Weber and Coulomb's MX-5

Post up a thread for your EV. Progress pics, description and assorted alliteration
User avatar
weber
Site Admin
Posts: 2598
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Weber and Coulomb's MX-5

Post by weber » Tue, 17 Jul 2012, 18:25

Thanks for that, BigMouse. I assume you mean "heatsink copper pours". There's no solder on them, only soldermask. But I take your point.

We were impressed by this Sparkfun tutorial regarding the merits of electric frypans ("skillets" to Americans) versus expensive reflow ovens: Sparkfun Reflow Skillet Tutorial

But clearly an oven makes double-sided reflow possible while a frypan does not, because in the oven the parts and board are heated from the top by radiation, while in the frypan only the board is heated, by conduction from the bottom. But the frypan is definitely kinder to the components. We'll let you know how the hot-air tool goes.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Weber and Coulomb's MX-5

Post by Nevilleh » Tue, 17 Jul 2012, 18:51

coulomb wrote:
coulomb wrote: What's left is about $1200 worth of parts (plus GST and import costs), excluding the piezos, the super-capacitors, and a few connectors.

In fact, buying $1200 of part from overseas in one order was a bit of a mistake. $1200 is over the threshold of $1000, under which I believe orders will escape duty and possibly even GST. When the order is over $1000, the order gets stuck in Sydney, and someone has to call you to authorise a credit card payment for the duty and GST (GST is paid on the duty and even on the shipping! Image ), and that cost us the parts getting here in time for an EV day (so it cost the project an extra week, effectively). Plus, the duty and GST came to hundreds of dollars. Of course, we don't mind paying our taxes, since it is always used for very worthwhile purposes...

So to avoid the delay, we could have broken the order into two parts, both under $1000. We'd still have qualified for free shipping (Digi-Key offers free shipping on orders over AU$200).


Oh yes, the free limit into NZ is $500 and I was very smart with some stuff I ordered in asking the supplier to split it into two $500 orders. But Customs added the postage on and hence the orders were just over $500, so SLUG! Didn't escape the gst after all. But I know for next time.
I bought a Sunbeam Mini Bake and Grill oven for my reflow work. It cost about $75 on sale, down from $99 and it lets me do double-sided components quite well. I do one side first, preferably the one with only passive components (to avoid doing the active ones twice), then lay the board that side down to do the other. The weight of the board and the flat metal tray under stops the first side parts from moving when they re-melt for the second go around.
I suspect that your modules are too big for my oven though, it measures 280 w x 240 d x 180 h and did all my bms boards.
The advantage of course, is that the components and board are all heated by the hot air and radiation in the oven and you are not relying on heat conduction through the board material. And the thing has 1400 watts of heaters so it doesn't take long to cycle a set of boards through - only a couple of minutes once its up to temperature.
Here's a photo of a set of boards just cooked:

Image

As an afterthought, it took me about 1 1/2 hours to do a set of 18 boards with about 25 components on each board. No stencil, just magifier, syringe and tweezers. That included cycling through the oven.
Last edited by Nevilleh on Tue, 17 Jul 2012, 09:14, edited 1 time in total.

User avatar
weber
Site Admin
Posts: 2598
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Weber and Coulomb's MX-5

Post by weber » Tue, 17 Jul 2012, 19:16

Does anyone have any ideas about how to support the board so it doesn't tilt or flex while solder-paste stencilling the second side, when the first side now has components of widely varying height all over it -- the highest being an electrolytic at 5.5 mm?
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Weber and Coulomb's MX-5

Post by Nevilleh » Tue, 17 Jul 2012, 19:26

Can you grip it between the jaws of a bench vice?

Another thought is get a tray, pour in some moulding resin (polyester), spary a board with WD40 (release agent) and stick it into the resin component side down. One the resin is set, lift the board off and you have a perfect jig for holding a board.
Last edited by Nevilleh on Tue, 17 Jul 2012, 09:30, edited 1 time in total.

celectric
Groupie
Posts: 56
Joined: Tue, 29 May 2012, 05:14
Real Name: Tim Bates
Location: Adelaide SA

Weber and Coulomb's MX-5

Post by celectric » Tue, 17 Jul 2012, 19:33

weber,
Drill holes in a base plate in 4-6 places that line up with blank spots in your PCB. Put nylon bolts through the holes and hold them in place with nuts. Turn the board so that the ends of the bolts are pointing up and rest your PCB on top. Voila, level PCB surface. (Cut the bolts short if it's too unstable.) Hope that makes sense...

