From: Larry Smith ([email protected])
Date: Fri Sep 22 2000 - 09:33:37 PDT
Chris - Robert Weber and Steve Weir have already posted good responses
concerning the stability of the feedback loop and the the fact that you
can leave in some high frequency, low ESR capacitors without major
problems. Your solution of a series resistor for the ceramic cap will
work. A further comments:
A 10 uF capacitor and 1 uF capacitor have slopes on a bode plot that are
20 dB apart. A 0.1 uF cap is 20 dB above that. The lower valued
capacitors affect the 'transfer function' of the network at a much higher
frequency than the 10 uF capacitor. The 'open loop' gain and phase
should have enough margin so that the high frequency capacitors do not
cause any harm to the stability of the loop.
I like to establish a target impedance (resistive) and then select
capacitors such that the parallel combination of all ESR's at each
capacitor's individual resonant frequency forms a fairly flat impedance
over a broad frequency range. This avoids stability problems in the
regulation loop as well as nasty surprises (anti resonances) when
inductive components bump into capacitive components on an impedance vs
frequency curve.
regards,
Larry Smith,
Sun Microsystems
> From: Chris Cheng <[email protected]>
> To: [email protected]
> Subject: RE: [SI-LIST] : ESR and Q factor
> Date: Thu, 21 Sep 2000 16:48:39 -0700
> MIME-Version: 1.0
>
> larry,
> here's the problem, i can certainly put series resistor in front
> of the bulk caps. however, i cannot control the q on the high
> speed decoupling caps which are needed for medium range decoupling.
> lt1086 is such a common part its hard for me to imagine this is
> not a problem people has seen before.
> chris
>
> -----Original Message-----
> From: Larry Smith [mailto:[email protected]]
> Sent: Thursday, September 21, 2000 4:17 PM
> To: [email protected]; [email protected]
> Subject: Re: [SI-LIST] : ESR and Q factor
>
>
> Chris - at frequencies near the bandwidth of the regulator loop, the
> open loop gain is usually going down at 20 dB per decade on a bode plot
> (assuming a single pole roll off). Therefor, the output impedance of
> the regulator (with no capacitor) is going up at 20 dB/decade. This
> looks very much like an inductor.
>
> If you put a capacitor in parallel with the (inductive) output of the
> regulator, you have a nice LC tank circuit. If the capacitor has high
> ESR (tantalum), there may be enough damping. If the capacitor has low
> ESR (ceramic), watch out! It can ring like crazy. You may even turn
> the regulator (which is really a DC amplifier) into an oscillator.
>
> The stability of the regulation loop is usually a strong function of
> the output capacitor. Too much low ESR capacitance on the output of a
> regulation loop will alter the phase margin and gain margin such that
> the loop becomes unstable (oscillates). Who ever is responsible for
> the loop stability of your regulator could probably re-compensate the
> loop so that it will tolerate low ESR capacitors on the output.
> Otherwise, stick to the tantalums.
>
> regards,
> Larry Smith
> Sun Microsystems
>
>
> > From: Chris Cheng <[email protected]>
> > To: [email protected]
> > Subject: [SI-LIST] : ESR and Q factor
> > Date: Thu, 21 Sep 2000 15:08:10 -0700
> > MIME-Version: 1.0
> >
> > dear friends,
> > this is a strange one. i am using a linear regulator to
> > generate 3.3V on one of my board. the app note calls out using
> > a 10uf tantalum for output stability. being a simple minded
> > person, i decided to use a 10uf ceramic cap instead, thinking
> > it has a better esl and esr value. to my surprise, the
> > output breaks into huge oscillation. when i switch back to
> > using tantalum caps, the output quiet down instantly. the
> > same can be achieve if i use a >1 ohm resistor in series
> > with the ceramic cap.
> > the question is, has anyone experienced similar sensitivity
> > in dc linear regulators (the part is lineartech lt1086)? why
> > do linear regulator has such sensitivity to q factors ?
> > thanks in advanced.
> > chris
> >
> > **** To unsubscribe from si-list or si-list-digest: send e-mail to
> > [email protected]. In the BODY of message put: UNSUBSCRIBE
> > si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
> > si-list archives are accessible at http://www.qsl.net/wb6tpu
> > ****
> >
>
> **** To unsubscribe from si-list or si-list-digest: send e-mail to
> [email protected]. In the BODY of message put: UNSUBSCRIBE
> si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
> si-list archives are accessible at http://www.qsl.net/wb6tpu
> ****
>
**** To unsubscribe from si-list or si-list-digest: send e-mail to
[email protected]. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****
This archive was generated by hypermail 2b29 : Tue May 08 2001 - 14:29:33 PDT