to tell the truth, I know more about Lishui controllers and Bafang motors than the TSDZ2. Andy is on holiday this week, I expect he'll need a couple of days to catch up with his intray before we can set up a test rig to figure out the way the TSDZ2 kit reacts to the battery voltage.

On the Lishui, the LCD controls the 5 battery segments. In the OSF TSDZ2 firmware, the code controls the 5 battery segments. I just don't know how Tongsheng does it, LCD or code? A test rig will work that out in a few hours.

So how does the six segment display of voltage operate on your rides? Does the rate of change of the display accelerate as the number of segments decrease and is there any correlation with what you see on your watt meter?

I use the VLCD6 display which only has a 4 segment voltage display and I tried several times via OSF to change the segment change points to give what I hoped would be a more proportionate indication of remaining battery charge, but without success.

 

So basically I now ignore the display totally. I check the voltge with a plug-in volt meter on my battery before setting off on a ride to ensure it's still fully charged (I re-charge after every ride of more that 10 miles) and I know from cycling more than 3500 miles on this battery what range I'll get.

 

As a 'belts and braces' measure, on re-charging the battery I always charge through a watt meter and take a note of miles vs Ah & Wh to detect any untoward behaviour.

 

Further, I ocassionally (once every 4-6 months?) fully charge and then discharge the battery though a watt meter into a resistive dummy load. In this way I can detect the gradual dropping of total capacity over time and charge/discharge cycles.

 

So far battery capacity has dropped maybe 5% or less but within that value there will be varyiations due to different ambient temperatures in my bike shed - but hey, I'm not looking for scientific accuracy here, just enough information for early detection of battery issues.

Edited April 15, 20233 yr by Bikes4two

to tell the truth, I know more about Lishui controllers and Bafang motors than the TSDZ2. Andy is on holiday this week, I expect he'll need a couple of days to catch up with his intray before we can set up a test rig to figure out the way the TSDZ2 kit reacts to the battery voltage.

On the Lishui, the LCD controls the 5 battery segments. In the OSF TSDZ2 firmware, the code controls the 5 battery segments. I just don't know how Tongsheng does it, LCD or code? A test rig will work that out in a few hours.

 

Crikey now I feel guilty it seems I've opened a Pandora's box . All I meant to do was express my surprise at how the display of battery state did not appear to be reacting to the use of assistance. If you do take a closer look I would be interested in how the four LEDs on the battery work. For example are they controlled by four zener diodes ?

 

(Just been reading on internet) Unfortunately any system that relies on voltage to estimate state of charge of a Lithium Ion battery is going to be inaccurate. A single cell will only drop 0.5v between 40% and 80% depth of discharge (https://www.powertechsystems.eu/home/tech-corner/lithium-ion-state-of-charge-soc-measurement/). I wonder if the following make any sense :

 

I don't know how my battery is built but a block of cells (21700) will be connected in series and then a small number of these series blocks will be connected in parallel. One method would be to connect 13 cells in series and connect three of these in parallel. This is based on a nominal cell voltage of 3.7 volts so 48/3.7 is just under 13 meaning that 13 cells have to be connected to achieve the nominal voltage of 48v. Each cell has a 5 Amp hour capacity so three blocks are needed to store 15ah. The maximum charge voltage is 4.2v which gives a maximum battery voltage of 54.6 volts for the nominal 48 volt battery being considered.

 

So if each cell drops 0.5v then the battery will drop 13* 0.5 volts or 6.5 volts between 40% and 80% depth of discharge (DOD) assuming a common drop across all cells.

 

I'm not saying this is how my battery is built as I am sure there are other methods but it suggests that using floating battery voltage as an indication of charge is dependent on battery configuration. It also suggests that a 36 volt system will have a smaller voltage drop for the same 40% to 80% DOD because fewer cells are used to achieve 36 volts. This implies that any software monitoring the voltage will have to account for this difference for it to be of value on both 36v and 48v systems (5 volts versus 6.5 voltage drop to indicate the same depth of discharge).

 

Its worth noting that at DODs > 80% the voltage drops off a cliff so it seems that it is important to consider the rate of change of voltage. When it accelerates its time to recharge.

 

Warning, the above may well be CUB!

 

S

Crikey now I feel guilty it seems I've opened a Pandora's box .

I need to know too.