User avatar
BigMouse
Senior Member
Posts: 600
Joined: Thu, 28 Oct 2010, 02:39
Real Name: Vincent Tannahill
Location: Silicon Valley
Contact:

Weber and Coulomb's MX-5

Post by BigMouse » Tue, 17 Jul 2012, 19:54

Yeah, I meant copper pours :-)

The place I've been looking at (http://www.goldphoenixpcb.com/) to get my boards assembled in China either let you supply the components, or they order them from Digikey/Mouser for you, so you're assured of the quality and legitimacy of the parts used. I haven't actually got a quote from them yet, but I managed to find a "price per pin" and my boards would come out to around $2 each to assemble. They make the PCB as well and I've read good reviews about their services in general.

As for a level surface for stenciling, the two suggestions above are both good. The moulding resin one could work, as long is it doesn't trap that capacitor. It wouldn't do well if parts aren't in exactly the same position on each board, each time though. What if you got some balsa wood or even some of that foam that people use to poke fake flowers in to for displays. Something that deforms easily when a point load is applied, but folds firm under a distributed load. Lay a completed board on it (you'd have to cut a hole for the cap), and apply an even force with a flat surface (hard cover book?) to create indents of the higher parts in the substrate. You'd probably have to repeat the pressure with the first few subsequent boards to account for variations in part placement, but once that's sorted, it should be quick and easy for the rest.

Celectric's suggestion is probably the most repeatable though, and cleanest.

Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Weber and Coulomb's MX-5

Post by Nevilleh » Tue, 17 Jul 2012, 23:12

Here's another thought:
Make a sort of mini-beanbag with tiny little polystyrene foam chips. Then you can just nestle the board into it and - well, you know how good a bean bag is!

User avatar
weber
Site Admin
Posts: 2598
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Weber and Coulomb's MX-5

Post by weber » Wed, 18 Jul 2012, 01:12

The stencils have arrived, so it's a little less academic now. They are currently underneath two volumes of the "Shorter" Oxford Dictionary to flatten them out, after being mailed in a tube. Actually they had very little curl when I took them out.

Thanks for all those great suggestions, guys. In the end I chose a solution that was phoned in to me by Jeff Owen. Partly because it didn't require me to go out and buy anything. In short: Print out the component overlay at true size (but mirrored), glue it to a piece of wood, and rout out the general areas where the parts are, with plenty of clearance.

At first I told Jeff I didn't have a router, so he suggested drilling an oversized hole for each component. Then I remembered I had the perfect mini-router for the job -- my Dremel.

It only took 3 prints to get the scale factor right (99.7%) on two sheets of A4 paper. I pasted them to an old laminex-clad kitchen cutting-board using a glue-stick. I used a 10 mm drill to make the holes for the 6.3 mm dia, 5.5 mm high electrolytics and used a 3 mm cylindrical burr set at 3 mm depth in the Dremel, with its router-base attachment, for routing the rest of the area (fuse was 2nd-highest part at 2.7 mm). Kneeling on a pillow (with a "Banana's in Pyjamas" pillow-case) on the workshop floor got my (protected) eyes close to the action, and let me keep puffing away the smoke and dust.

Image

After a trial fit I realised I should have included the paste-mask layer in the printout as some components' pins are not contained within their outline. But I managed to remember which ones they were and compensate from then on.

When it was done I tore off the parts of the paper that weren't glued, then used a damp rag to soak and rub the rest off. Then sanded it to remove burrs.

Image

Thanks Jeff.

[Edit: Added "(but mirrored)" in 2nd paragraph.]
Last edited by weber on Tue, 17 Jul 2012, 15:56, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

User avatar
weber
Site Admin
Posts: 2598
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Weber and Coulomb's MX-5

Post by weber » Sun, 22 Jul 2012, 05:14

Last EV day (Friday) I had my usual duo of volunteer geniuses. Electrical: Coulomb (Dr Mike Van Emmerik). Mechanical: Newton (Jeff Owen). But we also had the benefit of Mark Aylott from Brisbane AEVA whose own project is Peltier air-conditioning.

Mark was keen to see our frypan-based SMD (surface-mount-device) process in action, but while he was here he helped out with various other tasks both mechanical and electrical, including mounting and soldering through-hole components and helping Newton re-install a battery box to check on its lack of clearance to some existing 12 volt wiring (the wiring got relocated).

