Can this LDMOS be desinged to 70%+ efficiency at 887MHz?

[tony_lth]

Hi, ALL,
I need to design a PA at 887M with GSM signal.
1. I found a LDMOS which said it can have 77% efficiency at 870M, but my freq is 887M, is it possible to use the 870M match circuit for 887M TX? Attchment is the NXP LDMOS PA datasheet and 870M s-params and 887M s-params (In Amp-Angle format,). I compared 870M S-params and 887M S-params, they are almost same, only about 2% difference. I guess it can be ignored, but I don't know how this affect efficency. Any suggestions?
2. Since this LDMOS is a 16W PA, but I only need 10W, how this will affect efficiency?
3. Finally, I want to use LDMOS to design a PA with 70%+ efficiency at 887M, any comments?
Many thanks.
Cheers,
Tony Liu

PS:
Since *.s6p file extension is not accepted at this forum, so I change it to *.S6p.txt.

 

[vfone]

From the plots we can see that the gain power is dropping at 870MHz, but for sure you can get the desired power of 10W.
But if your main target is high efficiency, using a 16W transistor to deliver 10W you will get significant drop in PAE. Power added efficiency is maximum near the P1dB of the amplifier.
You will lose efficiency twice. Once because you use the transistor at higher frequency (see fig.11 in the datasheet), and second because you want to back-off the output power from 16W to 10W.

 

[tony_lth]

Hi, Vfone,
yes。
But I can't find a LDMOS PA with 10W P1dB at 887M for GSM signal, who can recommend me one?
Many thanks.
Cheers,
Tony Liu

 

[tony_lth]

Is it possible to combine two LDMOS together? I think it is ok for 0.2dB insertion loss, but the other GSM params, such as EVM, phase, etc., will be ok or not?

 

[mtwieg]

You can lower the P1dB by reducing the bias voltage. The output matching may need to be slightly adjusted, but you should be able to get 70% at 10W.

 

[vfone]

Unfortunately is not such easy. Reducing the PA bias it will damage the linearity (at all output power levels), and the base station PA's for GSM/EDGE requires prety good linearity to meet the specs.

 

[mtwieg]

I'm no expert in GSM, how good is "pretty good?" I see 4dB of gain variation in figure 13...

For combining LDMOS, you could try the old quadrature hybrid combiner trick.

I'm actually in a similar situation where I'm trying to find an LDMOS RFPA for 5-10W at 100-150MHz, but can't find anything specified for that specific range...

 

[tony_lth]

ok, thanks, mtwieg and vfone.
I heard by that GSM/PA doesn't requirement such good linearity, because it works in saturated region.
According to my simulation, OIM3 about 20dB is enough, but that seems easy to meet when PA works in OP1dB.

 

[vfone]

Standard GSM (GMSK modulation) PA can work near saturation, but GSM/EDGE (8PSK modulation) which use the same PA, needs a very linear PA.
From my experience I can say that not even PAs for standard GSM (GMSK) cannot work deep in saturation, because the system will never pass the burst mask and the switching spectrum.

 

[tony_lth]

Thanks, vfone.
Edge is really difficult, but I only use GSM/GPRS, which seems need 24dB ORR3 from my previous simulation of some successful design.
Cheers,
Tony Liu