l***g
发帖数: 1035 | 1 经常见到dsp或者cpld 的 dio 有resistor array 10ohm-50ohm。 记得是和噪声有关。
那位有相关的参考资料么？一般多高速的信号，时钟用的上？ |
g******u
发帖数: 3060 | 2 it's just a small low-pass filter. You set it depending on your frequency
and load, even though there is no on-chip resistors you may need to add some
yourself externally. |
l***g
发帖数: 1035 | 3 what kind of low pass ? lr or rc? |
t****1
发帖数: 827 | 4 有没有低通高通的目的，我不知道。我知道数字输入端口，出于ESD的考虑，片上要串联一个小电阻。 |
l***g
发帖数: 1035 | 5 any reference? every pin or just some particular type of pins?
串联一个小电阻。
【在 t****1 的大作中提到】
: 有没有低通高通的目的，我不知道。我知道数字输入端口，出于ESD的考虑，片上要串联一个小电阻。
|
g******u
发帖数: 3060 | 6 you can call it low pass, current limiting resistor, damper, anything.
串联一个小电阻。
【在 t****1 的大作中提到】
: 有没有低通高通的目的，我不知道。我知道数字输入端口，出于ESD的考虑，片上要串联一个小电阻。
|
t****1
发帖数: 827 | 7 输入端口在片子里面通常连接到 场效应管的门极，高输入阻抗，静电很容易造成高压
击穿，这个做IC的人都知道的。网上随便找了一个，你有兴趣就看看吧。http://www.dianzichan.com/anonymous/ic/ESD_Suggestion_for_HighSpeed_IO_Ker.pdf
传统的ESD， 有一个小电阻
缺点是在高速电路中这个电阻可能会影响电路速度(由于R和寄生电容的滤波效应)
为了解决这个问题，这位大侠搞了一个新方法，能否可靠工作我就不知道咧。
【在 l***g 的大作中提到】
: any reference? every pin or just some particular type of pins?
:
: 串联一个小电阻。
|
g****t
发帖数: 31659 | 8 这个文章很牛啊，收藏了。
输入端口在片子里面通常连接到 场效应管的门极，高输入阻抗，静电很容易造成高压
击穿，这个做IC的人都知道的。网上随便找了一个，你有兴趣就看看吧。http://www.dianzichan.com/anonymous/ic/ESD_Suggestion_for_HighSpeed_IO_Ker.pdf
传统的ESD， 有一个小电阻
缺点是在高速电路中这个电阻可能会影响电路速度(由于R和寄生电容的滤波效应)
为了解决这个问题，这位大侠搞了一个新方法，能否可靠工作我就不知道咧。
【在 t****1 的大作中提到】
: 输入端口在片子里面通常连接到 场效应管的门极，高输入阻抗，静电很容易造成高压
: 击穿，这个做IC的人都知道的。网上随便找了一个，你有兴趣就看看吧。http://www.dianzichan.com/anonymous/ic/ESD_Suggestion_for_HighSpeed_IO_Ker.pdf
: 传统的ESD， 有一个小电阻
: 缺点是在高速电路中这个电阻可能会影响电路速度(由于R和寄生电容的滤波效应)
: 为了解决这个问题，这位大侠搞了一个新方法，能否可靠工作我就不知道咧。
|
t****1
发帖数: 827 | 9 如果这个电阻在片外，那和ESD就没什么关系了，大概是和噪音滤波有关吧。
【在 g******u 的大作中提到】
: you can call it low pass, current limiting resistor, damper, anything.
:
: 串联一个小电阻。
|
g****t
发帖数: 31659 | 10 My two cents:
sigma delta A/D只能处理带宽有限信号,不然有可能会不稳定,性能也没法保证.
所以往往外面要加低通滤波.其他结构可能也有类似情况.
如果这个电阻在片外，那和ESD就没什么关系了，大概是和噪音滤波有关吧。
【在 t****1 的大作中提到】
: 如果这个电阻在片外，那和ESD就没什么关系了，大概是和噪音滤波有关吧。
|
|
|
g******u
发帖数: 3060 | 11 of course it does.
ESD can be generated outside board assembly, it can easily kill your on-chip
ESD protection.
【在 t****1 的大作中提到】
: 如果这个电阻在片外，那和ESD就没什么关系了，大概是和噪音滤波有关吧。
|
t****1
发帖数: 827 | 12 你assembly 这个 外接电阻的时候，就不会产生静电了？
当然你可以说，这个静电可以由on-chip ESD 保护。 问题是，如果on-chip ESD能够防止assembly这个外接电阻造成的静电，那怎么就不能消除assembly 其它原件造成的静电呢？事实上，我很少见到出于ESD的目的，而加外接电阻的。
chip
【在 g******u 的大作中提到】
: of course it does.
: ESD can be generated outside board assembly, it can easily kill your on-chip
: ESD protection.
|
g******u
发帖数: 3060 | 13 could be by assembly, by touching, contacts, anything.
they can exceed your onchip esd rating as it may not be designed to see sth
more than its rating. that's where you add another resistor.
for example a chip rated for Vcc+0.5V at input side, if the input is a
switch or an inductive source, it can generate way higher than the rated
input voltage, even though the IC has some esd protection, it's not designed
for that.
