After having the last couple weeks of my life stolen by first illness, then jury duty, I’m finally back in control and ready to get the stuff that needs to get done done to make the projects that need to happen happen- like the car and the kart. I didn’t shave until I was booted from the jury selection- It’s probably just superstition but hey, it can’t hurt to cultivate a “you don’t want me on your jury” look. What a relief to get out- the trial presentation for those selected will likely stretch into August. A real messy attempted murder case.

So, what’s up now? I’m making plans to get the electric MX-5 down to Monterey on Sunday for the Refuel expo/trackday/event at Laguna Seca. I’ve installed a set of MK3 Rudman Regulators to replace my home-built balancer setup as I discussed before. I’ve given the car a couple of test cycles to try to figure out what is really working and what’s not- the new regs look promising but need to be dialed in. There appears to be one really weak battery that is limiting the performance of the whole pack. And the brand new Belktronix DC/DC has something going on that I need to figure out. It’s blown a couple of HV fuses now (where do I recognize this scenario from?) If I can’t get it working it won’t ruin the weekend however, the aux. 12v battery seems to be up to the task of running the car. I got about 15 miles of driving today with heavy vacuum pump and headlight use before the aux battery ran hit the Zilla’s cutout point, and it wasn’t fully charged when I started.

It’s the little things that stack up and become seriously frustrating. The cycle analyst reports wildly varying and totally wrong current readings while charging the car, while seeming about correct under discharge. Even my Fluke 337 current clamp meter is acting sketchy and not entirely trustworthy, particularly for voltage or resistance measurements. There’s definitely something wrong with it.

So, like usual with the practice of trying to get stuff to work there is the sweet and the sour- some things go right the first time, more often things need more attention. Patience and perseverance.

I’m really looking forward to the event on Sunday, I know some people I want to see will be there and I expect plenty of people I will enjoy seeing but haven’t thought of will show up as well. Hopefully the Miata performs OK. Looks like my friends and I will also have the electric recliner in tow. After reworking the battery box today it is even more solid, and stealthy than before. Good stuff. Sadly we didn’t get a chance to set up the remote-controlled auto-recline feature. Another day.

Share on Facebook

Sunset in Southern Oregon (firefox kind of sucks the life out of this picture, unfortunately. Imagine it much more saturated...)

I’ve got three big areas that need work on the car:

1. Overloaded suspension
2. Plug-n-Play chargability
3. Amp-hour/range remaining instrumentation

Starting with #1

I’ve been researching and pondering suspension options. I was thinking about sticking some off the shelf eibach springs and nice shocks on it, but there are no aftermarket off the shelf springs that I have found for the Miata that have a higher spring rate but aren’t lowering springs. I don’t particularly want the car much lower than stock. I just think that option wouldn’t achieve what I’m trying to achieve. I thought about threaded coilovers, but was advised against the cheaper ones, the good ones are very expensive. So then, thinking about maybe combining custom heavy duty springs with OEM style shocks, I contacted Valley Spring Works and was quoted $328/pr for custom springs- nearly $700 just for springs all the way around. I also have been checking out air-shocks, and have found a couple packages- which are ~1500 and up. Similar cost to high end threaded coilovers. I’m not sure about the performance characteristics of the air springs, it seems like their primary purpose is for easily adjustable ride height. Seeing the price of air springs make the custom steel springs seem cheap… which they are not really.

And it just today occurred to me to check the GVWR of the miata- 2770, if you recall, dry it weighs about 2700 now. So it’s definitely over. Damn. Oh well.

