How to turn a high voltage DC-DC converter ON/OFF

[methods]

Rambling Brainstorm:

Big Picture: The rubber hits the road with me as customers may not already have a plan for boot-strapping their high voltage EV... I can ASSUME they will magically provide 12V to my BMS... and that may sometimes be the case... but either way I have to have a handful of solutions ready.

Examples:

* Directly switch the HV on and off (NO... for anything over 120V)
* Run an aux battery (YES, this is the best answer, switch 12V using the plethora of automotive options)
* Run an onboard DC-DC to allow users to switch me on and off... thereby bootstrapping their high power DC-DC (more expensive... but eliminates the NEED for an aux battery)

Solutions 2 and 3 are most likely
Solution 3 starts to force irritating decisions around specifying DC-DC input range
Solution 1 solves all problems across all platforms and is a superior path forward... be it lead... or more likely a little protected 4S pack...

So the REAL answer here is that CVS and K-Mart should have protected 4S lithium rechargeable batteries on the shelf like bottled water.
Whey they dont? Energizer, Duracell, and other Alkaline promoting profiteering shit-bags... but that too is a rant for another thread.

Back to the original brainstorm: How to bootstrap...

Wait - one more note:

Sevcon Gen runs an internal DC-DC step down... but requires user to switch HV for them.
Sevcon DC-DC can run pre-contactor as it has a pack level enable line (low current, high voltage)
Zero BMS runs an internal 12V step down... and I think they use this for boot-strapping. (3pcs DC to dc in the system)
Production cars all run 12V aux batteries along with a DC-DC that acts like an alternator
Ebikes switch HV at the controller and an internal switcher knocks it down to 12V for internals
Justin's LiGo obviously has an internal DC-DC
Most lighted vehicles must have a high power DC-DC... and whether that is post contactor or pre-contactor with enable... is case by case

Ok - now actually back to my thoughts

I know better than to think I can get good performance out of a reference DC-DC converter design that claims it can span 4V to 150V....
If it were so.... many SIP's and modules would be available yes?
Show me one that costs anywhere near parts cost.

The reality is that most DC-DC converters are tuned to a pretty tight input voltage range... about LVC to HVC of your traction pack.

For this reason, and for the fact that our BMS master needs to span 12V up to 400V+... I am putting the DC-DC off-board.
Effectively the BMS runs on a 12V buss which comes in on a pin referenced to a single point ground. (grounding theory is another topic... isolation brings problems... good grounding works)

So... a 4S user wont even need a DC-DC.
A user up to say... 120V... will need a class of DC-DC which is fairly easy to find but not any where near dirt cheap.
Anything over 120V starts to get real specific, real expensive, or otherwise a problem.

For this reason... most (ALL.... except maybe Zero Motorcycles) EV's use an Auxiliary 12V battery.
These days.... they are still lead acid for cost and complexity... but could easily be a 4S 18650 with a $7 balancing and protected BMS... or a 4S A123... or even a hobby king pack.
4 channel balance chargers ... or just integrated BMS... pretty easy to find - but I digress.

There are MANY reasons to run an auxiliary battery pack. The first, and most obvious you will run into while designing a BMS... is Boot Strapping.

How do you turn it on?

When analyzing any electrical circuit you start by addressing T0 (Time Zero), then Tsteady (Steady State operation), followed by transient response from T0 to steady, during any event, and at shutdown.

T0 for a BMS means that you have a 100V or 300V potential sitting there.
You cant reasonably just hook a switch to it and run it to the handle bars... I mean you can... but its retarded.
You can try to trigger a Sleep pin... something that takes the DC-DC out of a low quiescent current state and into Active... and this is likely the best option.
Reference this example:

4V to 140V, 400mA magic regulator (whether it works or not... separate issue... best bet is to copy the demo board EXACTLY so as to not upset the black magic)
http://cds.linear.com/docs/en/datasheet/7138f.pdf

On page 6 you see: RUN (Pin 14):
This pin puts the regulator into a 1.4uA sleep mode any time it is grounded.
Apply a voltage over 1.2V... and it kicks on.. but to take any input it requires full pack voltage. A voltage less than pack sets the Low Voltage Lockout.

Ok... so we can quickly see that without SOME sort of potential... even a pull-up resistor.. its very hard to turn a high voltage DC-DC on and off... unless it is equipped for such control.

