LVC / HVC / Parallel board run - open for suggestions

[GGoodrum]

Yes, it will, thanks. Also, can +12V be used to bias the HVC optos?

 

[methods]

HVC opto outputs are good from 0V - 18V
They can sink a few ma reliably at 12V as I have them configured (worst case)

-methods

 

[GGoodrum]

HVC opto outputs are good from 0V - 18V
They can sink a few ma reliably at 12V as I have them configured (worst case)

-methods

Perfect, thanks.

When you say there is a 1.2 second delay on the HVC, is this in the form of a hysteresis? What I mean is, if it gets set, after 1.2 seconds, and the voltage drops, resetting the opto, will it again take about 1.2 seconds before it is set again? If so, that will save me some parts, since we won't need a delay we currently have implemented.

 

[methods]

Pack hits 4.28V
Time elapses
Opto triggers after delay

Opto remains tripped while pack above 4.28V

Pack drops below 4.28V
Opto releases immediately

-methods

 

[methods]

There are tons of parts with a voltage window as well.... but I dont like them.
When I hear that something trips at 4.28V and wont let go until 4.1V I hear lock-up condition.
Time is the best way to dampen the signal IMHO

-methods

 

[GGoodrum]

There are tons of parts with a voltage window as well.... but I dont like them.
When I hear that something trips at 4.28V and wont let go until 4.1V I hear lock-up condition.
Time is the best way to dampen the signal IMHO

-methods

I don't like the voltage window ones either. We've been doing the same thing with the timed delay. This is for really errant cell failsafe anyway. Under normal conditions, This HVC line shouldn't trip. The shunt circuits will keep the voltages in check, up to a certain point, and once all the shunts are on, we shutoff the charge current.

I think I read where you will now use another male JST-XH connector on the board, for the balance tap outputs, and then I think you said that you are getting some "dual-headed" JST-XH female pigtails, is that right? Also, what is the connector you are using for the HVC/GND/Bias signals?

Thanks -- Gary

 

[methods]

I am going to use a double ended JST-XH-6/7 jumper pigtail

The 6S boards will get a 5S double ended jumper
The 5S boards will get a 6S double ended jumper
(or I may just settle on one or the other and hope numb-nutz dont blow things up)

My boards have space for a JST-XH-7 so that can be populated with either 5S or 6S board mount JST
The connections are documented in the schematic I posted

http://endless-sphere.com/forums/download/file.php?id=67824&mode=view

Pin
1 GND LVC
2 SIG LVC
3 VCC LVC
4 GND HVC
5 SIG HVC
6 VCC HVC
7 NC

I know it is a little awkward having the same connectors as the pigtails but we settled on it for several reasons:

1) We already have to buy the board mount JST connectors and the more we buy the cheaper they get
2) We also already have to buy the double ended jumpers - again - volume pricing
3) Super cheap and common for users to extend the cable. Any standard Hobby City extender can work to make the daisychain longer
4) Compatible with standard 2.54 spacing... so even 0.1 headers are compatible

I believe we are getting our jumpers in silicone as well... so that is nice to work with for stuffing extra slack.

I realize that it would be highly desirable to settle on either 6S or 5S for the daisy jumper and I might do that.
If I settled it would be on 5S
Most of the customers buying 5S are high dollar motorcycle guys who are (very generally speaking) a lot more careful and less likely to plug into the wrong spots.
Many of the customers buying 6S are building their very first Lipo packs so the chances of miss-wire are greater

Come to think of it... hrmmmm... I dont even know if it would be a problem if someone plugged a live balance tap into a board.
IF they were in a non-tripped condition then it would basically work out to an 8V VCC, ground, and 4V on signal....
BUT - if for some reason the opto tripped it would try to drive the cells on either 1 or 4 to short resulting in burnt up outputs.

yea... (thinking as I write - as usual)
I might settle on 5S for the daisy jumpers

Just lay your boards out for 6S and align to pin one and you can populate with either.

