Electric Unicycle's BMS problem and solution

[hobby16]

On 31/5/2016 at 10:33 AM, Chriull said:

Imho the shunt should be (shorting drain-source of the overdischarge protection Mosfets) between the heatsink and the source pin. Could be easily made as shown by the "faint" red line in the following picture:

http://imgur.com/n5SP3ho

 

I concur, it's the way to do.

An even better shunt would be to connect the heatsink directly to B1-, since the two 5 milliohms resistors for current sense have no purpose after the shunt (they are there to detect overcurrent and trigger shutdown but since the mosfet stage is shunted, no shuthdown can occur so they are useless).

[hyperair]

Hi,

I was just wondering if you considered the possibility of a voltage inversion in one or more of the cells when using a shunted BMS? http://liionbms.com/php/wp_lovtg_cutoff.php

I'm thinking that it's not altogether unlikely if the user has had numerous cycles (not sure how many) not been balance-charging the battery (charging it all the way up to 100% and letting it sit for a while), since the cutoff voltage on the main board is 45V for unicycles using 16S, and 13 x 3.7V = 48.1V, i.e. you could have up to three cells at 0V and still not trigger the board's low voltage cutoff, which then leads to a voltage inversion in these dead cells as the board continues to discharge the pack.

[hobby16]

On 4/6/2016 at 6:42 AM, hyperair said:

I was just wondering if you considered the possibility of a voltage inversion in one or more of the cells when using a shunted BMS? http://liionbms.com/php/wp_lovtg_cutoff.php

On electric unicycles, it's not possible to discharge deeply a shunted pack, the controller board won't let you do that, it will tilt-back the pedals to force you to climb down. And balancing does not start only when charging up to 100%, even charging only up to 90% can trigger balancing on most BMS, so the risk is low to have a cell so deeply discharged that the voltage is inverted.

If ever there was a voltage inversion on a cell, you would know it without doubt because the wheel's range would take a hit, for sure (and you will feel much less power from the wheel). The cell is dead anyway, I would then worry more about a faceplant (happened already on some GW or KS wheelers) than a dead cell.

[hyperair]

On 07/06/2016 at 4:09 AM, hobby16 said:

On electric unicycles, it's not possible to discharge deeply a shunted pack, the controller board won't let you do that, it will tilt-back the pedals to force you to climb down. And balancing does not start only when charging up to 100%, even charging only up to 90% can trigger balancing on most BMS, so the risk is low to have a cell so deeply discharged that the voltage is inverted.

I can see that being true for a healthy pack, but I'm uncertain about what happens as one of the cells approaches end-of-life. Does the cell still charge to 4.2V and simply discharge faster in such a scenario? If so, we're talking about a cell that will reach 0% SoC earlier than any of the other cells despite starting at 4.2V/cell. If one cell is at 0V and the other 15 cells need only be above 3V/cell to maintain a pack voltage above 45V.

Quote

If ever there was a voltage inversion on a cell, you would know it without doubt because the wheel's range would take a hit, for sure (and you will feel much less power from the wheel). The cell is dead anyway, I would then worry more about a faceplant (happened already on some GW or KS wheelers) than a dead cell.

Voltage inversion can very well lead to a fire when attempting to charge that dead cell back up, and I'm more worried about that than a faceplant (having gone through a number of them myself, including one at the top speed of a Microworks wheel).

[hobby16]

On 10/6/2016 at 0:24 PM, hyperair said:

I can see that being true for a healthy pack, but I'm uncertain about what happens as one of the cells approaches end-of-life. Does the cell still charge to 4.2V and simply discharge faster in such a scenario? If so, we're talking about a cell that will reach 0% SoC earlier than any of the other cells despite starting at 4.2V/cell. If one cell is at 0V and the other 15 cells need only be above 3V/cell to maintain a pack voltage above 45V.

Voltage inversion can very well lead to a fire when attempting to charge that dead cell back up, and I'm more worried about that than a faceplant (having gone through a number of them myself, including one at the top speed of a Microworks wheel).

Do you have or have seen any such battery pack and cell or are we talking about what-ifs and hypothetical scenarios ? If it's the latter, anything can be supposed, all the more when you don't know how the BMS handles cells' voltages mismatch and end of charge shutdown.

If you are so afraid of voltage inversion, buy a Charge Doctor, problem solved.

Here is a Charge Doctor's log of a faulty pack : the jumps in charge current & voltage is evidence something is wrong (a cell, several cells, the BMS...).

Personnally, I've seen packs with faulty cells, all I can tell you that you don't go far with them, either the wheel faceplant you, or its power & range are so low you won't be mistaken of the problem. I've been reported of thermal runaways (a recent example below) but never of fire. Maybe it's due to voltage inversion, maybe... not.

[hyperair]

10 hours ago, hobby16 said:

Do you have or have seen any such battery pack and cell or are we talking about what-ifs and hypothetical scenarios ? If it's the latter, anything can be supposed, all the more when you don't know how the BMS handles cells' voltages mismatch and end of charge shutdown.

No I haven't. These are hypothetical, and under the assumption that the BMS uses that shunted MOSFET to disable the pack if any one cell drops below threshold voltage (see the row on overdischarge protection in http://www.aliexpress.com/item/16S-15A-59-2V-li-ion-unicycle-BMS-PCM-battery-protection-board-bms-pcm-with-balancing/32414515852.html).

