The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> RF Design >> Inductor in PA design
https://designers-guide.org/forum/YaBB.pl?num=1140984093

Message started by jxbvt on Feb 26^th, 2006, 12:01pm

Title: Inductor in PA design
Post by jxbvt on Feb 26^th, 2006, 12:01pm

hi all....can some one explain this sentence to me.i have a fair idea but i was wondering exactly what is meant by:

"Output and load inductors will likely be off-chip due to high level of currents required". i know inductors are lossy and have low Q and that's why it's hard to implement them on chip but isn't it regardless of high levels of current or not. guess what i want to know is how current level prevents implementing inductors on chip. thanx

jb

Title: Re: Inductor in PA design
Post by ACWWong on Feb 27^th, 2006, 4:13am

metalization in silicon processes are typically ALuminum (although now copper is fairly common).
what limits the high current is electron migration... ie. too much current causes the metal to heat and eventually effectly distort/melt. to over come this one must make tracks wide or stacked to carry the current densities in high power circuits. When one considers that current of about 1mA/um width are typical, you can see even with thick top metal its becomes impossible/very difficult to make inductors take 100's mA and still meet the electron migration rules over tempertaure.

Title: Re: Inductor in PA design
Post by jxbvt on Feb 27^th, 2006, 10:07am

Thanks for the reply...makes sense. Practically, yes electromigration issues are very important and place a limit on conducting current but theoritically isn't migration issue a reliability issue? Meaning, let's say a circuit will be used once once (lol) then electromigration won't be an issue because the degradation is over time, right?

jb

Title: Re: Inductor in PA design
Post by milkdragon on Feb 27^th, 2006, 2:44pm

Electromigration is a physical issue. Once your current limit exceeds, the metal will melt. It is not something like it would degrade over time. There is another issue called mobility degradation, which is a function of time instead for PA design, this is especially true for CMOS process

Title: Re: Inductor in PA design
Post by Andrew Beckett on Feb 28^th, 2006, 7:36am

A useful paper I just found on Electromigration - http://www.cadence.com/whitepapers/4095_Electromigration_WP.pdf

Regards,

Andrew.

Title: Re: Inductor in PA design
Post by ACWWong on Feb 28^th, 2006, 7:57am

the electron migration effect is usually gradual effect over time, and of course depends on how much over the current density limits given by the foundry and the temperature (real chip temperature, not ambient and for a PA that'll be pretty hot!) as to how fast it degrades.
Of course if you measure something once and for a short period of time, you'll be ok if you are just outside the reliablity quoted by the foundry. But gradually the track resistances will increase, so over time your measurements will get worse. eventually you could get tracks that dislocate or even adjacent tracks could short.
I think the book by Hastings "Art of Analog Layout" covers the subject quite well.

jxbvt wrote on Feb 27^th, 2006, 10:07am:

Title: Re: Inductor in PA design
Post by totowo on Aug 31^st, 2006, 7:12pm

The current limit of inductor is quite right. But the answer to jxbvt's primary puzzle, i think, is not this.

The inductor of output match is offchip because of its High Q.
For example, if the on-chip inductor has a 3dB loss, that means half of the power is wasted.
at input port, the input power is, e.g. 0dBm=1 mW, so the loss is only 0.5mW. It's receivability.
while at output port, the output power is, e.g. 20dBm=100mW, so the loss on inductor is 50mW!!!
It degrades the efficiency too much! Although the loss is 3dB too.
Let's think in another way, the parasitic series res is, e.g. 100Ohm, when the current on inductor is 1mA, the loss is 0.1mW; while when the current on ind is 100mA, the loss is 1W!!!

In fact , the current limit of inductor is not the primary reason, for we can parallel several inductor to share the big current.

Title: Re: Inductor in PA design
Post by loose-electron on Sep 3^rd, 2006, 1:30pm

In the RF PA world, in additiion to electromigration there are a few other issues also:

Q of inductors needed.

PAE of the RF PA. Lossy inductors in a PA can kill your PAE numbers and make the product not viable.

Especially true of a PA in a cell phone.