Now we manually turn 120VAC and 240VAC directly on and off all the time...
But usually either inside or behind an IP rated connector.
This gets expensive... and for DC there are additional concerns.

To "do it right" you dont want to be switching anything more than say.... 48V. Yes you can... but it is not forward looking. Its... a hack.

So... most systems have a 12V source available to kick things off.
This battery can be slowly (or quickly) charged from a low power or high power DC-DC (Depending on accessory needs... and most EV's need at least a few hundred watts of lighting alone!)
With this aux source... no problems.

Without it.. eh... hrm...

well... lets get back to the benefits of having one:

* If the main contactor blows, and your high power DC-DC is on the switched side, you still have your 12V buss running.... at least for a while.
* Primary lighting is more reliable. I think it is irresponsible to run a headlight off of a DC-DC converter. Ever lose your lights going 70MPH? That equals death.
* The above mentioned ability to easily boot-strap a system into operating mode
* Did I say endless complexity is eliminated?
* Much easier to meet peak and even steady current needs. A 12V contactor needs an amp. Yes... you can run a 116V contactor... but to support the masses we will assume there are at least a few 12V relays or 12V contactors in the system. A light duty DC-DC wont handle these well... one that will gets expensive.

Opposition argument is that any EV will require a high power DC-DC to power headlight(s), tail light, brake light, blinkers, user interface, possibly ABS, onboard computer...
Rough math: 50W + 10W + 25W + 10W + 10W + ??W + 10W = a couple hundred watts.

200W / 13.8V = 15A
200W / 100V = 2A

So thats 2 amps coming off the traction pack (easy enough to control)
15A buzzing around the 12V system
200W total needed

Sevcon DC DC Converter 50-120 VDC to 13.4 VDC, 500 Watts, Isolated Output
Ebay.. $80
https://www.ebay.com/i/201994482352?chn=ps&dispItem=1

Ok... looks like a DAMN GOOD DEAL... how does the enable pin work?
Well... it too needs full system voltage applied... which is fine for 96V... but bullshit when you start to look at 300V.

You think Tesla has a Chinese push button in the dash board, inserted into a crookedly drilled hole, that snaps 420V on and off?
No
They have a 12V battery that controls relays and SSR's to control high voltage remotely and in an isolated manner.

Anyhow... the amount of Aux capacity is driven by the needs of the vehicle and the capabilities of the DC-DC.
FROM the BMS perspective...
I either need to ASSUME the vehicle can give me a 12V boot strap... and that I can draw some power off of that...
Or I need to pull down at least a 100mA of 12V off the primary... which costs a lot and opens up a lot of problems.

Historically I have resisted the temptation to ASSUME that someone will give me 12V... especially when I am the one in control of the contactor... and they may or may not actually have a battery on board.... like on an Ebike. Humpf... dropping that and directing folks to just run an AUX battery. Ironically even ebikes often have a pretty hefty aux pack to run high power lights...

For those who want to run strictly a DC-DC, powered on the battery side of the contactor, well - they can have the key switch that turns 100V on and off. Really there is no current on this pin (presumably) so it shouldn't be an issue. You can switch a thousand volts through a 1G resistor all day... its the current the wrecks switches... but its general principle and ability to scale which turns my nose up at switching full pack voltage.

The general solution scales...
And...

No investor who is paying attention would invest in something which is not at least acknowledging the general solution... especially if they are looking to lead the pack and set standards for the near future.

I digress...

lunch, 5yo's, eh... I think I got my answer.
I am going to punt on the DC-DC
Assume that magically this signal will get to me
Assume that it will be able to carry my contactor drive needs
And I can move on.

No onboard DC-DC

I will pin for B+ in, B post-contactor in, 12V in...
I will double those pins until we start seeing more COTS solutions.
We will solve it case by case... Ebike, Emotorcycle, Eplane, Eboat, E3wheeler, Egolfcart, Etarded.

Meh Meh.. I really need to drop the Ebike ethos and the sub-120V ethos and start thinking 420V.
As soon as I start thinking stupid.. 4 2 0 V

420...
420...