10 hours ago, hobby16 said:

If you are so afraid of voltage inversion, buy a Charge Doctor, problem solved.

Here is a Charge Doctor's log of a faulty pack : the jumps in charge current & voltage is evidence something is wrong (a cell, several cells, the BMS...).

Cool. I didn't realize charge doctors worked without balancing leads. Do you have a link to the one you use?

I ended up building a new 16S2P battery pack from a BMS with higher discharge current limits (80A, according to the specs in the link above) instead.

11 hours ago, hobby16 said:

Personnally, I've seen packs with faulty cells, all I can tell you that you don't go far with them, either the wheel faceplant you, or its power & range are so low you won't be mistaken of the problem. I've been reported of thermal runaways (a recent example below) but never of fire. Maybe it's due to voltage inversion, maybe... not.

It's not like they don't happen -- I saw a photo a couple of months ago of a burnt unicycle that had been left in a car trunk. The baby seat next to it was charred pretty bad. Probably not from voltage inversion though.

[esaj]

31 minutes ago, hyperair said:

Cool. I didn't realize charge doctors worked without balancing leads. Do you have a link to the one you use?

Hobby16 builds & sells the Charge Doctor (nowadays V2), the topic is here:

 

http://hobby16.neowp.fr/2016/02/06/charge-doctor-v2-10a-3/

 

[hobby16]

7 hours ago, hyperair said:

I ended up building a new 16S2P battery pack from a BMS with higher discharge current limits (80A, according to the specs in the link above) instead.

It's not like they don't happen -- I saw a photo a couple of months ago of a burnt unicycle that had been left in a car trunk. The baby seat next to it was charred pretty bad. Probably not from voltage inversion though.

Beware, there is much confusion about how the BMS cuts current (and faceplants you !) : that happens because of

1) maximum current limit

2) undervoltage on one or more cells

3) overheating

IMHO, 1) is more for regulations compliance than a real protection, since 80A would never happen in a wheel (because it's means 5kW) and the cables would have fused anyway at such high currents. On the contrary 2) happens as soon as you have a faulty cell, which is why a BMS with the possibility to cut off power is so utterly stupid and dangerous.on monowheels. So the only way to ensure no unexpected cutoff is to buy a BMS from Gotway or KS, or to shunt your BMS.

Re: battery fire, some wheel models (like "Pinwheel") use LiPo batteries. It's probably them which go on fire. Liion batteries, with 18500 cylindrical cells are more robust, billions of them are used, so one of them going on fire is probably anecdotal evidence. I mean going on fire while not on charge. LiIon on fire while charging is less rare though, I never let my wheels unattended when they are on charge.

[hyperair]

3 hours ago, hobby16 said:

Beware, there is much confusion about how the BMS cuts current (and faceplants you !) : that happens because of

1) maximum current limit

2) undervoltage on one or more cells

3) overheating

IMHO, 1) is more for regulations compliance than a real protection, since 80A would never happen in a wheel (because it's means 5kW) and the cables would have fused anyway at such high currents. On the contrary 2) happens as soon as you have a faulty cell, which is why a BMS with the possibility to cut off power is so utterly stupid and dangerous.on monowheels. So the only way to ensure no unexpected cutoff is to buy a BMS from Gotway or KS, or to shunt your BMS.

1) is for short protection, IMO, which can happen if your wheel shorts out. Here's an old Gotway MCM 14" I received after it went into thermal runaway. https://www.facebook.com/hyperair/media_set?set=a.10156323229550104.1073741834.516720103&type=3. I'm not sure it actually hit 80A, but as you can see, it the motor leads got hot enough to melt the PTFE insulation. Since one of the motor bearings have completely seized up, I'm thinking that the bearing getting destroyed caused the motor to get stuck and turn into a short circuit (those motor windings have really low resistance), which in turn caused way too much current to go through the entire circuit. The Gotway battery pack, built with Panasonic CGR18650CH cells, has a couple of vented cells, and the BMS has some lifted traces.

2) Frankly speaking, I think unicycle vendors would be in for a much more severe lawsuit if caught running with BMSes that can't cut power. On my list of do-not-wants, fires and thermal runaways still rank higher than faceplants.

3 hours ago, hobby16 said:

Re: battery fire, some wheel models (like "Pinwheel") use LiPo batteries. It's probably them which go on fire. Liion batteries, with 18500 cylindrical cells are more robust, billions of them are used, so one of them going on fire is probably anecdotal evidence. I mean going on fire while not on charge. LiIon on fire while charging is less rare though, I never let my wheels unattended when they are on charge.

Do you mean 18650?

[hobby16]

19 hours ago, hyperair said:

2) Frankly speaking, I think unicycle vendors would be in for a much more severe lawsuit if caught running with BMSes that can't cut power. On my list of do-not-wants, fires and thermal runaways still rank higher than faceplants.

BMSes in Gotway and King Song wheels (AFAIK) do NOT and cannot cut power! It's the only sensible thing to do on unicycles, whatever you might think about lawsuits.

