MC33771 cell measurement errors

electron_wrangler · ‎09-26-2023

I'm commissioning a build with the MC33771, and am noticing an offset of about +/- 20mV on all cell terminal measurements.

For example, I've built up an emulator board that I can tune a 'cell' to a specific voltage (let's say 3.700V), independently measured by a multimeter, and the reported measurement has a fairly consistent offset of closer to 20mV instead of the datasheet rated ~3-4mV [spec'ed error after soldering and "aging", even worst case 6mV for Verr_5a]. Some channels are reporting +20mV while others are reporting -20mV. If I measure next to the MC33771, I get measurements that match when I measure from the emulator board, so it's not an issue of voltage drop through any cabling (also true since I see cells that are reading +20mV of DMM measured).

What could explain this large deviation? I would think it could be a meter calibration problem, but we have two distinct DMM's that are seeming to agree with each other, and both are disagreeing with the MC33771 measurements.

electron_wrangler · ‎09-27-2023

Thanks for the message, Jozef.

We have capabilities built into our microcontroller to turn balancing on and off for individual cell bundles (with CBDRV_EN set to logic 1, raising bit 9 of the appropriate CBx_CFG register [$0C for CB1_CFG, etc] to turn on individual cell balancing) and when I do that I see about 100mV or more of drop, both reported from the MC33771 and on my DMM. After doing some of these tests I did see an individual cell reading within 5mV on one cell, but that ended up drifting back to about 20mV difference after some time.

So I'm not sure this completely explains what I'm seeing...

JozefKozon · ‎09-27-2023

Dear Brian,

this looks like some of the Cells are balancing. If I enable balancing for a Cell, than the voltage on the Cell drops approximately 20mV.

I enable the balancing for the Cell 1, the voltage drops approximately 30mV this time, but usually it is 20mV.

Please also note, that neighboring's Cell 2 voltage increases approximately 10mV.

With Best Regards,

Jozef

JozefKozon · ‎10-02-2023

Dear Brian,

please see below a reply from an application engineer I have contacted. Please answer his questions.

DESCRIPTION

1) can you exclude that the measured cell or one of the adjacent ones had closed balancing switch during the measurement?
2) which type of multimeter was used to measure the cell?
3) I couldn't grasp your argument about excluding errors introduced by connections, could you please provide more information?
4) last but not least, I'd like to get the measured value - with an accurate instrument having max 1 mV error - of the voltage across the adjacent CT pins belonging to a channel supposed to be inaccurate, and the corresponding measurement value provided by the MC33771.

With Best Regards,

Jozef

electron_wrangler · ‎10-03-2023

Hi Jozef,

To respond to the application engineer's questions:
1) I'm not sure what this is asking - to confirm that the nearby balancing switch(es) is(are) closed? If so, then I'll say that with the 20mV read error the balancing switches on all 14 cells are open - there is no balancing happening
2) I have both a Fluke 179 and a Rigol DM3058E that agree with each other within 1-2mV (both of which disagree with the reported cell measurement by up to about +/-20mV)
3) My statement was just to say that the error is within the cell controller, and not introduced by voltage drop over the harness connecting the emulated cells with the cell controller
4) With cell balancing off for every cell, and measuring directly at the copper floods labeled on the attached screenshot (Cell 1 reading is taken CT_REF to CT_1, Cell 2 is CT_1 to CT_2, etc.), I get the following data:

	Cell Controller reading	DM3058E reading
Cell 1:	4.050 V	4.0315 V
Cell 2:	4.003 V	3.9793 V
Cell 3:	4.007 V	3.9830 V

JozefKozon · ‎10-04-2023

Hi Brian,

thank you for your detailed answers. Please see a reply from the application engineer in the description. Please comment them. On how many MC33771 pieces have you seen this errors? Have you tried to replace the piece, which is showing this error for a new one? Do you see the same error on the new piece?

DESCRIPTION

that's very interesting, especially your answer to question 2) (two multimeters of different type used) and to question 4 (test performed) . I have questions about point 4):

Can we exclude there are voltage drops between measurement points and pins?
Were the tested boards new or already used in the field?
Were the large cell voltage errors detected while conducting a test?

If we can eventually exclude systematic errors or inaccuracy caused an abuse of the part -namely, exceeding max ratings- then I don't see any other option but the customer to open a CQC return.

With Best Regards,

Jozef

electron_wrangler · ‎10-12-2023

Sorry for the delayed response, I haven't been in office every day since last week.

We have 5 prototypes, and I was able to get 3 connected and checked against the multimeter. All 3 have approximately the same offset on the first few cells. I don't have new replacement cell controllers in house (these were built and assembled by a 3rd party vendor), but I can see about acquiring some spares to try that.

To reply to the application engineer,

A) there is a small drop from the measurement points at the cell to the pins, but there's still on the order of 15-20mV difference between reading on the copper attached to the pin and from the cell controller itself

B) they're new - we're still developing the product, so they were built and shipped from China to our office in New York

C) technically, yes. The test, however, was confirming the reading matched an independent measurement

Is it possible that the reflow process could have resulted in this deviation?

JozefKozon · ‎10-12-2023

Dear Brian,

can you please confirm, that all your MC33771 components were purchased from one of our authorized distributors? Please state which one. Please see this link for a list of our authorized distributors.

With Best Regards,

Jozef

electron_wrangler · ‎10-13-2023

It looks like we did purchase this lot through Arrow, we have a web order indicating the office through Centennial CO.

JozefKozon · ‎10-22-2023

Dear Brian,

I apologize for a late answer. Thank you for confirming the source of the MC33771 pieces. Please send the components back to the Arrow, of the vendor, from who you have purchased the pieces for analyzes by our quality engineers. The error is too large. Our engineers need to have a look at it.

The procedure is, that the authorized distributor will take the components, will let you fill a CQI (Customer Quality Incident) report and will send the components to NXP for analyzes.

With Best Regards,

Jozef