b********0
发帖数: 80 | 1 这不是好不好的问题，是能不能的问题。
有能drive both capacitive and resistive load的运放，但是成本高，设计复杂。如
果只是需要drive capacitive load的话，很多简单的opamp就可以了。可大部分简单的
opamp却不能drive resistive load，原因前面已经有人基本解释了。Maily gain and
output current issues.
OTA 根opamp 从严格的定义来说，是有区别的，就像前面有人指出的。但是在IC
design 领域，区别根本不严格，常常互用。opamp, amplifier, OTA may mean same
thing.
As for the stability when drive capacitive load, it really depends on the
location of dominate pole of amplifier. If it is at output, more capacitance
makes it more stable. If it is |
|
g*******r
发帖数: 140 | 2 In some applications, such as LDO, If you want to put the pole associated
with the capacitive load to very high frequency, a straight-forward way is
to cascode the opamp with a buffer.
If just for gain error, there is no difference between opamp and OTA, since
both of them have same DC voltage gain.
For slew rate, there are lots of technqiues using positive feedback to speed
-up the slew rate of opamp. |
|
c*******l
发帖数: 4801 | 3 说的意思是对于opamp,这个buffer级只在opamp用来驱动R的时候才有。如果是驱动
个C，那就很少用这个buffer级 |
|
h*******y
发帖数: 896 | 4 generally, opamp=ota+buffer
ota has large Rout, if driving low Rload, gain=Gm*Rout will decrease
opamp has low Rout, no problem drinve low Rload |
|
q******x
发帖数: 62 | 5 不错，很精辟的论述~~也就是说opamp驱动阻性负载的情况对于增益没有影响，不过如
果对于容性负载，opamp和OTA 分别有什么不同呢？
tx |
|
y*m
发帖数: 144 | 6 let opampA(s) be in a unit feedback config, the transfer function is A(s)/(1
+A(s)).
What is the transfer function if putting the impedence Z(s) between opamp
inverting input port and the opamp output? |
|
a*******e
发帖数: 62 | 7 我设计了一个OPAMP 最后调整了半天 相位裕度只有35度 快 tapeout了 感觉比较担心
倒不是担心settling time 主要担心产生震荡chip不工作 只要工作 性能差一点无所谓
大家能否根据tapeout的经验 来指点我一下 谢谢了 |
|
n******g
发帖数: 662 | 8 Not good. Fabrication also causes some phase margin drift.
Did you simulate your opamp? You would see the oscillation in the unit gain
configuration. |
|
b**s
发帖数: 231 | 9 Looking at the transient simulation is more meaningful than AC. AC is just
a double check.
See if you have a fall-back plan if 35 deg doesn't work on silicon. How
about increasing the overall biasing current of the opamp? Does it give you
more PM and better transient results by having more current? If so, have a
programmable pin to adjust the current. |
|
b********0
发帖数: 80 | 10 Most internal circuits (DAC, opamps) have high impedance output. They can't
drive resistive load. But if load is capacitive, they normally have no
problem to drive, so no need to add a buffer.
Capacitive load can help stability or cause unstable. It all depends on what
the output stage structure is. |
|
G******y
发帖数: 86 | 11 我觉得opamp一般都是designed to drive capacitive load，要是去drive heavy
resistive load（small resistor）的话，首先直流工作点就被破坏了，再加上output
resistance降低-》Gain降低。
可是不管resistive还是capacitive load的稳定性都要具体分析吧？大的cap load也未
必一定稳定吧 |
|
l*****i
发帖数: 296 | 12 depends on the input impedance of opamp.
suppose the input impedance is Zin
then
A(s)(vin-Zin/(Zin+Z(s))vout)=vout
H(s)=A(s)/(1+Zin*A(s)/(Zin+Z(s)))
(1 |
|
f*****0
发帖数: 489 | 13 most opamp will have no problem handling that, including the TDA. you just
don't need the input/output capacitors.
now, if you are very particular about output DC drifting, you should stay
with voltage feedback opamps (most opamps are like that). current feedback
opamps tend to have serious DC drifting issues.
