Back on track...
Mosfet Choice
IRLB4030
This is a logic controlled 100V 4mOhm HexFet
4.5V will turn it completely on - 4.2V will get it close enough
2 x CR2030 coin cells in series (
stacked in one holder) will drive in respectably across their full capacity
Driven directly off the uController pins - 85mA peak drive each - so a few uSeconds turn on time (~100nC Total Gate Charge)
On-Chip BOD will guarantee that we never drive the mosfets with less than 4.2V
Each mosfet is good for about 20A @ 2W of dissipation, so 40A average current with no external cooling.
I realize that there are no pull-down resistors on the Gates - I am going to handle that with the Tri-State/internal diodes of the ATTiny. I am pretty sure it will work - if it doesn't then I can always hang some post-facto. There are a ton of ways to dump charge off the gate and keep it off... like staying on 24/7 and actively holding it down.
Voltage Range
The mosfets are strapped with a 90V 1.5KW TVS Diode.
The actual DC voltage the mosfet will see is Vcharge - Vbat.
Code:
#_Cells LVC HVC Delta
1 3 4 1
2 6 8 2
3 9 13 4
4 12 17 5
5 15 21 6
6 18 25 7
7 21 29 8
8 24 34 10
9 27 38 11
10 30 42 12
11 33 46 13
12 36 50 14
13 39 55 16
14 42 59 17
15 45 63 18
16 48 67 19
17 51 71 20
18 54 76 22
19 57 80 23
20 60 84 24
21 63 88 25
22 66 92 26
23 69 97 28
24 72 101 29
25 75 105 30
26 78 109 31
27 81 113 32
28 84 118 34
29 87 122 35
30 90 126 36
31 93 130 37
32 96 134 38
33 99 139 40
34 102 143 41
35 105 147 42
36 108 151 43
37 111 155 44
38 114 160 46
39 117 164 47
40 120 168 48
41 123 172 49
42 126 176 50
43 129 181 52
44 132 185 53
45 135 189 54
46 138 193 55
47 141 197 56
48 144 202 58
49 147 206 59
50 150 210 60
51 153 214 61
52 156 218 62
53 159 223 64
54 162 227 65
55 165 231 66
56 168 235 67
57 171 239 68
58 174 244 70
59 177 248 71
60 180 252 72
61 183 256 73
62 186 260 74
63 189 265 76
64 192 269 77
65 195 273 78
66 198 277 79
67 201 281 80
68 204 286 82
69 207 290 83
70 210 294 84
71 213 298 85
72 216 302 86
73 219 307 88
74 222 311 89
75 225 315 90
76 228 319 91
77 231 323 92
78 234 328 94
79 237 332 95
80 240 336 96
81 243 340 97
82 246 344 98
83 249 349 100
This basically shows that we can control charge on nearly any pack up to hundreds of volts. The very high power TVS diode will sink all the nasty current spikes that happen at connection and removal. After that it is simply the DC voltage of the charger minus the DC voltage of the cells - SO SIM PO
Many folks overlook this options - and it is probably because they have seen many mosfets blow up in this configuration. The secret is the ultra high speed TVS diode that eats noise and shits happiness.
THE PCB
As I have stated - when it comes to PCB's (especially those I will be producing myself) I am a minimalist. I have the board down to only 9 parts! The secret here is looking at every single thing you can eliminate... because every one of those is a part you have to buy, test, lay out, procure in bulk, paste for, tweezer down, then inspect. Anyone who has not done medium scale production in a home environment knows nothing... until you do it... hundreds of hours of solder pasting, carefully placing parts with tweezers, baking, inspecting... you just can not appreciate what eliminating 1 part means. Later if I decided to send these out to a board house - yea sure - go crazy and focus only on parts cost - but for my production method - count is >> cost.
Any way, here is the full circuit as it sits right now. No details hidden.
Ok... Ok... before all you uController savvy guys start telling me that there are a bunch of parts missing - check this out holmes
Crystal and load caps
I eliminated the 3 parts by using the internal ossilator. Normally you need to place a crystal and two small caps. That is THREE PARTS! So the internal ossilator is a little sloppy - maybe +/-10% un-calibrated or or 2% if I decide to calibrate them. Totally fine.