 

So basically I now ignore the display totally. I check the voltge with a plug-in volt meter on my battery before setting off on a ride to ensure it's still fully charged (I re-charge after every ride of more that 10 miles) and I know from cycling more than 3500 miles on this battery what range I'll get.

 

As a 'belts and braces' measure, on re-charging the battery I always charge through a watt meter and take a note of miles vs Ah & Wh to detect any untoward behaviour.

 

 

Great post. Your use of a voltmeter appears to be the only meaningful way to assess the charge status of the battery i.e. measuring the floating off load value. On load voltages will be all over the place depending on the load.

 

I will be searching my parts bin for some connectors to add my cheap watt (coulomb ?) meter between the charger and the battery, (I'm hoping it can deal a continuous 2 amps). I see on-line that coulomb meters are available and that they utilise a clamp meter type method of measuring current meaning that the power line remains unbroken. This seems a safer method of measuring energy used than the in circuit ammeters. I'm also going to search for the Panasonic data sheet for the 21700 cells to try and determine the 0% to 80% depth of discharge voltages so that I can assess battery state mid ride.

 

As an aside I see that Grin Technologies of Canada produce a rather nice universal battery charger which allows various charging regimes to be pre-programmed and saved in memory e.g. Charge 48v battery to 80% capacity using a maximum of 4 amp charge rate. Its not cheap but could save money in the long run by helping keeping the battery charged in the ideal 80%-20% range and extending its useful life.

 

S

I will be searching my parts bin for some connectors to add my cheap watt (coulomb ?) meter between the charger and the battery

 

The Watt meter is fitted to the charger and I have recharged my battery to full as indicated by the LED turning green on the charger. To recap following my first charge I rode 76 flat Lincolnshire miles mostly on road and was wondering why the battery display was still showing five and half (pulsing sixth segment).

 

In an attempt to get warm I started a couple of rides with no power and I adjusted the wheel size to get the speed matching the existing cycle computer and a handheld GPS I am working on the principle that I covered 70 miles on the first charge.

 

At the time just before the second charge the main display showed five solid bars with the sixth flashing on and off.

The four segment LED display on the battery showed two bars and the rested battery open circuit voltage was 47.1 volts.

 

Recharging the battery to full took approximately five and half hours and the watt meter indicated the following statistics:

 

Initial charge voltage : 47.6 volts

Final charge voltage : 54.5 volts

Charge current : 1.87 amps (stayed constant throughout the charge)

Amp hours added : 5.329 amp hours

Watt hours added : 267.3 Wh

 

Initial no load voltage when back on the trike is 54 volts but the battery needs time to settle post charge.

 

The meter is completely un-calibrated so its not possible to tell how accurate the figures are. They suggest that I have 'only' used a third of the battery capacity and am using 4 W per mile.

 

Thoughts and comments welcome.

 

S

Thank you for the numbers. The first bar goes after 35% of charge used and you use 4wh per mile. I will check the LCD with lower battery voltages in the next two days.

Edited April 16, 20233 yr by Woosh

[mention=2387]Simon Knight[/mention] -

I see on-line that coulomb meters are available and that they utilise a clamp meter type method of measuring current meaning that the power line remains unbroken.

 

• I've used clamp meters for various AC voltage/current measuring uses and to get an accurate, consistent and reliable reading, the single wire passing through the measuring ring needs to be held steady and more or less in the middle of the ring.
  
• I borrowed a friend's clamp meter that also measured DC current and the ring positioning was even more critical to get a reliable reading. By all means give one a go but I suspect that any reading will not be consistent.
  
• And in practice you will need to physically separte the '+Ve' and '-Ve' wires enough to pass one of them through the ring.
  

As an aside I see that Grin Technologies of Canada produce a rather nice universal battery charger which allows various charging regimes to be pre-programmed and saved in memory e.g. Charge 48v battery to 80% capacity using a maximum of 4 amp charge rate. Its not cheap but could save money in the long run by helping keeping the battery charged in the ideal 80%-20% range and extending its useful life.

 

• That charger is a pricey piece of kit as you've noticed. There are far simpler ways of achieving 80% capacity and I'm curious as to why you'd want to do this anyway?
  