if you look for speed of a current feedback opamp but the DC performance of
a voltage feedback opamp, look into ths4012 from TI. |
|
f*****0
发帖数: 489 | 14
chips like 221 are supposed to do exactly like that.
in general, analog circuits (like an opamp) isn't best suited for such
high speed / high slew rate applications, especially VFB opamps. If you
really want to do it with an opamp, you can do some research and my
guess is that your best chances are with CFB opamp which tend to have >
100mhz bandwidth and slew rates over 1000 v/us. to achieve something
like that with a VFB opamp, you likely will have to rely on positive
feedback and that may in |
|
b*******a
发帖数: 78 | 15 IC前级
有人说6DJ8是为音响而设计的真空管，那么NE5534应该就是第一颗专为音响而设计的IC
。1981年对IC设计而言，尚不到发达的年代， Philips的子公司推出了NE5534 IC，宣
称特别为音响用途而设计，特点是采用双极性晶体差动输入，低阻抗输出，适合在前级
线路中使用。NE5534是一颗运算放大器OPAMP，它将放大器线路浓缩于一颗八支脚的IC
内，只要附加几颗电阻以及防止震荡的电容，就可以构成前级放大器中所需要的放大电
路。消息一出确实轰动业界，原本要使用不算少量零件构成的放大电路，竟然可以使用
一颗IC取代，不禁让设计师看了傻眼。不过当时大家普遍不相信IC的声音，总认为它的
特性甚差，声音不理想，因此并没有人愿意真正拿OPAMP来做前级的主要放大组件，除
了MBL 6010之外。
早期的OPAMP特性确实相当不理想，它的回转率低，杂音特性不佳，还得依照不同的电
路给予不同程度的补偿修正。但现代的IC性能可不能同日语，现代专为音响而设计的
OPAMP，具有如FET及真空管高输入阻抗的优点（具有数M奥姆的输入阻抗，其实比FET还
高），同时也有BJT低输出阻抗的优点 |
|
k********n
发帖数: 18523 | 16 ☆─────────────────────────────────────☆
oiltrader (布伦特之神) 于 (Sun Jan 11 18:26:40 2015, 美东) 提到:
再过几个月
杠杆做多原油和海上爆量囤油的hf 都会爆仓
如果没算错的话，会有一个单日10% 的brent drop。新闻报道到处都是negative，再也
没人说抄底。
这时候
如果之后能拉起来就safe了
☆─────────────────────────────────────☆
papyrus (小饭团) 于 (Sun Jan 11 18:28:46 2015, 美东) 提到:
10%....
☆─────────────────────────────────────☆
oiltrader (布伦特之神) 于 (Sun Jan 11 18:30:27 2015, 美东) 提到:
散户只有亏费的命
提醒了还是往火坑跳
觉得自己有无敌光环呢
[在 papyrus (小饭团) 的大作中提到：]
：10%....
：
：...........
☆─────────... 阅读全帖 |
|
f*****0
发帖数: 489 | 17
level
if it is a level shifter, it is dealing with a digital signal. then the
solution is too easy: find an opamp (any opamp), reference its inverting
end to a voltage reference (like a forward biased diode), and apply the
input to the non-inverting end. when the input is less than the voltage
reference, the opamp will output the negative rail. if the input is
higher than the voltage reference, the opamp will output the positive
rail.
if you need the output to be linear, just apply some negativ |
|
k*********3
发帖数: 257 | 18 经典的教科书上都讲述一个运放有2级或者三级 但是感觉凭空就增加了circuit的复杂
程度和设计难度 能不能先做一个一级的运放 然后在做一个一级的运放 两个串联起来
然后feedback loop 整个的加在两个小opamp的外边 就是说 用两个小的一级opamp形成
一个大的opamp？ 毕竟单级的电路好调啊 |
|
f*****0
发帖数: 489 | 19 you have two issues to solve, both are very easy.
a) use an opamp where you pick up the feedback via a current sampling
resistor in the same loop with the load. vs. the traditional topology, you
need just one more resistor.
b) find a high current output opamp. I would look into high speed dsl
drivers. ad815 (preferred in a sip package, not soic), tpa6120/ths6012, etc.
most of those adsl drivers are duo opamps with high slew rates and high
current capabilities.
now, if that's difficult for you (t |
|
f*****0
发帖数: 489 | 20 here is how "current feedback" works.