You can see the new hot-air rework station on top of the lab power supply in the following photo. It worked just fine to melt the stencilled-on lead-free solder-paste for the 1 watt 6332 size (2512 imperial) bypass resistors, without affecting what was already soldered on the other side. We set it to 260°C and maximum air-flow and used the largest nozzle. It was quicker than using a soldering-iron and solder "wire", and it did not require 3 hands. The first resistor on each BMU (Battery Monitoring Unit) took maybe 20 seconds, but then Coulomb could progress around the circle at maybe 5 seconds per resistor. So maybe 1 minute per BMU. We saw some resistors self-align as he melted the solder at both ends simultaneously.

Image

The plastic stencils work OK, but we screwed up when we accepted DesignSpark's default setting which reduces the size of the paste-mask holes by 75 um (3 thou) relative to the pads. (DesignSpark is the free schematic and PCB (Printed Circuit Board) software that we use.) This means that the microcontroller pads, which are only 380 um (15 thou) wide, end up with hardly any paste on them. We should have copied this setting over from our earlier Protel file, which enlarges the paste-mask holes by 50 um (2 thou) relative to the pads.

So I will perform some crude modifications with a sharp blade, to join up the 4 holes on each side of the micro into one big hole and rely on surface tension to separate them when the paste melts (as we did when applying the solder-paste with a syringe). And if that doesn't work I guess I'm up for another $82 and another week's wait for a new stencil.

But the general idea of the plastic stencil is good. It beats the hell out of placing paste individually on every pad. It is easy enough to register it under the magnifier-lamp and hold it in place with masking tape.

Image

It doesn't matter how many times you squeegee the solder paste backwards, forwards or sideways, so long as when you're done, you lift the stencil off cleanly in one go. However the choice of squeegee is critical.

First Goldilocks tried Mummy Bear's rubber kitchen spatula, but that was so soft that it gouged the paste out of the larger holes leaving them bear in the middle. Then she tried Daddy Bear's steel putty knife, and while that left the holes filled OK, it went bump. bump, bump over the slight thickenings at the edges of the holes due to the laser-cutting. Goldilocks just knew this was damaging the stencil. Then she tried Baby Bear's credit card, and it was juuuuust right.

The frypan wasn't running quite hot enough on its highest setting so I took apart the thermostat and found where to adjust the temperature offset. It is the tiny blurry silvery philips-head screw inside the brass tube in the photo. Anti-clockwise for hotter. But I couldn't figure how to do anything about the enormous hysteresis.

Image

The outermost BMUs on the string of 8 were taking much longer to melt their solder-paste than those in the middle, so I've tried adding some glass wool insulation to the corners, held in place by wire and masking tape. This gave only a slight improvement, so I plan to remove the oil from under the middle of the heat-spreader as well, leaving it only at the ends, to try to get a more even temperature along it.

Image

The four symmetrically-placed white streaks in the frypan above are the melted remains of thumb and index fingertips of these "100% Cotton" gloves we ordered on eBay, from a seller on the Sunshine Coast (Australia). Turning them inside-out revealed a tag saying "Polyester Cotton". Sigh. We emailed the seller and they responded with a full refund including postage, and did not ask us to return the unused gloves. They said they were not aware the gloves were not 100% cotton and will change their listing to avoid misleading others.

[Edit: Clarification. Added Mark's surname. Hot air temp changed from 320°C to 260°C.]
Last edited by weber on Mon, 27 Aug 2012, 06:41, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

User avatar
weber
Site Admin
Posts: 2598
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Weber and Coulomb's MX-5

Post by weber » Sun, 22 Jul 2012, 23:12

There has been a horrifying sight greeting visitors to the lab over the past year -- a pile of those copper straps for linking cells together, with evil-looking blue-green powdery patches all over them. "What happened there!?" they would say.

About a year ago we found that a large number of the straps, that were supplied with the batteries, had been left exposed to the air and were dark brown, which I understand is due to oxides and sulfides which are much less conductive than pure copper. These straps are made up of several thin layers of copper and presumably the tarnish was in between the layers as well. We didn't fancy taking them apart and cleaning every surface mechanically, e.g. with wire-brush, wet-and-dry sandpaper or abrasive paste. So I tried cleaning them chemically with hydrochloric acid, which dissolves the tarnish but not the copper. This was quite successful but I rinsed the acid off with water and did not immediately dry them and they turned a patchy brown again very quickly, depending on where water drops were sitting on them.

So I put them back in the acid, rinsed them with water, then gave them a final rinse in metho (denatured ethanol). This was disastrous! They dried with the aforementioned blue-green powdery patches! So I gave up and left them like that. But we can't ignore them any longer. So after sorting out the previously-mentioned battery-box cable-clearance problem, Newton moved onto the corroded-straps problem.

