Moving Away From Heavy E-Bikes

[PCeBiker]

We are probably riding e-bikes with quite different electronics and batteries. I could have never noticed any sag in any of my batteries for Specialized or Giant, and some of them are over 3 years old and heavily used.
Yes, I could notice the sag in my cheapest e-bike (which I don't ride anymore) but that e-bike is equipped with a cheap battery and electronics.

That has nothing to do with it.
All batteries have voltage sag while under load.

 

[Stefan Mikes]

That has nothing to do with it.
All batteries have voltage sag while under load.

EDITED:

This is the voltage chart from a Specialized Vado as the function of the battery percentage on a ride. Certainly, the voltage curve is not smooth. Don't you think good electronics can compensate for that, for instance manipulating the current sent to the motor?
It is a mid-drive, and the current sent to the motor is variable, also depending on the rider's leg power input at a given time. The objective of the electronics is to send as much power to the motor as to meet the criterion of amplifying the rider's leg input with a given Boost Factor. It is totally different on a hub-drive. With the hub-drive, the voltage is the driving force. With the battery sag, your motor gets less or more power, making the ride less smooth.

If you like, I can produce a chart of the power delivered to the motor at any time, and of the current sent to the motor.

Could you tell me why I need any voltmeter then?

 

[PCeBiker]

All batteries have internal resistance.
Some batteries more than others.
That resistance creates a voltage drop inside the battery reducing the output voltage, and heat inside the battery as well.

That heat can lead to battery fires.
Better quality lithium cells have very low internal resistance and can deliver power without getting as hot.
Same goes for charging batteries.

 

[linklemming]

Strange. Have never noticed that. Don't you think a good electronics could compensate for it, for instance sending higher current?

You likely wont see the sag as your battery indicator (assuming bars) isnt showing you the actual voltage but some sort of 'summarized voltage' averaged over time although it might be using other data depending on algorithm..

You would need to see actual RAW voltage to see sag

Good electronics could/does compensate depending on if they are trying to provide constant current (no) or constant power (yes). Cheaper controllers likely use constant current.

If your controller is providing constant power then the bars (assuming based on average voltage...preferably no load) will decrease faster as the battery voltage gets lower because it is using more amps. This would apply to a %display as well this has been 'proven' by many peoples observations on these forums....including me. Not a big deal, easy enough to compensate for but its there.

While cell resistance is a good indication of battery quality, it isnt the end all measurement for quality initially(in a simplistic sense). Cells designed to be higher output(30q) will definately have lower initial resistance but cells designed for more cycles but decent power (like the GA cells,,,used in older teslas) will have higher initial resistance. This will change over time with increasing resistance as the pack ages where the 30q resistance will outweigh the GA cells.

In the end, cell resistance(and resulting sag) is what defines the life of a battery (assuming no individual cells have died yet)

 

[PCeBiker]

Even the speedometer on my old e-bike is a computerized dummy gage.

As soon as I disengage the cruise control the speedometer instantly drops from 25 kph to 22 kph but my speed didn't change.

I was worried that my voltmeter was a dummy gage too, so I velcroed a voltmeter on top of the bike's voltmeter.
(note the remaining glue after I removed it. )

 

[Stefan Mikes]

You would need to see actual RAW voltage to see sag

See my edited post and read the explanation.
The chart is from the RAW data.

@PCeBiker: my EDIT also pertains to you.

 

[PCeBiker]

That's a computer generated graph.
It's too square and chunky and all the spikes are evenly spaced.
It's a list of samples and averages.

 

[Stefan Mikes]

That's a computer generated graph.
It's too square and chunky and all the spikes are evenly spaced.
It's a list of samples and averages.

This is from the e-bike RAW data. I could of course add the markers. Why not?

Battery Voltage On A Ride As Function of Battery Percentage.

Criteria for the graph:
-- Coasting filtered out
-- Single Assistance Level (Eco)

Unlike hub-drive motors, mid-drives operate with variable power. It makes all the voltage sag discussion irrelevant. (It is noteworthy the motor could draw even 280+ W at the 12% of the battery charge. Note the voltage was pretty low at that battery %).

 

[PCeBiker]

All ebikes vary the power to maintain a PAS level speed.
My geared hub motor was using ~170 Watts to go downwind and ~450 Watts to come back upwind at the same speed. Each way the Watt meter was bouncing around trying to maintain the set speed.

I'm actually trying to find a way to get my controller to put out a constant power but I'm finding it difficult to do.

I don't care what speed I'm going, I just want it to put out say, ~100 Watts continuously and I don't think I can set up my controller/display to do that?

That second red graph looks more like a real time graph, but the dots suggest sampling and the computer drew the curve.

 

[PCeBiker]

In the end, cell resistance(and resulting sag) is what defines the life of a battery (assuming no individual cells have died yet)

Can you explain to me how a bad cell can lead to a fire?
I assume it becomes overcharged then gets hot?

 

[linklemming]

This is from the e-bike RAW data. I could of course add the markers. Why not?



View attachment 150398
Battery Voltage On A Ride As Function of Battery Percentage.

View attachment 150399
Criteria for the graph:
-- Coasting filtered out
-- Single Assistance Level (Eco)

Unlike hub-drive motors, mid-drives operate with variable power. It makes all the voltage sag discussion irrelevant. (It is noteworthy the motor could draw even 280+ W at the 12% of the battery charge. Note the voltage was pretty low at that battery %).

Sorry but that is not raw data at least in the engineering sense but interpolated data over time. Raw data would occur at something like 60Hz (or higher) meaning there would be thousands of values. Simple computer sampling theory.

