methods
1 GW
Look closely and you can see the Arduino chip on the back 
-methods
-methods
Zero-Draw, Solid State Contactor w/Precharge (Arduino)
- Thread starter methods
- Start date
methods
1 GW
Although I love it when Luke is wrong.... 
in this case I think PWM is no good for a Peltier load
http://electronics.stackexchange.com/questions/28634/how-to-drive-a-peltier-element
-methods
in this case I think PWM is no good for a Peltier loadhttp://electronics.stackexchange.com/questions/28634/how-to-drive-a-peltier-element
-methods
Njay said:
methods
1 GW
So I modified the TO220 package to have no hole which opened up a bunch of space on the back of the board (we will solder them on the top). Seems the hottest spot will probably be under those heat sinks so... I will probably move the uController over by the legs and put some non-critical shtuf on the hot spot. Maybe some Tyler Durden quotes...
Tip: When you find a library part you like - but not in the package you like - you can go into the library and modify the part by adding an additional package option. For example, an n-channel mosfet is most commonly used (by me) in Vertical, Horizontal, and Horizontal with no hole. It is totally worth while to use the right package for the right job as it will save you from 153 stupid mistakes that are so easy to make while laying out boards.
-methods
Tip: When you find a library part you like - but not in the package you like - you can go into the library and modify the part by adding an additional package option. For example, an n-channel mosfet is most commonly used (by me) in Vertical, Horizontal, and Horizontal with no hole. It is totally worth while to use the right package for the right job as it will save you from 153 stupid mistakes that are so easy to make while laying out boards.
-methods
How about that !...
!...methods said:Although I love it when Luke is wrong....
http://electronics.stackexchange.com/questions/28634/how-to-drive-a-peltier-element
-methods
Njay said:
Hey Luke,
Do the new Zero bikes have 0 parasitic draw on the battery when the bike is "off" or is the BMS and other systems always on? Do you have a separate battery powering the BMS?
BRD motorcyles have decided to turn off the BMS and all electronics when the bike is off so the bike is truly off when you turn the key.
I'm curious about whether you think it is safe to turn off the BMS or not?
I like the button boob cell idea. Zero motorcycles does the same thing of sorts with an externally powered BMS for true 0 quiescent current in storage.
I agree with bigmoose that the all in one controller has to be the next step. The controller could have a built in battery for powering the BMS and contactor and more. We need to compete with the Russian sinus controller.
Do the new Zero bikes have 0 parasitic draw on the battery when the bike is "off" or is the BMS and other systems always on? Do you have a separate battery powering the BMS?
BRD motorcyles have decided to turn off the BMS and all electronics when the bike is off so the bike is truly off when you turn the key.
I'm curious about whether you think it is safe to turn off the BMS or not?
I like the button boob cell idea. Zero motorcycles does the same thing of sorts with an externally powered BMS for true 0 quiescent current in storage.
I agree with bigmoose that the all in one controller has to be the next step. The controller could have a built in battery for powering the BMS and contactor and more. We need to compete with the Russian sinus controller.
methods
1 GW
flathill said:
-methods
methods
1 GW
Ok... I suppose the BMS could sleep and datalog while in storage. Data is always useful.... but for the risk - I would just shut everything down.
Most motorcycles would just use a high quality switching regulator to power the BMS - it could run for a thousand years an never even start to eat up the battery pack. They have a fixed voltage pack and a ton of space (relatively speaking). The early Zero motorcycle packs sucked-ass in that they drew power directly from the first two cells in the pack (FAIL).
-methods
Most motorcycles would just use a high quality switching regulator to power the BMS - it could run for a thousand years an never even start to eat up the battery pack. They have a fixed voltage pack and a ton of space (relatively speaking). The early Zero motorcycle packs sucked-ass in that they drew power directly from the first two cells in the pack (FAIL).
-methods
liveforphysics
100 TW
methods said:
No amount of caps will help, may only make it harder on the switching circuit itself. Tiny inductor the size of a grape solved it.
liveforphysics
100 TW
flathill said:
BMS is always on, and personally I think it's a critical safety need. The draw in sleep mode is measured in uA though, and not a whole huge amount of them either. It's setup so that you can discharge the pack down as low as you're allowed to drain it, and then still the system can power the BMS for a very long time.
bigmoose
1 MW
Tiny, surface mount inductors like a DR73-221-R are friends of mine. They like to sit right next to my other friends, the cap-a-c-i-tators.
methods
1 GW
bigmoose said:Tiny, surface mount inductors like a DR73-221-R are friends of mine. They like to sit right next to my other friends, the cap-a-c-i-tators.