He still did it chemically, but with a couple of changes. No metho was allowed anywhere near them. The hydrochloric acid pickling stage was retained, but was preceded by a sodium hydroxide degreasing stage, and was followed by a hot-air drying stage and then a dip in the "grey slime" (explained later).

Image

The degreasing bath was 1 tablespoon of Sodium Hydroxide in 250 mL of water. The pickling bath was 250 mL of 33% Hydrochloric Acid in 250 mL of water, with additional acid added when it seemed to be weakening. You must use a face-shield, rubber gloves and tongs when working with these nasty chemicals.

Here you can see a strap being transferred from the blue degreasing bath to the green pickling bath. Notice the horrible blue deposits on it. They changed from their original blue-green to dark blue when they were put in the degreasing bath.

Image

Nearly done. The sink on the right is full of rinse water.

Image

The drying stage

Image

The "grey slime" was invented on the day. Newton suggested we should put the "grey goo" (electrical jointing compound consisting of powdered zinc in a mineral-oil-based grease) onto the straps as soon as they were dry. Mark noted that this wouldn't get between the layers of copper. Newton suggested liquefying the grey goo with a suitable solvent. I tried mineral turpentine. Utterly useless. Newton suggested WD-40. It worked beautifully.

Image

When it was all over we found that the stainless-steel sink had been copper-plated where splashes from the acid bath had been held under the trays by capillary action. It was surprisingly well stuck too. The photo shows it after I tried to remove it with steel wool. Ah well! That's life.

Image
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

User avatar
coulomb
Site Admin
Posts: 3685
Joined: Thu, 22 Jan 2009, 20:32
Real Name: Mike Van Emmerik
Location: Brisbane
Contact:

Weber and Coulomb's MX-5

Post by coulomb » Mon, 27 Aug 2012, 16:23

There hasn't been much visible progress on the car from Weber or myself, as we crank out the Battery Management Units (BMUs). However I should thank Jeff Owen for tackling mechanical issues. He helped install a replacement brake line that was nicked with an angle grinder years ago, spent hours visiting various places (with helpful advice from Graham of Suzi Auto) to see if they could put the ratio adapter part way up the speedo cable.

Image

Cables have been made up and this is now done, and he found a ratio adapter gear-lockup problem, and fixed it with some judiciously placed washers inside. There was an issue with wobbling of the speedo reading, which required much testing with an electric drill and fiddling. It all seems to be OK now; thanks, Jeff!
Last edited by coulomb on Mon, 27 Aug 2012, 08:07, edited 1 time in total.
Nissan Leaf 2012 with new battery May 2019.
5650 W solar, 2xPIP-4048MS inverters, 16 kWh battery.
1.4 kW solar with 1.2 kW Latronics inverter and FIT.
160 W solar, 2.5 kWh 24 V battery for lights.
Patching PIP-4048/5048 inverter-chargers.

User avatar
coulomb
Site Admin
Posts: 3685
Joined: Thu, 22 Jan 2009, 20:32
Real Name: Mike Van Emmerik
Location: Brisbane
Contact:

Weber and Coulomb's MX-5

Post by coulomb » Mon, 27 Aug 2012, 16:43

We are building our Battery Management Units (BMUs) on "boards" of 8 BMUs joined to each other with "squiggle joins".

There are several main steps: manual placement of the surface mount parts on the "red" side (red as in the colour of the tracks in the CAD package), bypass resistors on the "blue" side, hand soldering of the through-hole components, and testing.

Testing involves JTAG programming and verification of the various features of the BMUs. So we make each LED glow, check that the piezo is audible and possibly change the octave that we drive it at, and so on. Usually it takes around three minutes to test a board (8 BMUs).

The following should tile properly if your screen is wide enough and your browser is maximised:

ImageImage
ImageImage

Here you can see the test jig with its 16 clip leads for power; to its right a JTAG programmer interface (in a plastic bag), a 3.75 digit multimeter for voltage calibration, to the right of these a tested but failed board with coloured spots and brief notes about what failed; extreme right is the netbook we use for JTAG programming and serial comms; to the left of the screen a power supply; above that another multimeter to measure current (mainly for measuring bypass current); to the left of that a Digital Storage Oscilloscope showing comms waveforms with some spikes (these spikes disappear when the output is measured differentially); above that, our Tesla Roadster and some arty placements (Weber composed this image); to the left of the test jig is a non-contact thermometer for temperature calibration; above that our RS-485 interface, with some IFO electronics that isn't used just now; finally at the very bottom left, an new board ready for testing.

However, one string that was initially passed had to be checked again (because we changed the value of a resistor to make the yellow LEDs brighter), and we found some BMUs were intermittently not showing the blue LED after a command as they should. It happens to be the string in the photo.