What type of wheel do you have ? If you have a Gotway, by definition, you don't have faceplants by the BMS cutoff so you can't rank which is more important than what. If you rank faceplants lower risk than thermal runaways (what a strange idea), then avoid buying or using a Gotway since its has a BMS that can't cut power ! Try a TG wheel which has faceplanted so many people here, especially in cold weather and get back tell us about your priorities. And if you are still not convinced, try a Firewheel with its tremendously powerfull motor. A faceplant at 28km/h happening twice a day (it's not fiction tale, it can be reproducible as soon as there is a failed cell in the pack) will definitely reorder your priorities.

BTW, in your example, the fact there is no fire despite a catastrophic failure leading to short-circuit and cells thermal runaway and venting is evidence that fire is an exceptionnal event with LiIon cells, right ? Or am I missing something in the reasonning ? Once again, it's the motherboard that should take care of this situation, detecting overcurrent and forcing the rider to climb down by pedal tilt-up, it's simply a matter of firmware due diligence. There is no reason to let the BMS shutdown the power without warning, leading to a faceplant.

[hyperair]

3 hours ago, hobby16 said:

BMSes in Gotway and King Song wheels (AFAIK) do NOT and cannot cut power! It's the only sensible thing to do on unicycles, whatever you might think about lawsuits.

If you say so. Between fire and faceplant, I choose faceplant. I understand if some people feel differently.

3 hours ago, hobby16 said:

What type of wheel do you have ? If you have a Gotway, by definition, you don't have faceplants by the BMS cutoff so you can't rank which is more important than what. If you rank faceplants lower risk than thermal runaways (what a strange idea), then avoid buying or using a Gotway since its has a BMS that can't cut power! Try a TG wheel which has faceplanted so many people here, especially in cold weather and get back tell us about your priorities. And if you are still not convinced, try a Firewheel with its tremendously powerfull motor. A faceplant at 28km/h happening twice a day (it's not fiction tale, it can be reproducible as soon as there is a failed cell in the pack) will definitely reorder your priorities.

The only thing that remains Gotway about my wheel is the shell. The motor and board are Microworks parts, and I've used two different battery packs on this -- one Victpower 16S1P which faceplanted me numerous times, mostly on ramps or sudden potholes, or when attempting to accelerate too quickly, and one self-built 16S2P pack from the BMS I linked above and Samsung 25R cells.

I don't believe a Gotway has a BMS that can't cut power. It probably just has higher thresholds like the BMS I'm using with my 16S2P pack, and better 18650 cells with less internal resistance and hence less voltage sag at high currents.

Having gone through a faceplant-prone wheel for a couple of months, I'll agree with you that it's very tiresome to deal with. But rather than shunting the pack, I'd replace it (which I did).

3 hours ago, hobby16 said:

BTW, in your example, the fact there is no fire despite a catastrophic failure leading to short-circuit and cells thermal runaway and venting is evidence that fire is an exceptionnal event with LiIon cells, right ? Or am I missing something in the reasonning ? Once again, it's the motherboard that should take care of this situation, detecting overcurrent and forcing the rider to climb down by pedal tilt-up, it's simply a matter of firmware due diligence. There is no reason to let the BMS shutdown the power without warning, leading to a faceplant.

I think this wheel got lucky, and that these Panasonic CGR18650CH cells are decent (they vent but don't spout flames). The same can't be said for the average generic wheel one finds on alibaba/aliexpress/taobao/ebay, especially the cheaper ones. If you want to see an actual 18650 fire caused by overcurrent, here's one: https://www.youtube.com/watch?v=08BoXebt_pk

From what I can tell from various 18650 fire videos, 18650s tend to vent for 40-60 seconds before going up in flames, so that's how much time you have to work with if you see smoke.

[Chriull]

 

15 hours ago, hyperair said:

...

I don't believe a Gotway has a BMS that can't cut power. It probably just has higher thresholds like the BMS I'm using with my 16S2P pack, and better 18650 cells with less internal resistance and hence less voltage sag at high currents.

They can't cut power - at least the older models. There are enough pictures of BMS boards here that show clearly that there is nothing but the copper tracks between the batteries and the discharge cable.

KS imho had fuses in the output path.

15 hours ago, hyperair said:

Having gone through a faceplant-prone wheel for a couple of months, I'll agree with you that it's very tiresome to deal with. But rather than shunting the pack, I'd replace it (which I did).

That's for sure the number one solution. Once one or more cells are bad/dead that's the only choice one should make. (beside exchanging these cells, if one can safely do so)

If the pack is just underpowered (i.e. 16s1p or just too "small" cells) and the budget just does not allow an replacement and there is no way to get a refund i can understand why people decide to shunt instead of choosing a "neat/professional" solution which costs quite some money... ;(

15 hours ago, hyperair said:

I think this wheel got lucky, and that these Panasonic CGR18650CH cells are decent (they vent but don't spout flames). The same can't be said for the average generic wheel one finds on alibaba/aliexpress/taobao/ebay, especially the cheaper ones. If you want to see an actual 18650 fire caused by overcurrent, here's one: https://www.youtube.com/watch?v=08BoXebt_pk

This are two 18650 charged with 40A - so imho a different situation.

The charging protection of the BMS is not touched and in no way compromised by shunting. If the BMS allows overcharging (especially of "bad/dead" cells) it's dangerous crap anyhow...

