Chevy Volt battery GEM conversion

GEM Forum
1

I was asked to provide details of the Volt battery conversion to my GEM.
I am not finished, but the vehicle is operational daily. I will have pictures of the conversion later. Please ask away and I will attempt to answer. I will bore you with some theory and my thinking behind the conversion so keep that in mind.

I wanted to end up with a long-lasting conversion which was turnkey and easily done. I wanted to use as many stock parts as possible including the charger so the upfront costs were low and skill level to convert was minimal. From my research, this appeared to be a reasonable goal.

Mine is a 4 seater GEM 2002 with 5hp motor.

I initially bought 6 new US Battery 31DC batteries and drove around a bit. Well, OK, but from the very beginning I wanted to use lithium. You can buy the basically unused batteries from me- check the for sale section :slight_smile:

After reading, talking to folks, and checking forums (not necessarily in that order) I decided to use Chevrolet Volt modules. Mostly because I hadn’t seen it done before. Also, the NMC LMO or lithium nickel manganese cobalt oxide batteries are USA made by LG and have an amazing track record. Since 2010, Chevy reports battery degredation of 3% and they claim they have never made a warranty replacement for a battery. They have a pack they cycle and test frequently. Every once in a while they test the peak intermittent power output and it’s somewhere near half a megawatt. Well over 400KW. Good grief. Gotta get me one of those for the GEM, right? That’s over 500 horsepower…

Thousands of cycles and only minimal degredation? Serious power with minimal sag? Relatively safe chemistry with no random fires? Sign me up.

In terms of the long life and minimal degredation, it seems that the secret is how the battery is managed by the on-board battery management system. This isn’t a charge it up to 4.2v and run it down to 2.5v battery. Chevrolet intentionally takes around 17kwhr battery capacity and only uses around 11 kwhr (from memory, excuse if incorrect but you get the point). Chevrolet says they use 65% of the SOC. The strategy and set points are proprietary, and Chevy won’t divulge. However, by checking on Volt boards and reading what folks are posting about their aftermarket state of charge (SOC) software is reading, it seems the Volt never charges above 90% (varies somewhat based on environmental factors, age of battery, and probably tens of other parameters the Volt’s computer is using) and doesn’t go below about 25% SOC.

Translating to battery voltage per cell, that’s approx. 4.1 volts charged and 3.6 volts discharged. A NMC LMO battery is nominal 3.6-3.7v with max charge 4.26 (that’s max max) and discharged at 2.5. At 3.0 volts it’s essentially done and going to 2.5 really doesn’t get you much more energy at all. Remember, while lead acid charge vs volts is somewhat linear, the volt to charge on a lithium cell is not. That’s why it’s not as simple as placing a volt meter on a lithium cell and estimating state of charge.

I’d noticed on the lead acid packs that in the hilly area I live in, there was a really bad voltage sag even with freshly charged batteries. Say hello to the Peukert effect.
It takes time for a battery to sulfate its cells and make electricity, since it’s a chemical reaction. I figured if the batteries were to live, I’d never go below 50% SOC. Factor in the Peukert effect, which halved the battery capacity again, and I was looking at an effective 32 or so amp hours of usable battery capacity. Nuts!

Anyway, the first thing I did was to get a 72v Volt battery module. Each module is rated 54 amp hours. Each and every cell has Battery Management System monitoring using the stock plugs. The BMS is an aftermarket system, since at this point to use the CAN BUS system you need a working Chevy Volt. Nobody has cracked the codes to use the OEM BMS modules yet. Bummer. The aftermarket BMS is active 24/7, even with the cart off (you can’t turn it off unless you unplug from the battery). It will balance cells even when not charging. The only modification I had to do was to connect the negative lead of the Zivan charger to the BMS main board. So, the Zivan charges the battery through the BMS. THe BMS can accept a charge of up to 30amps, so it’s well within parameters. Connecting the BMS to the Zivan took literally 3 minutes and I fished the wires (each module has a BMS with a separate negative wire; right now only one BMS is wired to the charger which is risky and I will eventually wire all the BMS to the Zivan). through the base of the windshield. I forget about the car charging, fall asleep or whatever, the BMS will eventually shut off the negative lead of the charger and the charger will shut off. No harm is done and it cannot overcharge the battery. The one BMS shuts the charger off when all of its individual cells reach full voltage; I’m measuring that with cable voltage losses the pack in the rear is 0.5v lower than the front pack so I’m assuming there isn’t a problem. I will, however eventually put all BMS wires to the Zivan.

