Relation between slew rate and gain-bandwidth product?
Gundal21
Can someone tell me the relation between slew rate and gain-bandwidth
product?
Though the relation is not direct, I've been thought that the more the
slew rate is
the higher the gain-bandwidth product is. But I found that the MP103
from Apex(now Cirrus) has 167V/usec slew rate and 250kHz gain-
bandwidth product and the MP111 has 130 V/usec and 6MHz gain-bandwidth
product. Is there any relationship between these
two properties? I've been thought that the high gain-bandwidth product
means high speed,
so it also means high slew rate. It will be helpful to tell me any
kinds of materials on this.
Thanks.
George Herold
There is not any relation between these two. You might think of GBP
as the low (voltage) level frequency response and the slew rate as the
high level response. An opamp for low level signals might have a
great GBP but crappy slew rate.
George H.
MooseFET
In general faster op-amps have more of both but they are partly
disconnected issues. Slew rate is determined by how fast the op-amp
can charge and discharge some internal capacitor before something
clips. The "something" usually is the input transistors.
Look up the specs for the LT1028 and the LT1351 to see.
Phil Hobbs
Classically, bipolar op amps with no emitter degeneration in the input
pair have slew rates of about 0.3V*GBW. (See "The Monolithic Op Amp: A
Tutorial Study", http://www.national.com/an/AN/AN-A.pdf.)
Lower input stage transconductance allows higher slew rate, because the
op amp runs with more input error voltage (which introduces other
problems, e.g. distortion and crosstalk between summing inputs, that you
may not want).
Current feedback amps don't suffer from these limitations in the same
way, and other modern designs have more complicated tradeoffs, iirc.
Still, AN4 is a good place to start.
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal
ElectroOptical Innovations
55 Orchard Rd
Briarcliff Manor NY 10510
845-480-2058
hobbs at electrooptical dot net
http://electrooptical.net
John Larkin
Regular opamps use the input error to steer the diff pair current
source into later gain stages, so there's a limited amount of drive
available. CFB opamps steal the drive from the input signal and start
applying current gain instantly, and the bigger the loop error the
more they steal. This fooled me once recently, and required a board
spin: at higher speeds, the CFB amp non-inverting input was severely
loading my signal source.
Fet opamps, without input bias current concerns, have a different
tradeoff, but have still tended to have low slew rates.
John
Jim Thompson
<pcdhSpamM...@electrooptical.net> wrote:
Happenstance. Burn some current. See my venerable MC1530 for
example.
>
>Lower input stage transconductance allows higher slew rate, because the
>op amp runs with more input error voltage (which introduces other
>problems, e.g. distortion and crosstalk between summing inputs, that you
>may not want).
Indirectly. Lower input stage transconductance allows a smaller
pole-splitting capacitance, which, for a given current, slews faster.
There have been designs where there are three input stages in
parallel, one normal, two with skewed offsets... large input delta
turns on a high slew current.
>
>Current feedback amps don't suffer from these limitations in the same
>way, and other modern designs have more complicated tradeoffs, iirc.
>Still, AN4 is a good place to start.
>
>Cheers
>
>Phil Hobbs
...Jim Thompson
--
| James E.Thompson, CTO | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food.
Phil Hobbs
It isn't entirely coincidental, because the GBW and transconductance
both go as the input stage current, so cranking up the bias cranks up
both together. I'm talking 741 vintage, here--that app note is from
about 1970. National AN_4_.
Tim Williams
Think of it this way: fT is independent of voltage (or it's supposed
to be), but slew rate is in direct terms of volts. A 100V swing with
a slew rate of 100V/us is the same rise time as a 1V swing slewing at
1V/us. But both amps may have the same fT.
Tim
Tim Wescott
I'd say "little" relation, not "none".
But certainly little enough that you can't count on one given knowledge
of the other.
Tim Wescott
Frequency-bandwidth product is a measure of the behavior of the device
when it is acting in its linear region. Slew rate is a measure of the
point at which the device's behavior departs from the linear region, and
a (partial) measure of how it be behaves afterward.
There are no guarantees, but if you look at power consumption as well you
will find that low power devices will almost universally have low slew
rates relative to bandwidth while high power devices will mostly have
high slew rates relative to bandwidth. Amps designed for video will
almost always be fairly high bandwidth, fairly high power, and have
pretty darn good slew rates.
But if some criterion is important to you then you should always,
_always_ check the data sheet to verify that it is met. Further, you
should check with physical samples -- sometimes you'll find that a data
sheet parameter was established by calculation, or it is only valid for
test conditions which don't match your circuit, or that you just plain
misread the data sheet, etc.