Initially we got the idea that the program must be crashing, and the watchdog timer was resetting it. We came to this conclusion because while BMU 5 might refuse to obey a command, BMUs 6 to 8 after it would obey it no problem. We reasoned that it must be in a failsafe mode we designed for when the most recent downloaded program is not to be trusted. It just passes commands through and waits for a Bootstrap Loader password, which allows us to get a new program into it without having to physically access it when it is installed in a metal box crammed into a small car.

However, in-circuit debugging indicated that this was not happening. We noticed that the clock frequency was off a little on the boards that were playing up the most, so we adjusted the clock frequency (there is a calibration value supplied by the manufacturer that we can override), and that seemed to help a little, but by no means fixed the problem. I should point out that at this stage, only one board (string of 8 BMUs) seemed to have this problem.

Maybe it was the communications being corrupted; I did catch one of the misbehaving BMUs receiving a "v" (cell voltage) command as something like $C4. But on the DSO screen, after some futzing with the trigger levels, the comms lines looked quite good.

We wondered if the misbehaving BMUs were being reset part way through the command. We have a reset-on-break circuit; perhaps this was performing a reset when there was no break being sent. Perhaps Weber had loaded bypass MOSFETs instead of diodes into the reset-on-break circuit; that would explain why only this board was affected. Alas, no, they all checked out.

Perhaps one bad BMU was affecting all the others following it. One of the worst BMUs happened to be the second in the string and we need some strings of six anyway. So we cut the first two from the last six, but now we had a bad string of 2 and a bad string of 6. I re-soldered some of the components involved with comms handling; no improvement.

This was starting to burn up too much time; I'd already spent most of a day on this. So we decided to put that one string aside; if necessary, we could do without it. To our horror, the "dickiness" as we had started calling it seemed to appear on other boards when we started re-testing them. We marked the worst ones and kept pushing on.

Last Saturday Mark Aylott did a brilliant job, putting his air-force electronics training to work on the MX-5 project for a third day. He spent over 12 hours building BMU boards and repairing faults on some Weber had made earlier, without any need for supervision, just coffee, cake, meals and jokes. Thanks, Mark!

Image

So Mark freed Weber up to try to exorcise the demons of dickiness. Weber made a breakthrough that I failed to make: the "possession" went away after a series of characters, e.g. 5 carriage returns, went through, and the whole board would work happily until a certain amount of idle time, after which the possession would show up until enough characters had gone through again.

For those interested in the details, here is a schematic of the relevant communications path from one BMU to the next:

Image

It seemed to involve the 100 uF capacitor (C6) and Schottky diode (D6), both of which aren't used unless this BMU needs to drive an optic fibre. They form a voltage doubler since the fibre LED needs more voltage than the opto coupler's.

Our view had been that the capacitor and diode don't affect anything when the fibre LED isn't present, and they don't cost much, so why not populate them all, so it will be present wherever we need it. Weber noted that on good BMUs, the capacitor charged quickly, and stayed there, causing no problems, but on bad BMUs, it would charge very slowly, and would discharge slightly with each character. And it would reduce the amplitude of the communications signal until it had been discharged sufficiently. Aha! We both concluded that this sounds like the diode was in backwards. However, Weber checked this carefully, and reported that no, they were all installed correctly.

I puzzled and puzzled over this, and finally came up with an elaborate reason why this must be happening. It must be because the TX- output from the micro is open circuit. The voltage doubler circuit wasn't so much causing the problem, as imperfectly masking the real problem! I emailed my triumphant logic to Weber, but as fate would have it, he was away for a few hours (on non-EV business... shock!) and I had to wait for the result.

When it came, the result was an anti-climax. The board with the original dickiness did indeed have every one of the Schottky diodes (D6) inserted backwards, despite Weber having checked it carefully the day before. Rotating the diodes completely cured that board. Other boards that we thought were possessed turned out not to be bad. What was likely happening is that when re-checking a string of boards that had already been programmed, I did not leave enough time for the supercaps to charge, so these boards were operating initially at a low Vcc, mimicking the effects of the possession. When enough time was allowed for the supercaps to charge, they worked perfectly.

So the demons of dickiness have been exorcised. Hoorah!

Various edits for ease of reading and fewer acronyms, mostly suggested by Weber. Uploaded improved schematic.
Last edited by coulomb on Mon, 27 Aug 2012, 13:23, edited 1 time in total.
Nissan Leaf 2012 with new battery May 2019.
5650 W solar, 2xPIP-4048MS inverters, 16 kWh battery.
1.4 kW solar with 1.2 kW Latronics inverter and FIT.
160 W solar, 2.5 kWh 24 V battery for lights.
Patching PIP-4048/5048 inverter-chargers.