2

I first ran the GEM on a single battery (one 72v module at 54aH capacity) situated up front, which took the space of just over 1 battery. It was weird to have an empty rear battery compartment after seeing so many pictures of LiPO conversions with 24 batteries stuffed everywhere. As shipped it had a voltage of 72.2v, and I couldn’t wait to charge it, so I zipped around. Seat of the pants as well as speedo on the hills proved it was about as fast as the lead acid pack and 1 mph faster on hills (18 vs 16-17). I don’t have an accurate amp meter on the pack nor a voltmeter yet that works with the car running so I can’t tell you what the voltage sag is. I assume the difference in hill climbing is due to the lower weight. It’s pretty amazing that one of these modules supplies more usable energy than the effective capacity of the lead acid pack, taking into account SOC and Peukert limits.

I made an assumption (which was wrong) that the 72v module was based on the nominal voltage of 3.6v and therefore it was a 20S, or 20 cell, module. It sounds crazy that I would drop the kind of money (more on that later) I did and not know exactly everything about it, but it was a rookie mistake. Turns out, the 72v module (which is a 48v module and a 24v module placed in series) is an 18S module.

18S x 4.2v maximum is 75.6 volts at max charge
18S x 3.0v minimum is 54 volts at minimum.

With 2 of the 72v modules in parallel, (72v at 108ah), there is a slight improvement in performance. Another MPH up the hills. I didn’t expect much, because the pack doesn’t sag much and the additional ampacity wasn’t much of a performance boost.

Problem is, I can’t run the cart below 69v currently, because I’m using the GEM’s on board Zivan charger which will not charge a pack with a voltage lower than 68. Plus, I haven’t reprogrammed the controller (If that’s even possible to change the setpoint) and it starts up the low voltage buzzer at 69v or so. So I’m only using a fraction of the pack’s capacity. Plus, the GEM is very voltage dependent for performance. There is a significant difference in performance at 69v vs 75v. Which is to be expected, since the lead acid battery pack voltage change from charged to conservatively discharged is around 6v.

Ok, so to prove a point to myself, I put a 12v battery in series with the lithium pack at 75v. I triggered the high voltage fault, so put the lights on (which at the time were incandescent stock) and pulled the voltage down to where the fault went away.

Excellent! power was incredible- front wheels spinning on takeoff and zipped up the hill at 25mph where it was 18 max before. Call it the first lithium lead hybrid.

I only did this for a mile or so to prove a point since mixing different chemistries is sure to be a problem, but it was fun.

So, the cure to my problem is more battery cells. Now, mind you, the cart is completely usable with 1 or 2 modules at 72v nominal. Ideally, 20s would be perfect. The battery management system on the cells cuts off at 4.20+/-0.05v, so even at max charge of 4.25v you have 85v pack. Which is below the high voltage cutoff of 86v on my cart. I would not be able to charge the cart too high even if my safeguards failed. Using 3.5v as a conservative low cutoff, that’s 70v which is above the minimum voltage for the Zivan charger.

Well, there really isn’t any way to get individual cells out of the Volt pack. So a 20S is out. You can, however, rewire a 24v (6S) 54aH module to a 3S pack (12v at 108aH) so that is what we did. I don’t have the additional battery pack yet. I was waiting to make this post until after I had the pack finished, but :smiley:

Once that is placed in series, I’ll have a 21S pack. Running a max pack voltage of 86v puts the individual cell voltage at 4.09 which is right about where the Chevy Volt cuts off charge- about 90% SOC. Using a cutoff voltage of 3.5v per cell puts me at a pack voltage of 73.5v which is well above the Zivan’s minimum charge, and is right around where the Chevy Volt maintains minimum voltage. Voila! All I need is some way to reliably charge to 85v.

For charging, I have used the Zivan which has been modified by Zivan with some additional profiles. At the charge currents it’s capable of at 115v, it amounts to a trickle charge of about C/10, so whether it’s a CC or CV charge profile is probably irrelevant (could be wrong). It had no problem charging the 72v 108ah pack; it was just slow.

