KS14C Battery Charge/Discharge Curves

[DaveThomasPilot]

My 14C control board isn't measuring battery voltage accurately (see post about 62 year old fat guy for details).  Jason is having problems getting a good replacement board from KS.  While we're waiting, I want to ride my wheel.

But, I worry that since the wheel doesn't measure the battery voltage accurately, it could quit without the tilt-back warning if the battery voltage gets too low.  So, I decided to use the Charge Doctor to measure voltage as the battery discharges to something near the cut-off voltage and compare that voltage to what the wheel reports through the Android app.  I figured at least I could poll the app while riding to determine when I should get off!

First, I needing something for a load, so I bought some cheap light bulb holders and used 300 watt light bulbs:

 

 

 

I needed to connect the batteries to the Charger Doctor input (where the charger would normally attach) instead of its output, so I had to make an adapter.  I made two, so I could not only discharge the battery, but also drive the wheel's control board.  That way I could check when the wheel stopped tilting due to over-voltage and started tilting due to under-voltage.

Reached max file size..., so picture in next post.

 

 

[DaveThomasPilot]

I charged the batteries until the green LED came on, then measured a discharge curve.  The Charge Doctor website says a typical wheel cut-off voltage is 55 volts, so as the battery voltage approached 55 volts I decreased the load by unscrewing light bulbs. That way, I could get closer to 55 volts and eliminate more of the drop caused by the batteries internal resistance.

 

At 55.5 volts, the app reported 58.6 volts--more than 3 volts difference!

I then reconfigured the Charger Doctor for use with the fast charger Jason supplied with the wheel and got a battery charge curve to a full charge:

Interestingly, the charger put in less charge (540 watt-hours)  that what I took out (over 600 watt-hours!  Not sure what that's about.  I'll have to think about it.

The voltage difference between what the app reports and what the Charger Doctor reports (Charge Doctor correlates well with two hand held DVMS) is not constant versus battery voltage.  It varies from a little under two volts at full charge to over three volts when the battery is around 55 volts.

I used the deltaV and deltaI that resulted when the bulbs were unplugged to measure battery resistance at near 55 volts.  Also, used the difference in voltage at full charge when the load was first supplied.

Some questions:

1)  Does anybody know the voltages KS uses in the 14C to initiate tilt-back (over voltage and minimum voltage)?

2)  Is there something like low pass filtering used on the instantaneous battery voltage?  If so, time constant?  Is instantaneous (or filtered) current also used?

3)  Given the above discharge voltage and the wheel safety as the battery voltage decays, at what voltage should I stop riding?

Thanks!

 

[DaveThomasPilot]

Rationalized the difference between the charge and discharge curves...

I used the wheel's charge port when acquiring the charge curve.  So, the Charge Doctor was measuring the battery voltage AND the drop over BMS protection diode.  So, maybe charge was probably terminated prematurely. 

The charge plot shows 67.5 volts when the current is decreased to near zero.  The current is about 325 ma at that point. 

So, perhaps the charger cut-off prematurely, or at least at a different voltage than the first charge.  I saved the curve from the first charge too, so I'll review that.

[hobby16]

2 hours ago, DaveThomasPilot said:

At 55.5 volts, the app reported 58.6 volts--more than 3 volts difference!

It's probably the 14C control board's voltmeter function which has not been correctly scaled.

 

2 hours ago, DaveThomasPilot said:

Interestingly, the charger put in less charge (540 watt-hours)  that what I took out (over 600 watt-hours!  Not sure what that's about.  I'll have to think about it.
 

The difference is rather big and weird. Several causes :

- charging has been terminated a bit early. As you said, waiting until current drops to 0A would add some Wh to the total

- difference in discharge and charge currents accounts somewhat for the difference in Wh because energy loss (in the wires and inside the cells) is not the same depending on the current. I have seen it when comparing charges at 2A then at 4A.

- when charging at high currents, expect that the battery needs some "relaxation" time. If you let it rest some hour and plug the charger again, you can put in some more Wh.

 

2 hours ago, DaveThomasPilot said:

Some questions:

1)  Does anybody know the voltages KS uses in the 14C to initiate tilt-back (over voltage and minimum voltage)?

2)  Is there something like low pass filtering used on the instantaneous battery voltage?  If so, time constant?  Is instantaneous (or filtered) current also used?

3)  Given the above discharge voltage and the wheel safety as the battery voltage decays, at what voltage should I stop riding?

The tilt-back threshold is probably not much usefull here because your board does not measure correctly voltage anyway. You can detect the tilt-back by discharging down to 56V then ride your wheel back and forth on the same place until the tilt-back occurs.

As to the low voltage detection, from what I know on my wheels (GW, Firewheel, Airwheel), it is based on the instantaneous voltage, with no real filtering:  a big acceleration (when the battery is small or near empty, that causes a rapid votage drop) is enough to trigger instantaneously detection (tiltback on firewheel, beeps on GW). The stop voltage is very dependent on how you ride. If you keep the speed low and constant, the wheel will have hard time detecting any voltage drop as low threshold so you can ride very long time. If the ride is fast and/or with many acceleration, the tilt back will occur much sooner. Riders with many hours on their wheels can easily sense when they need to reduce speed for more safety because when the battery is emptying, the wheel becomes less powerfull and less responsive to accelerations. Anyway, your 14C has a quite big battery so it should be easy to find the max & min voltage to ride safely while keeping a decent range.