• A lot of folk like to charge to a little bit just under the max voltage put out by the charger - to achieve this you could for instance insert a suitably rated diode in series with one of the charger leads to reduce the voltage at the battery charging terminals by something between 0.1v and 0.6v depending on diode type. That does of course require some fettling with a soldering iron and 'stuff' but for a fraction of the cost of a major wallet extraction.
  
• Charging current - standard 'as supplied' chargers are 2 amp and it is cheap to get a 4 amp charger if there is a need. Apparently a 2 amp charge rate will get you more charge/discharge cycles from your battery than charging at 4 amps but in practice you need to wonder if this is worth the bother.
  
• For instance, if you are getting 50 miles from a full charge at 2A and the batteries have a 'life' of 500 cycles, then you're good to 25,000 miles.
  
• Let's assume at 4A you're only good for 400 charge/discharge cycles - that's 16,000 miles.
  
• Now maybe that's not a lot of miles for you but it'd take me 5 years to do that in which time battery tech will have inproved and I'll be happy to build a newer battery with greater capacity/lighter weight.
  
• On the other hand, you might be a Bosch battery owner and you NEVER want to fork out for a replacement :D:D.
  
• I charge at 4A and am happy with that.
  

Thank you for the numbers. The first bar goes after 35% of charge used and you use 4wh per mile. I will check the LCD with lower battery voltages in the next two days.

I am concerned that the 4wh per mile seems low as I do not class myself as a powerful cyclist. As the weather improves I will be using the e-trike more so will collect more data which I'm happy to report. I will try and collect more data points to enable the plotting of voltage to Wh used.

 

My search for a specific data sheet describing the Panasonic 21700 failed but I did find information that can be read across. Here are some the links :

Tentative Panasonic specification sheet for their NCR20700B cell : http://www.batteryspace.com/prod-specs/10873%20specs%20.pdf

"A Guide to Understanding Battery Specifications" http://web.mit.edu/evt/summary_battery_specifications.pdf

Lastly for those who enjoy detail "Energy Density of Cylindrical Li-ion Cells: A Comparison of Commercial 18650 to the 21700 Cells: :https://iopscience.iop.org/article/10.1149/2.0281814jes

 

It seems that cells are made in two forms: high power and high energy with the two types having different discharge curves. However, it seems that the chemistry means that all cells have the same basic discharge curve with a starting with a short sharp reduction in voltage, then a gentle decline before falling down over a cliff at near the 80% depth of discharge which is the point where deep discharge occurs. Taking cells into either end of the charge curve shorten the life of the cells with deep discharge being the worst.

 

It appears that a single cell exhibits a voltage drop in the range 0.5v to 0.7v depending on load and where you decide the "cliff edge" is. This was taken from the Panasonic data sheet and another graph in the data sheet shows similar differences depending on cell temperature. So voltage drop depends on both load and temperature. Note that Panasonic is calling a single cell a battery which is possibly confusing.

 

Applying my recorded data to the graph my battery dropped to 47.1 volts. Divide this value by 13 gives (I believe) the single cell voltage of 3.6volts. Reading across to the black line suggests that the cell has been discharged by 1600 mAh. Again I believe my battery is made up of three series wired packs of cells wired together in parallel so 3x1600 = 4800 mAh battery discharge. The watt meter reported adding 5329 mAh meaning the graph is in the ball park.

 

Note Woosh reported above that the lowest battery voltage should be 41 volts. This equates to 3.15 volts on the graph and is just after the point where the curve takes the plunge downwards.

 

One final point I read is that some cheaper cells claim a theoretical capacity based on chemistry and physical size and were found to be wildly inaccurate. Whereas the big name brands tend to quote measured values. You pays your money and takes your chance.

 

Enjoy!

• Let's assume at 4A you're only good for 400 charge/discharge cycles - that's 16,000 miles.
  
• Now maybe that's not a lot of miles for you but it'd take me 5 years to do that in which time battery tech will have inproved and I'll be happy to build a newer battery with greater capacity/lighter weight.
  

 

Trust me thats a lot of miles!

 

All the points in your post are excellent - thanks.

 

You are correct about the Grin charger being expensive but I do have several e-bikes in my household so it may be of value to me as it would simplify the charging equipment and possibly be more accurate than my cheap Watt meter.