#2

I have been planning modifications to the BMS master to allow the landline cord to be attached, the system to start charging then stop when it’s finished- I’ve picked up a few parts for that purpose. I will house the master board in an aluminum box with an AC solid state relay to provide/cut power to the charger. By default the charger will be unpowered, so in the case of a system failure or problem it will just be disconnected. I think I will set it up so that the master board can detect if itself or any of the slaves has crashed on account of software failure or EMI, and reset. There will also be a built in 12v supply to power the BMS off the landline-power, and circuitry for switching the power from landline to car’s power to entirely off. I have decided what sort of connector the charge inlet will use, and will be used in the in-car wiring and on the out-of-car cords and adapters- Neutrik PowerCon 32 amp:

Neutrik PowerCon 32

Power-in will use the chassis mount connector, Power-out will use the cable-mount. Unlike the PowerCon 20, there is no alternate style for power-in and power-out. So, on the long 10/3 cord I have to use with the car I will need to have both a cable mount and chassis mount connector on the ends of the cord- I think I will try making a round sleeve/strain relief on the lathe to mount the chassis mount connector on the cord instead of using a box- on the less prominent connectors inside the car trunk I will just use boxes. These connectors are expensive (best prices I found: here and here) and aren’t supposed to be used to break under load- but they are of very high quality and I feel better about using them than mis-using some NEMA twist lock that is actually supposed to be used with some particular H/N/G 120/240 circuit instead of this more flexible voltage input. These connectors also have a really high quality look and feel that I like. Note- if you screw the strain relief ring on a Neutrik connector down, then unscrew it again with great force, don’t expect the ratcheting action to be nearly as good subsequently- try to avoid taking these connectors apart after you’ve assembled them once!

#3

I ordered a 1000 amp shunt and an Albright SW-200 contactor to replace the Kilovac one on Otmar’s recommendation- in line with what I observed at UC Davis (four welded-shut kilovac contactors kicking around in the hybrid car lab), apparently the Kilovac contactors have an alarmingly high likelyhood of welding shut in break-under-load situations as compared to the Albright ones. Bonus with the Albright- you can inspect and clean the contacts if you’re worried about them. So, I’m going to rework the high-power wiring center in the front of the car to incorporate these new parts. I think I’m going to make interconnects out of copper busbar material instead of welding cable as well… Anyways, I’m working on figuring out how to amplify and detect the +/-50mv signal from that shunt in such a way that it can be used for peukert-corrected coulomb-counting. Which of course opens the door to some useful instrumentation. More on that as I get it working.

Oh, and the vacuum pump is getting to really annoy me. Every time I switch the car on it makes a huge noise, for a way long time. I think I might just get fed up with it- I’ve seen the YT STABLE product in person, I was pretty impressed- it’s powerful enough to replenish the vacuum fast and it’s not a diaphragm pump so you get much less vibration and noise. EV Source is carrying it now. It’s a little expensive but jeez that diaphragm pump is getting on my nerves. I turn the key and invariably whoever is around says- what’s that NOISE- man!?

Much more in the works. Stay tuned.

Share on Facebook

I visited ohio and drove my great-uncle Cal’s tractor on the old family farm:

The semester ended, I got good grades.

As I have hinted, I drove the car- three times now. I still haven’t edited the video from the first drive, maybe I’ll do that tonight. The second time was to the DMV to get the car “verified” and registered, and the third time was just last Sunday, down to the autocross event at Golden Gate Fields. I have been working hard on the BMS/automatic charge balancing system, I think it’s almost working. In fact, that ‘s what I’m working on right now. I’m doing the first run of the charging system with a very basic charge profile set up, just getting a feel for how (if) it works. I’m babysitting it here in the car right now, I figured I might as well use this time to write about what has transpired… as you can imagine I’ve been putting it off.

So, what first-

I am feeling more and more confident about the powertrain of the car, now having driven it three times, I have not detected any issues with the clutch, adapter plate/hub, motor mount, batttery mounts, battery cables or other electrics. The DC/DC converter (IOTA DLS 55) works just fine and quite unobtrusively once I upped the HV fuse for it (it instantly blows 6a fuses, works fine with a 17a fuse) I have a class J fuseblock and fuses installed presently, that’s what EVSource.com was selling at the time I was buying- now they’re selling smaller circular fuses and blocks, I may need to change to this new type of fuse- it’s surprisingly hard to source DC rated class J fuses for reasonable prices. Of course that’s another good $100 bucks on the table, so I’ll hold off for now.

