Designer’s Guide Community

ywguo

Community Fellow

Offline

Posts: 943
Shanghai, PRC

offset trim for opamp
Jul 12^th, 2008, 9:55am

Hello Guys,

I wanna trim the offset voltage for an opamp. The target is to make the input refered offset voltage less than 10~20uV. One possible way is to trim a resistive load, the other way is to trim a active current load. For both ways, I think that it is necessary to monitor the differential inputs of that opamp. Am I right? we need a very high precision comparator that can resolve such small difference, say 10uV, between those differential inputs.

If my understanding is right, how do I design such a comparator? Any literatures, any comments are appreciated.

Best regards,
Yawei

Back to top

IP Logged

HdrChopper

Community Fellow

Offline

Posts: 493

Re: offset trim for opamp
Reply #1 - Jul 12^th, 2008, 6:15pm

Hi Yawei,

What is the closed loop gain of your opamp? If you have a closed loop gain higher than 1x then it will be much easier to monitor the output offset voltage, since the input one will simply be the output offset divided by the closed loop gain. Therefore you are relaxing the resolution requirement for your comparator proportionally to the closed loop gain.

In case your amp gains 1x then the story is different. You will certainly need such high resolution comparator. I�m tempted to say you could chop the comparator for achieving such resolution, but if that was an option for you it would be simpler to directly chop the opamp...

Regards
Tosei

Back to top

Keep it simple

IP Logged

ywguo

Community Fellow

Offline

Posts: 943
Shanghai, PRC

Re: offset trim for opamp
Reply #2 - Jul 13^th, 2008, 2:03am

Hi Tosei,

Thank you for your kindly comments. I am sorry not to describe the requirement clearly. So I have to clarify the application here.

That is of 1x closed-loop gain. Sad

It is differential input and single-ended output. The last is that is a continous-time buffer, where the chopper technique can not be adopted.

Best regards,
Yawei

Back to top

IP Logged

Berti Community Fellow Offline Posts: 356	Re: offset trim for opamp Reply #3 - Jul 13^th, 2008, 10:36pm Just an idea: Can't you trim the offset disabling the feedback of the OTA (open-loop) and close it for operation. Cheers
Back to top	IP Logged

vivkr

Community Fellow

Offline

Posts: 780

Re: offset trim for opamp
Reply #4 - Jul 13^th, 2008, 11:39pm

Hi Yawei,

10-20 uV is a really tough spec.

The comparator scheme has no chance in this case because obviously, the same problem pops up in the
comparator. I remember reading some JSSC papers where they did achieve around 50 uV but I cannot find them. I will post the refs. I if do. I think the papers were from Quiting Huang et. al.

Trimming by itself is unlikely to help. There is always going to be the temperature-dependent offset drift which will be enough in most cases to scuttle your performance.

By the way, chopping is a technique that is used in continuous-time circuits. Usually, the upconverted flicker noise and offset are not of concern anymore, but if so, then you either configure your chopping in a way that the circuit in which the opamp is used automatically provides some filtering of this noise, or you build a dedicated RC filter. However, depending on the exact system requirements on your filter, you may sometimes not be able to use chopping. Just think about it.

By the way, do you have high quality BJTs available to you? And can you use these in your opamp (depends on input impedance requirements)? If so, then it would be far easier to reduce both the offset and the offset drift using trimming or other techniques in BJT opamps.

I will search for the references.

Regards
Vivek

Back to top

IP Logged

HdrChopper

Community Fellow

Offline

Posts: 493

Re: offset trim for opamp
Reply #5 - Jul 14^th, 2008, 4:00pm

Hi Yawei,

I agree with vivek's suggestions. Please consider chopping as an option (if you already did not do it) since it is really effective in removing offset and low frequency noise. Offset levels you are looking for are achievable by means of offset.

As for the BJT approach vivek also suggested, please take into account that although smaller input referred offset combined with trimming certainly is a better option than a pure CMOS opamp, drift might still hurt you since you cannot trim it (chopper does remove it too though).

