low resistance shunt issue

Hello, we've developed our own current measurement circuit for use with a large battery pack. There are several issues. A sketch of the circuit is shown--the dotted box is the shunt measurement board, which includes an ADC (not shown).

The digital output (14 bit resolution with 8x PGA active) always seems to be fixed at a small value (corresponding to a few mV) no matter how much or how little current is flowing through the shunt. Originally I was thinking we could use the 1k pot to balance the resistor ladder to account for resistance uncertainty to the shunt (100k 1% is still a rather large error for a 0.5mOhm shunt). Now it's clear a person's hand is not steady enough and instead it's probably better to clean up in software.

Something I was suspecting was that the ADC input (MCP3234, which has 3.2pf switched capacitor) is changing the voltage at the sense node because the ladder is fairly high-impedence. So I put a large 47uF cap on the inputs; it seems to work a bit better on the bench but not in the system. The only thing I can come up with is that there may be a voltage difference between the ADC ground and the pack ground (but it is differential measurement so should not matter??)

Any recommendations to improve the circuit? Right now hacks are preferred. In retrospect I should have just used INA148 opamp frontend and called it a day.

Thanks so much

The application demands for a dedicated highside shunt measurement amplifier IC or any kind of self designed high side measurement with level translation, typically using a current source to bridge the 100V potential difference with low common mode effect.

To design it yourself, think of a low power rail-to-rail OP, supplied through a simple R-Z circuit and a high voltage PNP transistor for the current source.

0.5m ohm seems extremely low, I wonder what is the level of current to justify so low resistance.

Zetex AN39 Current measurement applications handbook
www.diodes.com/_files/products_appnote_pdfs/zetex/an39.pdf

Features and Applications of the ZDS1009 Current Mirror/Level Translator
www.diodes.com/_files/products_appnote_pdfs/zetex/an32.pdf

alexan_e said:

0.5m ohm seems extremely low, I wonder what is the level of current to justify so low resistance.

Zetex AN39 Current measurement applications handbook
www.diodes.com/_files/products_appnote_pdfs/zetex/an39.pdf

Features and Applications of the ZDS1009 Current Mirror/Level Translator
www.diodes.com/_files/products_appnote_pdfs/zetex/an32.pdf

Click to expand...

The application is lightweight solar-powered vehicle, we've seen anywhere from 10 to 50 amps through the shunt depending on motor load. Unfortunately for this application power efficiency is paramount...

What also makes this very difficult is that there are not many active components suitable for our 90 to 110V battery pack

Thanks a lot guys, I'm gonna check out those resources

0.5 mOhm sounds reasonable. I have used 0.2 mOhm for similar current levels.

Thanks Alex for mentioning the Zetex devices. I think that 120 V rating may be a somewhat too small voltage margin for a 110V battery. In addition, OP based high-side shunt sensors have partly better offset voltage and -drift performance. Their voltage range is limited as well, but you can at least use the basic cicruit as a template for your designs. See e.g. LTC6101 as an example