I drove the car to the DMV with my Dad following me with a tow rope, just in case. The drive there and back was mechanically straightforward, and pretty fun. The inspection went fine, they verified the VIN and the electric-ness of it then took the old title, gave us a new registration sticker and sent us on our way. Unfortunately, as is described on the Wikipedia page on EV conversions, I came away with a Q motive power code as opposed to the rightful E power code- apparently that’s the default behavior of the computer system. Q designates hybrid power- as of now hybrids are not subject to smog checks, so this problem is not an imminent issue for me, but it is kind of dumb. I have heard that this is generally a pain and that one way to tackle it is to complain to the governor’s office. I have higher priorities on my list now but this is something I may take on later.

On the way back from the DMV Dad and I stopped by our mechanic’s shop to show off the car, they’ve heard about it plenty of times over the last two years and have been interested in seeing it. I wasn’t expecting all the mechanics from Steve’s Auto Care and the neighboring Albany Tire to come out onto the street and check out my car, but it sure was neat.

Then last Sunday I took the car down to the autocross at Golden Gate Fields parking lot- (1.5 miles away) I had heard they would have scales there, I have been quite curious how much weight the car gained and how my gut-feel engineering worked out for the weight balance-

2705 lbs, very evenly distributed- could be worse. I was hoping it wouldn’t weigh quite so much, but all in all it’s pretty good and I’m quite pleased about the weight balance. I definitely need to install stiffer suspension components- either OEM style shocks and custom heavy springs or fancy ground control coil-overs, depending on what sort of advice I get and what the budget is.

I gave one or two people rides through the parking lot although I didn’t attract that much attention as I was not officially part of the event- that’ll come another time. I think the car has potential to be a pretty competent autocrosser. When I arrived back home I tried the speedy launch technique with the clutch a couple times- it really is a rush to wind up the revs and then feather the clutch out. 0-40mph is stupid fast. Just for the heck of it I tried letting the clutch out a bit faster and did a tidy little burnout-

I boosted the contrast a bit to enhance visibility, the super crummy tires on the car now don’t really leave much behind. The tracks are about 15ft long and pretty even, I think it confirms the car has an LSD. It was super loud and pretty fun. I’ll have to refrain from being so obnoxious on my street in the future…

As for the BMS system, I haven’t got a lot of photos, but it is coming along well. Like I said, I am as I type watching the character VFD (vacuum fluorescent display) I programmed to display battery related data from the BMS bus as the PFC 20 charger is cranking about 13 amps into the battery pack. The battery voltages are within about .6v of each other, the top one is about 14.2 right now. At 14.7 the shunt resistor is supposed to start kicking in and the master node is supposed to start slowing the charger down via the REGBUS port. Since we haven’t hit 14.7 yet I’m not sure if all that is going to work… but I’m hopeful. Some things are already looking good, despite the charger generating an insane amount of EMI- it destroyed the radio reception here in the garage- the BMS data bus does not seem to be having any issues- now the PakTrakr is also not having issues, and it consistantly has problems whenever the Zilla is powering the motor. I feel hopeful for the bus though, it uses the differential-signaled CAN transcievers and operates completely over twisted pair wire. I was able to get the system to crash by cycling power to my old school flourescent desk lamp, but I already knew that it generates some pretty nasty EMI pops- it’ll make loud pops in headphones attached to the computer. Hopefully the system will be able to take it OK, it would be neat to have the individual battery volt readouts while driving.

The first battery has hit 14.7 and the shunt resistor duty cycle is about 1/5 right now, drifting up and down, but successfully clamping the battery voltage to 14.7. The real test will be when the duty cycle hits 245/255 and the master node tries to slow the charger down… so far so good at least.

