|
 |
regulator oscillating problem - RF Cafe Forums
|
pigger
Post subject: regulator oscillating problem
Unread postPosted:
Sat Jun 19, 2004 1:32 pm
our PLL (2GHz) is supplied by a voltage
output form a regulator, but the phase noise of this PLL is huge when
it is tied to the regulator. It is working very well when there is no
regulator. So we can guess it is the regulator oscillating that makes
the PLL working inproperly. We followed the regulator data sheet for
its peripheral circuit, but it did not work out.
can anybody give
me some idea on how to solve this prblem?
Top
Guest
Post subject: Regulator output
Unread postPosted: Sun
Jun 20, 2004 9:56 am
Did you check the regulator output on oscilloscope?
We have a couple of top notch PLL designers here, I will see if they
can recommend a regulator or remedy. Also, do you know what the pulling
or puhing spec to the VCO is? ( I forget which it is pulling or pushing
for Vcc sensitivity).
Top
Itay
Post subject:
Unread postPosted: Mon Jun 21, 2004 11:01 am
Have you put enough
decoupling capacitors on the regulator? Usually you will need to put
a super filter circuit on the supply of the VCO - this is extremely
important. Super filter circuit is based on an NPN transistor, which
is bypassed with decoupling capacitors at the collector and emitter
and its output at the emitter is connected to the Vcc pin of the VCO.
It drops the supply voltage from the regulator (Which needs to provide
a highr voltage than the supply voltage of the VCO) to the voltage level
required by the VCO.
In addition you will need at least one tantalum
capacitor with low ESR as part of your decoupling capacitors battery.
A value of 10uF is adequate for this purpose. You have to choose a regulator
with low output noise, for example: MICREL's MIC5205 is a good choice,
this device has a low output noise and by connecting an external capacitor
you can reduce the output noise even lower, this is excellent for low
current applications as VCO's/PLL's.
http://www.micrel.com/_PDF/mic5205.pdf
Good luck,
Itay
Top
Guest
Post subject:
Re: Regulator output
Unread postPosted: Mon Jun 21, 2004 4:12 pm
Thanks for replying. I checked the VCO spec, pushing figure
is 1MHz/V
Anonymous wrote:
Did you check the regulator output
on oscilloscope? We have a couple of top notch PLL designers here, I
will see if they can recommend a regulator or remedy. Also, do you know
what the pulling or puhing spec to the VCO is? ( I forget which it is
pulling or pushing for Vcc sensitivity).
Top
pigger
Post subject:
Unread postPosted: Mon Jun 21, 2004 4:15
pm
we put 2 tantalum caps, one is 10uF, the other is 0.1uF.
and there are 3 decouping caps for the VCO vcc pin, all are tantalum,
(1u, 0.01u and 10pF). But nothing helps
Itay wrote:
Have
you put enough decoupling capacitors on the regulator? Usually you will
need to put a super filter circuit on the supply of the VCO - this is
extremely important. Super filter circuit is based on an NPN transistor,
which is bypassed with decoupling capacitors at the collector and emitter
and its output at the emitter is connected to the Vcc pin of the VCO.
It drops the supply voltage from the regulator (Which needs to provide
a highr voltage than the supply voltage of the VCO) to the voltage level
required by the VCO.
In addition you will need at least one tantalum
capacitor with low ESR as part of your decoupling capacitors battery.
A value of 10uF is adequate for this purpose. You have to choose a regulator
with low output noise, for example: MICREL's MIC5205 is a good choice,
this device has a low output noise and by connecting an external capacitor
you can reduce the output noise even lower, this is excellent for low
current applications as VCO's/PLL's.
http://www.micrel.com/_PDF/mic5205.pdf
Good luck,
Itay
Top
Alberto
Post subject:
Unread postPosted: Tue Jun 22, 2004 8:23 am
Are the regulator
and the decoupling capacitors close enought to the VCO ? How did you
check the synthesizer without the regulator ? (I mean if did you use
the same PCB track to fed the VCO with or without LDO).
I think
you also have to check if your GND is good and well connected between
regulator and VCO (f.i. using a filled area with many vias).
Just another note: some LDO requires not too low ESR output capacitors,
otherwise they can become unstable: check the data-sheet of your regulator.
Top
pigger
Post subject:
Unread postPosted:
Tue Jun 22, 2004 11:48 am
we have decouping caps close to VCO
and regulator, but they are not quite close.
we have a test point
next to the regulator, so we just lifted the output pin of the regulator
and hooked up the test point to a power supply.
I checked the datasheet,
they ask for low ESR caps.
Alberto wrote:
Are the regulator
and the decoupling capacitors close enought to the VCO ? How did you
check the synthesizer without the regulator ? (I mean if did you use
the same PCB track to fed the VCO with or without LDO).
I think
you also have to check if your GND is good and well connected between
regulator and VCO (f.i. using a filled area with many vias).
Just another note: some LDO requires not too low ESR output capacitors,
otherwise they can become unstable: check the data-sheet of your regulator.
Top
Guest
Post subject:
Unread postPosted:
Tue Jun 22, 2004 11:38 pm
Hi,
Is your circuit already
exists on a final board layout, or is it still a prototype? If it still
can be modified, I suggest you will add a super filter to the VCO supply,
as I suggested. You can also put the VCO and PLL into an enclosure of
shielded metal can, which will be soldered to the GND plane of your
board. You will surround the VCO, PLL and the loop filter with a stripe
of GND stitched with via holes and to this stripe you will solder the
shiedling can. This will provide a high isolation to the sensitive circuitry
from the external noisy world. This is a good way to protect a sensitive
circuit.
Good luck,
Itay
Top
guest
Post subject:
Unread postPosted: Mon Jul 05, 2004 4:16 pm
hope you dont use LM317? Measurement-=> knowledge. If everything
is perfect, your system may oscillate.
Top
pigger
Post subject:
Unread postPosted: Sat Jul 10, 2004 2:00 am
no, we don't use national semi's stuff
guest wrote:
hope
you dont use LM317? Measurement-=> knowledge. If everything is perfect,
your system may oscillate.
Posted
11/12/2012
|
|
|
|
|