 

However, I'm don't think I'm going to add any extras to my set up for the time being as they are unlikely to provide any more useful information at present. I need to use the Tongsheng more to gain an understanding of how it performs. In short I shall follow your advice given above (and further above) which I can now do since adding the Watt meter to the charger.

 

Woosh will be looking at how the display operates on the 48v system and I am now in a position to note the voltage at which the display reduces by a bar for comparison.

 

S

Recharging the battery to full took approximately five and half hours and the watt meter indicated the following statistics:

 

Initial charge voltage : 47.6 volts

Final charge voltage : 54.5 volts

Charge current : 1.87 amps (stayed constant throughout the charge)

Amp hours added : 5.329 amp hours

Watt hours added : 267.3 Wh

5 1/2 hours at 1.87 amps (constant) looks like 10.3 amp hours.

How does that reconcile with the '5.329 amp hours added'

In low assist on a gravel style bike with touring gear and trailer and all up weight 130 to 150kg, I average 10Wh per mile in fairly gentle terrain, and I am almost all ths time below the cutoff speed.

 

So if it is just you and the trike with a fair amount of time above the cutoff in gentle terrain and no big head winds, 4Wh is a bit low but not impossible. It may be that as a newish e rider you are trying harder than you need to!

I set up a test rig this afternoon.

Here are the guide to the VLCD5's battery indicator, the TSDZ2 has factory firmware.

 

Voltage
Bars
State of charge


Above 48V
6
50%-100%


47V-48V
5
40%-50%


46V-47V
4
30%-40%


45V-46V
3
20%-30%


44V-45V
2
10%-20%


43V-44V
1
under 10%


Under 43V
0
0%

 

 

It follows from this table that Tongsheng programmed 43V-48V as the voltage range instead of 44V-54V.

You should charge the battery when the first bar is gone.

I won't blame you if you want to reflash your TSDZ2 with OSF firmware to rectify this issue.

 

PS: one annoying thing with the stock firmware: when you switch on the LCD, it shows all 6 bars first. You have to wait a few seconds (about 3-4 seconds) for the code to come round and update the bars.

Edited April 17, 20233 yr by Woosh

Well, that explains a lot! Thanks for doing this test, and makes a bit more sense why the 1st bar lasts for soooo long.

 

I won't blame you if you want to reflash your TSDZ2 with OSF firmware to rectify this issue.

 

I know why when I was on 3 bars the motor was struggling on hills. Unfortunately, flashing will void the warranty so I'm not especially keen to go down that route.

 

It appears I can get 25 to 30 miles in hilly area out of 1st bar with a fair amount of assistance and not hanging about. My initial assessment of 50+ out of a full battery looks about right if riding a little more conservatively.

 

C

Edited April 17, 20233 yr by Bogmonster666

With my OSF tsdz2 , I can adjust the bars to suit an approx. voltage range for my 44v battery pack.

6 bars full 96 -100% 4.16 -4.2v / 49.9v - 50.4v.

5 bars 80% 3.98v / 47.75v.

4 bars 65% 3.85% /46.2v.

3 bars 50% 3.70v / 44.4v.

2 bars 35% 3.55v/ 42.6v.

1 bar 15% 3.37v / 40.4v

0 bars empty 3.20v / 38.4v

 

I set 3.2v as my min lvc.

 

Today having towed my heavy trailer over 25 miles on a mucky old rail line turned over by horses in places. The battery bar was flitting between 1 or 2 bars , at home a voltage check showed 41.3v or 3.44v per cell group, a quick balance check on the extra external balance wire added did indeed agree and all were at 3.44v.

So the OSF settings are pretty good and work quite well.

 

My wh rate is in double figures somewhere above 12 wh but less then 14wh.

 

Battery pack is 12s2p of Molicell p42a, 44v/8.4ah for 369wh.

BMS is a same port 20a Daly.

*SJPT - no idea why the reported Ah is less than the constant current multiplied by the time. Possibly because its a cheap uncalibrated watt meter. I have confirmed what the units are suppose to display " The displayed value is the total charge in Amp-hours (x1000=mAh) delivered since the startup screen ended"

 

The amps value of 1.87amps is an average value calculated every 0.4 seconds. I checked the display from time to time and either 1.87 or 1.88 amps was always displayed.

 

I will look out a known resistive load and attempt to get a rough calibration as I agree the values do not appear to make much sense.