Good news- good good news- most of the batteries have hit 14.7 now and all the shunt resistors seem to be working perfectly. Even better, some of them have reached high enough duty cycle to make the master crank the charger down some, which seems to be working just fine. In fact, everything seems to be working exactly as it’s supposed to. I wish I had my camera out here… and I don’t want to leave this unattended.

Well I can use the iSight camera:

As you can see in that photo there are two batteries that seem to be lagging significantly. I’m not sure about them but I’m going to give it some more time.

Ah, well. So here’s the weird part- I noticed two batteries venting very slightly- number 10, which seemed to charge and act just like all the other batteries was hissing slightly from the (-) side vent. Number 6 which was the laggiest battery, its voltage would rise then fall then rise but I don’t think the shunt ever turned on for it was hissing more noticably from the seam where the top meets the casing. Pressing down on the seam would stop the hiss. I know for a fact that the datasheet figures were not exceeded, in addition to my BMS system which is reasonably consistantly accurate to .1v I checked all the batteries and these ones in particular several times with my fluke meter and found the voltages to be well within the expected range, 14.6-14.8v usually exactly 14.7v. I did not even attempt the Optima specified finish cycle (2 amps, constant current) where the battery voltage would certainly excede 14.8v. Hmmmmm. I’ll need to check these voltages tomorrow to get a good SOC approximation and call up the experts. During the previous two times I charged this pack, using the Husky brand 2/10/20 amp digital charger I never noticed these hissing noises, however the Husky charger fan is very noisy- although the hissing noises persist after removal of charge current and I never noticed it before.

I still count this test as a success, my BMS system performed exactly as I expected and intended. I have heard about how touchy these optima batteries can be, I hope I can figure out what exactly is going on here.

… so that’s the massvie update. I will attempt to keep more consistantly up to date in the future. It’s been a stressful and busy last few weeks though, believe me.

Share on Facebook

This is a work in progress- I am working on planning how the battery regulators for the car are going to work. One key part of that process is deciding on the actual optimal charging procedure for the Optima-type AGM battery.

One often-seen charging procedure is the “three-stage charging” process. It consists of three distinct phases:

1. Bulk, constant current. The charger puts out its maximum possible current until the battery voltage reaches a certain setpoint (usually 14-15v for 12v nom. batteries.
2. Absorption, constant voltage. The charger then holds the battery at that setpoint as the current declines, either for a certain amount of time or or until the current drops below a certain setpoint.
3. Float, constant voltage/current. The battery voltage is kept at a lower voltage (usually around 13.8v) where there is little change in the current.

This process is recommended for its speed and efficiency. It would be reasonably possible to implement with the battery regulators.

The Optima documents specify a different regimen, however, for cyclic applications (applications like in the car where the batteries are cycled significantly, often- as opposed to starting or UPS use)

Cyclic Applications:

• 14.7 volts, no current limit as long as battery temperature remains below 125°F (51.7°C). When current falls below 1 amp, finish with 2 amp constant current for 1 hour.

This is a bit different, and a bit strange seeming. I found the Exide Orbital tech sheet a little more descriptive and useful to clarify and reinforce this different idea:

2. IUI or Constant Voltage/Constant Current/Constant Voltage Recharge: Appropriate for more continuous
cyclic discharge/recharge applications
o Step 1: Maximum current limit of 20% of nominal capacity (approx: 15 amps) and voltage limit of 14.4 volts (A)
o Step 2: Hold at 14.4 volts until recharge current reaches 2% of nominal capacity (1.5 amps for most Orbital sizes)
o Step 3: Hold current at 2% of nominal capacity (1.5 amps) for 2 hours
o Discontinue charge if battery reaches 50oC (122oF)

Taken from Exide Orbital Technical Information PDF

While the numbers Exide chose for the Orbital charging recommendations are a bit different, the concept is very much the same. It’s definitely useful to see this kind of consistancy.

So, how will this work?

