I don't think it helps to run VCOs off of
noisy power supplies if you are trying
to minimize phase noise.
Zack KH6CP/1
>Zack KH6CP/1
I use a large capacitor (I.E. 470uf across 100n) in series with some
resistance and an ordinary regulator (78L08). This provides a noise free
supply in my designs.
From Steve Drury. Email st...@hamlab.demon.co.uk
G6ALU Milton Keynes England
: I don't think it helps to run VCOs off of
: noisy power supplies if you are trying
: to minimize phase noise.
Yep, it's a problem. In fact, it can even be a
problem for power or references for audio amplifiers,
if you are trying to get to really low noise performance.
It's a problem with voltage references as well as for
regulators. What I've done with some success is filter
the output of a voltage reference with an RC filter and
feed this to an op amp follower; the op amp might include
an emitter follower buffer in its feedback loop, in order
to get enough output current. You might also get by with
an LC filter on the output of a regulator, but I've never
had the space for large enough L's to do much good in the
10s of Hz range. If you use the op-amp follower approach,
use a low-noise op amp, and make sure its power supplies
are reasonably clean. If you use a single pole RC as the
filter between the reference and op amp, you can get it
to settle faster if you put a diode across the R. If you
aren't worried about filtering down to the 1Hz range, this
might not be a problem but it can be if the RC time constant
is several seconds. Of course, you can make the filter
a 2-pole active one, feeding the op amp output back
appropriately, and that helps a lot in getting good filtering
but reasonable turn-on times.
--
Cheers,
Tom
to...@lsid.hp.com
On 29 Dec 1995, Zack Lau wrote:
> Is there a better chip to use than the
> ancient 723 regulator chip? The National
> Databook lists its output noise as
> typically 2.5 uV rms between 100Hz and
> 10 kHz. The 78L05 and LM317 are much
> worse (40 uV, 0.003 %)
>
> I don't think it helps to run VCOs off of
> noisy power supplies if you are trying
> to minimize phase noise.
>
> Zack KH6CP/1
>
>
>
>
>
The lowest noise commonly available regulator is the TL431.
If you want to go lower than this, you have to use a 1N821 reference
diode.
Rick Karlquist N6RK
rka...@scd.hp.com
: I don't think it helps to run VCOs off of
: noisy power supplies if you are trying
: to minimize phase noise.
: Zack KH6CP/1
Zack,
I'm afraid not, at least for the moment. When I was a microwave
VCO design engineer at Omni-Spectra years ago, I was part of a long-term
test in which we evaluated SSB phase noise resulting from the use of IC
regulators. We found, after almost a year of on-and-off effort, that the
quietest regulator for a positive supply was the 723 with an NPN pass de-
vice, and for a negative supply the LM104 with a PNP pass device. I have
yet to hear anything different from those folks.
--
Regards,
Chris
------------------------------------------------------------------------------
Chris Trask / N7ZWY Circuit Design for the RF Impaired
ATG Design Services __ __ ____ ___ ___ ____
ctr...@primenet.com _~_ /__)/__) / / / / /_ /\ / /_ /
(@ @) / / \ / / / / /__ / \/ /___ /
----------------------ooO~(_)~Ooo---------------------------------------------
ooops, that should say electronic packaging!
RL
There's an ultra low-noise and stable
regulator design featured on page 375 of
"The Art of Electronics" by Horowitz and
Hill (2nd edition). It features the
following specs:
Noise and ripple < 1 uV
Temp coeff < 1 ppm/degree C
Output Z < 1 uohm
Drift < 1 ppm/day
At these levels, you'll need to pay
particular attention to circuit layout
and electronic package (crosstalk, EMC,
etc.)
Russ LeBlanc
> Is there a better chip to use than the
> ancient 723 regulator chip? The National
> Databook lists its output noise as
> typically 2.5 uV rms between 100Hz and
> 10 kHz. The 78L05 and LM317 are much
> worse (40 uV, 0.003 %)
>
> I don't think it helps to run VCOs off of
> noisy power supplies if you are trying
> to minimize phase noise.
Colin Horrabin, G3SBI, designed a low-noise 8V-9V regulator for some
experiments he's been doing with with low-phase noise oscillators. The
circuit is described on p 55 of the RSGB Radio Communication mag., July
1994. It uses a couple of VN10KM FETs and a BC184L. Zack should be able
to find a copy at the ARRL. Colin's oscillator circuits are also
described, which are well worth looking at; they use stripline
resonators.
Leon
--
Leon Heller, G1HSM | "Do not adjust your mind, there is
E-mail le...@lfheller.demon.co.uk | a fault in reality": on a wall
Phone: +44 (0)1734 266679 | many years ago in Oxford.
The Fairchild data sheet for uA723C lists 2.5 uVrms with Cref=5 uF in the
reference divider and 20 uVrms without Cref. It seems that the real culprit
is the voltage reference and not the error amplifier and output stage.
However, I could not find at what output voltage these noise voltages are
measured. If the main sourc of noise is the voltage reference, you get
much prettier figures at 2 V output than at 36 V output.
Tom Bruhns (to...@lsid.hp.com) wrote:
> What I've done with some success is filter
> the output of a voltage reference with an RC filter and
> feed this to an op amp follower; the op amp might include
> an emitter follower buffer in its feedback loop, in order
> to get enough output current.
The 723 contains a separate voltage reference and a differential amplifier
and an emitter follower in a smaller package count. Filtering can be
inserted between the reference voltage source and the differential amplifier.
> Ifou use te op-amp follower approach,
> use a low-noise op amp, and make sure its power supplies
> are reasonably clean.
Agreed. It is no use to clean up the refeence and then connect a noisy
op-amp (such as 741 or TL081) as the regulator, since the noise from
the op-amp alone can be greater than 2.5 uV at 10 kHz BW. Fortunately,
op-amps with noise densities below 3 nV/sqrt(Hz) are available at
reasonable prices, so you could get 0.3 uV at 10 kHz BW if you had a
clean reference.
Paul OH3LWR
--
Phone : +358-31-213 3657 Mail: Hameenpuisto 42 A 26
Internet: Paul.K...@cc.tut.fi FIN-33200 TAMPERE
Telex : 58-100 1825 (ATTN: Keinanen Paul) FINLAND
X.400 : G=Paul S=Keinanen O=Kotiposti A=ELISA C=FI
The TI and National Semi databooks both list the 723 as having a
typical output noise voltage of 2.5uV RMS in a test configuration
given as Figure 1 in both datasheets. The value of the reference
voltage divider resistors is given Table 1 in both datasheets;
the output voltage is 5.0V.
Interestingly, while both datasheets list 2.5uV RMS as the typical
noise voltage with a 5uF capacitor at the error amplifier input,
the National sheet lists the typical noise voltage as 86uV RMS
without one, and the TI sheet lists 20 uV RMS. While the Motorola
MC1723 sheet lists the same conditions and values as the TI sheet,
it omits the table specifying the resistor values used to achieve
the 5.0V output.
--
* Dana H. Myers KK6JQ, DoD#: j | Views expressed here are *
* (310) 348-6043 | mine and do not necessarily *
* Dana....@West.Sun.Com | reflect those of my employer *
INPUT -+------- ------ OUTPUT
| C\ /E
| ------- MPSA18 or 2N5089
| |B
+--/\/\/--+
6.8K |+
--- 10uF
--- Tantalum
|
---
///
It has worked well on 3 VCOs so far.
73,
John, KD6OZH @ KO-23