R1 is your load, R2 is the sampling resistor that converts the current going
through R1, I1, into a voltage signal to be fed back the inverting end of
the opamp through R6/R3.
so it follows that
I1*R2*R6/(R6+R3)=Vin.
or I1=(1+R3/R6)*Vin/R2.
I typically choose R2 to be < 1/10 of R1, and R3/R6 will depend on what
opamps you use and what input impedance you need.
if you use a regular voltage feedback opamp, I would chose R3=10k.
if no DC blocking capacitor is u |
|
f*****0
发帖数: 489 | 21 the output of the opamp will be either +max if the voltage on the non-
inverting end is greater than that on the inverting end (in this case, it is
zero), or -max if otherwise.
the voltage on the non-inverting end is the weighted average of Vin and Vout
. so the opamp will flip to +max if Vin is such that it causes the weighted
average to be positive; or the opamp will flip to -max if the weighted
average goes negative.
this causes the system to have a small hysteria (thus noise resistance)
depe |
|
n***r
发帖数: 25 | 22 Let me go over with you again.
;the gain of the amp is 500k/10=50k x, approaching most opamp's open loop
;gain, if not more.
Let's replace the 10 ohm with 0 ohm. Will your inverting amplifier theory
work? will the transimpedance amplifier work? Admit it, you were misled by
the 10ohm resistor then you came up with a wrong AC bandwith analysis.
;at that king of gain, the bandwidth of the opamp is very low, likely <10hz -
;check out the opamp's datasheet.
Wrong! The bandwidth of the Transimpedance |
|
ET
发帖数: 10701 | 23 我们的目标就是让后辈少误入歧途，或者让道中人早日迷路知返。
lol...
其实我觉得根本原因是，analog /mixed-signal design的要求越来越高了。 而市面上
能做相对知识面不需要那么宽的designers,可能已经日趋饱和。可以类比为，当初能网
络编程做网页做网站就能称之为程序员，找份工作，到现在当个程序员也需要这算法，
那算法。
还在琢磨中，不知正确与否。
就看ucb的ee140 course project的变化趋势，早些年设计个opamp就是个project， 无
论是telescopic, folded, two-stages,很少有gain-boosting的吧。 现在的projects就
是个whole sensor，这个sensor要有voltage reference, current reference, ADC, M
UX (VGA), 还有啥switched-cap amplifier. 虽然是个team project.
市面上要求的analog/mixed signal的designers, 基本都是要做high speed adc & ... 阅读全帖 |
|
ET
发帖数: 10701 | 24 啥叫cmfb问题多多？
一个对称的电路，比起一个不对称的电路，哪个对common mode的变化处理的会更好些？
opamp大多会用来驱动下一级，这个opamp都有个differentinal 的input, 下面那集自然
也希望有个differenital的input, 意味着prefer这个opamp有个diff. output.
diff. in. diff. out的总体性能一定好与diff.in, single-ended out, 但一定是牺牲
了某样来获得diff. in, singled out.. |
|
s******n
发帖数: 11 | 25 大多数的面试都只会考基本功，除非面试者目前遇到跟你做科研时的题目一样，才会去
问你的论文内容。
Even a simple one like ideal OPAMP with feedback resister has various of
fundamental but tricky questions. For example, simple question like input
impedance, output impedance, then noise contribution of OPAMP input referred
noise and the feedback resistor, and which one would become the dominant
source under which circumstances, how non-linearity got suppressed, why not
an open-loop amplifier then switch to OPAMP design, why Ahuja compensation
is b... 阅读全帖 |
|
h*******a
发帖数: 9 | 26 你已经知道是线上阻抗比较大，又用了open collector，有没有试过把RX端的上拉电阻
加大一点，相应的输出低电平会低一点
仔细选一下RX的logic part which has higher VL threshold.