-methods

There are tons of parts with a voltage window as well.... but I dont like them.
When I hear that something trips at 4.28V and wont let go until 4.1V I hear lock-up condition.
Time is the best way to dampen the signal IMHO

-methods

I don't like the voltage window ones either. We've been doing the same thing with the timed delay. This is for really errant cell failsafe anyway. Under normal conditions, This HVC line shouldn't trip. The shunt circuits will keep the voltages in check, up to a certain point, and once all the shunts are on, we shutoff the charge current.

I think I read where you will now use another male JST-XH connector on the board, for the balance tap outputs, and then I think you said that you are getting some "dual-headed" JST-XH female pigtails, is that right? Also, what is the connector you are using for the HVC/GND/Bias signals?

Thanks -- Gary

 

[GGoodrum]

Got it. I went back and reread page 11.

So, for a 6s board, which normally uses a 7-wire pigtail for the cell connections, you will use a 6-wire (5s...) jumper for the LVC and HVC signals. That's easy enough. I will use one 20-pin plug that uses 2 pins for the main negative charge lead, 2 pins for the main positive charge connection (each pin is good for up to 9A...), 13 pins for the cell connections and three pins for the HVC signal, HVC ground and the +12V bias signal. I'll also add the current limit resistor for the LVC, so that there can be three input pads, for the throttle, and three output pads to the controller.

-- Gary

 

[methods]

Errrr... ok
I have not followed any of the BMS work in a long time so I have no idea what is going on with it.

-methods

 

[GGoodrum]

Errrr... ok
I have not followed any of the BMS work in a long time so I have no idea what is going on with it.

-methods

Actually, there's two parallel efforts going on. We have redone the full "Zephyr" BMS, to use our latest charge control scheme, and to make use of the resistor array idea that I stole from you (it only seems fair... ), which drastically reduces the build time. This latest version supports up to 32 channels and is primarily targeted towards larger LiFePO4 setups. I started a second effort to do a complete 12s LiPo solution that basically splits the BMS functions into a pack resident, 12-channel LVC/HVC/parallel adapter board, and then put the balancing circuits and the same new charge controller into one of the small Hammond boxes. I've been testing this concept, which uses a "lite" version of the cell circuits that can manage about 250mA of shunt current, just off a higher power 431 regulator, along with the new charge control circuit for some time now. In my previous test versions, I've included the 12-channel LVC/HVC board, but now that you have your boards ready, I've eliminated this portion, and have added the simple adapter below:



This board has, as its input, three JST-XH connectors, two 7-pin versions for the two 6s balance taps, and one 6-pin version for your "daisy chain" LVC/HVC plug. It also has two 3-pin pads for throttle in and throttle out, and has the 4.7k current limiting resistor. There are two connections as well for the main pack charge leads.The output is a single 20-pin VAL-U-LOK plug. It's mate sticks out one end of the balancer/"CMS" box. The only other connection to the balancer/CMS box is a two-wire connection to the "dumb" CC/CV supply/charger. In any case, there is just a single charge plug coming out of the pack that needs to be connected to charge and balance the 12s pack.

-- Gary

 

[methods]

Ah ha.

Oh boy oh boy oh boy... Just got back from Tap Plastics. I got a set of pigments and some cabosil so now I can control the color, appearance, and viscosity of my epoxy. With my new design I am going to be able to make silicone molds then pour tinted epoxy - yellow, red, black, and a bunch of other colors.

Still have not decided now to pot the part in the middle. Worst case I will make a doughnut ring of silicone and then tape the edges of the connectors with tape that has release on it.

I did not submit my PCB design yet due to a distributer Fing me around on an order for the detectors.... As soon as I get those locked down I will order the PCB's (the last thing I need is 240 PCB's and no parts to populate them with

-methods

 

[amberwolf]

When I still made models and sci fi props, I used to go to a distributor of theirs here in the Phoenix area. I always found it funny that they call themselves TAP Plastics, since the TAP stands for Taylor and Art Plastics. :lol:

 

[methods]

Ok - after battling vendors to get my detectors I have finally locked them in. I ordered 3,000pcs - enough for 500 to 600 boards. I was waiting to lock those in to submit my PCB's - they are now submitted. I ordered 240 boards.