I actually made a joke earlier and deleted it where I said that the only Inductors I can even tolerate are those in an 0805 package
I pretend they are small tiny gray resistors.-methods
"I like the button boob cell idea. Zero motorcycles does the same thing of sorts with an externally powered BMS for true 0 quiescent current in storage."
edit!!!! that should have read mission motors not zero! I wasnt sure about the zero. I am about the mission
sorry!
edit!!!! that should have read mission motors not zero! I wasnt sure about the zero. I am about the mission
sorry!
methods
1 GW
I dont know why I had it in my head that these IRFB4115 fets were logic level. Someone else already tried to point out that they were not - then Luke came over a couple hours ago and he brought it up again. Damn... need like 7V for these 150V fets. So basically that means that now I have to run 3 coin cells in series and add a gate driver.... so... now I am off to go research the quiescent current for the gate drivers I have sitting around here.
Good news is that now we can run the Arduino down to 3.3V
Might still run it at 5V tho for compatibility
While Luke was here we argued about current control for at least 2 hours. In the end it works out that he is right... I will have to run a separate precharge circuit and it will require an inductor... BUT... I originally rejected the idea thinking that I would have to bring in V+ and PWM that out... but really all I have to do is just string a fet and an inductor right across the main discharge fets... so it is actually pretty easy.
Luke pointed out that if we run separate paths (one for discharge and one for charge)... and we put an inductor in the charge path... then the charge fets can be used as the Precharge controller and as a secondary effect we could then have variable current control for the charge current. You can charge control without an inductor but it takes an ass-load of caps and then it wont serve as a Precharge.
Luke brought over a bunch of nerd stroke material. Pretty impressive batch but I would not let him know that I was like "yea whatever - throw that stuff in the creek" 
-methods
Good news is that now we can run the Arduino down to 3.3V
Might still run it at 5V tho for compatibility
While Luke was here we argued about current control for at least 2 hours. In the end it works out that he is right... I will have to run a separate precharge circuit and it will require an inductor... BUT... I originally rejected the idea thinking that I would have to bring in V+ and PWM that out... but really all I have to do is just string a fet and an inductor right across the main discharge fets... so it is actually pretty easy.
Luke pointed out that if we run separate paths (one for discharge and one for charge)... and we put an inductor in the charge path... then the charge fets can be used as the Precharge controller and as a secondary effect we could then have variable current control for the charge current. You can charge control without an inductor but it takes an ass-load of caps and then it wont serve as a Precharge.
Luke brought over a bunch of nerd stroke material. Pretty impressive batch but I would not let him know that
I was like "yea whatever - throw that stuff in the creek" -methods
methods
1 GW
Oh yea... all of my gate drivers are isolated
http://www.silabs.com/Support Documents/TechnicalDocs/Si8220-21.pdf
That is actually a pretty sweet part. It is an opto-isolater and gate driver in the same package so you can use your TTL circuit to drive your high power and not even fret.
Ok... off to research efficient gate drivers.
-methods
http://www.silabs.com/Support Documents/TechnicalDocs/Si8220-21.pdf
That is actually a pretty sweet part. It is an opto-isolater and gate driver in the same package so you can use your TTL circuit to drive your high power and not even fret.
Ok... off to research efficient gate drivers.
-methods
methods
1 GW
Ok... here is a nice one.
http://cds.linear.com/docs/en/datasheet/lt1157.pdf
Will work at 3.3V but it will cost 80uA. Ugh!
Lets round that up to 100uA and assume it is on 24/7
100uA/hour
2.4mAh/day
72mAh/month
So a couple of coin cells in series will last a month or so
If the circuit is only run 2 hours a day you can multiply that by 12 - so a year
Now that sounds a little more reasonalbe
Bike sits "off"
When bike turns "on" it costs 100uA @ 3.3V so 330uW. The relay was taking about 3W, so four orders of magnitude less holding current
Bike stays on for 2 hours a day and the rider gets a year between battery changes/charges
Yea... that sucks having to give up an extra 80uA
Back to the drawing board.
-methods
http://cds.linear.com/docs/en/datasheet/lt1157.pdf
Will work at 3.3V but it will cost 80uA. Ugh!
Lets round that up to 100uA and assume it is on 24/7
100uA/hour
2.4mAh/day
72mAh/month
So a couple of coin cells in series will last a month or so
If the circuit is only run 2 hours a day you can multiply that by 12 - so a year
Now that sounds a little more reasonalbe
Bike sits "off"
When bike turns "on" it costs 100uA @ 3.3V so 330uW. The relay was taking about 3W, so four orders of magnitude less holding current
Bike stays on for 2 hours a day and the rider gets a year between battery changes/charges
Yea... that sucks having to give up an extra 80uA
Back to the drawing board.