最保险的方法：可以在RX端把输入信号用opamp处理一下，比如： 输入 2V～5V，opamp
输出0V～5V，opamp周围几个电阻和bias就可以算出来了。 |
|
V*******5
发帖数: 842 | 27 具体的真记不住了。大概就是60%拿着简历问做过的，10%直接拿电路图问什么功能，剩
下的主要集中问cascode opamp，folded opamp，2-state opamp， 米勒补偿原理及如
何消除，current mirror，mismatch，slew rate，LDO，各种DC-DC converter，BJT的
电路，以及device的一些知识。都是针对fresh的问题，估计不入大牛你法眼~ |
|
f****i
发帖数: 20252 | 28 我来从放大器的角度说说，有不对请纠正
先做两个假设：
假设每通道瞬态输入最大为1/4W，而且耳机够好，1/4W输入的时候，工作在线性模式。
对于32Ohm阻抗的耳塞，瞬时电压为2.8V，瞬时电流为88mA，用OPAMP加3.3V或5V电源就
能做到。
对于300Ohm阻抗的大耳机，瞬时电压为8.6V，瞬时电流为29mA，用OPAMP加10V电源可以
做到。
便携式的MP3 Player要设计成驱动32Ohm的耳塞很容易，要设计成驱动300Ohm的耳塞就
麻烦了，跟走集成化降低成本这条路线刚好相反。
开。 |
|
g*********r
发帖数: 543 | 29 glovekeeper，您好：
您转给 OpAmp,现金(伪币):10,收取手续费:0.1
同时附加了如下留言给 OpAmp.
IB还可以这样低，good to know.
站务
在哪里可以设置？ 可以随时转换吗？ |
|
k********n
发帖数: 18523 | 30 ☆─────────────────────────────────────☆
oiltrader (布伦特之神) 于 (Fri Jan 16 17:52:41 2015, 美东) 提到:
几大公司隔夜濒临破产倒闭
最近阶段，做空方一定会更加谨慎
原油反弹可期
下周 all eyes on EU QE and US earnings data
2项都顺利的话，美股创新高。
☆─────────────────────────────────────☆
xcy999 (蛋定) 于 (Fri Jan 16 18:03:00 2015, 美东) 提到:
“原油反弹可期”。。。。谢谢布神正式通知我等青蛙。。。那下周，我就可放心加1/
4缸油啦。。。祝布神长周末愉快。。。
几大公司隔夜濒临破产倒闭最近阶段，做空方一定会更加谨慎原油反弹可期下周 all
eyes on EU QE and US earnings data2项都顺利的话，美股创新高。
☆─────────────────────────────────────☆
oiltrader (布伦特之神) 于 ... 阅读全帖 |
|
|
t******t
发帖数: 248 | 32 ☆─────────────────────────────────────☆
gojeremylin (*_*) 于 (Sat Mar 3 20:17:08 2012, 美东) 提到:
照片都没要到
☆─────────────────────────────────────☆
miny (bee happy) 于 (Sat Mar 3 20:22:10 2012, 美东) 提到:
人家都有家庭了
☆─────────────────────────────────────☆
miluo (络络) 于 (Sat Mar 3 20:31:01 2012, 美东) 提到:
没关系，加油，还有很多available的，我加你了，等着你呢
照片都没要到
☆─────────────────────────────────────☆
wrongest (You could not be wronger) 于 (Sat Mar 3 20:31:31 2012, 美东) 提到:
Which red skirt stood you up?