Jim Thompson
<pcdhSpamM...@electrooptical.net> wrote:
I'm from further back than that ;-)
Bob Eld
"Gundal21" <hoche...@gmail.com> wrote in message
news:93a9fc82-b3d8-4fd6...@y10g2000prg.googlegroups.com...
The slew rate required is determined by the peak voltage in a circuit and
the maximum frequency of that voltage. It is only indirectly related to
GBWP. The slew rate, p (ro) = dV/dt, the rate of change of the voltage. The
voltage is V = Vp*sinwt and the first derivative, dV/dt = Vp*w*coswt. Vp is
the peak voltage. w (omega) = 2*pi*f and the maximum cosine value = 1
Therefore the slew rate, p = dV/dt = Vp*2*pi*f.
As an example take a 10 Volt peak voltage at 1 MHz, the slew rate p =
2*Pi*10*1*10^6 = 62.8*10^6 volt per second or 62.8 volts per microsecond.
This says that for an amplifier or other device to provide that voltage at
that frequency, it must have a slew rate of at least 62.8 Volts per
microsecond.
Slew rate is determined by how fast the available current in an amplifier or
other device can charge and discharge the internal capacitances. In most
amplifiers, there is a dominant capacitance usually used for compensation
that must be charged and discharged by the local current. The rate this
happens determines the slew rate: dv/dt. The change in voltage, dv =
(1/C)int*i*dt determining the slew rate from the current, i, and the
capacitance, C. The break point or "Pole" related to this capacitance
determines the GBWP but is only a lose relation to the slew rate.
George Herold
<pcdhSpamMeSensel...@electrooptical.net> wrote:
> MooseFET wrote:
> > On Oct 6, 3:15 am, Gundal21 <hocheol....@gmail.com> wrote:
> (See "The Monolithic Op Amp: A
> Tutorial Study",http://www.national.com/an/AN/AN-A.pdf.)
>
> Dr Philip C D Hobbs
>
> - Show quoted text -
Thanks for the link to the app note Phil. I had always thought these
were unrelated.
'Silly me'
Also great (simple) explanation of what's going on inside a 'classic'
opamp.
George H.
Robert Baer
product.
That being said, *IF* the slew rate of an op-amp is high, its GBW
will (usually) be high.
Phil Allison
"Gundal21"
>
> Can someone tell me the relation between slew rate and gain-bandwidth
> product?
> Though the relation is not direct, I've been thought that the more the
> slew rate is
> the higher the gain-bandwidth product is. But I found that the MP103
> from Apex(now Cirrus) has 167V/usec slew rate and 250kHz gain-
** Try reading again.
Think you will find 250kHz is the full power bandwidth.
The MP103 is a power amplifier.
..... Phil
Jim Thompson
<rober...@localnet.com> wrote:
>Gundal21 wrote:
>> Hi~
>>
>> Can someone tell me the relation between slew rate and gain-bandwidth
>> product?
>> Though the relation is not direct, I've been thought that the more the
>> slew rate is
>> the higher the gain-bandwidth product is. But I found that the MP103
>> from Apex(now Cirrus) has 167V/usec slew rate and 250kHz gain-
>> bandwidth product and the MP111 has 130 V/usec and 6MHz gain-bandwidth
>> product. Is there any relationship between these
>> two properties? I've been thought that the high gain-bandwidth product
>> means high speed,
>> so it also means high slew rate. It will be helpful to tell me any
>> kinds of materials on this.
>>
>> Thanks.
> There is *NO* relation between slew rate and gain-bandwidth
>product.
Of course there is... it's just not an absolute relationship. Slew
rate, in an OpAmp, depends on first stage current. This current
generally defines first stage transconductance, which critically
influences the size of the pole-splitting capacitor (loop
compensation). Capacitor size and input stage current define slew
rate. So there!
> That being said, *IF* the slew rate of an op-amp is high, its GBW
>will (usually) be high.
Maybe ;-)
...Jim Thompson
--
| James E.Thompson, CTO | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
We right-wing conservatives go so far out of our way to be polite
that we wouldn't deign to use such words as "ignorant" or "yellow-
belly". Instead, to keep everything warm and fuzzy, we substitute
the politically-correct synonyms "leftist weenie", "liberal" and
"Democrat" ;-)
Jim Thompson
I erred.
Suppose we have an ideal OpAmp, bipolar (so I can easily calculate
input stage transconductance ;-), input stage biased by current,
Ibias, "pole splitting" capacitor of value, C, then...
GBW = (q*Ibias)/(4*k*T*C)
k = Boltzmann's constant
q = Charge on electron
T = temperature in �K
Slew Rate = Ibias/C
This is for a classic two-stage OpAmp, with simple "pole-splitting"
compensation.