-methods
Below is a generic diagram of how to make a low quiescent linear regulator work at higher voltages. This is trick Otmar gave us. The CPC5602 is rated for up to 350V. Pick your favorite LDO and add this transistor to make it work over an incredibly large input voltage range. FET adds nothing to the quiescent drain and supplies about 2v higher than the output to the input.
Maybe you still like coin cells. I like the boob shaped ones for sure, but this cheap trick would be great if you want to run off the pack voltage. Just do the math on the maximum dissipation and stay under.
On precharging, I'm not sure about a series FET switch and inductor. Seems like the inductor current wouldn't have a path when off (major voltage spike)? I need to draw it out and think about it. I think it needs a diode somewhere.
Maybe you still like coin cells. I like the boob shaped ones for sure, but this cheap trick would be great if you want to run off the pack voltage. Just do the math on the maximum dissipation and stay under.
On precharging, I'm not sure about a series FET switch and inductor. Seems like the inductor current wouldn't have a path when off (major voltage spike)? I need to draw it out and think about it. I think it needs a diode somewhere.
methods
1 GW
fechter said:Below is a generic diagram of how to make a low quiescent linear regulator work at higher voltages. This is trick Otmar gave us. The CPC5602 is rated for up to 350V. Pick your favorite LDO and add this transistor to make it work over an incredibly large input voltage range. FET adds nothing to the quiescent drain and supplies about 2v higher than the output to the input.
Thanks man... I am basically losing my mind working the circuit over and over and over trying to make it into what I want
You know what I have been thinking.....
I have been thinking about sneaky ways to snitch power from the open relay while in the OFF state. Like... I snitch 10uA of current across the open relay to charge my battery or super cap - then when it closes I use the supercap. Simplest visualization of this would be a 3.3V Zener + huge resistor across the gap, then tap the zener with a blocking diode that allows it to charge the cell, but the cell wont discharge backwards once the relay closes.
Just an idea... anything, anything, anything to avoid bringing in that wretched V(+) line.
Maybe you still like coin cells. I like the boob shaped ones for sure, but this cheap trick would be great if you want to run off the pack voltage. Just do the math on the maximum dissipation and stay under.
I am considering it.... Especially now that my total system current is more like 200uA instead of 200mA. Easy cheesy to make a 150V regulator that can handle that. 200mA on the other hand... that is what drove me away from the Contactors.
On precharging, I'm not sure about a series FET switch and inductor. Seems like the inductor current wouldn't have a path when off (major voltage spike)? I need to draw it out and think about it. I think it needs a diode somewhere.
Yea I totally oversimplied it - you would most definitely need a flyback diode if we are going to switch an inductor on and off
-methmouth
liveforphysics
100 TW
Supercaps are awesome for short-term small quantity energy storage, but not so good in long-term energy storage from self-discharge bleeding them down while sitting charged. Some are better than others about it, but definitely check the datasheets carefully.
It does seem like it might be possible to use a tiny rechargeable coin-cell battery, and only when the switch is on running the ebike does it have some little current limiting resistor and a Zener to clamp from over-charge or whatever that slowly trickles juice back into the stack while the bike is running to keep them topped-off to last through long periods of sitting or whatever. But that adds all sorts of complications to the circuit, including drawing power from it while it's on, which I realize is unattractive for a variety of reasons. Just an option to think about if you're really struggling to get a low enough power budget.
It does seem like it might be possible to use a tiny rechargeable coin-cell battery, and only when the switch is on running the ebike does it have some little current limiting resistor and a Zener to clamp from over-charge or whatever that slowly trickles juice back into the stack while the bike is running to keep them topped-off to last through long periods of sitting or whatever. But that adds all sorts of complications to the circuit, including drawing power from it while it's on, which I realize is unattractive for a variety of reasons. Just an option to think about if you're really struggling to get a low enough power budget.
methods
1 GW
Here are a couple of coin cell options:
Rechargeable
24.5mm x 5.2mm
6.4g
3.6V
110mAh
396mWh
https://www.sparkfun.com/products/10319
Disposable
20mm x 3.2mm
2.81g
3.0V
250mAh
750mWh
https://www.sparkfun.com/products/338
When I close my eyes and think about using this product the thing that excites me most is the fact that it requires no wiring... it just has two tabs... you cut your cable and put this inline... and it just works. Plug in a power switch if you want to control it manually. Plug in a BMS line if you want it to protect you at the cell level.