• At first, all the regulators will not shunt, voltage will be under 14.7 volts at all batteries and current at all batteries will be the same, up to 20 amps (charger limit)
• As some batteries start to approach 14.7v the regulators will start regulating, up to their maximum shunt current, at which point the master regulator will start scaling back the charger current. Batteries that have not reached 14.7v yet will continue to charge at the lesser amperage.  Charger current will continue to decrease to avoid pushing the strongest batteries over 14.7v.
• The charger output will go as low as 3.5 amps, since the regulators sink about 2.5 amps at wide open, that will allow the strongest batteries to reach 1 amp at 14.7v while the weaker ones continue to charge at up to 3.5 amps
• Once the strongest batteries have reached 1 amp at 14.7v the regulators can go into “extended absorbtion” phase for those batteries, with the charger still at 3.5 amp output can hold those batteries at 2 amp constant current for an hour, while the weaker batteries continue to do whatever they need to do.
• Once the strongest batteries have been charging at 2 amps for an hour, the charger will reduce to 2.5 amps to allow those batteries to stop charging entirely or 2 amps if all the other batteries have reached “extended absorption” phase. The regulators on those batteries that have completed an hour will shunt all the charge current.
• A low current float phase could be implemented after all batteries have completed an hour at 2 amps.

The individual regulators can be aware of the voltage of the battery they’re regulating and the current they’re sinking in the shunt by way of ohm’s law and the duty cycle the shunt is running at. The regulators must be able to be aware of the current the charger is putting out for a constant current mode. Temperature compensation capability would be desirable. The master charger regulator must be aware of the duty cycle of the regulators and the voltage they’re sensing to decide how to scale back the charge current. Alternately the regulators could ask the master to reduce current, although the latter setup could be less versatile.

I think that accounts for all the provisions laid out by the Optima charging specs, and this hardware setup. I’m sure I will continue to think about it and how it’s going to work though– of course!

Share on Facebook

I haven’t tested everything about every board, but as far as I can tell, so far everything seems to be working as it should.

What’s Next?

Let’s review the theory of operation of shunt regulators- The charger fast charges (constant voltage) the series string of batteries until a regulator detects that a single battery has reached a voltage setpoint, indicating that it is full. The regulator starts shunting the charging current off that battery. The charger then throttles down to an “absorbtion” rate charge (constant current), and as other batteries fill, their regulators shunt off them too, until every battery has reached the voltage setpoint as detected by the regulators. At that point, the charger enters a float charging mode until it is unplugged from “land” power.

I now have regulators that detect voltage, are able to shunt current and communicate with a master via one serial port of currently undefined protocol and two digital lines that can be pulled low or high by the µC pins. I need a way to instruct the charger to enter the second and third stages of charging. The charger (Manzinita Micro PFC20) has a 6 connector RJ12 port called REGBUS which is designed to allow the charger to work in sync with Manzinita Micro’s Rudman Regs. It’a an analog interface, electrically specified pretty well via the files on this page. There is no specific information or How-To on faking the interface with a microcontroller though- it’s a simple bus though, so I decided to figure it out myself. I have a current clamp meter to monitor the charger’s output, and some potentiometers and RJ12 plugs and wires to generate signals on the REGBUS, I’ll see what happens. I was thinking of picking up a pair of L16 batteries to test with, which I can then integrate into my long neglected solar-electric power system.

I can envision this working one of two ways. Either the charger current response is proportional to the analog signals on the REGBUS, or the current response is preset and the analog signals are used as triggers to move the charger into different preset modes. I hope it’s the latter, so that I can generate the analog signals with a few resistor networks and optocouplers- if it’s the former, I might have to look into digipots or DACs.

No use thinking too much about it before actually doing the testing and figuring out what I’ve actually got to work with here.

