How to reduce drift after soldering shunt?

I have a crystalyte 24-72 20a controller.. I soldered the shut a bit and now I'm up to 36-38 amps.. My dp ca is calibrated using I charger and following methods instruction and help of the spreadsheet. I got it dialed in perfectly but Sometimes when I start up the controller my CA is displaying huge negative amps - 200+ and after after some undetermined amount of time 1 or so minutes it functions normally to a degree but this throws off my CA results which is unacceptable when you use your card as a gas Gauge.. Anyone care to provide insight on how to eliminate this.

Most likely a bad connection between the shunt and the CA. If you have a connector in the path, the contacts/wires may not be well-crimped/soldered, or they may not be making good contact with each other.

(I have this problem intermittently with my V3 CA and I think it is a wire *almost* broken off at one of the two shunt wire connections due to my poor handling of them; just a touch on the connector and it then works fine).

Im hitting this thread again. Not with more info but a better understanding of the problem.

I have a 72v 20A Crystalyte controller. According to my calculations by me soldering the shunt Ive bumped the current limit from 20 to 35-40A. This was my target.

The problem I'm having now is especially when the bike has been sitting cold when I turn the controller on the CA sometimes has a negative current showing. This throws off my readings cause the CA things its in regen and counts down the AH sometimes extremely fast. This is causing me to loose track of forward consumption data. I assume since I have a geared motor I can look at the FWD AH which should be pretty close to correct.

Ive consider getting a second meter such as a Watts up and putting it in my battery bag to monitor one string of my lipo so I can get a true reading when it comes down to figuring out how much AH I really have left.

I soldered the controller shunt up with Silver Bearing Solder. And this messes with the temperature coefficent of the shunt. So lets say I calibrated the shunt at 50F if i go check it again at 30-40F Im going to get a different reading. As the soldered shunt gets colder the resistance decreases. As it gets warmer it increases.

So in theory if I calibrate when its cold and it warms up I will have a tendency to have a positive drift where it counts off more but if I calibrate it when its warm and it gets cooler I would tend to get a negative drift I would assume. Zeroing the amps does temporarily solve this but is there something I can do like insulating the shunt so that the temperature stabilizes.

I also notice if I do a couple of intense rides up and down the street it immediately solves this issue without zeroing the meter but If I leave the controller on and walk away for a few moments after this there is a chance or possibility the current will start drifting again once the temp of the controller and the shunt drops back down to open case temp. Im considering maybe putting a cotton ball between the heat sink plate and shunt to stabilize these fluctuations a bit better what do you thing.

The more i surf the forum and read posts here and there the more i realize that I might not get the CA data to match perfectly so I maybe switching my DP CA to a SA or just run that backup meter on one of my pack sections.

I dont think soldering the shunt is a curse. If indeed it is variable because of the temp ... If im beating up on the controller and the motor and the shunt heats up the current drops down thus providing less current protecting the motor and the controller so thats a plus but im unsure of how much swing in voltage it is.

I will however test my theory on this temperature shunt value changing. I will wake up and calibrate the shunt one cold morning in the garage and then leave my by bike in a room with a heater going for a few hours to bring the temp up and check again and see what the % swing in the value is.

Proper shut wire is made from alloys that have a very low temperature coefficient (home much the resistance changes with temperature). Solder does not. Also you may be getting a thermocouple effect due to the dissimilar metals in contact. But I don't think that is the reason for what you are seeing... those effects should not produce a -200A amp reading.

Hey, seeing like 200A is almost always the result of a bad connection on one of the shunt leads, but seeing much small amounts (<0.2A usually) can happen from thermal effects, especially if you don't have the white and blue sense leads for the CA soldered RIGHT at the shunt itself, sometimes people think they can take advantage of the ground trace on the PCB and solder one of the taps to S- elsewhere on the board and that will totally mess up the low current offsets and amplify any drifts. Another big no-no that people do is combine the ground wire with one of the sense leads.


That said, even if you take care of the offset drift, the overal accuracy with a solder blobbed shunt is never going to be all that good, both due to the high temp-co of the additional metal and also the thermoelectric effect from the dissimilar metal junction. If you want a more precision setup, my real recommendation would be to instead get one of the precision 1.00 mOhm manganin shunts that we offer:
http://www.ebikes.ca/store/photos/Shunt_Raw.jpg


Then you can crimp this inline with the battery lead inside the controller housing, and you'll get a known and exact RShunt to use with the CA which won't be subject to thermal effects and calibration uncertainty like your solder blob.


-Justin

Well that shut me up. Of to ebikes Ca to get a shunt... I wonder if the shunt will fit in an Anderson Powerpole Crimper. It looks like it does.

I did a quick ride today after taking my controller apart and checking everything. The solder joints are solid for all the Ca connections. Put it back together took a quick ride and i didnt see any issues. I dunno maybe this whole thing was a fluke.