User avatar
woody
Senior Member
Posts: 1713
Joined: Sat, 21 Jun 2008, 02:03
Real Name: Anthony Wood
Location: Mt Colah

Weber and Coulomb's MX-5

Post by woody » Mon, 27 Aug 2012, 17:28

coulomb wrote: The following should tile properly if your screen is wide enough and your browser is maximised:
Awesome hack.
Planned EV: '63 Cortina using AC and LiFePO4 Battery Pack

User avatar
coulomb
Site Admin
Posts: 3685
Joined: Thu, 22 Jan 2009, 20:32
Real Name: Mike Van Emmerik
Location: Brisbane
Contact:

Weber and Coulomb's MX-5

Post by coulomb » Mon, 27 Aug 2012, 18:28

Now that the dickiness demon has been exorcised, testing can continue on our pile of boards to be tested:

Image

The black cylinders in the middle of the BMUs are the piezo transducers (alarms). It looks a little strange because many of them are covered by the "ears" of the adjacent board (protrusions for battery terminals).

Under the boards is the engraver that we use to identify the boards. You can appreciate that with eventually 240 BMUs in various states (tested OK, no bypass, needs yellow LED resistors replaced, still need blue side components installed, and so on), we need to keep careful track of board and even individual BMU status. We use a low tech notepad (paper variety) for this.

Ultimately they join our collection of boards with white dots (at left), indicating that they are tested OK:

Image

These are in a box we dubbed "the coffin", because of our habit of using the red error LEDs and piezos to discharge the one farad super-capacitors as soon as possible. It still took several minutes, and the sound is quite distracting (by design, of course Image ), so we wanted to "put the screaming, bloody mouthed vampires back in their coffin" to minimise the noise. The noise is especially shrill because their frequencies are not exactly the same, so there is a beat frequency adding a warble to the sound.

Alas, the noise was annoying enough even from the coffin that we added a "Q" (Quiet) command to the debugger, so we can put them away with red LEDs but no piezo alarms sounding.

Edit: we've completed 21 of the 30 boards, so they are 70% assembled, but only about 23% tested. Testing should catch up with assembly soon. The failure rate, after a shaky start, has been quite low.
Last edited by coulomb on Mon, 27 Aug 2012, 08:33, edited 1 time in total.
Nissan Leaf 2012 with new battery May 2019.
5650 W solar, 2xPIP-4048MS inverters, 16 kWh battery.
1.4 kW solar with 1.2 kW Latronics inverter and FIT.
160 W solar, 2.5 kWh 24 V battery for lights.
Patching PIP-4048/5048 inverter-chargers.

PlanB
Senior Member
Posts: 465
Joined: Sat, 16 Jan 2010, 15:24
Real Name: Kris McLean
Location: Freemans reach
Contact:

Weber and Coulomb's MX-5

Post by PlanB » Mon, 27 Aug 2012, 21:46

One of the great ironies of my failing eyesight has been the steady shrinkage in electronic component sizes. The infinitesimal band of white paint that passes for a polarity indicator these days on things like SM diodes is just cruel.
In this regard I don't understand why Mark is not looking THROUGH the maggy lamp? Oh wait is it that he is a person on the RIGHT side of 60?

User avatar
weber
Site Admin
Posts: 2598
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Weber and Coulomb's MX-5

Post by weber » Mon, 27 Aug 2012, 22:05

PlanB wrote: One of the great ironies of my failing eyesight has been the steady shrinkage in electronic component sizes. The infinitesimal band of white paint that passes for a polarity indicator these days on things like SM diodes is just cruel.
In this regard I don't understand why Mark is not looking THROUGH the maggy lamp? Oh wait is it that he is a person on the RIGHT side of 60?

Hee hee. Let me set the record straight immediately. The dicky board was the 3rd one made, well before Mark even volunteered. It was most definitely me wot dun it. Coulomb was just being kind, not spelling that out. And I was most definitely looking through the maggy lamp at the time. And although you're right that the polarity marking is atrocious (not even paint, just a slight crease very close to one end), I apparently could see them because I managed to get every one (all 8) backwards on that board. Musta just been brane fade guvna. Image

Yes. Mark is in fact quite a way on the right side of 50 and can work faster without the maggy lamp, with most components. The wider field of view means you spend less time moving the board back and forth, and your regular hand-eye coordination works just fine, rather than being like the first time you tried to shave in a mirror. But Mark does use the maggy lamp to check everything after he has placed it, and has never placed a part wrong, as far as I know. All his boards have tested OK.
Last edited by weber on Mon, 27 Aug 2012, 12:06, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

User avatar
coulomb
Site Admin
Posts: 3685
Joined: Thu, 22 Jan 2009, 20:32
Real Name: Mike Van Emmerik
Location: Brisbane
Contact:

Weber and Coulomb's MX-5

Post by coulomb » Mon, 27 Aug 2012, 23:58

I forgot to include this short video in the long post about testing Battery Management Units (BMUs).

