Post subject: unbanlance of ACPR Posted: Wed Nov 16,
2005 11:15 pm
Joined: Tue May 24, 2005
When testing the ACPR of CDMA PA, the difference
between upper frequency and lower frequency was a little big. For
example, the amplitude of +750KHz was -46.6dBc, while that of -750KHz
But normally, these two value were almost same.
Could you tell me what would cause this and how to avoid? Many thanks.
Post subject: Posted: Wed Nov 30,
2005 3:53 am
This is a common problem when designing PAs for
relatively wide band waveforms (>1MHz BW). It's called a memory
effect, and is basically a voltage ripple on your power supply feed
(drain for LDMOS). Can be remedied with larger values of capacitors
(10uF to 220uF) on supply feed, or de-coupling network. Also, try
minimizing length of all DC lines or cables.
Post subject: Posted: Thu Dec 01, 2005 11:30 pm
Are you sure that this is called memory effect?
If I recall well this is just called signal envelope decoupling.
I tought memory effect involved instrinsinct semiconductor issues
and not only surrounding circuitry consideration.
Post subject: Posted: Sat Dec 03, 2005
Joined: Mon Jun 27, 2005 2:02
VIP is right
in he post. It is a memory- effect related. Memory effect has to
do with working on a wideband frequency range under which the amplifier
behaves differently. The amplifier behaves somewhat as Filter -
has different AM-AM & AM-PM behavior over the entire band and
this is shown as different linearity (ACPR) performances. VIP is
also right in his suggestion for remedy to this problem. The bypassing
network should contain several values of capacitors and often ferrite
beads and resistors - it is always useful and helpful to consult
the data sheet of the specigfic device.