Nobody wants to switch 420V on and off.
Everybody wants to scale
Any contactor needs to be supported - 12V, 24V, 48V, 86V, 116V... for retrofitting legacy designs (I have that one dialed)
We need to identify good, protected 4S solutions anyway... and there are at least 50 BMS's out there that will work... and at a minimum lowest form we can run 4pcs 4.2V DC-DC and just parallel charge them then watch full voltage for LVC.. (that gets dangerous... but not with LiFe.. better to have a little BMS on there with fets that watch each cell)

Oh bother...
Off to the coal mine to figure out where next to dig.

Sorry I post all this shit.
I dont exactly have a peer review team here at my house... so if I dont post my assumptions while brain storming I get stuck in constant reconsideration. By posting my thoughts I can set it aside... assume at least 4 people who are useful read it... and that I did not completely miss the forest for the trees... and keep moving.

Hit target price
Hit timeline
Have solid arguments for dropping features
Have alternate Systems Level solutions ready for dropped features
Have fun
Pay the rent
Make a BMS that does not suck

Raise the 5yo
Dont get caught by the cops (lol.. when I am driving my Golf Cart 0-60mph in 1.2 seconds on city streets under 35mph...)
Light things on fire every once in a while
Stand up to authority figures when they are not operating on logic and reason
Accept nothing less than what is best for the whole
Dont get personally greedy
Get laid sometimes
Go to the beach

Most important... never be embarrassed to let people see your thought process. You cant improve it if you are not getting negative feedback on it.
Been doing this since the 90's
It works... if you have peers who are skilled at negative feedback looping.

-methods

 

[bigmoose]

How about the DC-DC converter being a "plug in module?" Where the main board is scarred for it? You could originally start out with no DC-DC then do a module for charging your independent 12V source based on the first or largest sale of primary battery voltage and 12V battery chemistry.

 

[Inwo]

New here, don't know the back story, of what you build.
Are you saying no to the normal ev dc to dc, because of the idle current in "not" enabled mode?

For a zero quiescent current solution, you might use a high voltage fet.
Driven with a sealed piezo switch.
"Turn off I haven't addressed. No sense thinking too far ahead.

 

[Lenk42602]

Dunno.

whattabout "tapping" the equivalent of 12-13.7 volts out of a through bms utility where consumption is lower than a couple of amps?
build the utility into the bms with a 12v out put. activate it either switched or CAN.
would take up less space than a small buck
anything requiring big watts on a larger ev, move to a watercooled converter (sevcon,astrodyne, etc) and aux 12v battery.

 

[methods]

How about the DC-DC converter being a "plug in module?" Where the main board is scarred for it? You could originally start out with no DC-DC then do a module for charging your independent 12V source based on the first or largest sale of primary battery voltage and 12V battery chemistry.

Thats great advise BigMoose.
If I am reading you clear it will go like this:

DC-DC is not included on V1. Pins are mapped at the connector in a way that supports internal or external DC-DC. Inside is a header that is daughter board ready mapped to said pins. On an as-needed basis daughter boards are installed, matching system requirements, and the off-board DC-DC can be eliminated in most cases.

That is a viable path forward. Not much lost if I dont ever get around to the internal DC-DC. Allows others to develop a board if they want to. Might be some diode OR'ing to consider.

I am packaging with a box with a gasket and a 35 pin connector with a gasket - so no potting on this one - so can be changed later.
IP66

-methods

 

[methods]

Thanks for chipping in.

I am basically trying to build to a customer requirement (specific solution) while trying to generalize enough that the solution can be useful for our DIY activities. On Ebikes... we often dont have a dedicated DC-DC in the system... so in my last design I put it in the requirements. That effort was never funded, and was a much lower power target. This solution may be funded, and in this solution we have a green light to burn as much current as needed while "awake" and we can assume 99.9% of the time a high power External DC-DC will be present in the system to support other functions.

One of my directives is to create a BMS that can drop into an existing system (the Zero Motorcycle) which.. unlike most EV's... does not run an onboard 12V battery. They sneak the voltage off of the balance tap lines via the BMS (I only know that because I heard so much about popped BMS boards where the DC-DC went). I dont think this is a great way to go about it... but I have to at least put in hooks to be viable for that system... so I am trying to figure out how to work around the fact that I will have only a high power DC-DC which is enabled by a high voltage enable line... and bootstrapping becomes a chicken before the egg problem.

I like the idea of placing the high power DC-DC AFTER the contactor... for general protection.... and to allow for cheaper solutions that dont have a built in enable.

