1994 US Electricar Repair Blog by Mike

Tonight I finally got this system debugged. Turns out it had a bad transformer on phase A. Runs smooth as glass now. I left it outside since it’s so cold here now. This will cold soak the Dolphin and help to test if it’s prone to classic dropouts or not. Cold and humid weather pushes any marginal board right into classic dropout mode. Clunk clunk!

Got another customers board today from Hawaii. It needs the dropout mod and also I’ve been requested to add the 70kw upgrade. Busy weekend.

I have have gathered up all of my individual schematics of the Dolphin system and am drawing them up to make a master schematic.

Since I’ve been hammering away at debugging this east coast system, I built up another system to help me. So now I have another completely assembled system that runs.

I’ve been working like a dog on making an electric vehicle system for a guy on the east coast. It’s been assembled for days. Runs too. But it has this kind of miss to it. Almost like it’s got one bad spark plug wire. It just runs rough. I’ve not seen that kind of problem before with an AC electric vehicle system.

Also have the spot welder nearly assembled. It just needs some finishing. I hope to create a process to spot weld braided flat cable to lithium battery cells. The cable allows flexing without breakage. Mostly for our combat robot buddies. For automotive use it’s not really necessary. But it will be a great process to have under my belt.

After this system gets shipped, I have gathered all of my hand drawn individual schematics, and will create one large schematic for the entire Dolphin main board. I have almost every square inch documented. Just a little remains. But even a mostly completed schematic will be helpful in debugging. I’m still deciding what cad package to use.

Today it was time to get the Dolphin chargers that are not running back into running condition. The first step was to remove all of the bad parts. Most get thrown away, but the T1 transformer gets rewound. Four of them were rewound today.

I’m building a super charger for a customer. It runs cooler and does a much more reliable job at charging the battery pack.

The control board and the fet boards are nearly done. The fet board is probably going to get some modification. There are no gate drive resistors installed. I think this is a bad idea. Also another fellow added ferrite beads to his gate drives and it really quieted the ringing he saw on his scope. I’m not sure about using ferrite beads like he did. I am more inclined to use a diode like the OSMC design has on it. Not sure what the diode does. Fairly sure it’s for shutting off the gate quickly. I need to send a few emails to find out. Once this mod is finished I can then add the 18 fets.

The chassis I have for mounting all of the spot welder hardware is a bit small. So I think the initial power up will be done on a piece of plywood the same footprint as the chassis. That may help me to layout all of the components.

I finally started a blog for my 2002 Honda Insight. It’s my daily driver but also a car that will get lots of testing so that it can keep up the gas mileage that it is so capable of delivering.

Yesterday the Insight cell testing was finally completed. I put 4 chargers and 2 constant current power supplies to work individually charging 6 sticks at a time. That way the discharger was always testing a stick of cells. It took about 5 hours to do the remaining 15 or so sticks. The diode balancer didn’t work fast enough. So I just used the shorting clips from the flooded nicads to discharge each cell in a stick on those sticks that looked like it could help equalize. I found 3 sticks out of 20 that were able to sustain 6 minutes at a 50 amp (8C) discharge load. The rest fell under that time. That’s 5ah of capacity. Since nicads have a bit of a Peukert factor, then I think an 8C discharge will show a lower capacity than the nominal 6.5ah that they are rated at. My constant current discharger is so incredibly useful for testing these nicad cells.

The next step is to remove the pack from the Insight and do the same testing to it. I am not sure with 80k miles on the pack where the testing times will fall compared to the used pack. I still have to decide what is more useful, fully charging or discharging the Insight pack before I remove it. In order to duplicate the tests performed on the used cell, I would have to charge the Insight’s cells until very warm like I did the used ones.

Here are some Insight nicad sticks. After being tested there is usually 1 or more weak cells. I add a jumper across the weak cell so testing can continue on the rest of the cells.

Here is the ‘production’ area for screening the replacement cells just in case my Insight has some weak cells. I will remove the Insight’s pack after all of these get tested. I really need to get the spot welder assembled so I can repair any weak cells within a stick. The mill will come in handy for milling out the spot welds to separate the cells.