☆────... 阅读全帖 |
|
o******1
发帖数: 12177 | 33 假ee，opamp为反馈而生，open loop opamp会导致根本无法控制屁股的大小。 |
|
h**g
发帖数: 94 | 34 1) Transistor small signal model, gm, ft
2) Simple amplifer Gm, Rout: common source, common gate, source follower
3) Current mirror, Layout matching
4) Cascode amplifer, Cascode current mirror
5) Differential pair: gain, bandwidth, swing
6) Opamp with miller compensation
7) Opamp with feedback: resistor, capacitor
8) Bode plot, phase margin, pole shifting
9) Transistor noise model
10) Basic equations for diode and BJT
I took many phone/on-site interviews recently. All these questions were asked |
|
f**u
发帖数: 559 | 35 The point is how do you calculate the snr? because it is tricky. it is signal
and noise.
uaually ppl forcus on the int1 noise which comes from two sides: switch noise
and opamp noise.
switch noise is kt/c, so big cap helps.
opamp noise you can run hspice and later refer all them back to your input
side.
also noise and distortion are two different issues. good settling help
distortion. big dc gain at int1 help its distortion, also help to attenuate
int2, int3 etc noise and distortion (that is the |
|
f*****0
发帖数: 489 | 36 the amp is picking up straight signal. maybe your inputs are not properly
shielded or terminated. as to the square wave, that's the opamp clipping
when you dial up the gain.
many ways to deal with it, using a small capacitor is one but none of that
is foundamental.
your approach of using one opamp may not work, depending on your circuit.
the inverting end and non-inverting end in your configuration may have
different input impedance, which may or may not be an issue for your design.
You may want |
|
C*******g
发帖数: 82 | 37 because CS can provide both current and voltage gain. It will be helpful to
have max. voltage and current gain at the last stage of an opamp if the
amplifier is configured in feedback configuration.
Besides, for the max. driving capability, the last stage of an opamp is
sometimes class A/B, if not all the times. I saw a lot more CS configured
class-AB than CG, maybe because it is easier to implement the bias? My guess. |
|
h**g
发帖数: 94 | 38 1) Transistor small signal model, gm, ft
2) Simple amplifer Gm, Rout: common source, common gate, source follower
3) Current mirror, Layout matching
4) Cascode amplifer, Cascode current mirror
5) Differential pair: gain, bandwidth, swing
6) Opamp with miller compensation
7) Opamp with feedback: resistor, capacitor
8) Bode plot, phase margin, pole shifting
9) Transistor noise model
10) Basic equations for diode and BJT
I took many phone/on-site interviews recently. All these questions were asked |
|
g*******r
发帖数: 140 | 39 需要的精度高吗？
如果简单来做，可以用f1(d1) (d1 means duty) 通过一个低通滤波器，得到一个Vo1 =
d1*Vdd. (假设f1的Swing is [0,Vdd]).
同样，将f2(d2)通过同样的滤波器，得到Vo2=d2*Vdd.
将Vo1,Vo2 feedback 到一个high gain opamp, modulate d2, 迫使Vo2 follows Vo1，
这样就得到d2 = d1.
当然，opamp的offset是一个问题，还有feedback loop可能需要compensation如果低通
滤波器高阶。 |
|
g*******r
发帖数: 140 | 40 My two cents:
1) It seldom to make a decision on topology between LC-Ladder and Cas-Biquad
according to power concerns.
2) The chocie of implementation is uaually related to DR. If specs of DR is
not high (40dB~60dB), Gm-C is the most power&area efficient implementations
for both topologies (LC-Ladder and Cas-Biquad). Opamp-MOSFET-C is usually
good for DR around 60dB ~80dB, even 90dB with good nonlinear cancellation
techniques. Opamp-RC is easy for DR >80dB.
Welcome comment, thanks! |
|
f*****0
发帖数: 489 | 41 quite a few problems:
1) you don't need a virtual ground. since you are using four batteries, you
can just ground the circuit to the true ground - center of the batteries.
and if you really really want to use the virtural ground for whatever reason
, put a small eletrolytic cap on the resistor and lower their value to no
more than 1k.
2) you absolutely need to have power supply decoupling or your opamp will
likely oscillate. the particular opamp is a pretty fast one (40mhz gbp), so
you may want |
|
n***r
发帖数: 25 | 42 the gain of the amp is not 500K/10. You think the current source impedance
is 0 ohm? LOL. Primary school students in china know that.
Stupid jerk!