Steve sent me 50 JST-XH-6 (5S) silicone jumpers. These are about a foot long and very high quality - much nicer than the PVC junk from HK. I think I am going to settle on 5S for the jumpers. I also have some 6S jumpers on order for those who want to expand beyond 8P ( >40Ah ) - regardless these 6S jumpers are super useful for all sorts of things and I will sell them as optional items.

I have 3 new silicone molds ready for the new boards. I still have not figured out how to pot the connectors side, but the backs are covered. I am going to try pouring a mold right over a populated board to see how that goes. The JST connectors set very consistently so I think I can just make them part of the mold - pins and all.

-methods

 

[amberwolf]

If you have to, I guess you could put a dab of vaseline into the connectors on the PCB, so the potting doesnt' get into them. It'll just squish out of the way when it's plugged in, and FWIW would also be a bit of additional moisture-proofing in actual use.

I don't know if the JST connectors have enough contact force to guarantee to squeeze the vaseline away from the contact surface when plugged in, but Anderson SB50s, PP15/30/45s do, as do old IBM AT and PS/2 keyboard/mouse connectors, and the good brands of DB-xx style pins, as well as RJ45 and RJ11 connectors. Those I've used that way before for water resistance/moisture proofing.


Alternately, if you can take a test board and coat it with mold-release (usually PVC/alcohol mix, sometimes wax/alcohol), then fill the JSTs with it, plus a little bit extra over the top (kapton tape around the edges to let it build up, perhaps), you could make a silicone "plug" mold to cover the JSTs with, then put the whole thing in your regular mold. Pull the plug out when potting is cured.

 

[methods]

If you have to, I guess you could put a dab of vaseline into the connectors on the PCB, so the potting doesnt' get into them. It'll just squish out of the way when it's plugged in, and FWIW would also be a bit of additional moisture-proofing in actual use.

Dude... that is a really good idea. The JST's should be packed in dialectic grease anyhow... but I dont know how I would sell customers on that. Most people are not aware of the properties of dialectic grease - I think most people assume that things are either a conductor or an insulator... and it is a slippery greasy mess.

I would be a little bit worried about conductive particles getting into the grease too - like aluminum or steel dust. The grease is at its best in a sealed connector.

I will give that a think though as it would certainly make them more water resistant.

-methods

 

[jonescg]

Patrick, to save me from scanning the whole thread again , these little boards allow for paralleled balance leads AND are able to output a signal should a single cell drop below 3.5 V or exceed 4.2 V right? This signal can be a relay or a buzzer or something right? So it's sort of like a Battery Monitoring System?

I'm interested in them for my next bike should I go LiPo. Still unsure what the best way to go is - pre-built 6S packs or a bunch of individual 40C Turnigys.

Cheers mate,
CHRIS

 

[amberwolf]

Patrick, to save me from scanning the whole thread again , these little boards allow for paralleled balance leads AND are able to output a signal should a single cell drop below 3.5 V or exceed 4.2 V right? This signal can be a relay or a buzzer or something right?

Basically, yes. Although I think it's 4.3V HVC, IIRC (basically just an emergency HVC to prevent cell damage). The signals for LVC and HVC are separate, and they can be used to trigger whatever you want, with the appropriate circuitry between the signal and what you are triggering.

So it's sort of like a Battery Monitoring System?

Personally I'd call them PackWatchers.

Don't want them to get confused with the Battery Murdering Systems usually associated with "BMS".

 

[amberwolf]

I would be a little bit worried about conductive particles getting into the grease too - like aluminum or steel dust. The grease is at its best in a sealed connector.

Well, I gotta say that if people get that in their connectors, they probably haven't got their pack protected well enough. That, or they ought to take the pack off the bike before they start hacking on the frame. :lol:

That said, I guess it could be a problem, since these things are designed to be environment-resistant. Not sure what to do about that.