Regards
Tosei

Back to top

Keep it simple

IP Logged

HdrChopper Community Fellow Offline Posts: 493	Re: offset trim for opamp Reply #6 - Jul 14^th, 2008, 4:01pm HdrChopper wrote on Jul 14^th, 2008, 4:00pm: Hi Offset levels you are looking for are achievable by means of offset. I'm sorry, I meant by means of chopping...
Back to top	Keep it simple IP Logged

vivkr

Community Fellow

Offline

Posts: 780

Re: offset trim for opamp
Reply #7 - Jul 14^th, 2008, 11:41pm

Hi Yawei,

A point which I think is not clear from my previous post:

If you are using a BJT input pair for your opamp, and you trim out the offset by whatever means (adjusting resistor, current etc.), then you will also have minimized the offset drift at the same time. This special property exists only for the BJT.

Regards
Vivek

Back to top

IP Logged

vivkr Community Fellow Offline Posts: 780	Re: offset trim for opamp Reply #8 - Jul 15^th, 2008, 2:30am Hi Yawei, For a good example application, please refer to the JSSC paper by Burt & Zhang (Dec. '06). Regards Vivek
Back to top	IP Logged

HdrChopper

Community Fellow

Offline

Posts: 493

Re: offset trim for opamp
Reply #9 - Jul 15^th, 2008, 6:37am

vivkr wrote on Jul 14^th, 2008, 11:41pm:

Hi Yawei,

A point which I think is not clear from my previous post:

If you are using a BJT input pair for your opamp, and you trim out the offset by whatever means (adjusting resistor, current etc.), then you will also have minimized the offset drift at the same time. This special property exists only for the BJT.

Regards
Vivek

Hi Vivek,

What you suggested as for the offset drift is true for a bipolar diff pair. However one must be very careful in order to guarantee that happens.
For example a bipolar pair with a resistive load will definitely have an offset contribution from the resistors. That will introduce a different
offset drift component which certainly be different than the one contributed by the bipolars. This is even
worse if a MOS active load is considered.
Therefore, the only way to guarantee the offset drift will be totally proportional to the offset value at the reference temperature is by using
a bipolar active load.
In this way "purely-vbe" related offsets are involved and therefore their drift can be predicted and minimized.

Regards
Tosei

Back to top

Keep it simple

IP Logged

ywguo

Community Fellow

Offline

Posts: 943
Shanghai, PRC

Re: offset trim for opamp
Reply #10 - Jul 15^th, 2008, 7:11am

Hi Vivek, Tosei, and Berti,

Thank you very much. Your suggestions throw light on my topic.

I study the application again. That amplifier is used as an output buffer for a DAC. The offset is to be minimized to reduce the offset current. However, at least the offset voltage does not affect the linearity, then it is not necessary to reduce it to 10~20 uV. It is enough to be of 1~2 mV offset voltage. Then the design constraint is relaxed very much. Thank you again.

Yawei

Back to top

IP Logged

vivkr

Community Fellow

Offline

Posts: 780

Re: offset trim for opamp
Reply #11 - Jul 15^th, 2008, 7:45am

HdrChopper wrote on Jul 15^th, 2008, 6:37am:

[

Hi Vivek,

What you suggested as for the offset drift is true for a bipolar diff pair. However one must be very careful in order to guarantee that happens.
For example a bipolar pair with a resistive load will definitely have an offset contribution from the resistors. That will introduce a different
offset drift component which certainly be different than the one contributed by the bipolars. This is even
worse if a MOS active load is considered.
Therefore, the only way to guarantee the offset drift will be totally proportional to the offset value at the reference temperature is by using
a bipolar active load.
In this way "purely-vbe" related offsets are involved and therefore their drift can be predicted and minimized.

Regards
Tosei

Hi Tosei,

I agree that the drift due to resistors will be likely different, and needs to be accounted for, but I would be wary of using a BJT active load, simply because the noise performance will be terrible. Whatever you do, the load will contribute as much noise at the input of the amplifier as the input devices, a very bad situation.

Regards
Vivek

Back to top

IP Logged

HdrChopper Community Fellow Offline Posts: 493	Re: offset trim for opamp Reply #12 - Jul 15^th, 2008, 7:21pm Hi Vivek, I agree and my point was that if offset and offset drift was the main concern, the best way to predict its drift and at the same time minimize it by trimming it is by using bipolar active load, at the expense of noise as you stated Certainly offset vs noise performance is the trade off involved here. Regards Tosei
Back to top	Keep it simple IP Logged