In order to make the charging system work without intervention, I need this master to monitor the “Reduce” and “Finish” digital signal lines on my “BatBuss” connector and generate the analog voltages needed by the PFC charger. In future versions, I’d also like to integrate…

• CAN Bus for reporting state and voltages, AH counts and SOC
• battery current sensor, charging current sensor
• Solid state relay switching of AC side of charger for protection or automatic time of day charger on/off switching
• one RGB LED PWM channel for making a Tesla-style “fuel filler light ring”
• data logging

Sounds good, right? For now, I just want to get the batteries charging. Speaking of batteries, mine are on order! Should be arriving Apr. 8!

Share on Facebook

The circuit boards came in from Gold Phonenix, it really is very easy to order from them, and I am pretty happy with the quality. I first contacted them on March 13th (Friday), asking for quote confirmation for a 155 in^2 “Special Price” order, including the gerbers in a .zip file and my location (California). I was given a quote on the 15th (Sunday). I sent the money via PayPal, with my shipping address and the name of my .zip file. I was notified that the order was processed on the 16th (Monday). A shipping notice with tracking number was sent on the 23rd (the next Monday), and I recieved the boards this morning. That’s roughly 12 days.

The boards look quite good, all the traces and through plated holes stand up to visual inspection. The silkscreen is not very well alligned, and has minor smudging in a couple places.

I was quoted for 2 8″x8″ panels, with 4 boards each, and recieved 3, for a total of 12 boards, 168.75 in^2. This is in line with other reports I have read of GoldPhonenix delivering more than you might expect. Fine with me! These boards ended up costing about $10 each, $.71 per square inch. Really good deal.

My order of parts from Digi-Key arrived over the weekend. It has now taken the record of biggest Digi-Key box yet from the prototype parts shipment.

Better yet, I managed to order all the parts I need! Planning pays off.

So far so good, the power supply section of the board works just fine-

and the ICSP works just fine

The power transistor footprints look good, I picked up some 3/8″ long 4-40 machine screws and nuts- looks pro

So, where’s the dramatic tension? Why is this post named “FAIL”? Because I screwed up the circuit- take a good look at that first image, DUH, I swapped the connections to C and E. It’s so clear now. Oh well, I can live with a little re-work. Gives it character. Everything else seems to work well. I’ll fix that, and fix my minor screwup with the silkscreen labels of the faston terminals- CONN_1, CONN_1, CONN_1… is not very descriptive. I thought I had fixed that- KiCad un-fixed it when I reread the netlist. CONN_1 is the default value of that schematic element, and by default it is not printed in the schematic- so you have to double click on the schematic element, go to “Fields” then select “Value” to change it. You can change the text in pcbnew (be sure you change the value, not the reference), but it will revert to whatever Value is in the netlist if you read the netlist in.

I also replaced the current limit resistors on all the LEDs with 470ohm instead of 220ohm, except for the (blue) power LED.

I think we can make this work.

Share on Facebook

I finished the new board design a few days ago- I don’t think I’m an expert or anything, but definitely did a better job this time than I did on the prototype- Just to begin, I checked my schematics and footprints carefully. That in itself will make a big difference! I also got the overall size way down- from almost 24 in^2 to about 14 in^2. I spent some time refining and fixing up the silkscreen for usefulness and legibility, and ran DRC tests. I have at least 8 mil clearance throughout the board- It’d be nice to have more, but it is sufficient for the requirements of the board house. I ended up running a bunch of traces between pins of headers and DIP chips.

As with many projects, in order to move things along it’s now time to buy stuff.

