Designer’s Guide Community

newic

Senior Member

Offline

Posts: 138

Stable Iref across PVT
Jun 09^th, 2010, 2:33am

Want to generate a stable across PVT Iref to opamp and other circuits.
What are the alternative?

Bandgap reference Vref?
However, it is still sensitive to Vdd unless voltage regulator is used, right?

The generated Vref from bandgap is stable across temperature (Vref variation=+/-0.5%). However, when converting vref to current Iref, the Iref becomes sensitive to temperature due to Vt variation, resulting in about +/-5% Iref variation or total 10%.

Any guidance?

Back to top

IP Logged

XY-oriented Junior Member Offline Posts: 23	Re: Stable Iref across PVT Reply #1 - Jun 9^th, 2010, 2:39am I would say +/-5% is pretty decent ... your design should be robust enough to stand 10% bias current variation. any circuit that can� t stand +/-10% bias variation is a bad one in my opinion.
Back to top	IP Logged

newic Senior Member Offline Posts: 138	Re: Stable Iref across PVT Reply #2 - Jun 9^th, 2010, 2:42am that is just the temperature variation. it does not account for the process and voltage variations.
Back to top	IP Logged

XY-oriented Junior Member Offline Posts: 23	Re: Stable Iref across PVT Reply #3 - Jun 9^th, 2010, 3:02am then you should think of trimming your bandgap so to limit process variation to +/-5%
Back to top	IP Logged

AnalogDE Senior Member Offline Posts: 137	Re: Stable Iref across PVT Reply #4 - Jun 9^th, 2010, 9:16am Just curious, how are you doing a BGVref to Iref conversion?
Back to top	IP Logged

Mayank

Community Fellow

Offline

Posts: 334

Re: Stable Iref across PVT
Reply #5 - Jun 9^th, 2010, 9:27pm

For Temp,
Instead of making a BGap Voltage Reference, generate a Zero Temp. Coefficient Current from BandGap Core.
For Process,
Trim BG resistors to get desired current value.
For Voltage,
Keep PSR of the Loop High Enough. V variation should not be a problem then.

Result : - You get A Current Ref. independent of PVT variation.

--
Mayank.

Back to top

IP Logged

newic Senior Member Offline Posts: 138	Re: Stable Iref across PVT Reply #6 - Jun 9^th, 2010, 11:56pm Hi Mayank, it makes sense to make deriv_Iref/deriv_temp = 0 instead of deriv_Vref/deriv_temp = 0. what if I need both Vref & Iref independent PVT?
Back to top	IP Logged

Mayank Community Fellow Offline Posts: 334	Re: Stable Iref across PVT Reply #7 - Jun 10^th, 2010, 1:11am Quote: Hi Mayank, it makes sense to make deriv_Iref/deriv_temp = 0 instead of deriv_Vref/deriv_temp = 0. what if I need both Vref & Iref independent PVT? Use Different Circuits. That's the Common practice
Back to top	IP Logged

newic Senior Member Offline Posts: 138	Re: Stable Iref across PVT Reply #8 - Jun 10^th, 2010, 9:59pm Use BG Vref to convert to Iref. Is it a common practice?
Back to top	IP Logged

newic Senior Member Offline Posts: 138	Re: Stable Iref across PVT Reply #9 - Jun 10^th, 2010, 11:53pm how to trim the resistor? adding a fuse? (resistor that parallel with a transistor)
Back to top	IP Logged

Mayank

Community Fellow

Offline

Posts: 334

Re: Stable Iref across PVT
Reply #10 - Jun 11^th, 2010, 12:02am

Quote:

Use Different Circuits. That's the Common practice

Be advised.
If you want to use BG_VREF to create I_REF, it is difficult to get exact temperature compensation, leave alone process. For BG_VREF / Res technique, you require Res. trimming to eliminate process variation.

Better Sol. --
Build a BG_VREF BandGap Circuit with Zero Temperature Coefficient Voltage. (Generally, PTAT current * CTAT resistance = ZTC voltage)
Build another IREF BandGap Circuit with Zero Temp. Coeff. CURRENT, not ZTC voltage. (Generally, CTAT current +PTAT current= ZTC current )

Then apply trimming after fab to get precise PVT compensated VREF & IREF.

--
Mayank

Back to top

IP Logged

newic Senior Member Offline Posts: 138	Re: Stable Iref across PVT Reply #11 - Jun 11^th, 2010, 12:13am Thank you very much!
Back to top	IP Logged

newic

Senior Member

Offline

Posts: 138

Re: Stable Iref across PVT
Reply #12 - Jun 15^th, 2010, 10:16pm

Quote:

Build another IREF BandGap Circuit with Zero Temp. Coeff. CURRENT, not ZTC voltage

What is the normal current value in the bandgap core and Iref?

Normally the core current is very small <50uA. I read from papers, is it true? will it suspectible to interference?

How about Iref value as a global current distribution.

Back to top

IP Logged

Mayank

Community Fellow

Offline

Posts: 334

Re: Stable Iref across PVT
Reply #13 - Jun 16^th, 2010, 2:41am

Quote:

What is the normal current value in the bandgap core and Iref?

Inside BGR core -- < 50uA
Inside Current References core -- < 100uA

Quote:

Normally the core current is very small <50uA. I read from papers, is it true? will it suspectible to interference?

Current being small is true. Layout of these blocks is done with extreme care which takes care of averting susceptibility to substrate noise, etc.

Quote:

How about Iref value as a global current distribution.

As raja quoted somewhere, dont keep current multiplication ratios bigger than 100x. It increases the noise in the current.

--
mayank.

Back to top

IP Logged

RobG

Community Fellow

Offline

Posts: 570
Bozeman, MT

Re: Stable Iref across PVT
Reply #14 - Jun 16^th, 2010, 2:12pm

Mayank wrote on Jun 11^th, 2010, 12:02am:

Quote:

Use Different Circuits. That's the Common practice

Be advised.
If you want to use BG_VREF to create I_REF, it is difficult to get exact temperature compensation, leave alone process. For BG_VREF / Res technique, you require Res. trimming to eliminate process variation.

Better Sol. --
Build a BG_VREF BandGap Circuit with Zero Temperature Coefficient Voltage. � (Generally, PTAT current * CTAT resistance = ZTC voltage)
Build another IREF BandGap Circuit with Zero Temp. Coeff. CURRENT, not ZTC voltage. (Generally, CTAT current +PTAT current= ZTC current )

Then apply trimming after fab to get precise PVT compensated VREF & IREF.

--
Mayank

FWIW, you can do it with just a voltage source and a ZTC resistor. if you don't have a ZTC Resistor available you can create one by using two resistances of opposite temp co. You also get less process variation using two resistances, assuming they are statistically independent. For giggles I wrote up a paper on how to do it and maintain the same temp co even if the two sheet resistances change in opposite directions. http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.76.2458&rep=rep1&type=p...

But like you say, the big issue is the variation in sheet resistance, which you will need to trim out or use an external resistor.

Back to top

IP Logged