 

*Mathewslack I agree that 4Wh per mile is rather low. I need to gather more data and where possible confirm the watt meter is working o.k.

 

* Woosh Thanks for publishing this table which answers my original questions about the display. You say "You should charge the battery when the first bar is gone." Did you mean this as it means recharge once the battery voltage drops below 48 volts? My 15Ah battery has a four bar display built into the side, three green LEDs and one red. At 47.1 volts this displayed 2 bars of charge. I am inclined to keep using the battery until only the red LED is illuminated, note the off load voltage and then recharge.

 

I have just reread your post and suspect you meant when bar number one goes out i.e. all bars out and a battery voltage of below 43 volts

 

* Bogmonster666 "Unfortunately, flashing will void the warranty so I'm not especially keen to go down that route." I'm with you. I will only look at flashing if I have issues with a rough power cut off which has been reported elsewhere. However, so far it has always been a smooth transition.

 

* Nealh Extrapolating your numbers the equivalent deep discharge voltage on my 48v battery would be 44.5volts (give or take), which I suspect will be close to my batteries single red LED or 2 bars on the main display.

 

On reflection having the display working in this slightly unexpected is not terrible. Bar six switches off at half charge and then each subsequent bar indicates a drop of 1 volt all the way down to 43 volts. In effect the display has finer granularity in the lower voltage region where it is most useful.

 

S

 

One annoying thing with the stock firmware: when you switch on the LCD, it shows all 6 bars first. You have to wait a few seconds (about 3-4 seconds) for the code to come round and update the bars.

I run OSF on a vlcd6 display and at power initially only just one battery bar shows then after a few seconds the true state shows. I always treat this time period, rightly or wrongly, as the boot up period during which time I keep my feet clear of the pedals whilst the torque sensor circuits are calibrated.

 

Might this behaviuor be what is happening with the OEM firmware?

Might this behaviuor be what is happening with the OEM firmware?

yes, same.

Just an update. Going purely on the display bars (the 1st of which I think we have established goes out at 48v, here are my findings:

 

Yesterday I did a moderately hilly 36 miles. Some distance from the end the 1st bar momentarily flickered out, only to return for the remainder of the trip - including some hills. Recharged when I got home.

 

This morning I did 37 miles, probably with less hills, but still hilly. Today I was really finding it hard going, I think I am overdoing it as not cycling fit - my legs felt like concrete. Anyway, being very lazy on the last couple of hundred feet, the 1st bar flickered as I was practically turning into my drive.

 

I actually think my 10ah battery maybe not quite 10ah given the 50g cells used but close enough so let's ignore that. Looks like the voltage goes off a cliff shortly after 4000mah at around 3.3v which I suspect the lvc is?

 

Looking at an average discharge rate of 2a (per cell string), 54v to 48v delivers ~4000mAh(2000mAh per cell). Then the remainder of the battery delivers another ~4000mAh until 3.3v.

 

This was all based off cell discharge graphs I found on the interweb.

 

Realistically 5000mAh isn't achievable with the cells and getting to 4800mAh you are down to 2.4v per cell.

 

So roughly speaking the 1st bar gives 4Ah and the remaining gives 4Ah with a very linear discharge resulting in a usable 8Ah.

 

The 1st bar (goes out at 48v) gives ~200wh so at 37 miles that's 5.5 wh a mile.

 

The remainder of the battery should give ~180wh (due to lower voltage) so I might expect no more than 32 miles out of that.I will need to test the theory (have not felt confident enough to embark on a long ride with a half depleted battery yet) but I think 60+ miles is achievable for me in my surroundings, possibly approaching 70 miles, but no more. I expect that to be somewhat less in the colder months, although it was cold yesterday evening with a heavy frost. Playing it safe, I think 50 miles range is the maximum I would consider in these parts. That's quite good imo, but I guess I need to test the theory out in the real world as I made a lot of assumptions and there are a lot of variables.

 

Realistically I am unlikely to want to go much further than 50miles in a trip - and today I was definitely regretting the last detour and wishing I'd headed towards home sooner. The 10ah bag battery is a good size for my needs.

 

Early on I had gotten 50+ miles and down to 3 bars so the achievable range is proving remarkably consistent.

Edited April 22, 20233 yr by Bogmonster666

Your never get the Mah rating unless only drawing 0.2a per cell.