For shunt mod I prefer the copper wrap technique...ie wrap a portion of the shunt's length with thin strand copper and soak some solder into it. The point is to make a portion of the shunt length have effectively 0 resistance. It's easier to be reasonably accurate, and more importantly, no big blobs of solder that easily melt and cause problems. You also won't get the shunt value drift with temperature. Sure copper resistance varies with temp, but even big % changes of have little effect due to such low resistance.

John

Yes. Best to make a mechanical connection, and then solder to keep contact area hermetic and less sensitive to stress / vibration / corrosion.

Sounds to me you may have a cold / disturbed solder joint. Silver will greatly raise the melting temps most times, and require much more heat from the iron. Use eutectic solder 63/37 to minimize the slushy phase when cooling /heating it.

I have posted the 4W surface mount shunts that I use. They come in a variety of flavors from miliohms to hundreds of microohms. My choice is to solder one of those in parallel with the existing shunt on the back of the board. This increases power handling, has no negative temperature effects, and is reasonably calculable. Exact current limits are over-rated and can be tuned with the CA-V3 if desired.

Remember to calculate the constant and dynamic current that your shunt will be conducting so that you can assure you dont have heating inaccuracies. On the 24 fet controllers we target 500uOhms and usually hit between 470 and 520 without trying.

8W = I^2 * 0.5mOhms

That is good for about 100A continuous - which is way more than any ebike will actually draw. The 8W is the sum of the power handling of the existing shunts and the new shunt.

We never see negative currents or drift currents.

-methods

So Methy, that would be one of these then that you solder in parellel? (Searching through your posts, I couldn't find a part number for the actual shunts you use)

http://www.digikey.com.au/product-detail/en/CSS4527FT1L00/CSS4527FT1L00CT-ND/3883178
http://www.seielect.com/catalog/SEI-CSS_CSSH.PDF

So, the standard shunt wires work out at around 1milliohm, and you get your targeted 500microohm usually by parelleling another 1milliohm?

Those are similar but not the exact ones. The units I use are 4W
Search digikey for Resistor/surface mount/4W
http://www.digikey.com.au/product-detail/en/CSS2725FT1L00/CSS2725FT1L00CT-ND/1923237

I dont like throwing around numbers about shunts in Infineon type controllers. There are a dozen shapes for those shunt wires the Chinese builders like to use and they can populate 1, 2, 3, or 4 of them. I have seen anything from 1mohm to 5mili ohms.

Generally speaking tho - most 24 fet controllers come with a 1mOhm shunt. If memory serves we put about 1mOhm in parallel with that to come up with apx 500uOhms.

There is nothing special about those particular shunts other than that back in the day they used to fit nicely right under the stock shunt, they are reasonably priced and available, they have good power handling, a low profile (to fit under the board), and come in many flavors down to 200uOhms.

If anyone is serious about modding controllers there is no reason not to "do it right".
One should have on hand a few shunts, some nice 250V ceramic caps, some mylar caps, some lower ESR 100V 1000uF caps, some good cabling and connectors, heat transfer paste, board cleaner (like heavy duty flux remover), some silicone conformal coating... yea... ok... I guess not everybody wants to get that geared up - but then I have probably modded 100 controllers so...

-methods

EDIT: I see the datasheet you posted actually includes the shunt series that I use. So yea - those are the ones.

but the 4w units are only rated to 31.62A continuous and 63.25 peak....

Just for info if you hadn't seen it, I presented a different way to manipulate the shunt issue, and get precise tuning without "soldering" the existing shunt. This thread has the specifics:

http://endless-sphere.com/forums/viewtopic.php?p=494927#p494927

boostjuice said:

but the 4w units are only rated to 31.62A continuous and 63.25 peak....

Click to expand...

4W
31.62 ^2 * 4mohm = 4W

For that row you selected the smallest shunt they have is 4mohms so the highest current is in fact 31.62A
Please look up 2 rows at the CSS2725 series

Range 250uOHms to 3mOHms
Cont current is 126A for the 250uOhm shunt

100A*100A*1mohm = 10W
That power is split between the existing shunt and the new surface mount

You have to calculate the power - not just look at the data sheet because that number on the data sheet (for starters) only relates to the lowest shunt value... and worse it assumes perfect heat sinking.

-methods

That is a clever modification. The only thing I don't like about it is that it does not increase power handling. Many of the 12fet and 18 fet controllers start with only 2W or 3W and often times we significantly increase the current.

Back in the day Doc experimented with bringing out a pot for real time adjustment. Dont remember why but it was fail. Could have been the wrong circuit or could have been noise - not sure.

-methods

bigmoose said:

Just for info if you hadn't seen it, I presented a different way to manipulate the shunt issue, and get precise tuning without "soldering" the existing shunt. This thread has the specifics:

http://endless-sphere.com/forums/viewtopic.php?p=494927#p494927

Click to expand...

