmission quality in the other bands better/worse?

 

[lyxiao653]

Schematic is attached. It is a simple 3stage cascade PA.

The frequency of the PA is from 1710MHz to 1910MHz. I swept the frequency when measuring the EVM and ACPR and I could not see any difference between channels. The EVM is really bad for all the frequencies.

 

[:Dark_Force]

Come on, that's it? No values or part names? And for god's sake, it's 2014. The times when 640x480 were state of the art are definitely over!

 

[lyxiao653]

My bad. Is this OK?