hi,
now I am designing a mixer in TX for GSM, and I want to know how to get the spec(linearity and gain) from the TX specification by GSM 05.05.
actually, to calculate the gain is an easier job, since the total gain is defined in the spec of TX (GSM), but there seems to be little information on the TX linearity as the output signals are constant envelope theoretically.
Well, for GSM there really isn't a linearity spec on the transmit mixer in terms of IP3/IP2/P1dB. This is because the input to the transmit mixer is a clean one from the modulator, so you shouldn't see much in the way of interfering tones.
I would worry about linearity leading to spectral mask issues, but you'd have to simulate that with a GSM signal. This is slightly different because in this case, you are dealing with the modulated signal being distorted by the nonlinearities in the mixer. You need to start with a GSM signal to figure this one out.
yes, the linearity of TX will pose an impact on the spectral mask.
and I wonder whether there is any mathematical relationship between the mask and the linearity. or we can only try to define specs of the TX components by simulation?
Just to clarify, you are talking about a mixer that upconverts a modulated carrier from an Transmit IF to a carrier frequency right?
In that case, you should probably be more worried about LO phase noise and wideband noise floors. Linearity spec's on a mixer I would use simulaiton to look at.
If you are talking about a transmit mixer in an IQ modulator, then you've got IQmismatch things to worry about, but linearity again isn't really an issue. Since you have control of the levels, must make sure that the Mixer is spec'ed to operate in a linear range.
Well, I got it. thank you very much.
I will simulate the mixer in a system level with IQ mismatch.
besides, I found the measurement of noise floor of VCO(PLL) phase noise at TX output is much too high. the curve turns flat when frequency is higher than 2MHz with much higher phase noise than expected.
how did this happed?
Yup, Out of band noise floor will probably be more of a problem. Somewhere in the GSM spec is a specification for TX output noise @ 20 MHz. Dig that up and think about what that spec means to noise floor's through the system.
yeah.
20MHz is important in GSM spec. as it is the frequecy space between TX and RX. higher noise floor may affect sensitivity of RX.
I have tested vco individually, and it appeared to be not so bad.
I guess bad out-band noise results from the poor performance of on-chip loop filter.
well, I thought loop filter supressed outband noise in the loop,
and vco was just responsible for it inband noise.
is that right?
Sorry to say that I am not quite familiar with PLL.
and hope that you will clarify it for me.
thank you very much
Nope, you've got it backwards. Inside the loop filter bandwidth, the loop filter suppresses the VCO phase noise, and the Reference oscillator/Divider/Jitter performance dominates. Outside the loop filter bandwidth, the VCO performance dominates.