Just reporting back. My calibration is pretty much dead on its right at 2% over which is totally fine with me Basically my CA says I drained 2.604 out my pack and my Icharger was putting in roughly between 2.555 to 2.525 into my pack sub sections. So at the end of my ride When I think Ive drained 20AH out of m pack i will prob be around 19.5AH so there is a safety cushion built in along with the LVC designated for my pack we should be golden. I still have shunts from Ebikes CA on the way. Not sure If im going to use them tho.

It always dives me crazy when someone calibrates a shunt in that way I guess it is a pretty reasonable way to do it - I am just bugged out about calibration.

If you have an iCharger then you should be able to calibrate it tits-on. I know you know the procedure - but it is about time to repeat it for the new folks.

How to calibrate the shunt value in your CA using a constant current source:

Collect an iCharger
Choose foam cut from the special modes (constant current)
Set the current as high as it will go - 20A or 30A
Set the voltage to something nominal - like 3V

Devise a way to Y into your controller ground wire
Hook your controller to your pack (remove the motor)
Turn the controller on with the CA hooked up

Hook the iCharger red wire to the controller ground Y (between the pack and charger)
Hook the iCharger black wire to one of the three phase wires (which are not connected to a motor - I always pick yellow)
Turn the iCharger on

As soon as the icharger settles on 20A or 30A write down the value on the CA
Shut off the iCharger

Go into the CA and find the shunt setting

Go find the CA Shunt Calculating Excel spreadsheet that is bobbing around here someplace
Enter the iCharger setting in the first column
Enter what the CA reads in the second column
Enter the CA shunt value in the red box
Then - enter the value that is calculated (the new shunt value) into the CA

Repeat the test and the CA should read +/- 5% or 10%

For increased accuracy use a Fluke DMM inline and keep the currents below 10A

What is happening is that the iCharger is driving a constant current in through the ground of the controller, through the shunt, up through the source of the lower mosfets, through their body diodes, out their drains, and out the phase wires. This is a perfect way to drive a known current through the shunt without taking the controller apart, without poking around taking bad voltage measurements with a DMM, etc. Originally posted by Justin - he suggested using a C battery

-methods

-methods

methods said:

It always dives me crazy when someone calibrates a shunt in that way I guess it is a pretty reasonable way to do it - I am just bugged out about calibration.

If you have an iCharger then you should be able to calibrate it tits-on. I know you know the procedure - but it is about time to repeat it for the new folks.

How to calibrate the shunt value in your CA using a constant current source:

Collect an iCharger
Choose foam cut from the special modes (constant current)
Set the current as high as it will go - 20A or 30A
Set the voltage to something nominal - like 3V

Devise a way to Y into your controller ground wire
Hook your controller to your pack (remove the motor)
Turn the controller on with the CA hooked up

Hook the iCharger red wire to the controller ground Y (between the pack and charger)
Hook the iCharger black wire to one of the three phase wires (which are not connected to a motor - I always pick yellow)
Turn the iCharger on

As soon as the icharger settles on 20A or 30A write down the value on the CA
Shut off the iCharger

Go into the CA and find the shunt setting

Go find the CA Shunt Calculating Excel spreadsheet that is bobbing around here someplace
Enter the iCharger setting in the first column
Enter what the CA reads in the second column
Enter the CA shunt value in the red box
Then - enter the value that is calculated (the new shunt value) into the CA

Repeat the test and the CA should read +/- 5% or 10%

For increased accuracy use a Fluke DMM inline and keep the currents below 10A

What is happening is that the iCharger is driving a constant current in through the ground of the controller, through the shunt, up through the source of the lower mosfets, through their body diodes, out their drains, and out the phase wires. This is a perfect way to drive a known current through the shunt without taking the controller apart, without poking around taking bad voltage measurements with a DMM, etc. Originally posted by Justin - he suggested using a C battery

-methods

-methods

Click to expand...

It was pre calibrated on the Icharger in foam cut and I used the cheater spreadsheet. I did follow your instructions O Mighty Methods and got it within 2%. I just used the Icharger to confirm.

I know you did - I was just posting it again for others who are fumbling about. Signal to noise ratio getting higher and higher on this forum (and i am totally guilty) so I make up for it by spewing out random useful bits of information from time to time to keep them near the top. Stickies do not work - they are the last thing I ever look at on a forum... search doesn't work.... spoon feeding doesnt work... the forum is the most awesome thing that has ever happened except for the S/N ratio and the propensity for outdated information to continue getting propagated.

-methods

Thats' why we started the wiki:
http://www.endless-sphere.com/w/index.php/Main_Page
If you have a moment, you could make a quick article on how to calibrate the shunt in a controller.