Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronic 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.

Register Log in

[SOLVED] is length important in current mirrors?

Status
Not open for further replies.

polygon pusher

Newbie level 5
Joined
Sep 15, 2010
Messages
10
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,281
Activity points
1,349
Hi,

By boss tells me that to get good matching in current mirrors they should be designed where L is large. My own understanding that as long as the devices are large in area and laid out well then they will match. So my question is length important or is making the area of the mosfet large sufficient?

Thanks
 

erikl

Super Moderator
Staff member
Joined
Sep 9, 2008
Messages
8,112
Helped
2,687
Reputation
5,354
Reaction score
2,288
Trophy points
1,393
Location
Germany
Activity points
44,153
... my question is length important or is making the area of the mosfet large sufficient?
Length is always important ;-). But you're right saying that device area is decisive for matching. The reason for designing current mirror transistors with (relatively) large length is: usually you don't want to spend large width just for current mirroring, so for matching accuracy you need larger length. And low W/L ratios achieve better simulation accuracy and lower Vds,sat , AFAIR.
 

ninju

Full Member level 3
Joined
May 14, 2011
Messages
189
Helped
49
Reputation
98
Reaction score
48
Trophy points
1,308
Activity points
2,398
Its related to the lambda parameter. Generally, in analog, longer MOSFETS are preferred.

Longer length also means r0 (r0 = 1/gds) is higher. Which is good for a current source. Also, if the resistance changes , it will be a much tinier fraction of what it would've been if the MOS were shorter. eg, if r0 is 1k and resistance changes by 100 ohm, its a 10% change. If its 10k, and it changes by 100 ohm, its 1%.
 

Junus2012

Advanced Member level 4
Joined
Jan 9, 2012
Messages
1,297
Helped
45
Reputation
94
Reaction score
43
Trophy points
1,328
Location
Italy
Activity points
12,552
long channel length is related to both layout and the electrical properties of the current mirror.
for the electrical, long L will reduce the channel length modulation and hence increase the output resistance of the current mirror, as a result the current will be more stable.
for the physical (layout) properties, the large channel match better than the small channel because the variation on the wafer will averaged for the long channel while it is concentrated in the small channel, this make large different properties between the small channel transistors.

However we have ratio that is (W/L) and we have size (W.L), having long channel doesnt mean that you have small ratio because you can change the W as well.

for the current mirror matching you need large size transistors and small ratio to increase the overdrive voltage for better matching

- - - Updated - - -

I think Erikl you wanted to say larger Vds

And low W/L ratios achieve better simulation accuracy and lower Vds,sat , AFAIR.[/QUOTE]
 

sainiparvesh

Newbie level 6
Joined
Jul 1, 2009
Messages
14
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,281
Location
noida
Activity points
1,371
Hi,
For high speed circuits(>10Gbps) in tech. like 45nm or lower, we need to mirror current of the order of 3mA using programmable current steering DACs(which update at very low freq.) with the number of bits of the order of 7 or 8. Is it a good idea to have length of the main mirroring MOS of the order of 0.5um or even less with some cascade MOS with length 0.1um to save area? What else can be the smart ways to save area ?
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Top