 

[methods]

I finally had time to go back and design an HVC interface to control charge. I went with a universal 12V output that can control any relay and the idea is that this relay could either break the 110V AC side or the DC charge side. If you recall the concern was that if a relay was directly controlled from the HVC output the circuit would oscillate. There is a second or two of hystrysis built into the HVC detector (it will allow the cell to break 4.29V for a second or two before latching) but that was not quite enough. I set up an additional 5 seconds of "hold time" so that if the circuit trips it will keep the charger off for about 5 seconds.

Basically I am supplying a 12V DC-DC that powers the HVC opto output and the gate of a fet. There is a 1uF cap on the gate and it charges through a 10M resistor. When the circuit is energized it takes about 5 seconds for the cap to charge to the point where the fet will turn on. Once that fet is on current can flow through the coil of the relay and everything turns on. The HVC Opto Output is in parallel with this gate capacitor - if it closes the gate cap is quickly discharged which opens the relay. The relay remains open until the HVC opto releases. Once it releases it takes an additional 5 seconds for the charger to come back online.

For the final circuit I might switch to a 0.1uF cap and a 100M resistor to reduce stress on the opto (it has to dump all that energy). Alternately I may end up adding a small current limiting resistor between the capacitor and the opto output - we will see. I dont like to ever use resistors larger than 10M because things like finger grease, epoxy across temperature, and dust start to affect the circuit :x


Anyhow - the circuit will be compatible with any 12V relay - be it a small 10A for the AC side of the charger or a huge 300V 2000A contactor. The DC-DC can take 110V AC, 220V AC, or any DC over 50V. I will make the relay an option... there are tons of possibilities here -> Some folks could bypass the relay and directly control their DC charge current with the fet (though I don't recommend that due to the noise that bounces around when big DC currents are broken).

I will pot up 10 of these and make them available at a discounted price to those who participated in my early board runs.

Basically it will look like this:

12V Wall Wart
Small cube with 6 wires

12V bias in (connected to wall wart)
12V GND in (connected to wall wart)
12V out to relay
return from relay
HVC control in
HVC GND



Folks who want to use a relay that works on a voltage other than 12V will just tie into the 12V ground (with their ground) and drive their relay with whatever voltage they want (say 72V). I will populate these with IRFB4115's so you will be good up to 150V. So 72V into coil, coil return goes to relay return, 72V ground goes to 12V ground.

-methods

 

[methods]

The more I think about it the more it makes sense to just bypass the relay and do an inline mosfet with the charge current. I have a lot of IRFB4115 fets - that should cover 99% of ebikes out there. If we do it this way folks can build this into their pack and have it much more plug and play.

I am moving forward with the relay version but I am going to see what it takes to do a 48V to 150V 20A package. The only trick is controlling power on the DC-DC - can be done with a giant Schottky diode inline with the charge current or with a 3 prong charging plug where the charger has GND, DC, and a shorting plug to link the DC to the switcher.

-methods


Edit -> Seems better to replace the Schottky diode with an inverted IRFB4115 - hook the fets up Source to Source and they can both be controlled by the isolated 12V. The DC-DC can be hooked up to the charger side all the time - no need for a shorting plug. User just hooks up + and - to charger. Now to see how many amps I can put through the 4115's before they need heat sinking. Probably going to need a few of them (lowering IR being preferable to heat-sinking). Maybe make a 4110 version for those under 100V.

 

[methods]

Sweet - got the second circuit working.



This circuit is intended to be built into the bike or battery. It goes between the battery and the charger and ties into the LVC/HVC boards. When the charger is hooked up this little circuit turns on and allows charging. If the HVC triggers it cuts charge and holds it off until the HVC condition clears. It then delays 5 seconds and resumes charge.

The circuit draws nothing when the charger is not attached. I use a mosfet to isolate the small DC-DC from the battery so it can only draw from the charger. Active draw is about 6.6mA and inactive read 0uA on my Flukes lowest setting.

The fet that isolates the DC-DC from the battery gets turned on instantly and the fet that isolates the charger from the battery is on the delay.

I tested this with my 66V 10Ah lipo pack and a 75V 17A charger. The fets used in this example were IRFB4115 150V fets and they got a little hot. Dropping down to an IRFB4110 will probably keep it cool enough to not need double fets. Even so - I might double up on the fets to halve the rdsON resistance.