So clean and simple
High side or low side (though it is directional)
Obviously works for any battery under the rated voltage of the mosfet
It must be battery operated!!!
I would probably be more apt to run 4 batteries than to try and snitch power from the open tabs... if for no other reason than it will be impossible to do well with an input range of 1V to 150V.
With those 250mAh batteries we can run from 6 months to 6 years even with the added cost of a gate driver
I am going to build this thing so that it can run any mostfet or any IGBT!
-methods
Rechargeable
24.5mm x 5.2mm
6.4g
3.6V
110mAh
396mWh
https://www.sparkfun.com/products/10319
Disposable
20mm x 3.2mm
2.81g
3.0V
250mAh
750mWh
https://www.sparkfun.com/products/338
When I close my eyes and think about using this product the thing that excites me most is the fact that it requires no wiring... it just has two tabs... you cut your cable and put this inline... and it just works. Plug in a power switch if you want to control it manually. Plug in a BMS line if you want it to protect you at the cell level.
So clean and simple
High side or low side (though it is directional)
Obviously works for any battery under the rated voltage of the mosfet
It must be battery operated!!!
I would probably be more apt to run 4 batteries than to try and snitch power from the open tabs... if for no other reason than it will be impossible to do well with an input range of 1V to 150V.
With those 250mAh batteries we can run from 6 months to 6 years even with the added cost of a gate driver
I am going to build this thing so that it can run any mostfet or any IGBT!
-methods
methods
1 GW
Can someone please help me find a gate driver with an ultra-low quiescent current?
* It does not have to be isolated
* Low side driver
* 3.3V or 5V input is fine
* Integrated boost or an input that will go down to 9V (three coin cells)
* quiescent current in the tens of uA - lets target <100uA for starters (<50uA is better)
* If it has a really clever sleep mode this may go a long way - many can sleep for like <3uA
It is a little hard to find them with the common search terms that I use like
* micropower gate driver
* ultra low power gate driver
* low quiescent current mosfet driver
Too many mixed words - since most people are concerned with the parts ability to *drive current* not conserve current...
Usually good engineering companies will make app sheets that will hit on phrases like that.. but not so much for gate drivers... or their idea of ultra low power is a lot different than mine.
The idea of a gate driver with a built in boost sounds f'ing awesome... but that comes at quite a cost.
This is an amazing example of something that *almost works*
* 50uA burn current
* Built in boost for 3.3V operation
* Active low input (we dont want this... easy to address tho for about 1uA)
* High side (not what we want... but sometimes they offer a kluge mode... not with this one it appears)
http://www.irf.com/product-info/datasheets/data/auir3240s.pdf
-methods
* It does not have to be isolated
* Low side driver
* 3.3V or 5V input is fine
* Integrated boost or an input that will go down to 9V (three coin cells)
* quiescent current in the tens of uA - lets target <100uA for starters (<50uA is better)
* If it has a really clever sleep mode this may go a long way - many can sleep for like <3uA
It is a little hard to find them with the common search terms that I use like
* micropower gate driver
* ultra low power gate driver
* low quiescent current mosfet driver
Too many mixed words - since most people are concerned with the parts ability to *drive current* not conserve current...
Usually good engineering companies will make app sheets that will hit on phrases like that.. but not so much for gate drivers... or their idea of ultra low power is a lot different than mine.
The idea of a gate driver with a built in boost sounds f'ing awesome... but that comes at quite a cost.
This is an amazing example of something that *almost works*
* 50uA burn current
* Built in boost for 3.3V operation
* Active low input (we dont want this... easy to address tho for about 1uA)
* High side (not what we want... but sometimes they offer a kluge mode... not with this one it appears)
http://www.irf.com/product-info/datasheets/data/auir3240s.pdf
-methods
I haven't read the mass of text lately, so this may have been covered. I don't think it's hard to make a charge pump based gate driver. You need about 3 pins from the MCU and some small components. Or do you want the IC just to decrease part count?
methods
1 GW
bearing said:
Yes
I have developed many products but only a few of them have successfully made it into continuous production... this is because every part counts... every part has to be placed with tweezers
Cost is not the issue at all
It is all about minimizing parts count and power draw.
