LVC / HVC / Parallel board run - open for suggestions

[methods]

I am going to be doing a another run of LVC / HVC / Parallel boards for a customer. I am going to spin off a few hundred extra for you guys. I am pretty firm on how they are going to look but if you have any ideas for making them better I am all ears.

This is basically what they will look like

1S - 6S (populated to suit)
1P - 8P (maybe less... like 4P expandable)
JST-XH vertical board mount connectors
Lipo or LiFe HVC
Lipo or LiFe LVC
Hopefully 2 layer with no silk or mask
foam sticker on the bottom for insulation/padding
Dip/spray/shrink to cover the components on the top
double-ended JST jumper for expansion

I am going to use the same basic design that Fechter/Gary/Kendrick have all been using that was originally introduced by Silicium (IIRC) back in the day. These will be fully populated and tested. Price will be similar to what you have seen offered by the above Gentlemen - perhaps a bit more since I am not interested in going broke.

No kits - takes longer to kit them up than to just populate them
Raw boards will be available to DIY'ers of course!

Not sure how I am going to handle the eBrake out line yet... will probably just leave two empty holes on the board. I have never cared for the 2 pin RC type JST connectors

Anyhow - if you know what I am talking about and you have been mulling this over you have a week or so to sound off. I am going to start parts selection and layout tonight.

Please dont ask "are they dont yet? are they done yet?" This may take 5 weeks and it may take 5 months - all depends on how busy I get with other things. I will not be taking any pre-orders but I am looking for an "angel investor" as I am pretty broke these days.

This will be "open source" so I will post the design and population techniques as I go.

-methods

 

[methods]

Thoughts

1) Is anyone really using the HVC feature? Prices could come down significantly without it.... In my experience most people are simply hooking the opto-out to the throttle for LVC, right?

2) What if I plugged JST extensions into the board and then encapsulated the boards in epoxy? They would be shock/water/short circuit proof. Maybe a silicone... maybe epoxy.


-methods

 

[methods]

reserved

 

[bigmoose]

Methods, did you find any TC54xx equivalent parts with 3.4 to 3.6 volt thresholds? Wondering what LVC voltage detectors your experience dictates for LiPo and LiFePO4 when using fixed voltage detectors?

 

[methods]

Methods, did you find any TC54xx equivalent parts with 3.4 to 3.6 volt thresholds? Wondering what LVC voltage detectors your experience dictates for LiPo and LiFePO4 when using fixed voltage detectors?

The TC54 comes in more flavors than the datasheet lets on - what do you need 3.4V or 3.6V for? Of course you could use a divider on the front - but parts count parts count...

I am going to stick with the TC54 for LVC on this design (since it is a 1 part wonder) but I am hopping mad trying to replace the HVC section. The original design uses an LM431 - but that requires a divider to trigger the input to the LM431 and a current limiter to allow it to pull the opto down. That is 1 part to handle all of the LVC and 4 parts to handle the HVC :?

The trick is to find a supervisory chip (reset chip) that has an active high instead of an active low - and one that will trigger around 4.2V. The active high part eliminates 1 resistor - so parts count drops to 3. Finding one that will trigger at 4.2V would drop parts count from 4 to 1

I have a few prospects... one will trigger at 4.38V and I am thinking of ways to trick it with a diode to cut the parts count down to 2 from 4.


-methods

 

[methods]

Example:



U15 will pull its pin 1 low if its pin 3 drops below 3.0V
This activates the opto

When 4.17V is applied to the divider R24/R23 then U14 will kick on and also activate the opto

Now in your head replace U14 and R21 with something that acts like a TC54 only active high instead of active low... now we can remove R21 - yea! There are things like this available - off the top of my head an MPC810. So this puts us at 3 parts instead of 4....

I just started thinking about the next step of ghetto which would be to replace the voltage divider with an inline diode - by playing with different diodes we may be able to drop just the right amount of voltage to have it trigger at 4.2V, dropping us down to 2 parts per channel for the HVC.

I think I can find one that will trigger at 4.2V though... Not too worried about hysteresis since this is not part of a BMS that is driving shunt resistors. This is really an "OH CRAP" protection circuit. Heck - I probably wont even be populating the HVC section for most people if that will allow me to shave a few bucks off the price.


-methods

 

[bigmoose]

Methy, good to have you back, designing and posting! Congrats on your designs running on the circuit last weekend!

I don't like a voltage divider in front of the TC54xx's because of parasitic current draw.
I don't like the TempCo on a PN junction... like 50mV over 25 deg C for use in a voltage reference application.

