The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> RF Design >> Vdd Gnd Bonding
https://designers-guide.org/forum/YaBB.pl?num=1182887584

Message started by fz2101 on Jun 26^th, 2007, 12:53pm

Title: Vdd Gnd Bonding
Post by fz2101 on Jun 26^th, 2007, 12:53pm

Hello everyone,

I'm using a QFN package with exposed paddle, i.e., the bottom of the package is a big ground plane and can be directly
contacted to the PCB ground plane. This allows me to downbond the "Gnd" pin to the ground plane and thus achieve a
small bondwire inductance, L_Gnd, of 1nH, let's say.

On the die, Vdd and Gnd pins are right next to each other. If I don't downbond, then the Vdd and Gnd bondwires
run in parallel to each, and because of the opposite direction of the current flow, the inductance of both are reduced (by
the mutual inductance).

So my question is, which bonding scheme is better?
1. Downbond Gnd ----> large L_Vdd, small L_Gnd
2. Do not downbond -----> medium L_Vdd, medium L_Gnd

Thanks,

frank

Title: Re: Vdd Gnd Bonding
Post by ACWWong on Jun 26^th, 2007, 2:49pm

I would go for downbond for ground for shorter ground inductance. Although its true that adjacent bondwires carrying complementary currents will have mutual inductance as to reduce the effective self inductance of the pair, will your gnd and vdd really carry differential RF currents (at frequencies the inductance becomes noticable) ? what's the pcb arrangements for vdd and ground square ? I think its only safe to assume differential RF currents in things like differential VCO with equal bondwires to a common landing point/package pin.

Anyway you can do some simulations or hand calculations, as it'll depends on the lengths and spacings of the wires... assuming 25um diameter then 1mm down bound ground is ~ 1nH, and say vdd is 2mm (~2.2nH).
Two parallel 2mm tracks spaced at 0.4mm have a mutual ~0.6nH, so in hairpin mode the series inductance of gnd & vdd is 3.2nH (assuming IDEAL differential currents), i.e more than 3nH.

Title: Re: Vdd Gnd Bonding
Post by fz2101 on Jun 26^th, 2007, 4:01pm

Yes, the current through Vdd and Gnd will not be differential at RF frequencies for most circuits, that was my mistake.

I will go with the downbonded Gnd.

Thanks

Title: Re: Vdd Gnd Bonding
Post by mg777 on Jun 26^th, 2007, 7:12pm

ACWWong's advice is bang on the money. Also use multiple down bonds for the GND's - pepper your pad ring with GND pads.

BTW, anyone know of a situation where GND bond inductance improves signal integrity? Not referring to amplifier source degenration here.

M.G.Rajan

Title: Re: Vdd Gnd Bonding
Post by fz2101 on Jun 27^th, 2007, 8:32am

Hi MG,

I think for a differential circuit, the ground bondwire inductance improves the common-mode rejection. For example, we can compare the idealized pseduo-differential amplifier on the left and the one with ground bondwire inductance on the right. The one on the right is essentially the differential L-degenerated LNA, with the real impedance introduced at the input for matching as an added bonus.

frank

Title: Re: Vdd Gnd Bonding
Post by fz2101 on Jun 29^th, 2007, 10:48am

The above statement about the L-degenerated LNA is wrong, since differential signals do not see this common-mode inductor. The real impedance introduced by the ground bondwire is only for the single-ended L-degenerated LNA case. To achieve input impedance matching for the differential LNA, on-chip inductors are needed. Sorry for the confusion. :P

frank