防止assembly这个外接电阻造成的静电，那怎么就不能消除assembly 其它原件造成的
静电呢？事实上，我很少见到出于ESD的目的，而加外接电阻的。
【在 t****1 的大作中提到】
: 你assembly 这个 外接电阻的时候，就不会产生静电了？
: 当然你可以说，这个静电可以由on-chip ESD 保护。 问题是，如果on-chip ESD能够防止assembly这个外接电阻造成的静电，那怎么就不能消除assembly 其它原件造成的静电呢？事实上，我很少见到出于ESD的目的，而加外接电阻的。
:
: chip
|
g*z
发帖数: 124 | 14 the resistor is used to kill the ringing, overshooting and undershooting of
high speed digital connection.
please note that here, high speed has nothing to do with the data rate. Even
the data rate is low, if there's sharp rising edge or falling edge, it is
still high speed circuit. |
t****1
发帖数: 827 | 15 Although it is a common issue, especially in motor driving circuit,
overshoot due to the inductive load is not an ESD event. This kind of
overshoot usually lasts longer and has more energy than that in an ESD event
.
sth
designed
【在 g******u 的大作中提到】
: could be by assembly, by touching, contacts, anything.
: they can exceed your onchip esd rating as it may not be designed to see sth
: more than its rating. that's where you add another resistor.
: for example a chip rated for Vcc+0.5V at input side, if the input is a
: switch or an inductive source, it can generate way higher than the rated
: input voltage, even though the IC has some esd protection, it's not designed
: for that.
:
: 防止assembly这个外接电阻造成的静电，那怎么就不能消除assembly 其它原件造成的
: 静电呢？事实上，我很少见到出于ESD的目的，而加外接电阻的。
|
l***g
发帖数: 1035 | 16 comparing to adding a for example 100pF at the end of the IO line, which way
is preferred? i recall that adding a small cap can be treated as 'impedance
matching', which way is more appropriate? and what about analog inputs/outp
uts?
of
Even
【在 g*z 的大作中提到】
: the resistor is used to kill the ringing, overshooting and undershooting of
: high speed digital connection.
: please note that here, high speed has nothing to do with the data rate. Even
: the data rate is low, if there's sharp rising edge or falling edge, it is
: still high speed circuit.
|
g*z
发帖数: 124 | 17 damping should be placed as close as the io pins, overshooting is bad for
both sender and receiver, cuz it may cause latch-up |
t**********1
发帖数: 26 | 18 I agree with ghz's comment.
This is the terminating resistor. When a line is long, meaning the cable
length exceeds 1/6 of the electrical length of a rising edge, the cable
needs terminating resistors. Without terminators, reflections may occur
at
either end of long cable(or PCB trace).
For further information, please check Howard Johnson's book, high-speed
digital design, a handbook of black magic, Chapter 6.
undershooting
of
rate.
Even
it
is
【在 g*z 的大作中提到】
: the resistor is used to kill the ringing, overshooting and undershooting of
: high speed digital connection.
: please note that here, high speed has nothing to do with the data rate. Even
: the data rate is low, if there's sharp rising edge or falling edge, it is
: still high speed circuit.
|
l***g
发帖数: 1035 | 19 thanks for the reference. by electrical length, do you mean wavelength of
the fundamental?
【在 t**********1 的大作中提到】
: I agree with ghz's comment.
: This is the terminating resistor. When a line is long, meaning the cable
: length exceeds 1/6 of the electrical length of a rising edge, the cable
: needs terminating resistors. Without terminators, reflections may occur
: at
: either end of long cable(or PCB trace).
: For further information, please check Howard Johnson's book, high-speed
: digital design, a handbook of black magic, Chapter 6.
:
: undershooting
|
t**********1
发帖数: 26 | 20 There is an equation in this book (page 7) as shown below.
l=Tr/D
where l=electrical length of rising edge,in.
Tr=rise time,ps
D=propagation delay, ps/in.
This book also gives an example of a 10kHz ECL signal with a rising edge
of
1ns. This rising edge has a length of 5.6 inch when propagating along an
inner trace of an FR-4 PCB. The electrical length will help you understand
if you need to treat a trace as a lumped or distributed component(
transmission line) for a digital signal. |
l***g
发帖数: 1035 | 21 thanks for sharing.. interesting... no wonder it's called black art..
【在 t**********1 的大作中提到】
: There is an equation in this book (page 7) as shown below.
: l=Tr/D
: where l=electrical length of rising edge,in.
: Tr=rise time,ps
: D=propagation delay, ps/in.
: This book also gives an example of a 10kHz ECL signal with a rising edge
: of
: 1ns. This rising edge has a length of 5.6 inch when propagating along an
: inner trace of an FR-4 PCB. The electrical length will help you understand
: if you need to treat a trace as a lumped or distributed component(
|