I like your drive to reduce parts count. I like a hardware OVP as backup protection, as certain manufacturers a hardware base supervisor over a smart BMS if the micro/firmware based BMS "hangs."

The mid 3.x V LVC I was thinking of was for application to nano LiPo packs.

You buying you TC54xx's in 50,000 quantities again to get any voltage you want?

BTW take a look at NCP303LSN42T1 and see if you can use pin 5 in your design. Currently stocked by digikey, arrow and avnet.

 

[liveforphysics]

I like it!

This is going to be a must-have board for everyone building new RC LiPo packs, or looking to update a pack they slopped together before and would now like to add some safety, neatness, and super ease of combining all the cell taps without spending the hours soldering them up that I always end up doing while cursing and wishing a board like you're describing existed.

 

[EBJ]

What would the HVC be used for?
I imagine to keep one from over-volting when regen-braking? If so I wouldn't mind that feature missing, as I plan to just charge to 90% w/ the Hyperion. (not sure if other chargers allow you that freedom, but it seems like a simple fix to eliminate an HVC need, while also extending battery cycle-life)

I'd also like an LVC of around 3.6, maybe 3.5.... as I've noticed my cells differentiate greatly after that 3.66V mark.

 

[methods]

Methy, good to have you back, designing and posting! Congrats on your designs running on the circuit last weekend!

I don't like a voltage divider in front of the TC54xx's because of parasitic current draw.
I don't like the TempCo on a PN junction... like 50mV over 25 deg C for use in a voltage reference application.

I like your drive to reduce parts count. I like a hardware OVP as backup protection, as certain manufacturers a hardware base supervisor over a smart BMS if the micro/firmware based BMS "hangs."

TempCo has been a problem in the past - especially when sitting right next to 5W shunt resistors This caused all sorts of trouble with early BMS designs around here.

The mid 3.x V LVC I was thinking of was for application to nano LiPo packs.

You buying you TC54xx's in 50,000 quantities again to get any voltage you want?

BTW take a look at NCP303LSN42T1 and see if you can use pin 5 in your design. Currently stocked by digikey, arrow and avnet.

That NCP certainly is a nice part. Too bad everybody wants to use a 200mV hysteresis. If I selected the NCP 4.0V part it would *almost work*.... as my voltage crept up to 4.0V + 0.2V it would display the correct behavior - but then it would not let go until voltage dropped below 4.0V creating a lock-up condition between 4.0V to 4.2V. Even if I could get past that the part is only available in active low for the open drain config - if I want active high I have to go to the push-pull which adds another layer of complexity / one more part.

Ima keep on looking though :wink:

-methods

 

[methods]

What would the HVC be used for?
I imagine to keep one from over-volting when regen-braking? If so I wouldn't mind that feature missing, as I plan to just charge to 90% w/ the Hyperion. (not sure if other chargers allow you that freedom, but it seems like a simple fix to eliminate an HVC need, while also extending battery cycle-life)

I'd also like an LVC of around 3.6, maybe 3.5.... as I've noticed my cells differentiate greatly after that 3.66V mark.

HVC is tied to a relay that can turn off the charger. With a Zivan you can actually tie directly into the charger. The idea is that you can balance every once and a while, but most of the time you bulk charge right up till the first cell hits HVC. Worry free charging with a power supply.

Changing HVC is trivial at this point as it is only a divider. Changing LVC is limited to what is available - namely:

2.7V
2.9V
3.0V

I suppose I could put hooks in the board to do other voltages but 99% of people will be able to use those. I dont think setting hardware LVC to 3.6V is a wise idea. A better method is to use the Cycle Analyst to monitor your Ah (and plan accordingly) and maybe set a pack level LVC with the CA (that can be defeated on the fly)... but to leave the actual cell level hardware protection down at the actual LVC. You never know when you will be in an intersection near the end of your ride - punch it - and sag below 3.6V under load.

Resting LVC is very different than dynamic LVC.

-methods

 

[Toorbough ULL-Zeveigh]

Methods, did you find any TC54xx equivalent parts with 3.4 to 3.6 volt thresholds? Wondering what LVC voltage detectors your experience dictates for LiPo and LiFePO4 when using fixed voltage detectors?

not sur if it would meet all ur requirements but i've been using an ltc1540 as lvc that i lifted straight off their app note.
for my modest requirements the added board space & parts count of some external resistors was a easy trade off.
i get the freedom to be able to set whatever voltage & hysteresis i want without having to buy a different part each time.
plus going with some blu banded resistors got the power consumption down under the TC54's 1µA draw.