It shows me trying to test the monitor program, which is what ultimately runs in the BMUs, with a supply voltage from a power supply of just over 3.6 V. Power is connected to the eight BMUs under test through clip leads with thin wires; the bypass current of some 400 mA is enough to cause a voltage drop across these leads, so the micros see lower than 3.6 V and the bypass current turns off. Under some circumstances, there is a sort of shift register effect, where the bypass LEDs (on the left in the video) seem to move towards the peg.



Unfortunately, the pretty colours of the LEDs have been washed out in the video, but the chaos may be amusing to some.
Last edited by coulomb on Mon, 27 Aug 2012, 13:59, edited 1 time in total.
Nissan Leaf 2012 with new battery May 2019.
5650 W solar, 2xPIP-4048MS inverters, 16 kWh battery.
1.4 kW solar with 1.2 kW Latronics inverter and FIT.
160 W solar, 2.5 kWh 24 V battery for lights.
Patching PIP-4048/5048 inverter-chargers.

User avatar
weber
Site Admin
Posts: 2598
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Weber and Coulomb's MX-5

Post by weber » Tue, 28 Aug 2012, 00:05

Guess who these dudes are:

Image
Last edited by weber on Mon, 27 Aug 2012, 14:50, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

User avatar
woody
Senior Member
Posts: 1713
Joined: Sat, 21 Jun 2008, 02:03
Real Name: Anthony Wood
Location: Mt Colah

Weber and Coulomb's MX-5

Post by woody » Tue, 28 Aug 2012, 03:48

Before you frame them, I like these portraits better...
Image
Image
Planned EV: '63 Cortina using AC and LiFePO4 Battery Pack

User avatar
coulomb
Site Admin
Posts: 3685
Joined: Thu, 22 Jan 2009, 20:32
Real Name: Mike Van Emmerik
Location: Brisbane
Contact:

Weber and Coulomb's MX-5

Post by coulomb » Sun, 02 Sep 2012, 02:26

I think the guy in the uniform is by far the more handsome Image

Mark came over today again to load components on the last of the BMU boards! We also now have 15 in the "tested and ready to use" box.

The fail rate has been a little disappointing. There have been the expected niggling details like LEDs not soldering properly; they're easy to find, and easy to fix. [ Edit: But there have been too many cases of weird faults that take a long time to track down. ]

There was a bit of a fright today when I adjusted the voltage of the power supply up a little (using a screwdriver adjustment we added to the side of the power supply for fine voltage adjustment), and suddenly the current went to 2.5 A and stayed there. I can imagine that changing the voltage across effectively eight farads of capacitors would draw more current for a while, but I expected the current to come down quickly, but it didn't. Things started smelling hot, so I turned everything off in a hurry. It seems that four of the 8 voltage regulators fried in that little episode; I still really have no idea why. Fortunately we had enough spares to replace them, and all was well with that string again. I'll refrain from adjusting the voltage while the BMUs are connected in future.

There is also a frustrating software or hardware problem (I can't see how it could he hardware, but I also have trouble seeing how it could be software) where the frequency calibration value is wiped to $FFFF. When that happens, the processor runs way too quickly, and serial comms is not possible until the frequency calibration value is at least set to a default value. This is a cumbersome process that soaks up some time.

I've been way over optimistic with my figure of 3 minutes testing per board of 8 BMUs. It's more like 10 minutes, and of course some boards have to be tested twice and even more times (after fixing faults). I'd say there is another thee or so days of solid testing and repairing, and finally the cell-top battery management units will be done and ready for installation in battery boxes.

Today we came up with a new name for our debugger (software we use temporarily in our BMUs; ultimately they run a thing called the monitor, which we may rename to the manager). It was getting confusing because the IAR software suite (assembler, editor, compiler, and debugger) also has a debugger. Ours does testing, sets the IDs, and performs calibration, so we're calling it TestICal (pronounced test-eye-cal, how else?).

