| Author |
Message |
jodenma
Joined: 21 Oct 2004
Posts: 48
Location: deep in IC
|
 16 Dec 2004 15:00 precise resistor how design? |
|
|
|
hi ,now i need design a precise resistor to get a precise voltage with VDD change. the VDD is 2.5v and V1=0.195V, V2=2.305V. i use serise resistor(total 250kohm )to get these two voltage.
the technolog of resistor is 500 ohm/square with +-10% error.
now how can i design the layout to get the precise voltage?
|
|
| Back to top |
|
 |
chinito
Joined: 05 Jul 2004
Posts: 133
Helped: 4
|
| 16 Dec 2004 18:38 Re: precise resistor how design? |
|
|
|
| If I were you, I'd layout the resistor with N-Well because sheet resistance of N-WELL is in k-ohms/square. Next, I'd do matching of these resistor (interdigitize them) so if one resistor varies the other varies by the same amount. So it's the ratio that's important to me not the exact values. I hope it doesn't confuse you. Thanks.
|
|
| Back to top |
|
|
serdes
Joined: 18 Nov 2004
Posts: 10
|
| 17 Dec 2004 8:54 Re: precise resistor how design? |
|
|
|
| Ratio but the accurate value is important under such condition.
|
|
| Back to top |
|
|
Humungus
Joined: 10 Jul 2001
Posts: 417
Helped: 25
|
| 17 Dec 2004 9:38 Re: precise resistor how design? |
|
|
|
If your using a resistive divider, only the ratio between resistors matters.
Go to the process specification sheet and get the matching parameter of the resistive material you are planning to use (say Ao). Then you compute the standard deviation of each resistor, if proper layout techniques are used, as
sigma=Ao/SQRT(W.L), where L is the total equivalent length of your resistor and W the stripe width.
Propagate the error to obtain the error of the resistors ratio.
Remember that the total error is 3 or 4 sigma (3 is generally enough).
Or, just determine the total error you con tollerate, and compute the W.L product that you need, under the constraint of your total resistor value.
|
|
| Back to top |
|
|
jodenma
Joined: 21 Oct 2004
Posts: 48
Location: deep in IC
|
| 17 Dec 2004 17:48 precise resistor how design? |
|
|
|
Thank all of you very much!
To Humungus: will you be so kind to tell me the specific meaning of "the matching parameter of the resisitive material" ? does it mean the +-10% error? can you tell some layout techiques of the matching resistor?
Thanks&Regards
|
|
| Back to top |
|
|
gevy
Joined: 17 Nov 2004
Posts: 377
Helped: 41
Location: Russia
|
| 18 Dec 2004 16:21 Re: precise resistor how design? |
|
|
|
The methods of matching resistors are in the book
Hastings "The Art of Analog Layout".
It is the very useful book about analog layout.
|
|
| Back to top |
|
|
Hughes
Joined: 10 Jun 2003
Posts: 712
Helped: 84
|
| 19 Dec 2004 2:02 Re: precise resistor how design? |
|
|
|
| jodenma wrote: |
Thank all of you very much!
To Humungus: will you be so kind to tell me the specific meaning of "the matching parameter of the resisitive material" ? does it mean the +-10% error? can you tell some layout techiques of the matching resistor?
|
The +-10% error isn't the matching parameter. It is the discrepancy between different wafer runs.
|
|
| Back to top |
|
|
lakeoffire
Joined: 10 Dec 2002
Posts: 54
Helped: 2
|
| 19 Dec 2004 7:28 precise resistor how design? |
|
|
|
| The error of the absolute value of resistors is out of your control and is determined by the process. However, the error of the ratio of two or more resistors can be significantly reduced by careful layout.
|
|
| Back to top |
|
|
terryssw
Joined: 18 Apr 2004
Posts: 176
Helped: 14
|
| 04 Jan 2005 17:50 Re: precise resistor how design? |
|
|
|
So, Can someone tell me that the matching parameters (Ao in previous posts) is determined under what conditions? When the resistor match very precisely (using common-centroid techniques), or just placing closer to each others?
For example,
Case 1:
in datasheets Ao=1.2% um is referred to just closely placed resistors? So that means if I use common centroid techniques better matching can be achieved (e.g. Ao=0.6% when using common-centroid)?
Case 2:
in datasheets Ao=1.2% um is already referred to Very precisely matched resistors? So that means this is the maximum matching performance that I can achieved?
Which case is corrected? Thanks for any replys!
|
|
| Back to top |
|
|
Colbhaidh
Joined: 10 Aug 2004
Posts: 151
Helped: 16
|
| 05 Jan 2005 11:33 Re: precise resistor how design? |
|
|
|
+/-10% is the absolute worst case wafer to wafer ; this week to next tuesday difference that the Foundry will guarantee. Within a single die location, resistors are probably within 1% especially if they are poly resistors arranged in a common centroid or other layout.
Anyway, always use ratios to avoid absolute matching issues.
|
|
| Back to top |
|
|