 

[bigmoose]

Methy, I missed the hysteresis on the NCP part... sheesh. Good to have another set of eyes on this. Thanks for the advice on the LVC, I was looking for that based on your racing and usage experience. It is invaluable advice.

Toorbough the ltc1540 is a nice part! LTC always makes nice parts... but too bad it is a $2 part; at $0.50 it would have possibilities. For information a similar is TI TLV3011/TLV3012, they spec a bit cheaper, but still well above $0.50

If they would only make a 5 terminal TC54xx part, but bring out a terminal for the voltage divider, and the comparator output, then we could trim it to where we wanted it. ... then it would be the LTC1540 at $0.35!

 

[GGoodrum]

Like method says, under load voltages are a lot different than "at rest". From what I've seen, most of today's LiPo cells are pretty well depleted when the resting voltage gets down near 3.5V, or so. Under load, however, a cell down that far can be well under 3.0V. Using the 3.0V version of the TC54s has worked well. Typically, with most of my packs, when the LVC first hits the throttle, there's been about 10% capacity left. By backing off the throttle you can still have a "limp home" capability, for another couple miles (this is with 10Ah setups...), before giving it any throttle at all will cause continuous "hits" on the throttle. At that point my packs will take in their full capacity. I would also get the same sort of capability with a123-based LiFePO4 packs using the 2.1V version of the TC54.

Cell level LVC should be the last resort type of protection. I think it is wise to have a higher pack level LVC set in a CA or the controller. The cell level LVC is simply to keep the cell voltages from reversing, which kills the cell.

I am going to use the same basic design that Fechter/Gary/Kendrick have all been using that was originally introduced by Silicium (IIRC) back in the day.

Actually, this is not an accurate statement. The original TC54-based LVC design is what Bob Mcree and I did back in October of 2007. Silicium's BMS design was much later.

Not sure how I am going to handle the eBrake out line yet... will probably just leave two empty holes on the board. I have never cared for the 2 pin RC type JST connectors

Just make them with the standard .100 hole spacing, so a 2-pin header can be used.

Regarding the HVC function, I'd vote for leaving it out. What is really the only function that needs to be resident within a pack, in my opinion, is cell-level LVC protection. Unless you are going to use some sort of charge control circuit, or know how hack into a Zivan, or something similar, you probably aren't going to use this function. This is going to be true for most out there how will want to use this pre-made circuit. To me, it is better to either do a full BMS, which includes cell balancing, or use a balancing charger, like the Hyperion 1420i. Just my opinion, though.

-- Gary

 

[NeilP]

OK, probably limited uptake of people using Nano Tech Lipo, but the 3.6 is about right LVC for them. I try and keep mine up at 72 or absolute min of 70 volt on a 20 series pack. Get below 3.6 and the balance really starts going way ou.

Like the idea of them boards encased in epoxy thou...bu maybe not with JST connectors encased. If it is going to be encased into the thing, would like to see something a bit more robust

 

[methods]

Regarding the HVC function, I'd vote for leaving it out.

I agree with Gary here - I have never used the HVC function and I am pretty sure most users wont either. The customer wants HVC though so that's the way I have to do it. I am thinking at this point that I will just not populate that part of the circuit.

Or....

I am closing in on a "4 part HVC" replacement. The secret is finding a part that uses *** TIME *** to create hysteresis instead of a voltage window. There are actually quite a few parts out there that look just like a TC54 except that they have active high output (what we need for HVC) and they have a much smaller 50mV hysteresis. These will be a second "one part wonder". The design would then be

1 part for HVC
1 part for LVC
1 resistor for current limiting the opto
1 opto

The final trick is going to be selecting an opto that does not need a current limiting resistor. They are out there... there are a million flavors of digital isolators.
That will be 3 parts per channel. That is what I am going to aim for.

OK, probably limited uptake of people using Nano Tech Lipo, but the 3.6 is about right LVC for them.

Right - but like we were just saying... hardware-resting-emergency-LVC (these boards) is totally different than dynamic-everyday-in use-LVC. If 3.6V is the magic number then multiply it times the number of cells in your pack and set the controller or CA LVC at that level. Set it up to allow a few seconds of overshoot too :wink: Keep your hardware LVC at the true chemistry LVC of the cells. You will be thanking yourself the day your pack hits LVC 10 miles from home


-methods


P.S>

I am going to use the same basic design that Fechter/Gary/Kendrick have all been using that was originally introduced by Silicium (IIRC) back in the day.

Actually, this is not an accurate statement. The original TC54-based LVC design is what Bob Mcree and I did back in October of 2007. Silicium's BMS design was much later.