I had considered holding a charge for boot strapping in a little capacitive circuit.
Using a piezo is actually brilliant way to ensure a system can spark even after sitting for extended periods. You could easily charge-pump it...

On that sort of thinking... incorporating an input for a solar cell... from the size of your hand to the size of the roof... is also a good deal.

The scary part....
Whatever we build... be it smart or stupid... has to some degree an impact on what others build... so I am trying to think about including features which will be awesome and deleting old methods and ways which are not useful anymore moving forward.

I have not decided yet whether to make the code or the hardware open source. Probably the code this time... tho I do fear that students will just nab the code to skate on senior projects. If I become funded to the level of being able to pay the bills... I will make both the code and hardware open for others to check out, improve on, and check back in.

-methods

New here, don't know the back story, of what you build.
Are you saying no to the normal ev dc to dc, because of the idle current in "not" enabled mode?

For a zero quiescent current solution, you might use a high voltage fet.
Driven with a sealed piezo switch.
"Turn off I haven't addressed. No sense thinking too far ahead.

 

[methods]

Woha... Lenk back from the dead.
Where you been man? Beam me up Scotty.

I am not super clear on what you are proposing - but one hitch is matching the DC-DC to the pack voltage. I had incorporated at tiny 4V - 150V DC-DC on the BMS PCB... but requirements came down the pipe to support higher voltages... and above about 150 the windows for acceptable range tighten up and it gets very "not general solution 'y" as we end up needing 4 or 6 different versions of the DC-DC... and as voltage goes up, power goes down, and it turns to shit.

background
I was always annoyed when buying aftermarket parts that required me to select from a pull-down menu that dictated early on in a project what some parameter would be. Early ebike controllers were this way and it causes much rework with the soldering iron. Later controllers moved from a linear regulator to a basic switcher... and after that almost any input voltage was supported... leaving only LVC resistors to tune.... which later became software programmable.

Now like any engineer... I love to make a "learning platform" for others to cut their teeth on.. but this late in my career and being a parent... I now have to shoot out more finished designs aimed at more of a production target.

Anyhow... big picture I am not going to be able to support ebikes very effectively.
I will however... be able to support just about everything bigger.

Ebikes really should have a DC-DC on them to power lights, phones, and other as yet added items... so maybe that will develop in an organic way... in which case we may become a financially viable option. Pricing for "the DIY crowd" is going to have to look something like $10/cell in series... so a 24S pack would be on the order of under a few hundred bucks... PLUS externals. For an EV this is nothing. For an ebike its a harsh.

-methods

Dunno.

whattabout "tapping" the equivalent of 12-13.7 volts out of a through bms utility where consumption is lower than a couple of amps?
build the utility into the bms with a 12v out put. activate it either switched or CAN.
would take up less space than a small buck
anything requiring big watts on a larger ev, move to a watercooled converter (sevcon,astrodyne, etc) and aux 12v battery.

 

[bigmoose]

Yes, you interpreted my suggestion perfectly methods. I think you have around 10 extra pins, and this might be a good use for 4 or 5 of them.

 

[Inwo]

Once the dc to dc latches on, turn off and reset should be trivial. I'm a relay guy. My ideas run with relay logic, but there may be simple ss solutions.
An scr is a good choice for something like this. Normally a relay contact to commutate . Anything dropping current to zero may work. A NC device in series.??

 

[Inwo]

Have you looked at the high end active balancers for larger projects. At 10 amps, they are specifically designed for non symmetric battery loads. Such as running lights from 12v tap.
Completely negate the need for bms and all the associated wiring and problems. Lvc and hvc are easier to deal with than the other bms functions.

Downside is $20 per cell. I don't know if they are potted or lend themselves to reverse engineering.
I'm onboard with any effort. Most common for me is 24s. Doable but much larger starts to pinch.

 

[Inwo]

After reading your other thread, it seems you looking for some of the same solutions I am. But on a larger scale. I'm more in the 6s to 24s range.

The modular high power balancers seem a logical part of the solution. I have not researched the upper cell count level.
They are offered by usa sellers for about the same price as China, so the $20/cell can be beat.

If I wasn't invested, in developement of the bms-24, I would be dissecting some.