Ps.: Generic wheel with bad BMS and bad cells should be banned imho anyhow... They are just a waste of money and dangerous

 

[hobby16]

20 hours ago, hyperair said:

I don't believe a Gotway has a BMS that can't cut power. It probably just has higher thresholds like the BMS I'm using with my 16S2P pack, and better 18650 cells with less internal resistance and hence less voltage sag at high currents.

Once again, you are making hypothesis instead of checking. Pictures of Gotway BMS can be found everywhere, look at them, there is no mosfet to cut power. There is only one small mosfet to cut charging current to protect against overcharge, that's all.

And I repeat, it's the only sensible thing to do, whatever you believe or not. It would be totally crazy to let the BMS cut power on a wheel that can go up to more than 30 km/h. Maybe you don't mind faceplants but you are the exception, not the norm. Any normal person would prefer to be protected against the certainty of faceplants than against the hypothetical and near zero risk of battery fire. End of debate.

[smallexis]

On 15/06/2016 at 0:21 AM, hobby16 said:

Once again, you are making hypothesis instead of checking. Pictures of Gotway BMS can be found everywhere, look at them, there is no mosfet to cut power. There is only one small mosfet to cut charging current to protect against overcharge, that's all.

Just to confirm what's hobby16 said, please find below picture from GOTWAY BMS (from MCM4 680Wh), POWER wire are directly soldering at cells pack on + & - poles

[tjcooper]

Does anyone have full schematics of these Samsung BMS packs?

When I was doing all the quad copters and high current main-liners for airplanes, we always could find someone (typically in China) who would take an injured electronics box and strip all parts and do a "bed of nails" to find out full schematic.   Has anyone done this for the KS or Ninebot battery packs?  Would allow us to answer most of the BMS shunting questions very quickly.

Side thought:  done this a number of times in high power helicopters.  When any cell gets low (below 3.0 vDC) a pulse line from each cell goes to a small micro that does three things:

1) start PWM of the power output drives (easy to do for outrunner motors)

2) turns on a beeper or bright LED with lockup to warn the flyer that shut down is eminent but not shut off power for at least 10 seconds

3) monitor temp sensors across battery pack and FETs.  It would store away which area was in excess heat for future readout.

This could give EUC rider ability to sense the "shutter" in forward power and dismount safely.  The LED would show user that undervolt or over current had happened.

I have had many LiIon and LiPoly batteries go to "vent and flame" while in use (one happened while the battery was in my pocket).   There is no measured case where a LiIon or LiPoly when to flames without at least 1 minute of excess over temperature.  I think if more temp sensors were put on the battery packs, you could completely remove the case where flames would occur......it might even give the added parametric assistance to the mainboard to indicate that the battery was setting itself up for overcurrent conditions when strong loads occur.   Did 15 years of LiPoly at Sony, in the old days, when flaming airplanes were part of your basic training.

    tjcooper

[Chriull]

22 minutes ago, tjcooper said:

Does anyone have full schematics of these Samsung BMS packs?

When I was doing all the quad copters and high current main-liners for airplanes, we always could find someone (typically in China) who would take an injured electronics box and strip all parts and do a "bed of nails" to find out full schematic.   Has anyone done this for the KS or Ninebot battery packs?  Would allow us to answer most of the BMS shunting questions very quickly.

The BMS circuitry has the charge side protection (Mosfet for overvoltage shutoff, some/?most? have the reverse polarity diode) and the discharge side protection (Mosfets for overcurrent shutoff) as hobby posted here in this thread in the first post. KS and Gotway's BMS did not have the discharge side protection (KS had an fuse instead of the Mosfets, Gotway nothing) - no idea how the BMS for the latest modells are in detail. Imho there where no pictures posted here till now... Just the new KS from the german KS importer ( @1RadWerkstatt) have an own BMS - imho the number one contact if you are replacing your batteries and look for a state of the art BMS.

These to Mosfet groups are controlled (in most BMS's) from an special IC's which controll each battery cell (under/overvoltage, temperature) and these ICs also controll the balancing of the cells (some with external smaller mosfets and some just with the internal circuitry of the IC). This used IC's where identified by @Craniumfor the Ninebot BMS (including some more details of the BMS circuitry) in this post + the following in : 

Imho noone so far tested how this controller IC's are programmed in detail for the BMS's (threshold voltages, temperatures and delay times)

22 minutes ago, tjcooper said:

Side thought:  done this a number of times in high power helicopters.  When any cell gets low (below 3.0 vDC) a pulse line from each cell goes to a small micro that does three things:

Except for Firewheel's no wheel has a communication between the BMS and the mainboard. The mainboard just measures the total battery pack voltage itself. For low voltages most mainboards start speed reductions, limp home mode and finally some continous beeping mode.

Additionally the current is also measured and the motor power reduced in case of overpower (imho especially regarding the threads with the latest firmware updates, where many people moarn about the reduced performance of the wheels...). And which leads to some overlean scenarios (without BMS cutoff)

The BMS itself checks the voltage/temp of each and every cell and the total output current and shuts off, once certain (preprogrammed) thresholds are exceeded for some (preprogrammed) time.

Normally the mainboard thresholds (should) have a high enough margin to the BMS thresholds, so no unpredictable shut-off (should/)can happen.

