Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

gate connection using poly vs metal

Status
Not open for further replies.

Junus2012

Advanced Member level 5
Joined
Jan 9, 2012
Messages
1,552
Helped
47
Reputation
98
Reaction score
53
Trophy points
1,328
Location
Italy
Activity points
15,235
Dear friends,

When I do layout I usually connect the gate of the single matched array by extending the poly. it is easy to implement and I think that extending poly might be better than pitting via per every gate to move to metal connection as the via also introduce resistance,

however in the book of the Art of layout design he suggested to connect the gates using metals

I would like to have your opinion according to your experiance

thank you very much


gate_matching.png
 

Dear friends,

When I do layout I usually connect the gate of the single matched array by extending the poly. it is easy to implement and I think that extending poly might be better than pitting via per every gate to move to metal connection as the via also introduce resistance,

however in the book of the Art of layout design he suggested to connect the gates using metals

I would like to have your opinion according to your experiance

thank you very much


View attachment 153596

You can't route with poly unless the technology is ~20 years old. It's a non-starter.
 
Dear NotSam,

I am using the 0.35 technology.
 

There will be a "master" and N "slaves" in the current mirror
rack and you want to pay attention to where the reference
current flows. For example if you push the current in at the
left end of a S-S-M-S-S rack, the M device's drain current
makes a successive I*R drop between S1.G and S2.G, S2.G
and M.{G,D}, and then there is no further drop to S3.G and
S4.G; so you would have three slightly different (or not so
slightly, depending) current outputs with a systematic mismatch
(S1>S2>S3,S4).

You should understand the poly resistor options and for an
analog layout, attempt to make the poly-route look more like
some number of poly resistors so effects are represented that
you might have overlooked.
 
Dear freebird,

As I understood from you you are not in favor of connecting the gain using the poly, surely it gives more resistance no doubt about it.. but I I am matching 10 gates for example then I need to put 10 via to make the connection with metal one or 20 via if I use metal 2 which I usually use in the horizontal routing,

20 via would also add contact resistance but I am not sure if it is still less than comparing if I use the poly,,,, I need to know from your experience please.. if you confirm me that metal+via is better than only poly then I will not use the poly contact any more for connecting gates

Note: I am only using poly for connecting gates, all other circuit parts are connected with metal
 

There are good reasons for and against, as the engineer
the value judgment is up to you.

For example maybe it's better to route poly in a standard
cell logic library, than "bake in" metal2 obstructions that
cost you routing efficiency and time. More true back in
the few-metal-layers days (but these are not entirely
over - especially if you live on the trailing edge).

Via resistance has a spec. It's going to be lower than a
square of low doped or undoped poly; maybe more,
maybe less than silicided. You can parallel vias.

Whatever you do, you can insert presistor elements to
assess the effect and use that feedback to decide.
Since poly has the highest capacitance to bulk, of the
conductors, you might look into that as well.

One other impact of routed gate poly is the elevated
effective Rg. That can impair high frequency PSRR and
gain, allow Miller kick to more easily / significantly perturb
current mirror output (presuming any decoupling / stiffness
on the reference gate node is remote). Again realism in
modeling at the schematic would show this, if you looked.
 
Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top