;I always say that Chinese schools should teach students how to ask
questions
;. you are just articulating as a rock.
;the gain of the amp is 500k/10=50k x, approaching most opamp's open loop
;gain, if not more.
;
;so for an input differential of 0.06v-(-1.24v)=1.3v, your opamp would have
;clip'd long ago.
;
;at that king of gain, the bandwidth of th |
|
f*****0
发帖数: 489 | 43
in your circuit, the output should have a dc content of 1/2Vdd because
the opamp is powered by a single rail, and you shouldn't read 0 on the
scope.
a regular DMM shouldn't read out a 500Khz noise so that's OK.
that's never the issue.
the most likely issue is the circuit's output impedance mismatch vs.
that of the probe, or a defective probe. if the signal you are
amplifying is really a high frequency signal, opamp's output impedance
does go up with frequency. but unless you are using some weir |
|
f*****0
发帖数: 489 | 44 "另外，如果用电压源和电阻，运放，如何做电流源？"
an opamp's based CCS may not be useable here, due to the high minimum Vdd/
Vss an opamp needs to operate.
if you insist on doing that, you are far better of with a 3-pin voltage
reference, like lm317lz or tl431: a couple of capacitors, plut one resistor
+ your variable resistor, and you have created a voltage source whose output
changes with the variable resistor.
if you have to go down the CCS route, you may explore discrete CCS,
especially cascoded jfets: they are inc |
|
r*****s
发帖数: 262 | 45 谢谢回答。
我的问题是
switched capacitor integrator需要settle到1/1000以内？
还是要settle到1/10000以内？或者更精确？
因为我已经给出了half clock cycle的时间长度--》50ns
如果需要settle到1/1000以内，tao=1/(2*pi*f)=50ns/7=7ns
f是integrator的GBW。所以就可以定出opamp的最小带宽
但是如果需要settle到1/10000以内，opamp的最小带宽就是另一个数值了
所以我实际上在问 16bit 3阶modulator的第一级switched capacitor integrator
需要settle到何等精确。
integrati
t
sho |
|
ET
发帖数: 10701 | 46 credit to:
razavi's book :
opamp 一章
stabiligy response 最有一道problem
还有berkeley ee240 讲义。
我只是copy & paste (kind of)
珍惜生命，原理analog...
opamp
/ |
|
b*******2
发帖数: 2121 | 47 最近需要一个非常narrow的filter,最好RLC 电路就能搭起来的简单电路,或者有其他的
方法也非常感谢.
我自己的想法如下图左显示,就是简单的LC 串联filter,然后接opamp负端输入.因为
opamp虚短的特性,负输入端(图上点2)电阻应该很小,这样的RLC应该可以产生非常
narrow的filter. 简单run了一下pspice,结果看起来很好(如图右).可是实际操作连接
中,相差巨大,根本看不到这样的效果.
小弟学艺不精,请轻拍.
请大家指教,问题出在哪里?这样的方法可行否?
有没有什么好的方法可以设计非常narrow的filter? 中心频率要求不严格,几百到几kHz
都可以.
非常非常感谢 |
|
b**********g
发帖数: 141 | 48 比如说一个damped RC integrator，
R1 is between Vin and the negative input terminal of opamp.
R2 and C in paralled, between Vout and the negative input terminal of opamp.
请问R2的noise的contribution怎么算呢？
谢谢！ |
|
l***g
发帖数: 1035 | 49 not quite possible unless you have the transfer function of each opamp. if y
ou need something close, simulink control tool can plot the bode. you may ne
ed some simscape toolboxes for the opamp. If u need something acurate, build
the circuit and run spectrum analyzer bah.. much cheaper than the sw. |
|
g******u
发帖数: 3060 | 50 Your load is loading the opamps which it cannot take.
What you need is a BJT output,with the opamp configured as voltage regulator,
like the bottom picture of this link.
http://en.wikipedia.org/wiki/Voltage_regulator
现在调试中发现最后输出电压(Vfout)随load电阻变化，大侠帮忙看看是什么问题？谢
谢。 |
|