The BMS and the mainboard should be well coordinated with the overcurrent protection so that the mainboard gives overpower warning, (reduces power -> danger of overlean!) before the BMS cuts off. But as soon as the battery pack ages and the first cells start to get worse they trigger the undervoltage warning of the controller IC and the BMS shuts off, while the motherboard still saw a high enough voltage for normal operation. Secondly imho the internal resistance of the battery cells rise once the age (?maybe combined with stress/high-load situations?) so also "quite normal" high current situations can lead to undervoltage cutout.

This risks can be eliminated by the shunting - but one should be aware, that with an battery back/wheel that cuts off often one has presumably some bad/aged cells in the pack. In the best case these cells deteriorate so far, that the pack cannot be charged anymore - in the worst case they could vent/burn...

22 minutes ago, tjcooper said:

1) start PWM of the power output drives (easy to do for outrunner motors)

2) turns on a beeper or bright LED with lockup to warn the flyer that shut down is eminent but not shut off power for at least 10 seconds

3) monitor temp sensors across battery pack and FETs.  It would store away which area was in excess heat for future readout.

This could give EUC rider ability to sense the "shutter" in forward power and dismount safely.  The LED would show user that undervolt or over current had happened.

...

This communication between BMS and mainboard should be the next goal for every unyciyle manufacturer - would be imho a big leap for security increase! A BMS and a mainboard working side by side without "knowing" each other just leads to confusion and unnecessary cut-offs.

Also your mentioned memory features (which cells overheated, etc...) sound very interesting and should be implemented with EUCs. Perfect would be an own BMS controller with intelligend firmware, so the thresholds for the single cells would adapt over time regarding their condition. So the speed restriction modes would not start with some abritrary voltages, but just once there is not enough power from the cells anymore. And additionally the cell status could be checked via the smartphone app...

[tjcooper]

UPDATE OF UPDATE:  see my next note that real battery count is 16 cells:

 

I tore apart the BMS from the MoHoo to learn more about why it stops so easily.  It is a Samsung unit with top grade 18650 units...but only 15S.  I only see one set of 4 big FETs to shunt the system.  Does this mean I cannot disable the low voltage/high current shutoff?  What is interesting is that +C and -C go to the Deans connector and is the same 15S that the main XT-60 connector has but has three diodes going from +B to +C connector.  Not sure what that is for unless it "pre-warns" the motherboard that reduced voltage is happening.  C+ is just 0.7 volts below what B+ measures.  I don't know what is under the "black tar" beside the big FETs but I am assuming it is just more smaller MOSFETs to protect some individual cells.  After a short run on the training wheels of 5 minutes, I measured each battery.  All of them (disconnected from motherboard) measured in at 4.04 voltsDC.  So no "weak sisters" were found.

QUESTION: I would like to take another set of 15 of the high quality 18650 units and wire them one at a time across the 15S of the Samsung BMS?  This should roughly double my current capacity.  I am assuming that the individual MOSFETs could handle the extra capacity because the load should still be the same as the original set.  I would just run 16 silicon wires from the Samsung BMS to my remote pack with a quick-release magnetic connector.  Is there any reason to believe this would not stop my abrupt stops because the individual cells cannot deliver the needed current for normal operation?  I want to do this more to stop the "abrupt stops" than to increase driving range.  Any comments would be appreciated.

   tjcooper

 

 

 

[esaj]

As usual, take this with a pinch of salt, as I'm just a hobbyist when it comes to electronics, but:

7 hours ago, tjcooper said:

I tore apart the BMS from the MoHoo to learn more about why it stops so easily.  It is a Samsung unit with top grade 18650 units...but only 15S.  I only see one set of 4 big FETs to shunt the system.  Does this mean I cannot disable the low voltage/high current shutoff? 

Typically the BMSs seem to have 3 or 4 mosfets, out of which 2 or 3 are for the discharge side and 1 is for the charging side. I earlier thought that the discharge-side mosfets were in series (so there'd be one mosfet for each protection, and turning any of them off would cut the power), but it seems that they all get their gate voltage from the same line, so when the gate voltage drops, they all cut off. Probably they're set in parallel to better handle the large discharge currents.

 

Quote

What is interesting is that +C and -C go to the Deans connector and is the same 15S that the main XT-60 connector has but has three diodes going from +B to +C connector.  Not sure what that is for unless it "pre-warns" the motherboard that reduced voltage is happening.  C+ is just 0.7 volts below what B+ measures. 

Probably nothing to do with the mainboard, unless there's some extra connector going to the mainboard for data transfer. Likely the diodes are there for reverse polarity protection, and they've placed multiple diodes in parallel to handle the charging current (small diodes would burn if say 2A would be pushed through a single one).

 

Quote

I don't know what is under the "black tar" beside the big FETs but I am assuming it is just more smaller MOSFETs to protect some individual cells.  After a short run on the training wheels of 5 minutes, I measured each battery.  All of them (disconnected from motherboard) measured in at 4.04 voltsDC.  So no "weak sisters" were found.

No idea what the black "tar" is for, the cells sound to be good.

Quote

QUESTION: I would like to take another set of 15 of the high quality 18650 units and wire them one at a time across the 15S of the Samsung BMS?

AFAIK, placing more cells in parallel with the existing ones should be ok, of course take care to check that the voltages are the same before putting them in. Not sure if using different cells would be ok, although I believe someone said it should work just fine.