• I contacted Gold Phoenix’s sales rep to arrange an order of the final circuit boards. I think I should be able to get 10 or 11 boards for $99- an excellent deal.
• I recompiled my BOM and made a digi-key order for all the parts necessary for populating another 5 boards- $68 dollars worth of wirewound resistors! Over $250 total.
• I purchased LEDs, fuses and heatshrink from MPJA, I still haven’t decided exactly what LEDs are going to be what colors, but I knew I needed a bunch of reds and greens.
• I bought a buch of 1/16 aluminum sheet on eBay, I can use a shear at school to cut it to make base plates for the regulators- I’m planning a setup kind of like this:

• I also ordered four 80mm fans on eBay to mount over the resistors. I decided it was totally worth it to get the blue LED versions.
• And, I ordered a Fluke 377 current-clamp multimeter. Besides always wanting one, I will use it to monitor the charger when I figure out how to control it via the REGBUSS port. I considered cheaper clamp meters, but the conclusion I drew from reviews online was that the Fluke was really the way to go. Expensive, yes.
• Oh, also I ordered a new FTDI board from sparkfun that will go right on the header I have on the reg board, as well as some little board with a RJ12 pigtail- meant for use with the PIC ICSP system, I will use for the REGBUS. And a Salae Logic analyzer. I think it will be useful.

So, am I some kind of high roller here? Spending all this money? No, but I have to finish this project right, and I am very lucky that my parents are supportive of my ambitions.

I am looking forward to testing this regulator system on a real battery pack- I think that’ll be coming pretty soon.

Share on Facebook

I finished testing the battery regulator board over the last two days, as I planned I did a full-current “burn” test and tested the MCP2551 (CAN transciever) based communications circuit. Things went mostly according to plan.

I had planned to burn in the power circuit for ~2hrs, but I got all the information I needed in about 15 minutes- the transistor is fine, by my calculations it should be burning no more than 1w (less with higher hfe optocoupler) which the datasheet specifies is OK without a heatsink up to ambient temperatures of 40-60 C. So, I will make thermal pads and bolt the TO-220 pads to the board, I don’t think a heatsink is necesesary.

The power resistor on the other hand didn’t fare so well. Just because a power transistor is rated “50w” doesn’t mean it can take a power close to that limit for a sustained amount of time without any thermal considerations. That resistor got crazy hot. 33 watts is a lot of power when it’s all going to heat in such a small area. The anodizing changed colors, the plastic plugs at the ends of the resistor expanded outwards and the solder melted causing the wires to slide around.

I cut the power when I noticed this, once it cooled down again the plastic plugs retracted slightly, but you wouldn’t mistake it for new- and the anodized surface is still a little darker than new resistors. It still reads an acceptable resistance though, so I’m not going to toss it.

When revising the schematic earlier I decided to nix the “Fan” output transistor, which after seeing this I have decided to add back in. I think if I mount the resistors on some aluminum plate and use some forced air everything should be OK.

About the MCP2551 CAN transciever, which I added to the schematic as an easy robust way to implement an EVILbus style communication bus (BatBuss) is a complex little guy- read that datasheet closely. I assumed it wouldn’t care how slow the baud rate on the bus was, so I could use it for even DC signals- that’s not the case though.

1.5 TXD Permanent Dominant
Detection
If the MCP2551 detects an extended low state on the
TXD input, it will disable the CANH and CANL output
drivers in order to prevent the corruption of data on the
CAN bus. The drivers are disabled if TXD is low for
more than 1.25ms (minimum). This implies a
maximum bit time of 62.5μs (16kb/s bus rate),
allowing up to 20 consecutive transmitted dominant bits
during a multiple bit error and error frame scenario. The
drivers remain disabled as long as TXD remains low. A
rising edge on TXD will reset the timer logic and enable
the CANH and CANL output drivers.

When I used another MCP2551 to check the onboard one by making a little sketch to toggle the “TX” pin every couple of seconds I got no output on the second transciever. When I set the sketch to toggle with 2ms cycle delay (effectively probably 300-500 hz) my multimeter read ~2.5v on the output of the second chip. This would idicate a duty cycle of 1/2 and is enough to make me think the chip is working right. This means my bit-banged “BatBuss” must run at at least 16kb/s- more challenges. At least I can leave that until later.

And voilà- I think I have enough information to redesign the board. When I have finished, and verified correct operation in service I will post the schematic and software, or at least software specifications. I have considered making this a kit also, since I wasn’t able to find these regulators on sale when I was looking, someone else may be interested in skipping this arduous design process!

