The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> about constant current reference
https://designers-guide.org/forum/YaBB.pl?num=1152179400

Message started by caosl_zju on Jul 6^th, 2006, 2:50am

Title: about constant current reference
Post by caosl_zju on Jul 6^th, 2006, 2:50am

hi,all
I need a constant current reference in my circuit, it's value is about 1uA.I summed a PTAT current from bandgap and a CTAT current to achieve this target, it's TC performance is ok, but the resistance's value will departure 20%, so the current value will change at different corners.Anyone has an idea to design a current reference which only relevant to resisitance's ratio more than it's absolute value.Any advice is welcome,thanks!

Title: Re: about constant current reference
Post by Jerome on Jul 6^th, 2006, 5:41am

Hello,

To my knowledge, such a block does not exist (without trimming of course).
I remember searching for such a block through the IEEE litterature without
success. This was 4-5 years ago.

If you find one, I am interested.

Good luck, Jerome.

Title: Re: about constant current reference
Post by vivkr on Jul 6^th, 2006, 7:35am

Hi,

There is no fundamental current reference. So you will have to live with the variation, unless you can use an external R, or if you can trim out the process variation of the on-chip R. Both techniques will require use of extra pins.

Regards
Vivek

Title: Re: about constant current reference
Post by mikki33 on Jul 6^th, 2006, 12:44pm

The use of external resistor is a common practice. In some cases it requires only 1 pin (if your groungs inside the chip and out are practically the same). In this case you may use 1% accuracy external resistor and get 3-5% accuracy of the "constant" current.

Title: Re: about constant current reference
Post by ycj on Jul 11^th, 2006, 8:46pm

Several papers have targeted this issue.

1. Stephen Tang, Siva Narendra, Vivek De from Intel proposed a method for sub-1V operation and file a patent on it.

2. S. Sengupta, K. Saurabh, and P.E. Allen ( Yes, that P.E. Allen of GATech ) proposed another method on ISCAS 2004.

I myself tried Sengupta's method and got good result (+/-9%) in simulation, but failed in Stephen's method.

Title: Re: about constant current reference
Post by loose-electron on Aug 1^st, 2006, 7:54pm

To quote the cited article:

"Measurements on a prototype chip, implemented in a 150nm logic process technology, validate the theories for two sub-1V MOS reference current generator circuits and show that 2X reduction in current variation is achievable across extremes of both process and temperature"

Not too terribly impressive. 2X of what? 40% down to 20% or?

To the best of what I know, the other opinions stated prior are accurate. Accurate voltage, not a problem. Accurate current requires a trim method or an external reference resistance. If anyone knows of something else, I would be very interested.

Jerry

Title: Re: about constant current reference
Post by ycj on Aug 7^th, 2006, 11:09pm

loose-electron wrote on Aug 1^st, 2006, 7:54pm:

So you can try Sengupta's method if you don't like Stephen's.
It is true that accurate current source is difficult to be implemented without trim or external reference resistance. However, good current source still can be achieved with some circuit skills.

ycj

Title: Re: about constant current reference
Post by ywguo on Aug 15^th, 2006, 10:45pm

Hi guys,

Sometimes I wonder if we should seek constant current or not. For eg., a constant GBW and phase margin must be the final goal for designing the amplifier. However, GBW and phase margin vary as the PVT even if there is constant current reference.

If we try to stabilize GBW and phase margin as PVT vary, should we use a constant gm current source? But a constant gm current reference is inversely proportional to R. R varies with PVT. It's near 40% including 100 degree Celsius temperature variation and +/-20% resistance value variation due to process.

Any comments are appreciated.

Best regards,
Yawei

Title: Re: about constant current reference
Post by caosl_zju on Sep 4^th, 2006, 7:02pm

hi,ywguo,
There are some other situations need constant current,for instance,in oscillator,i need a current to charge the capacitor.If this current is not constant well,the frequency will be vary too much that cannot be accepted.