I’m currently using the on-board BMS system as a hard stop on the cell voltage max. However, I have the Zivan’s cover drilled to uncover the potentiometers and plan on using the voltage control to set max voltage of 85-86v. If it messes up and overcharges, the BMS will cut off but I’ll have to drop the pack voltage before operating. To make the Zivan accurate, I will have to modify it by cutting off the temp probe and substituting a resistor. I have an extra Zivan I’m working on in case I release the smoke. Eventually, if this works, I plan on isolating the chargers and will install the other Zivan to increase charge current.

I have plans to install a J1772 charging receptacle with the logic board that allows operation. Then with 230v, putting a Zivan on each leg of the power. That’s down the line and I think I’ll run into problems isolating the chargers from each other and I do not want to make a cart that I have to switch this and unplug that to charge. I want it completely automatic; just like a stock GEM. I have budgeted money for a new 230v capable charger (Elcon).

3

I’ll get some pics of the install so you can visualize what I’m saying. It’s incredible how small the modules are. I am not finished with the wiring; I have the 1/0 cables and connectors but am waiting for the 12v battery module to arrive before finalizing. My apologies for the roughness of the install.

4

Wow. Sorry I found the thread from Kingfisher and Inwo and others about using the Volt batteries. Sounds like I could have saved a bit of grief by reading it first! Thanks guys…sorry I didn’t find your thread I was looking for Volt keywords in the titles…
Sounds like we’ve had similar experiences. Can you guys post up exactly how you’ve mounted yours? I figured if I posted my batteries held down with ratchet tie downs I’d be pilloried but I’m in with kindred spirits!

BTW I’ve had no problem running full Zivan output through the factory BMS plugs.

5

I tried re-configuring for different voltages, but decided that the simple way of using 24s, works best for me.
Tapping the cells is not a safe or easy task.
It was easier to spoof controller into accepting higher voltage. I charge to just under 100v, using DQ on highest lithium setting.

An aluminum angle work good for mounting. Picture of one on the shelf.

http://www.electricforum.com/cars/gem-nevs-forum/10006-another-lithium-thread-9.html

6

1/0 is way overkill. Stock cables worked for me. A little resistance is good to limit fault current.

7

Here are some products tailored to Volt.
Allows use of factory cables.

8

David

Give us more words on spoofing the controller

Rodney

9

The die is cast, so to speak. Having a 1KWhr module tapped to 12v. I’m having it done rather than doing it myself. If I pop it back to 24v, I’d need to buy another one to get 100aH and with the BMS I have that gets expensive; I should have just got 4 of the 48v modules off the bat. If it doesn’t work out, can you give us more info on how exactly you’re spoofing the controller? Op amp? You’re letting it see a fixed voltage less than pack for a time until pack voltage drops below 86 then disabling spoof?

I know it will run to voltages around 96 in operation but not on startup so I’m assuming that’s how you’ve done it.

In my very specific instance, I have to be careful. in my commute, I go up a hill gaining maybe 70’ altitude right off the bat then drop down a few hundred feet over the next 2 miles; there’s a lot of regeneration. I’d hate to overcharge the batteries so I figured with my plan of charging to 4.09 even if they get a lot of regen current (some folks have posted up to 200A of regen with Li batteries) they won’t be damaged.

10

Any progress to report?
Is this a similar bms? It’s from Ebay seller. I’m concerned about there being no connection shown to B+.
Bms port only supports 5amps.
Same seller makes a tapped Volt battery. I’m wondering if he did a better job designing the jumpers.

11

I might not understand how it’s connected. The charger neg. goes to red?:confused:
Charger pos. to B+?

12

I’ll find a link to the BMS so I don’t post misinformation but that looks exactly like the project boxes enclosing the BMS on my batteries. The BMS are Chinese and wired similarly; I don’t know the seller on eBay though. Could be my guy buys from them. I should have waited a bit and researched it longer, frankly. I suppose I could have done a lot more myself. However, I’m happy with the performance of the system and no smoke has been released so far.