 

[Lenk42602]

Woha... Lenk back from the dead.
Where you been man? ...... had incorporated at tiny 4V - 150V DC-DC on the BMS PCB... but requirements came down the pipe to support higher voltages... and above about 150 the windows for acceptable range tighten.

im around

what you've described above is what i was envisioning. must have missed that post..

spent a good bit of the last few years bringing up a large commercial ev platform - 300 - 700v range systems - we ended up with an external 12v battery and large liquid cooled dc-dc's to support our 12v needs. liqid cooled LTO's, & liquid coolant was subambient-ly conditioned in order to optimize discharge and regen for 60,000 lb applications..implemented an ecu so CAN was integral, but also included some manual redundancy to meet functional requirements. Then we changed the entire design of the system...

outta that for now, working on autonomous vehicle platforms here in the burgh while its hot.

looking forward to peering into your mind regarding system grounding. always something to be learned there for me.....

cheers

len

 

[methods]

I will trade you one bucket of system ground for one thumb drive of CAN implementation...

Still looking for the turn-key CAN board that's as cheap as a Bluetooth ($10)

-methods

 

[bigmoose]

...Still looking for the turn-key CAN board that's as cheap as a Bluetooth ($10)
-methods

I'll bet I am missing something with respect to modern CAN standards... but 12 or so years ago I rolled a CAN system based on the MCP2510 from Microchip, then rerolled some nodes with PIC's that had CAN interface imbedded. It was a real simple protocol. It worked. The end.

That said, I think you have said that the IXXAT CAN interfaces have internal routines, much like modems of the past? That's a lot smarter than just the simple CAN layer I did.

 

[jonescg]

Yes a common problem I encounter with EV projects. Allow me to type myself smart

Cars are big - you can have an auxiliary battery which does most of the work, and you can even charge it with an onboard 12 V charger, offering complete isolation from the traction pack.

But a DCDC converter would be required to charge the 12 V battery if you went down that road. And in the case of Voltron, 700 V to 12 V doesn't really exist outside of the railway infrastructure. We just use a 12 V auxiliary battery in a run-to-waste manner, and top it up between races.

On e-bikes, there is no 12 V auxiliary battery, so a DCDC converter is required. I find those CU-INC or XP-Power converters to be pretty good - wide range, 500 mA 12 V output. As they aren't normally powering much you can switch the HV input side easy enough when it's under 100 V. Like this one: https://www.youtube.com/watch?v=2r1UceB8K3s
The DCDC converter on my E-max scooter is a bit like this - the key switch literally switches a 48 V supply (at very low current - a relay coil).

But for motorcycles where you have a sizeable 12 V DC load, like water pumps, headlights, sat nav, air conditioners these can represent a sizeable current. And the supply is hazardous - over 100 V DC. I guess you could leave the DCDC converter on the whole time and only switch the 12 V stuff on when needed. But DCDCs draw something just by being connected. A contactor on the HV+ would work, but what do you power the contactor with? An auxiliary battery I guess. Which is charged by the DCDC converter and only as small as needed to turn a relay on.

 

[fechter]

Inside the dc-dc converter there are switching transistors feeding the transformer/inductor. Seems like you could just make those turn off and leave the thing connected all the time. Many dc-dc converters have this feature built in (enable line).

 

[Lenk42602]

capacitors?

once Pack HV and dc dc contactors are closed, dc dc should be able to supply constant 12v power to the contactors.

might be able to energize the caps from in rush current.

or a few pedal strokes on a pedal bike.

or a faux kick starter on a motorcycle. or hand crank.

cap (s) woud have to sustain 12v output long enough for contactors to close and dc dc to start outputting 12v supply

 

[methods]

Inside the dc-dc converter there are switching transistors feeding the transformer/inductor. Seems like you could just make those turn off and leave the thing connected all the time. Many dc-dc converters have this feature built in (enable line).

Trouble is chicken before the egg...
For a solution that works with 12V, 120V, 1200V... getting even the bias voltage for the transistors inside can be a challenge.
It sounds easy till you draw it up on the board and start trying to put part numbers on it :|

I cant see a scale-able solution (which is not Rube Goldberg) that does not incorporate at least a small battery on-board someplace.
On that route of thinking... it could be a coin cell... like those found on a PC board
Just enough to bias a fet or otherwise bootstrap the system without touching HV.

-methods