微波EDA网,见证研发工程师的成长!
首页 > 研发问答 > 微波和射频技术 > RFIC设计学习交流 > MOST和Bipolar做输入管的摆率比较

MOST和Bipolar做输入管的摆率比较

时间:10-02 整理:3721RD 点击:
在sansen书上p149,0717段有句话,“只有电流源采用双极型器件,输入器件采用MOST比较好,因为他们能提供很小的输入偏置电流和很高的转换速率。”
偏置电流很好理解,而摆率方面,为什么MOST比双极做输入管摆率高呢,难道是因为在同等gm情况下MOST需要更高的电流么、

correct

那如果我用BJT做输入,采用同样的偏置电流,岂不是既可以得到大的摆率,又能增加gm,一举两得啊;有什么缺点呢。

That's the basic of microelectronics.
Compared to old CMOS technology, BJT is better to be used in analog circuit, it can offer lower power, lower noise, and fast.
But as CMOS scaling, MOSTs are fast enough.
Initially, CMOS only used in digital circuit, because its near zero static power consumption, low cost and high integration density.
Now MOSTs are good enough now, and it offers another design freedom : Vgs-Vt, so in many cases, it's actually better than BJT.
Most analog circuits are implemented in CMOS because of lower cost. BJT's area is much larger than MOSTs.
There is a saying in microelectronics : MOS transistor is Bad, but MOS technology is excellent.
Hi, buddy, Sansen's book is much, much harder than Gray's, Razavi's or Allen's book to understand thoroughly.
Please start with these three books, not Sansen's.

嗯,谢谢。我以前用的一直是CMOS工艺,也就是因为便宜以及数模混合电路兼容性好吧,对BJT几乎没多少认识。现在因为要做一个高速宽带运放,所给的工艺是Bicmos工艺,里面只有npn管,其特征频率大约是MOST的10倍。而运放的GBW要求几乎和MOST的Ft差不多,因此不得不使用BJT。第一次用两种类型管子设计电路,很不适应。
我有一个疑问,对MOST和BJT的取舍应该遵循什么样的原则或规律呢。
下面是我的一点认识,请您帮我看看有什么不妥或者补充的:
由于没有pnp管,因此有些地方必须要使用pmos来代替,比如某些电流镜;
npn和nmos做电流镜尚不知如何选;
在信号通路上尽量不采用MOST,因为会影响带宽;
输出级在输出电流较大(几个mA)时,采用BJT,感觉MOST流过的电流一般在uA量级;

Sorry, I don't have much experience inBJT either

btw, how large is your GBW requirement ?

1)MOS 栅极不需要偏置电流;2)MOS的漏衬电容比Biplor的集电极-衬底电容小---》Higher SlewRate;MOS没有栅极电流,差分对的偏置电流全部镜像到输出节点的电阻做SlewRate---》Higher SlewRate;

Those you talked about are minor reasons.
The fundamental one is that BJT's gm/Id is much larger than MOST

你现在在鲁文大学?还在在那边工作?


咳咳,是否是在GBW相等这个前提下,差分输入级的gm要相等,要达到相同的gm,MOS需要更多偏置电流;因此SR更大。
如果输入级的电流相等,MOS差分输入,Bipolar差分输入的SR 应该差别不大。

GBW≥7GHz

Yes

Ok, 7GHz is much much smaller than the transit frequency of CMOS technology. Then I think it's possible.

But in your another post, you said your GBW should be THz, that's not possible.

Ahha, you checked my IP, hehe~

I study at KULeuven

呵呵,加油

为什么不全用cmos? 参看下图:


完全足够,在7g的时候。

给的是0.35um的工艺,MOST如果要达到较高特征频率过驱动就会很高,然后就进入速度饱和了。即使那样还是没BJT快,实现起来有难度。

Copyright © 2017-2020 微波EDA网 版权所有

网站地图

Top