Quote

This should roughly double my current capacity.  I am assuming that the individual MOSFETs could handle the extra capacity because the load should still be the same as the original set.  I would just run 16 silicon wires from the Samsung BMS to my remote pack with a quick-release magnetic connector.  Is there any reason to believe this would not stop my abrupt stops because the individual cells cannot deliver the needed current for normal operation?  I want to do this more to stop the "abrupt stops" than to increase driving range.  Any comments would be appreciated.

PROBABLY the mosfets would do just fine, although the extra cells make it possible to draw more amps with less voltage drop from the cells, the motor shouldn't draw more than before... maybe?   Most power mosfets can handle very large currents (tens to a hundred of amps, with very high currents usually the packaging / legs become the limiting factor, rather than the mosfets themselves?), as long as the temperature doesn't raise too high.

Another option would be to build a separate pack with its own BMS, and wire it in parallel to the discharge- and charge-connectors. This might be a better option if you plan on housing the cells in a separate pack anyway, as you wouldn't need to run many wires between the packs, but just make Y-connectors between the charge-port and the packs and the mainboard and the packs.

If the "abrupt stops" are indeed the BMS cutting the power due to too much voltage drop, adding more cells should take care of that, as less amperage is drawn per cell-series / pack, causing less voltage drop.

Yet another option is to bypass the discharge-side protection mosfets (the "shunt" that this thread talks about), but you must first identify which one is the charge-side overvoltage protection mosfet, so you don't shunt that on accident. BUT, as your pictures show the P- (discharge side negative) and the B- (negative of the last cell in series), you can just shunt straight between them, and it will bypass the discharge side protection mosfets without needing to connect the drain and source of the individual mosfets. At least I'm fairly sure of that, maybe better check with someone who knows for sure 

[tjcooper]

UPDATE on MoHoo Samsung power supply.  In my previous note I originally said Samsung was a 16S unit.  Then I did an "edit" later after I opened the shrink wrap and saw the batteries listed as B1 to B15.....so I changed my update to say this was only a 15 cell 18650 configuration.  Tonight I looked at it again...under a large magnifying glass.  They had 16 cells.....but only in 6 point type they made a small single label of B0.   Only a mathematician would label their batteries from B0 to B15.  The rest of the world labels countable objects from B1 to B16.  Anyway, the layout is 16 cells.

I still want to consider putting 16 additional cells external to the onboard pack and connection each cell one-by-one in parallel to the original 16 cells.  I think I will begin wiring up the system.  I will use a 7 ohm power resistor to test out if I can get and keep a sustained 10amp draw until the cells get down to 3.00 vDC per cell.  This would mimic the worst case load on the MoHoo for a 120 KG load going up a 10 degree hill for a long time.   If the battery does not "buckle" on that kind of load, then it should give a real BMS solution to my "abrupt stop" problem.  Any comments?

    tjcooper

UPDATE: can anyone verify that by shunting the -P to -B terminals would disable power shutdown MOSFET circuit?

[Chriull]

 

3 hours ago, tjcooper said:

...  I will use a 7 ohm power resistor to test out if I can get and keep a sustained 10amp draw until the cells get down to 3.00 vDC per cell.

I hope you try this somewhere outside or at a very good vented place. I assume you'll produce some very nasty scent by doing this...

This would be 700W power dissipation - also power resistors need cooling in this range. I'd recommend you do it like @esajwith the "glowing heating wires"....

3 hours ago, tjcooper said:

  This would mimic the worst case load on the MoHoo for a 120 KG load going up a 10 degree hill for a long time.   If the battery does not "buckle" on that kind of load, then it should give a real BMS solution to my "abrupt stop" problem.

Not necessarily - with the resistor one has a nice and neat constant load. While driving there could be quite some peaks leading the BMS to a shutoff.

3 hours ago, tjcooper said:

  Any comments?

Try to get a nice state of the art BMS board since you are already investing in this project...

3 hours ago, tjcooper said:

    tjcooper

UPDATE: can anyone verify that by shunting the -P to -B terminals would disable power shutdown MOSFET circuit?

Hard to say from this pictures - at your BMS it seems that from B- go the shuntresistors (current sense) to the source of the 3 Mosfets (overdischarge protection). Which is quite strange compared to the normally used layout like shown here:

http://forum.electricunicycle.org/topic/459-electric-unicycles-bms-problem-and-solution/

I cannot determine where the drain (pin 2, heatsink) is connected to...

... And the mosfets used could be this: http://www.aosmd.com/pdfs/datasheet/AOT470.pdf? 

Anyhow - seems like this is not the normal standard schematics used (or i mixed something completely up?) - you'd have to work the schematics of your board out...

If you'r sure that the used mosfets are this  http://www.aosmd.com/pdfs/datasheet/AOT470.pdf you can shunt by making a connection between the pins 3 (source) and 2 (drain)

PS.: Are the soldier joints of the Mosfets really so horrible or does it just look like this in because of the a bit unsharp pictures?

[esaj]

6 hours ago, tjcooper said:

Only a mathematician would label their batteries from B0 to B15. 

Or a programmer

[tjcooper]

esaj,

every programmer I know counts :"one banana...two banana...three banana...four banana".  Never heard one say "I have ZERO BANANA on my plate so now I can eat my lunch.   LOL.