Reset Pull-up
I am using the internal 40K-60K internal reset pull-up. It draws basically nothing so long as nothing is connected. The fear people have here is noise, and the solution to that is a solder-jumper that allows me to hard-wire the Reset to VCC if noise becomes an issue. Solder jumpers are little pads that are spaced perfectly to either bridge (or not bridge) at the time of manufacture. They work by sizing the solder paste window. Bridging the RESET to VCC just means that to program one has to touch it with a soldering iron. The main power switch will work JUST FINE to cause a reset condition... as every AVR powers on into reset.
HVC Trigger Interrupt only 1 pin
Basically this has an internal pull-up to VCC. Static draw is sub 1uA 99.999% of the time. In the event that an HVC edge comes we wake up on that pin asynchronously. Current draw spikes when the opto's are tripped - but this can be instantly killed by turning the pull-up off. SO SIM PO.
Power - 2 x CR2030 Coin Cells in series in one stacked holder
So... the absolute maximum voltage for the chip is 6V. I am totally pushing that limit, which make the chip suck power like crazy, and it is quite possible that I may put a schottky diode inline with the battery to drop anywhere from 0.3V to 0.6V. This would get me down to the totally reasonable 5.5v... but for starters I am going to go big or go home. Direct 6V power to the chip FTW. The chip will work down to at least 4.5V before the BOD kicks in - so that is 2.25V per cell... which is basically when they are dead - so if we drop another half of a volt total that puts us at 2.5V per cell... not ideal and I dont want to do it but I will if I have to. For starters - lets see what happens if we just run at 6V
Noise
I am going to roll the dice on this one. I have the standard 0.1uF on the chip and i am adding a very large 22uF next to the coin cell to handle the big inrush blips (like mosfet turn-on, chip wakeup, etc). I think this combination of caps will provide enough stability (considering our power source is like 2mm away) to keep the boogieman away. Only testing will tell - and only rich guys can afford to dangle every bit of protective crap all over the place. Better to try to do what you can and fail, than to see that you cant do it the "best way possible" and never try.
Power Switch
The circuit shows a power switch. That really will be a 2-hole header that goes off to an optional handlebar power switch. This is not useful for normal HVC breaking, but for those who want to use the circuit to do other things... I want to keep the option open. The circuit as it stands could go between 100V and the controller regulator to implement an LVC that would not require wiring into the throttle and would totally shut off system draw. In this case we really are limited to 100V - unless we want to start getting clever
So for some users they may put a switch on their handle-bar to act as the controller main switch. Instead of physically bringing the switch to the outside of the controller box, or doing it Lyen style, they could do it this style
ISP
I am not sure exactly how I am going to implement this - but the circuit will have an In Circuit Serial Programming port. This will allow for end user reprogramming if they have a programmer. I will probably just make an ultra-cheap ($5) Arduino based programmer that folks can use if they want.
Programming Functionality
By using the HVC Trigger Input we can program in any number of user preferences. Simple things like Charger Timeout, Soft Start, etc will surely come up later and I am ready to implement fun functions
Expanded Use - PWM etc
I have selected the two mosfet pins to be PWM pins. This means that we *may* be able to hang a big DIY ghetto inductor off the front of the board and use it for discharge. As it sits it is probably good for about 40A average - which in many cases could be a 100A current limit (depending on the KV match of the voltage/motor, wheel size, etc, etc).
So... that is where I am at today.
This is intended to fill the hole that my HVC breaker has left open. I think it will be able to do everything the HVC breaker could do but for cheaper, lighter, smaller, less current, wider voltage range, MUCH simpler production, wider function range, easier expansion, and just cooler. I can then pass my savings on to the members and try to get more people into LVC/HVC protection. I am considering a package deal for the future that will have as many LVC/HVC/Parallel boards are needed, one of these for charge breaking, and one of these for Discharge breaking, and all the balance taps etc for a reduced cost. I want more people to have cell level protection - that is how you win, and that is how you preserve batteries and have peace of mind.
Possible Changes
I may reconsider the power switch. The circuit draws such little current (it will last YEARS) that I may just want to leave it powered all the time and have the power switch run into one of the ports. The "power state" could be recorded in FLASH and in that case we would actively hold down the gates of the power mosfets. This is a good compromise... as the amount of power I would burn staying powered 24/7 is probably LESS than I would burn hanging resistors off of the mosfet gates.
Ok - I have to go take care of some other business... I am sure I forgot a bunch of stuff but this is where I am at
-methods