 

[DaveThomasPilot]

Your comments about the differences in charge and discharge curves make sense.  I checked the accuracy of the Charge Doctor voltage measurement using DVMs, but not the current.  However, given how close the charge current was to an even 5 amps, gives me confidence in the current measurement accuracy. 

I didn't check the math for running integral of volts * amps for the watt-hours, but I'm sure you got that right.

I'm very happy with the Charge Doctor!

Quote

The tilt-back threshold is probably not much usefull here because your board does not measure correctly voltage anyway. You can detect the tilt-back by discharging down to 56V then ride your wheel back and forth on the same place until the tilt-back occurs.

Agree.  But, the idea is to test the tilt back thresholds on a replacement control board

Quote

Riders with many hours on their wheels can easily sense when they need to reduce speed for more safety because when the battery is emptying,

Good to know.  That says you don't really need a "fuel gauge" once you have enough experience?  But, besides being just a beginner, I'm trying to keep my speed down to just above walking speed.  I really don't trust the wheel's overall reliability when there's what's probably a design issue with an unknown (to me) root cause.  I'm not sure one could detect battery voltage at slow speeds?

 

[esaj]

On 5/9/2017 at 2:36 PM, DaveThomasPilot said:

Good to know.  That says you don't really need a "fuel gauge" once you have enough experience?  But, besides being just a beginner, I'm trying to keep my speed down to just above walking speed.  I really don't trust the wheel's overall reliability when there's what's probably a design issue with an unknown (to me) root cause.  I'm not sure one could detect battery voltage at slow speeds?

I'd say that once you know (at least roughly) the range of your wheel, the battery gauge becomes less important, but it's still nice to know how much juice there's left now and then, especially if riding less familiar routes and not being that sure how many kilometers you've ridden. Also the consumption may vary depending on conditions (temperature, headwind, terrain) and ride-style. Even though the Firewheel would let the battery voltage (momentarily) drop to about somewhere around 46...48V before forcing you to stop, with my weight, I never felt it was less powerful/more "sluggish" in acceleration, until there was just a kilometer or two at max left in the wheel, which could leave me stranded far away with a 15kg wheel to carry   I've always thought the "percentage" gauges aren't that useful in estimating how long I could still go, especially with the Firewheel's original 00-99% two-digit percentage gauge, which towards the end (after 15+km or so, I had about 25km max range on the original batteries on good conditions, but less than 20km with heavy headwinds) could jump up tens of percents while descending a hill or braking hard and could show 0% at steady speed on level ground for 5km or even more before really running out of juice, and even after that, you could usually wait a while for the battery to "recuperate" a bit and then ride on a little more after the battery warning. Once I ended up carrying the damn thing in full protection gear on a hot summer day...

So what I did was to add a separate voltage/current -meter to my Firewheel at one point:

The white thing hanging from the charge port is an old Charge Doctor (V1), in this case the battery BMSs didn't have protection diodes, so the voltage could be read from the charge port also.   The current was measured by a large metal shunt I had attached on the mainboard steel-plate. The current measurement was not really useful in the end, I doubt it was really accurate (0-50A detected from a very small voltage drop of the shunt)  and would go up and down all the time while riding, so reading it was a bit too difficult. Knowing the voltages where the wheel tells you to stop, and (at least roughly) how fast it starts dropping towards the end, I had (IMHO) a much more reliable way of telling how much juice was left (of course to get a real accurate number I'd have to stop and let the voltage "settle" a while).

I didn't dare trying to power the meter from the board itself at that point (especially since I have at least one board that blew it's step-down circuitry on it's own), so there was a small box (the see-through -one with red button) to power it from a separate 9V battery. I think there are the kinds that can be powered directly from higher voltage lines, so those could run straight off the wheel battery.

Later on I did this:

It's just a small 100V max voltage display, powered directly from the battery (from the charge port in this case) through a TL783 regulator, which can handle up to 125V max difference between input and output ( http://www.ti.com/lit/ds/symlink/tl783.pdf ). I don't remember what voltage exactly I took out from it, probably something like 12-15V? The display should work between about 5 and 30V, if I remember correctly. Even though the regulator's linear, without heatsink and dropping up to around 55V, because the display draws only about 17mA (conveniently right above the minimum 15mA recommended for steady regulation), the power dissipation hasn't been a problem... the datasheets claim between 19-25 degrees C / W junction-to-ambient thermal resistance. I didn't use it much in the end, don't know if it would need a heatsink for longer term use. A small fuse in case of malfunction wouldn't probably hurt either

Newer wheels usually show the voltage in the app, but Firewheel had no app or much else, plus I'd prefer to ride without holding a smartphone in my hand (and as your case shows, can you really always trust the on-board voltage measurement? ). Although who knows how trustworthy these cheapo displays are in the long run either... Maybe I'll encase a similar style voltage-meter to my next wheel at some point...