Churill,

my 7 ohm resistor is wire wound on ceramic and supports almost 1000 watts.   It is used by my battery tester to validate if my car/RV battery is weak.  Glows in the dark and keeps me warm in the winter time.

Keeping the load near 10 amps for long periods of time (like 20 minutes) should mimic the upper limit of battery drain as people have reported in the EUC forum.  I have a second 7 ohm wire wound resistor I could use to emulate peak current ( the copper traces on the BMS are in the 30 amp load limit).

What would I have to ohm out on the B- side to know that I could successfully shunt out the over discharge?  You talk about 3 Mosfets, but the picture shows 4 and just looking around at their connections they all appear to be in a parallel configuration.  I thought that Samsung made only top quality BMS systems (which this one is) or do they make many different models with different safety factors?   The only connections to the mainboard are the main discharge of the XT-60 and the Deans connector (which I believe to be the charge port for the 16S configuration.  There are no other wires to mainboard (or anything else).  And yes, the wave solder on the board is good for most items, but the big Mosfets appear to be hand done to get the solder to "flow uphill".

I worry that if I used any other BMS on the MoHoo system that subtle interactions with the mainboard could really mess up the rider.  It would cost me less that $100 to get the 16 cells of 18650 from Panasonic.  Putting those in parallel with the existing 16 cells seems like the cheapest test I could make to find out if current "drop out" is the basic problem for the "abrupt stops".   I assume a BMS would be in the $350 range?  Am I missing something here?  Much thanks for all your comments.

    tjcooper

[Chriull]

3 hours ago, tjcooper said:

esaj,

every programmer I know counts :"one banana...two banana...three banana...four banana".  Never heard one say "I have ZERO BANANA on my plate so now I can eat my lunch.   LOL.

That's right for bananas, but at least in C indexing of an array always starts with zero (or one is wasting memory or computing time).

Quote

...

What would I have to ohm out on the B- side to know that I could successfully shunt out the over discharge?

Sorry - i don't understand the question. Or do you mean by "ohm out" using the multimeter in resistance measuring mode? If so i would not advise doing this (or you should know for 100% sure what you are doing) ...

Do know if you can shunt B- to P- you'd have to get the connections of the PCB by visual inspection (best or only possible by viewing both sides...) as it could be that the schematics used for your BMS deviates from the "standard" schematics I have seen by now. 

Quote

  You talk about 3 Mosfets, but the picture shows 4 and just looking around at their connections they all appear to be in a parallel configuration.

The three Mosfets Q47, Q46 and ?Q44? on the "top" aside each other are in parallel. From the photos i can only see, that the fourth mosfet (Q48) has the Drain connected together with the drains of the other three. The Gate seems to be connected to a different "path" (? to upper leg of R105 and then to a leg of Q43?) as the gates of the three other mosfets ( ?they are connected to R102 ?)

The sources of the three upper Mosfets are connected to a leg of R111 and R109 (presumably the current sensing shunts), the source of Q48 is connected with the other leg of R105 and continues to somewhere on the other side of the PCB.

So only three of these Mosfets are in parallel (presumably the "overdischarge protection" Mosfets) and Q48 seems to be alone as "overvoltage protection"

To shunt this BMS one could try shunting Drain and Source of these upper 3 "overdischarge protection" Mosfets. Should work, but without knowing the exact schematics of the "power" part of this PCB always something unforseen could happen.

Is P- connected to the Drain's of all the Mosfets (the heatsinks soldered to the PCB)? If so then it seems that they use the Mosfets switches with invers Drain Source polarity...

This could lead again to the possibility of a shunt between P- and B-....

Quote

  I thought that Samsung made only top quality BMS systems (which this one is)...

The main problem is that the BMS protects the battery. It can be that this BMS is top quality and protects perfectly the battery - but many of this BMS are not really adequate for usage in an EUC. There the specifications have to be adjusted to ensure also a rider protection is in place to prevent cutouts leading to nasty faceplants.

But it could be that "everything" is alright and just the 16s1p Battery cell configuration cannot deliver enough peak current and once you established the 16s2p battery pack everything is fine again and no need for shunting.

Quote

...  And yes, the wave solder on the board is good for most items, but the big Mosfets appear to be hand done to get the solder to "flow uphill".

It's a bit disturbing to see solder on the top part of the Mosfet legs - thats a place where normaly no sodlering should happen. Or at least just very carefully to not overheat the silicon die inside the Mosfet... Imho not really professional for a series production.

Quote

I worry that if I used any other BMS on the MoHoo system that subtle interactions with the mainboard could really mess up the rider.  It would cost me less that $100 to get the 16 cells of 18650 from Panasonic.  Putting those in parallel with the existing 16 cells seems like the cheapest test I could make to find out if current "drop out" is the basic problem for the "abrupt stops".   I assume a BMS would be in the $350 range?  Am I missing something here?  Much thanks for all your comments.

    tjcooper

BMS should be much cheaper from what i remember to have seen around here in the threads - but don't know if they are more adequate for use in EUCs than yours. For $350 you should almost get a 16s2p battery pack including BMS.