This design will scale to pretty much any voltage.
No relays or mechanical contacts
No need to run LVC Opto wires all the way over to the charger
No fooling with AC lines
No shorting plugs or silly charger plugs - just Charger(+) and Charger(-).... Exactly as you would normally hook up.

If I put a JST-XH-6 connector on here it will plug-n-play right into the existing LVC/HVC/Parallel boards. I will leave bare 12awg wires so folks can use whatever connector standard they have settled on.

RUN A DAMN FUSE BETWEEN THE BATTERY AND THIS CIRCUIT. ALWAYS A FUSE! ALWAYS!
Some people dont like to run a fuse on their main power - I understand that - but there is never an excuse to not fuse the charging line. That should always be fused with a fast blow ~30A fuse just for the case where the shit hits the fan..... and it does from time to time.... so FUSE UP.

-methods

 

[amberwolf]

What about these? Could they help?
http://www.endless-sphere.com/forums/viewtopic.php?f=2&t=33850

 

[methods]

Those look cool - voltage is a little low.

I think I have it licked with the fet pair.

Working on a version now that uses a resistor-zener to drop pack voltage down to 12V then goes through a super small VBSD1 isolated DC-DC. This would eliminate the big bulky switcher and increase reliability.

-methods

 

[RallySTX]

Sweet Jesus what a read! I learned a few things, and a LOT more went over my head, but that's me. Looking forward to ordering six of these for my bikes. I'm interested in harness solutions like a jumper wire for each connection, so I don't have to ruin what I have, and to aid plug and play, for my one armed self. I like lvc and hvc for safety's sake. The charger minder on this page is a great idea, I would also be interested in. If I remember right, you are also offering balance lead jumpers to add multiple packs in parallel to the boards? I'm just not sure what to hook the boards too. Either the CA, or the controller is my guess. An adapter pigtail with matching connectors would work for me. I'm reluctant to cut off connectors or open up new and expensive components. If you already addressed that issue, please forgive my ignorance. I have been reading too long. Took me almost four hours to read this, and the only issue I have is the lost pics. Happy Thanksgiving Pat, and everyone else too.
Brian L.

 

[dougnutz]

I got three of these boards from Methods for my upgrades over the last month. Firstly these things are tiny and beautiful. They are about the size of the end of a 6s nanotech. I think you could actually mount them right on the end of the pack, so it’s super easy to find room for them in your pack.
Anyway hooking those up is pretty straight forward. Just pay attention to where the resistor goes. I recommend making a little jumper harness to make things a little more plug and play, but that’s totally optional.Here is what I did.I traced the all white wires with a marker to make it more clear where they go. The connector mid line it to make a clean break so I can unplug the pack from the bike without takeing the main 6pin connector appart.

But that said I did run into a snag when using this board with my new Lyen controller. This isn’t a problem with the boards at all or with the controller but rather the combination of the two. I thought I would share the details; when I hooked things up the throttle only responded from the half-way point on. And I had a seriously nerfed top speed. I’m a bit of an amateur when it comes to diagnosing electrical circuits so I had to lean heavily on Methods (via email) to help trouble shoot. It turns out the Lyen 12 Fet extreme edition controller has only about 5kOhm between the throttle signal wire (green) and ground. For comparison the Ebikes.ca controller I have has 68kOhm.

I can’t explain the nitty gritty of why, but some resistance between signal and ground will help ensure the throttle stays at 0 volts when disconnected (a good thing).

In the end I need a much smaller resistor, something like 1kohm or possibly 500 ohm instead of the 5kohm recommended. I didn’t have any so I twisted several 5.6kohm resistors together in parallel to get down to about 1.4kohm and the throttle worked normally. I don’t think it had 100% but it was close and I probably would not have noticed it if I had not been checking. Tonight I’ll head out and grab some smaller resistors to confirm the fix.

Oh yea, the recommended 5kohm resistor worked fine with the controller from ebikes.ca.