I will do it if I have to... but the all in one chips offer a lot of really nice features... like under voltage protection. Looking at this one right now:
http://www.maximintegrated.com/datasheet/index.mvp/id/7913
And I know it is a horrid mass of text. I am just thinking out loud
It usually results in lots of awesome help - like what you are trying to offer.-methods
methods
1 GW
bearing said:
I have through some more about this. I worry that I will not be able to turn on/off the mosfets fast enough with a setup like this. In the past most of my high power mosfet issues have been due to my failure too turn them on or off fast enough. Especially off.
I considered just poking a tiny (linear technology) inductorless boost setup-up on the pin output and that suffers the same issue - slow gate turn-on. I think it is best to design something that can SLAM the gate on and then slow that down with a resistor if needed. In this way I can expand out with mosfets in parallel... or expand up with higher voltage/power fets (more gate charge).
I am considering high side switching - but ugh - I really dont like the implications. Low side is just so much simpler..... and I have a rare opportunity for simplicity with this battery that is usually only afforded when an isolated driver is used.
I will find something. I have already looked through at least 50 data sheets... my experience is that at the end of one of the binges someone just pops up with some dream device that slipped through the net
-methods
methods
1 GW
I think I found it!
I had looked over Farchild... when I remembered them I used the parametric search to choose the crappiest gate driver they had
http://www.fairchildsemi.com/ds/FA/FAN3111C.pdf
We are looking at the FAN3111E
Dont get confused by the FAN3111C.... (which has input thresholds that are a percentage of Vdd and not xref)
The FAN3111E can take either 3.3V or 5V input by using a reference input (2V to 5V, sets the threshold levels)
On the driver side it can take 4.5V to 18V
Icc is 5uA with no inputs connected
Iin LOW or HIGH are both 50uA max (though I bet they run way lower at ambiant with a reasonable xref)
So call it a 50uA part worst case(ish)
It can drive over an amp into the gate - plenty
Output has a 100K pull down - so: 9V/100k = 90uA
hmmm...
Ok - now we are talking about more like 150uA operation.
Considering the higher rate disposable batteries
150uA/hour
250mAh battery
1,667 hours
69 days in full blast
833 days on for 2 hours per day (2.3 years)
Getting there... Getting there...
And that is just one cell.
Sticking features up your butt does not make you a chicken!
-methods
I had looked over Farchild... when I remembered them I used the parametric search to choose the crappiest gate driver they had
http://www.fairchildsemi.com/ds/FA/FAN3111C.pdf
We are looking at the FAN3111E
Dont get confused by the FAN3111C.... (which has input thresholds that are a percentage of Vdd and not xref)
The FAN3111E can take either 3.3V or 5V input by using a reference input (2V to 5V, sets the threshold levels)
On the driver side it can take 4.5V to 18V
Icc is 5uA with no inputs connected
Iin LOW or HIGH are both 50uA max (though I bet they run way lower at ambiant with a reasonable xref)
So call it a 50uA part worst case(ish)
It can drive over an amp into the gate - plenty
Output has a 100K pull down - so: 9V/100k = 90uA
hmmm...
Ok - now we are talking about more like 150uA operation.
Considering the higher rate disposable batteries
150uA/hour
250mAh battery
1,667 hours
69 days in full blast
833 days on for 2 hours per day (2.3 years)
Getting there... Getting there...
And that is just one cell.
Sticking features up your butt does not make you a chicken!
-methods
methods
1 GW
I am now considering different coin cell configurations
* One coin cell to directly power the uController
* Two or three coin cells to power the mosfet only
I know that does not sound like a good idea... but the idea of running two different battery voltages to avoid the need of a regulator + caps is interesting.
I may do this:
3 cells in parallel - 750mAh
Power the chip directly off of these cells
Run a boost regulator to pump the voltage up to greater than 7V, and hang a huge cap out there to supply the gate driver
The uController could control the boost regulator - shut it down to save power when the system is shut down
Just thinking...
Just thinking...
Would normally rather just have 3 coin cells in series than a boost regulator - but Linear makes some really nice units that require no outside components other than a couple caps (not even an inductor).
-methods
* One coin cell to directly power the uController
* Two or three coin cells to power the mosfet only
I know that does not sound like a good idea... but the idea of running two different battery voltages to avoid the need of a regulator + caps is interesting.
I may do this:
3 cells in parallel - 750mAh
Power the chip directly off of these cells
Run a boost regulator to pump the voltage up to greater than 7V, and hang a huge cap out there to supply the gate driver
The uController could control the boost regulator - shut it down to save power when the system is shut down
Just thinking...
Just thinking...
Would normally rather just have 3 coin cells in series than a boost regulator - but Linear makes some really nice units that require no outside components other than a couple caps (not even an inductor).
-methods
Similar threads