I’ve ordered and will characterize and evaluate a minimal bms system called MiniBMS. Dimitri sells it from Florida. It’s all analog and uses a continuous loop of wire for signaling that one cell has gone too high or too low, by breaking the loop. A buzzer or any other indicator can be used to notify the driver. This single wire concept is suppose to have next to zero noise issues. I would be concerned that such a long length of wire could have inductance sufficient to create a high voltage spike when ever the loop is broken.

The only drawback is that the idle current is rated at 10ma. That uses 87.6 amp-hours per year from the pack. I consider that high. My own design uses 0.75ma at idle, or 6.57 amp-hours per year. I suspect it’s keeping an opto coupler enabled that accounts for the current draw. It won’t tell you what cell is having trouble, only that one of them is in trouble. Several loops could be used instead, breaking the pack into groups. This would at least narrow it down some. My pack will require 100 cells. So it’s important to me. It is suppose to have thermal protection on it as well to keep it from bypassing more than 1 amp.

The BMS board pictured here is $12 for each cell.

The control board pictured here is $30 for each vehicle.

The control board has an on board timer that allows an adjustable delay of the alarm enable. So if the driver momentarily loads the pack past the low voltage threshold, the alarm won’t sound. If the pack has a cell that is low for longer than the delay, then the alarm sounds. The control board uses a simple resistor and capacitor to achieve the delay.

The bms boards can fit any of the large format cells that have 6mm screws. The bms board is small and bolts to the negative terminal only. A length of wire is used to reach the positive terminal. The downside is that during assembly the board could be rotated into a neighboring cell and create a serious short circuit. So assembly has to be more carefully done than usual. More on this system when it arrives.

Lead acid battery’s are the most expensive technology in our vehicles on a $/mile basis. If a good lithium pack could be assembled, it would make for a lighter and far more efficient vehicle.

In the near term I’m building Lithium modules from ten A123 cells in parallel. That’s a 23ah module. Not much harder to make a 20 cell/46ah module. I’m guessing based on my records, that my truck could go 40 miles max on 23ah since the pack would be so much lighter than even the nicad that I currently have. 70-80% of 40 miles is 28-32 miles. That’s a reasonable range for my city driving needs. Each 10 cell module will run about $125 in just cell costs. That’s $3250 for 154lb bare pack. There still has to be a housing of some kind designed to hold the module. It will use up just one half of the battery box. Leaving the other half for another pack as funds allow. Factory packs weigh from 1400-2200lbs. Can you imagine the range and efficiency increases from loosing up to a ton of pack weight??

CAD model of 20 cell A123 module. Displaces 2 flooded BB600 nicads.

Right now the large format cells that most EV folks are using, have very poor regen capabilityof around 0.3C. On a 100ah cell, that’s only 30 amps of regen allowed. Our vehicles have up to 200a with stock settings. 250 amps with modified settings. A 10 cell module can handle 100 amps of charge/regen current for 15 minutes at a time. So I can easily see them handling 200 amps for 5-10 seconds when regen kicks in with modified software, under specific conditions. In other words they are very low impedance cells that can handle our regen needs nicely.

As for discharging, the A123 10 cell module can handle 700 amps of discharge continuously, and up to 1200 amps for 10 second bursts. Our vehicles can only discharge at 200 amps stock or 250 amps in a modified configuration. Here is the data sheet from A123 on this cell.

These cells have a very long cycle life. 1000 cycles at the max ratings listed. Double that with any care at all. The factory told Bill Dube that if a person can tolerate a loss of 50% capacity over time, then these cells can go 10,000 cycles. Most lead packs lose 50% capacity at 200-300 cycles.

For all of these reasons I’m building an industrial spot welder for making A123 modules. Not just for EV’s, but for the Combat Robotics crowd as well. They use these cells without a BMS and they still last a decent amount of time. Far less than they could, but those folks are happy with the performance and longevity.