[ Edit: Hopefully reduced confusion over what our debugger is, and which one I am talking about. ]
Last edited by coulomb on Sun, 02 Sep 2012, 05:27, edited 1 time in total.
Nissan Leaf 2012 with new battery May 2019.
5650 W solar, 2xPIP-4048MS inverters, 16 kWh battery.
1.4 kW solar with 1.2 kW Latronics inverter and FIT.
160 W solar, 2.5 kWh 24 V battery for lights.
Patching PIP-4048/5048 inverter-chargers.

Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Weber and Coulomb's MX-5

Post by Nevilleh » Sun, 02 Sep 2012, 14:44

You guys are certainly making a marathon out of this!
Hope you don't have a complete TestICalsUp.
(Shoulda bought an oven Image )

User avatar
weber
Site Admin
Posts: 2598
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Weber and Coulomb's MX-5

Post by weber » Sun, 02 Sep 2012, 22:02

Nevilleh wrote: You guys are certainly making a marathon out of this!
Hope you don't have a complete TestICalsUp.
(Shoulda bought an oven Image )
Hee hee. The frypan isn't the problem. It's the stencils. At first I ordered one where the apertures were reduced from the pads by 75 um (3 mil) (the DesignSpark default) and that was too little paste, particularly for those things with small pads, like the microcontroller and the dual LEDs. Then I ordered one expanded from the pads by 50 um (2 mil) (the Protel default) and that has been too much paste for some things, causing a few resistors and caps to tombstone and occasional shorts between micro pins. It seems we need one that is the same size as the pads. Oh well, we'll know next time.

We had a record turnout for EV day at the C&W lab on Saturday. Newton (Jeff Owen) took this shot (and so managed to avoid being in it). We joked that he was the only one doing any visible work (that dirty noisy mechanical stuff). The evidence of this is the earmuffs around Coulomb's neck. From left to right (or top to bottom): Weber, Joule (Mark Aylott), Coulomb.

Yes Woody, as you can see, the scruffy bearded photo of that dude is far more apt.

If you maximise your browser window, and your screen is big enough, or you zoom using a two-fingered salute on your trackpad or something, then these four images will tile into one big panoramic one.

ImageImage
ImageImage

[Edit: Reduced images so they will tile correctly on more screens, and added to the viewing instructions.]
Last edited by weber on Sun, 02 Sep 2012, 17:23, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Weber and Coulomb's MX-5

Post by Nevilleh » Mon, 03 Sep 2012, 01:44

I liked the last one of the bearded one studying a sheet of ancient hieroglyphics, but I think he had it upside down!
I made 50 of my bms modules, no stencils, all done with a syringe, but mine I think had less parts. Anyway, they all worked and testing was quite straightforward as a result. I wouldn't have liked to make - what, 135 or something? I admire your persistence!
Its not easy, is it?
One of these days I'll post a readout of all my cells under both charge and discharge from the serial link I've added to my bms master.

User avatar
weber
Site Admin
Posts: 2598
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Weber and Coulomb's MX-5

Post by weber » Mon, 03 Sep 2012, 03:40

Nevilleh wrote: I liked the last one of the bearded one studying a sheet of ancient hieroglyphics, but I think he had it upside down!
Hardy har har. Image

I've just reduced the size of those images, and added to their viewing instructions, so if you couldn't get them to pop into the full tableau before, reload the page, scroll back and give it another go. It's really one big photo carved into four parts, to work around the 150 KiB limitation of the forum, but with each part making sense on its own. I really love this photo because it was not posed at all. We had no idea that the reason Newton, off in the far corner, had stopped making grinding noises and showering sparks for the moment, was that he was setting up this shot.
I made 50 of my bms modules, no stencils, all done with a syringe, but mine I think had less parts. Anyway, they all worked and testing was quite straightforward as a result.
Aw c'mon Neville, There's no doubt you did well, but I seem to remember some early photos of tombstoned resistors or something. Image
I wouldn't have liked to make - what, 135 or something? I admire your persistence!
240, and thanks.

[Edit: Added "reload the page".]
Last edited by weber on Sun, 02 Sep 2012, 18:04, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Weber and Coulomb's MX-5

Post by Nevilleh » Mon, 03 Sep 2012, 14:15

Yes, you are quite right about the "tombstones" and I found that was caused by too much solder paste. I borrowed a little gadget from a mate that fitted to the syringe and gave the same sized squeeze every time you pushed the button and once I got it set correctly, it made applying the paste much easier and much more consistent. But 240 modules? I find that quite daunting and had I needed that many I would've investigated the stencil route, for sure.
Have you had any further thoughts on why your regulators blew up?

Post Reply