Ah - noted.
I think the patent is run out on that one

I am fairly certain that we are going to come up with a new HVC/LVC combo here that will be the new normal. After we get it done you can steal it back from me You and I know that the guy who DOESN'T produce these boards is the one who will be making the most profit I am pretty much just doing this as practice for bringing simple PCB's to market... with the help of the community. After all the help I got on the 18 fet infineon thread I realized that development time can be cut by orders of magnitude by letting everyone see the design and comment. Sure someone might steal it... but if they do... they will just go broke that much faster

 

[chroot]

Sweet, I glad to see you again. I don't need a HVC and I would buy a few premade LVC boards. I hope this is going to be good product.

Either way using a epoxy or silicone the lvc board should be able to prevent the shock, vibrate, and water proof. It also provides an opportunity to prevent the steal your product (protect your design.).

*subscribed*

 

[liveforphysics]

I personally really like the HVC function, because I made a charger that just uses a bridge rectifier and 4x halogen flood lamp bulbs (to handle the voltage drop). It's just for opertunity charging. It weighs about 1/2 a pound, and is the size a candy bar, and lets me steal a charge at 15amps from an outlet, but if left alone would charge to ~156v and smoke my pack, so you have to babysit it. Something with an HVC would be nice, so I could have tiny high power opertunity charging without burning our bikes down if we get distracted and forget it's plugged-in.

 

[GGoodrum]

I think the patent is run out on that one

I am fairly certain that we are going to come up with a new HVC/LVC combo here that will be the new normal. After we get it done you can steal it back from me You and I know that the guy who DOESN'T produce these boards is the one who will be making the most profit

You got that right!

Regarding reducing the parts count, have you looked at the Seiko chip? Richard and I were going to try playing with this at one point, but I think you need to order like 3000, or so, in order to get them. Anyway, it does both LVC and HVC and I think it has active high (or low) outputs.

-- Gary

 

[nicobie]

I personally really like the HVC function, because I made a charger that just uses a bridge rectifier and 4x halogen flood lamp bulbs (to handle the voltage drop). It's just for opertunity charging. It weighs about 1/2 a pound, and is the size a candy bar, and lets me steal a charge at 15amps from an outlet, but if left alone would charge to ~156v and smoke my pack, so you have to babysit it. Something with an HVC would be nice, so I could have tiny high power opertunity charging without burning our bikes down if we get distracted and forget it's plugged-in.

That would be a nice feature.

 

[NeilP]

I already have my CA LVC set at 70v, so got that covered, and understand the point of the lower level....final hardware cut off...would be happy with that.

I would like the option of HVC cut off, I have one of Garys Meanwell limiter boards, so it would be nice to couple th HVC cutout to the board. I do have Garys old hVC/LVC boards, built 4 up with 5 per channel for my 20 s pack... but after two rides, I removed it and it is now gathering dust. It was just to big and delicate to have on the top of the pack, transistors kept getting bent and shorting out on the legs. I just run with the CA at 70v for overall cut off and just keep an eye on the balance a each charge

 

[grindz145]

HVC is great is you have a circuit to drop the charger. Especially when dealing with some of the crappier chinese chargers. 200mV tolerance just isn't good enough

Sweet.

 

[methods]

I will check out the siko chip. I have been looking at other dual-channel parts. There really are hundreds out there and most of them dont turn up on a search. You really have to dig down into each manufacturer website. Buying 3kpcs is not really a big deal IF we can get the price down low enough to justify it.

3k/6 = 500 boards
.... BUT ....
If we can cut assembly time down and per/ch cost down on every board I can just throw away 2,000 of them and still be better off :wink:

Here is one I came by that could be used for HVC:
http://www.intersil.com/data/fn/fn6141.pdf

* no voltage window
* selectable trigger voltage in 100mV steps

The bads are that it is active high output, hard to find, and a little expensive. Also it would only do the high or the low - of course combining those would be optimal.

New goal:
1 part for both HVC and LVC
1 part per every 2channels for opto (no current limiting resistors)

-methods

 

[methods]

OMG... too many to choose from :x
monitor circuits

Now I am starting to feel even more greedy... why not just use 2 or 3ch chips?
Ok - off I go to read 30 more datasheets

This one is easier to read Product Tree

-methods

 

[GGoodrum]

Here's the Seiko part datasheet:

View attachment Seiko S8209A_E balancing.pdf

Richard spent several days/weeks doing just what you are doing, and said the Seiko chip had the best features, including cascading outputs so that no optos are required at all. It even has the time-based hysteresis delay you mentioned.

Anyway, happy hunting.