Imho the main problem is to have not enough battery cells in parallel. I would never use a wheel with 16s1p since by the peak currents there can be too much voltage sag so the BMS cuts out (or even the mainboard gets not enough supply voltage anymore?). 16s2p should work normally well and starting with 16s4p battery burden should be no real problem anymore. All presuming the use of high quality "high discharge" cells.

[esaj]

7 hours ago, tjcooper said:

esaj,

every programmer I know counts :"one banana...two banana...three banana...four banana".  Never heard one say "I have ZERO BANANA on my plate so now I can eat my lunch.   LOL.

Like Chriull mentioned, most programming languages use 0-based indexing (or offsets). For C, the arrays are actually just continuous blocks of bytes, and you can do stuff like type casting and pointer arithmetic to move around the array, thus 0-offset points to the start of the array (basically the array operators ARE just syntactic sugar for pointer arithmetics, using the array type sizeof() to move the pointer along the memory block)... But yeah, that goes waaay beyond the point, and it was meant as a joke anyway.  

 

7 hours ago, tjcooper said:

my 7 ohm resistor is wire wound on ceramic and supports almost 1000 watts.   It is used by my battery tester to validate if my car/RV battery is weak.  Glows in the dark and keeps me warm in the winter time.

Keeping the load near 10 amps for long periods of time (like 20 minutes) should mimic the upper limit of battery drain as people have reported in the EUC forum.  I have a second 7 ohm wire wound resistor I could use to emulate peak current ( the copper traces on the BMS are in the 30 amp load limit).

With two 7 ohm power resistors in parallel (3.5ohm total resistance), you certainly can get a lot of current out of the pack, but the resistors might not take it, also since the cells DO have internal resistance, they will also dissipate heat (if the cells would have for example 40 milliohm internal resistance each, the 16S pack itself already has 16 * 0.04ohm = 0.64ohms of resistance + how much ever the BMS and wires/connectors add, might need to calculate how much power they will dissipate to make sure they don't overheat?). Also take into account that the resistors need to drop A LOT of voltage too, so leaving out the internal resistance of the cells themselves, at 60V you get:

60V / 3.5 ohm = 17.14... amps

The power dissipation over both resistors will then be

17.14A * 60V = 1028.57... or roughly 1 KILOwatt!   Of course, since there are two similar resistors in parallel, the current divides in half among them, so "only" 500W per resistor   If your ceramics can really handle a kilowatt each, then it's probably not a problem, though..?

I've understood that the 50W / 100W metal plated power resistors can't really withstand even the 50W / 100W power dissipation without being bolted to a separate heatsink? Like the ones I have here:

Wouldn't try to dissipate tens or a hundred watt with those without a heatsink, but yours may be different.

Chriull also mentioned my "glowing heating wires", what I did to test the packs was use Kanthal (Nichrome would work too) heat resistor wire:

Details here:

7 hours ago, tjcooper said:

What would I have to ohm out on the B- side to know that I could successfully shunt out the over discharge?  You talk about 3 Mosfets, but the picture shows 4 and just looking around at their connections they all appear to be in a parallel configuration.  I thought that Samsung made only top quality BMS systems (which this one is) or do they make many different models with different safety factors?   The only connections to the mainboard are the main discharge of the XT-60 and the Deans connector (which I believe to be the charge port for the 16S configuration.  There are no other wires to mainboard (or anything else).  And yes, the wave solder on the board is good for most items, but the big Mosfets appear to be hand done to get the solder to "flow uphill".

I still suspect that 3 out of those 4 mosfets are for the discharge-side protections and 1 is for the overcharge (overvoltage) protection on the charge side. If there's only discharge wires going to the mainboard, then I don't think there's a way for it to "know" anything about what's going on at the BMS, so only measures the voltage (and probably current) coming out of it.

 

7 hours ago, tjcooper said:

I worry that if I used any other BMS on the MoHoo system that subtle interactions with the mainboard could really mess up the rider.  It would cost me less that $100 to get the 16 cells of 18650 from Panasonic.  Putting those in parallel with the existing 16 cells seems like the cheapest test I could make to find out if current "drop out" is the basic problem for the "abrupt stops".   I assume a BMS would be in the $350 range?  Am I missing something here?  Much thanks for all your comments.

    tjcooper

The 16S -BMSs I've seen run about 20-50€ / piece, you can get good quality BMSs from 1RadWerkstatt for example (they're located in Germany). The fourth mosfet is likely the overcharge (charging-side) protection, rest are probably the discharge side protection mosfets. But like said, I'm no expert

[Chriull]

32 minutes ago, esaj said:

...

I've understood that the 50W / 100W metal plated power resistors can't really withstand even the 50W / 100W power dissipation without being bolted to a separate heatsink? Like the ones I have here:

Wouldn't try to dissipate tens or a hundred watt with those without a heatsink, but yours may be different.

...

I'd assume the same - i found a 1kw ceramic resistor with cement coating and in the datasheet is mentioned a maximum temperature of 350 degrees celcius and water/oil cooling.

but imho thats just normal with all power components - they can dissipate a lot of power, but only if properly cooled. If not they just "burn through" like nothing.

like the mosfets used in the euc's - from the datasheet specs they all should handle driving with the euc quite without problems - but since they have no proper heatsink/cooling they just die....(beside the ones dying by bridge shootthrough's)