Most likely, it keeps track of voltage when not supplying power and averages that over time. If advanced, it could determine cell resistance based off voltage at various loads, in any sense it would still be averaged.

All this can be found in the TSDZ2 open source software code, I doubt specialized/brose has gotten around well understood physics

Your last statement is completely false and actually moreso on your(and mine) 36V bike compared to most chinese 48V systems.

Voltage sag under load is identical for my BBSHD, GMAC and DD motors using the same battery under the same load

 

[linklemming]

Can you explain to me how a bad cell can lead to a fire?
I assume it becomes overcharged then gets hot?

I havnt looked into it much so Im definately not the expert on cell failure, especially when charging.

That being said, I am concerned about failure that can occur due to shorted cells. For a parallel section, if a cell shorts, all the other cells dump their current into the shorted cell causing it overheat and potentially catch on fire. No BMS can account for this. I try to only use batteries that have individually fused cells like em3ev and luna(wolf), pretty sure the big 4 are using fused cells as well.

I sold a Juiced CCX to a neighbor last year with upgraded GMAC motor and he has ridden the snot out of it and the battery probably has 9k miles on it now. Just a few weeks ago, it stopped balancing charge anymore as noted by his satiator charger. Not sure what that means but I convinced him to get a new battery

 

[Stefan Mikes]

Sorry but that is not raw data at least in the engineering sense but interpolated data over time. Raw data would occur at something like 60Hz (or higher) meaning there would be thousands of values. Simple computer sampling theory.

Most likely, it keeps track of voltage when not supplying power and averages that over time. If advanced, it could determine cell resistance based off voltage at various loads, in any sense it would still be averaged.

All this can be found in the TSDZ2 open source software code, I doubt specialized/brose has gotten around well understood physics

Your last statement is completely false and actually moreso on your(and mine) 36V bike compared to most chinese 48V systems.

Voltage sag under load is identical for my BBSHD, GMAC and DD motors using the same battery under the same load

What you are saying here is a pure sophistry (translated to Simple English: Bullshit).

An industrial Distributed Control System (DCS) has a resolution (sampling frequency) of 1 s. The sampling frequency of the Specialized system is 1 s, too. It seems you know very little of Measurement & Control.

Are you trying to tell me your voltmeter gives your eyes the sampling frequency of 60 Hz? Ha. Ha. Ha.

 

[PCeBiker]

That being said, I am concerned about failure that can occur due to shorted cells.

So I assume it's possible for a cell to short itself out internally?

 

[PCeBiker]

Are you trying to tell me your voltmeter gives your eyes the sampling frequency of 60 Hz? Ha. Ha. Ha.

That raw data could be fed to a voltmeter and it will show you the "average", but the average is done very fast, like 2 or 3 times a second or more.

My needle voltmeter samples continuously but is a tiny bit "averaged" due to the mass of the needle and the time it takes to accelerate it.

Some needle meters are in fluid to smooth out the movement so they aren't so twitchy.

 

[PCeBiker]

An industrial Distributed Control System (DCS) has a resolution (sampling frequency) of 1 s. The sampling frequency of the Specialized system is 1 s, too. It seems you know very little of Measurement & Control.

Your graphs aren't showing every 1 second sample.
I count about 90 dots on your red graph.
You didn't drain a fully charged battery to 10% in a minute and a half.


If you hooked up a voltmeter directly to your battery you would see the real time voltage as it is happening, not what the computer decides to show you on a screen.

 

[linklemming]

What you are saying here is a pure sophistry (translated to Simple English: Bullshit).

An industrial Distributed Control System (DCS) has a resolution (sampling frequency) of 1 s. The sampling frequency of the Specialized system is 1 s, too. It seems you know very little of Measurement & Control.

Are you trying to tell me your voltmeter gives your eyes the sampling frequency of 60 Hz? Ha. Ha. Ha.

Are you actually trying to imply that all sampling in the world occurs at 1Hz. Thats an insane and completely false claim. I have worked in the embedded software industry for well over 35 years including avionics flight controls, measurement, test, simulation and most recently spacecraft attitude and control.

They all use sampling frequencies based on the desired inputs, performance, accuracy and hardware used. I can provide examples all day long.

https://www.sciencedirect.com/topics/engineering/nyquist-theorem#:~:text=Nyquist's%20theorem%20states%20that%20a,higher%20than%20this%20is%20necessary.

I never implied anything about my eyes and 60Hz, you need to stop being so emotional when presented with facts

Most likely the display has a refresh rate lower than 60Hz anywho

 

[linklemming]

Your graphs aren't showing every 1 second sample.
I count about 90 dots on your red graph.
You didn't drain a fully charged battery to 10% in a minute and a half.

Its pretty common knowledge in the RC industry that 18650 cells shouldnt be totally discharged in less that 1/2 hour due to their high internal resistance.

I have used High Current RC Lipo packs for FPV drone flying and discharged them in a minute...they usually puff and dont work anymore..sometimes catching on fire

 

[linklemming]

If you hooked up a voltmeter directly to your battery you would see the real time voltage as it is happening, not what the computer decides to show you on a screen.

Most likely, the data being shown while sampled at 1Hz(by the app) is sampled at higher frequencies within the embedded firmware within the motor and then averaged according to some criteria (of which I have mentioned previously). That data (before being averaged) is what is considered RAW data in any engineering discipline.

I cant believe we are even debating this

 

[PCeBiker]

I've got to reduce my sampling rate.

I spend more time looking at the display than looking at the road.

I think I'll cover it with duct tape once I get it set up. Lol