So GBW and Slew Rate are exactly related.
Add extra stage(s) and/or zeroes, and it'll change, possibly getting
more GBW without a consequent increase in Slew Rate.
"Real" OpAmps won't be quite as fast as predicted with this simplistic
analysis.
...Jim Thompson
--
| James E.Thompson, CTO | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
JosephKK
<To-Email-Use-Th...@My-Web-Site.com> wrote:
What!? And the universe didn't end? Gosh, i guess i won't feel so
bad when i make an error. Not that they are frequent any more.
>
>Suppose we have an ideal OpAmp, bipolar (so I can easily calculate
>input stage transconductance ;-), input stage biased by current,
>Ibias, "pole splitting" capacitor of value, C, then...
>
>GBW = (q*Ibias)/(4*k*T*C)
>
>k = Boltzmann's constant
>q = Charge on electron
>T = temperature in ºK
Jim Thompson
>>T = temperature in �K
>>
>>Slew Rate = Ibias/C
>>
>>This is for a classic two-stage OpAmp, with simple "pole-splitting"
>>compensation.
>>
>>So GBW and Slew Rate are exactly related.
>>
>>Add extra stage(s) and/or zeroes, and it'll change, possibly getting
>>more GBW without a consequent increase in Slew Rate.
>>
>>"Real" OpAmps won't be quite as fast as predicted with this simplistic
>>analysis.
>>
>> ...Jim Thompson
Throughout my life I have thrown away more designs than survived, by
at least an order of magnitude.
To this day I often offend clients by tossing out everything 2/3 of
the way to completion, "... because it just doesn't feel right".
Thus my finished projects _always_ work.
Unfortunately there lurk here "designers" who run something up the
flag pole, swear by it, and then defend their errors as being of good
design practice... despite definitive criticism :-(
JosephKK
<To-Email-Use-Th...@My-Web-Site.com> wrote:
>>>T = temperature in ºK
>>>
>>>Slew Rate = Ibias/C
>>>
>>>This is for a classic two-stage OpAmp, with simple "pole-splitting"
>>>compensation.
>>>
>>>So GBW and Slew Rate are exactly related.
>>>
>>>Add extra stage(s) and/or zeroes, and it'll change, possibly getting
>>>more GBW without a consequent increase in Slew Rate.
>>>
>>>"Real" OpAmps won't be quite as fast as predicted with this simplistic
>>>analysis.
>>>
>>> ...Jim Thompson
>
>Throughout my life I have thrown away more designs than survived, by
>at least an order of magnitude.
>
>To this day I often offend clients by tossing out everything 2/3 of
>the way to completion, "... because it just doesn't feel right".
Hmmm.
>
>Thus my finished projects _always_ work.
>
>Unfortunately there lurk here "designers" who run something up the
>flag pole, swear by it, and then defend their errors as being of good
>design practice... despite definitive criticism :-(
>
> ...Jim Thompson
Interesting, in my current workplace that describes the little new
work done by most electrical engineers in my organization. Of course,
none of them has built anything (except course projects) with their
own two hands.
Jim Thompson
"JosephKK"<quiett...@yahoo.com> wrote:
>>>>T = temperature in �K
>>>>
>>>>Slew Rate = Ibias/C
>>>>
>>>>This is for a classic two-stage OpAmp, with simple "pole-splitting"
>>>>compensation.
>>>>
>>>>So GBW and Slew Rate are exactly related.
>>>>
>>>>Add extra stage(s) and/or zeroes, and it'll change, possibly getting
>>>>more GBW without a consequent increase in Slew Rate.
>>>>
>>>>"Real" OpAmps won't be quite as fast as predicted with this simplistic
>>>>analysis.
>>>>
>>>> ...Jim Thompson
>>
>>Throughout my life I have thrown away more designs than survived, by
>>at least an order of magnitude.
>>
>>To this day I often offend clients by tossing out everything 2/3 of
>>the way to completion, "... because it just doesn't feel right".
>
>Hmmm.
>>
>>Thus my finished projects _always_ work.
>>
>>Unfortunately there lurk here "designers" who run something up the
>>flag pole, swear by it, and then defend their errors as being of good
>>design practice... despite definitive criticism :-(
>>
>> ...Jim Thompson
>
>Interesting, in my current workplace that describes the little new
>work done by most electrical engineers in my organization. Of course,
>none of them has built anything (except course projects) with their
>own two hands.
I think it is a shame that students no longer build anything. I
really feel blessed to have grown up in a radio/TV repair shop, with a
father who funded my experimenting.