The 21 cell system works very, very well. The stock charging profile for the Zivan cuts off at 85.4v (flooded cell profile) per the sticker on the modified charger and per my Fluke meter it is accurate. That gives 4.06volts/cell max.

I have been charging through the BMS ports with >5A without any problems. However, I’m going to change back to the stock charging connection. I note that there is a voltage drop at the controller, where the original negative lead from the charger connects. IF you take this voltage drop into account I’m thinking it will drive the pack voltage higher closer to 4.11v/cell? I have run the pack down to around 70v. I have the luxury of making short trips so I can get voltage down to where I need it. Like you noticed, there is a pretty knobby knee at around 3.6v/cell so going down below 74v doesn’t get me much.

Charging takes forever at Zivan rates. Seriously thinking about a new 230v charger but realistically, it’s not good money spent since I don’t use more than a small part of the battery capacity with each trip and I have the luxury of time to recharge. SO the Zivan makes more sense. I would only get the charger if I could figure out a way to (legally) get out of the bottlenecked area I’m in so I could do longer trips. unfortunately, I’m surrounded on all sides by roads with posted >50MPH speeds. I can go shopping and take kids to school and do almost every errand; but I can’t go to work. I live right next to Leisure World in Southern CA which is the hotbed (so to speak) of golf carts cruising city streets. If I can just make it over there; per law they can use within 1 mile of a golf course and lots of otherwise >45mph roads are marked with a golf cart lane.

I’ve not seen more than 2 volts pack voltage drop under load. Controller gets “choppy” at over 80v; I think it’s how it was reprogrammed after the blue motor install since it wasn’t doing it before. Switching to Turf mode gives me a bit more power than stock 5HP motor/Lead acid combination had and better modulation without choppiness. I leave TM on for the wife to drive it. Also, there is some arcing I hear of the motor contacts (per seller, brushes need to bed in) at speed and load. It’s better with time, but I cannot do a full power run up the hills due to it yet. Almost. Ive not seen any talk about this on the boards before so hope I don’t have a bad unit but it does appear to be improving. I can tell you the controller does not like it at all and will stutter if I let it go on too long or get too loud.

So far, so good. 12v tapped battery mod has lived. 21S seems to be just about perfect in terms of usable capacity and not having to modify any of the other systems.

I

13

Thanks for information.
Keep it coming. I’ve been waiting for another Volt user. :clap2:
I would like to see a picture of how battery was tapped.
I run stock motor and programming with no choppiness, except when foot bounces during initial start from stop. Never have to run over 1/2 throttle. Over 30 uphill.

Just notice the ebay seller has a 96v pack with bms.
Do you use separate bms on each?

I paralleled Volt packs on my 48v Tomberlin. Made a breakout board that parallels each cell thru bms ports. Run two 72v packs. Stock Volt orange jumpers. Wireless power monitor shown on B-.
Complete Volt battery was <$2500. Enough for Tomberlin and two 100v Gems. :slight_smile:

Charge Tomberlin with DQ72 and Gem with DQ72 and ~98v profile. Bms is absolutely not needed with proper charger.

14

Any updates?/

For followers. 1 Volt assembly is enough batteries to do 4 -50ah Gems or 2- 100ah Gems.

I went the junk yard route myself.:wink:

15

Being a noob to all of this I would be curious as to how you verify the battery pack in the salvage yard is working? Besides the obvious signs of physical damage of course.

Thanks

Dino

16

Luck I guess.
Ask about model year and mileage.
Hope for an honest seller.
If not a gambler, buy new on Ebay.:wink:

17

[quote=Inwo;33106]Luck I guess.
Ask about model year and mileage.
Hope for an honest seller.
If not a gambler, buy new on Ebay.;)[/quote]

Fair enough. Just thought there might be some things to look for that you might want or want to avoid.

18

can you post some pictures of the install ?

I also started with 20 cells (leaf modules) 83volts and this weekend well be going to 22 cells and 90volts . the stock motor loves the higher voltage!

19

Just bought my first 2002 gem 4 seater. I’d like to learn more about the lithium packs and cost. I work in OC but live in Corona. if you can get me pointed in the right direction that would be great. Troy
Not sure how to send a private message.

20

troy ,

I was born and raised in oc . glad to help u with what ever u need .