I have been meaning to publish some of my older projects, as well as to make an about page, put up some links, and change or customize the CSS- look for all that soon.

Share on Facebook

I have confirmed that the board programs and communicates correctly via serial with the Arduino environment. Almost everything is working as planned.

The timer that drives the PWM on pins 11 and 12 (ATMega168) or 5, 6 (Arduino), OC0A and OC0B, is used for system timing stuff too, so it doesn’t ever go completely “off”. I moved the optocoupler I noted was giving me problems with the ICSP to one of these PWMs- so once the PWM was enabled, it was never at duty cycle 0. That’s not OK, because then the shunt is always sucking a little current. So I need to move it. I think I’m going to move it to PD3 (OC2B) and move what was on there somewhere else. Makes stuff a little less neat, but I think it’s the best option.

Things that need doing in order to order more circuit boards:

• Verify correct operation of communication circuit
• Run long term (2hr) burn test at full shunt power to make sure nothing overheats, gauge the need for heatsinking the power transistor
• Fix power transistor footprint and redesign board and traces, make it smaller. Use 1/8 watt resistors where appropriate

Things that need doing after that:

• Test and plot linearity of analog optocoupler, adjust trimpots
• Get a feel for characteristics of pwm cycle control over shunt current
• Plan firmware

There’s a lot to do. I’m especially concerned with the eventual need to develop the “BatBuss” serial interface, loosely based on “EVILBus”. It’s going to need to be Bit-Banged, and I really don’t have a good idea about how to implement that yet. I’ll figure it out, and I actually don’t even need to to just get the regulators working, but I’m worrying about it anyhow. I also need to get an idea of the characteristics of the REGBUSS on the Manzinita Micro charger, and figure out how I’m going to effectively trigger it to go into Float and Finish modes. I think it will consist of some optocouplers and resistor networks, but I’m not sure yet. Need to test it, and to test it I need some guinea pig batteries and a current meter. An excuse to pick up a nice clamp meter? Possibly.

Share on Facebook

I continued troubleshooting the board yesterday, and came to a few conclusions- The gain of the power transistor is indeed high enough to be driven directly from the optocoupler. In order to get the C-E voltage drop nice and low (.35 v or less) I am driving the base with about 20 ma- it’s definitely saturated. The optocouplers I ordered have CTR 40-80%, so as it is right now, it’s a bit of a balancing act with keeping the OC pulling less than ~30 ma from the AVR pin while maintaining enough C-E current to keep the power transistor saturated. I think next time I order parts I will order higher CTR optocouplers (there are some better ones in the same series of parts) and maybe tweak the resistor values just for a little headroom.

After having gotten that squared away (whew!) I took a quick trip to Fry’s in Sacramento (got very lost on the way. Why are there two different freeways with such similar names?? I80 and Business 80?) to pick up some new resistors for the power transistor base and the optocoupler LED. Was also able to find pinheaders (Yuss!) and a 5v linear regulator. Having all the parts, I finished populating the board.

(Note the upright metal film resistors placed in the TO-92 footprints)

Looks pretty good, right? Well then I couldn’t connect to the microcontroller via my USBTiny and ICSP header. Frustrating. I thought about it overnight and then earlier this morning I confirmed my suspicion that one of the ICSP pins (MOSI in fact) was also being used to drive one of my optocouplers- I assume that was putting enough of a load on the pin that the ICSP was unable to connect.

So, I would say that most of the kinks have been worked out of this hardware! I’m going to take some time now to fix up the schematic and do a little bit more trace reworking on the board so that it’ll reflect the schematic changes. A note- in KiCad eeschema, using wire labels is useful- but if you have labels like Sense 1, Sense 2, Sense 3, they will all be assumed to join together. So don’t include those spaces- I didn’t get this when I was first trying to layout the board so I made wires in the schematic between everything- messy. I’ve fixed it now.

Share on Facebook