Each bare 10 cell module will be 66mm wide x 260mm long. That’s 2.6″ x 10.24″. There has to be some kind of structure built to house the module and some stout connections designed so they can be connected together into a pack. I’m thinking of making the structure look a lot like a flooded nicad cell. In fact I wonder if a nicad cell housing could be adapted as the structure. These modules won’t need to be strapped tightly together like the large format cells require so they don’t over expand and die. These cells will never expand.

In theory, a 23ah pack would be 100 modules. This could fit into the Prizm battery tray. My truck holds 126 nicad cells in each half of the battery box. So it could hold 126 A123 modules per side, for a max of 252 modules per truck. The stock charger could not charge them up past 400v so only 100 modules would be fitted per side. A second 100 module pack could be added to the remaining side if desired, and wired in parallel.

I have a couple of dozen A123 cells and the 10,000 amp (5 ms) microprocessor controlled spot welder is under construction. I plan to make one 10 or 20 cell module to take the place of 1 or 2 nicads in my current pack, for testing.

BOTH 2006-2009 Prius D4R headlight bulbs replaced for $126 total, period. That’s right. Both headlight bulbs replaced for $126.

There are several brands of the D4R bulb. The factory Prius bulbs are from Philips. Since the D4R is a commodity part, in other words it’s made by many manufacturers in quantities in the millions, then Toyota just puts out for a bid and picks the lowest bidder with the best reputation. So Philips is nothing special.If you want to replace your factory Prius bulbs with the factory Philips part, then you pay far more. Fine by me. It’s your money. The issue is still that even after being replaced, they are ‘burning out’, again. Think it’s still the bulb?? The Prius is not the only car to run the D4R. Not by a mile. Do you hear all of these same headlight issues with other brands of cars? Not even! Here’s how industrial design electronics work. A spec is generated and then engineering designs the circuitry to that spec. In good companies they test the daylights out of the part to make sure their spec is sound. If it fails, they make the improvements, and test the daylight out of the improved part. Bad companies will just lower their part price to dump the defective design onto the market. Happens all of the time.

Here’s my take on Toyota’s headlight issue. If the bulbs repeatedly die in a Prius, but not in other brands of cars, then it’s not the bulb, it’s the control module that drives it. When a spec is written, a rookie circuit designer will make sure his circuit meets the spec. He’s fresh out of collage and has a lot to prove. So he works hard. What he is missing is that time is the great educator. He doesn’t know if his circuit design will still work once it ages, or once the bulb ages. Since both the control module and the bulbs are active elements of the design, they are prone to wear. So they can drift out of spec. An experienced designer knows this, and makes sure his circuit is made to work with parts that are very well worn or seasoned. Yes I know they do accelerated age testing. But that ain’t real life. Only real life can show how well something works over time. Everything else is just a guess. Otherwise there would never be a recall on anything. In theory, the design should work. In practice, the theory doesn’t work.

Here is what I am offering. Since the bulbs will most likely burn out again in the future, read about it online if you don’t believe me, then just put in something equivalent until Toyota gets their act together. The lower cost bulbs that are available are not that much lower in cost than the factory Toyota bulbs. Remember any good car maker designs in a significant amount of profit for the parts they sell.So of course they will cost far more than the equivalent parts made by other manufacturers. Is it really worth leaving your car at the dealer, arranging another ride back to work, then arranging yet another ride back to the dealership?? Over light bulbs?

Email me and I’ll schedule you in. I’m at mikep_95133 at yahoo.com. If you want me to come to your work and do it there, it’s another $25. I can take paypal, cash, or even checks if you don’t look like a slime ball. If your check bounces I’ll repo my bulbs 😉

If my distributor runs out, there may be a delay. So far so good. Same with the prices. So far so good on them too.

If you absolutely want the Philips bulb, then pay me in advance and I’ll order them for you. The price will be somewhat higher.

If you already have bulbs, and want me to install them. I’ll charge you $60 total for my labor. I’m just trying to pay the mortgage.