Weber and Coulomb's MX-5

[weber]

Have you had any further thoughts on why your regulators blew up?

No. Except that presumably the lab supply that was powering them, briefly put out significantly more than the 6 volts they are rated for. Its voltage was being adjusted at the time, but downwards. Attempts to reproduce this, without BMUs but with the scope watching in single-shot mode, failed.

Our BMUs are designed to withstand 900 V in reverse (full pack voltage in case a cell goes high resistance under load) but in the forward direction all we can do is try to ensure they fail without making a spark. They have a 1 A 100 Vdc fuse (max difference between charger open circuit voltage and battery voltage in case a cell goes high resistance on charge) but these did not blow in this case as the power supply was current-limited to about 1 A. The four regulators just quietly turned into resistors of 0.7 R, 1.4 R, 1.6 R and 9.1 R.

But I am very pleased to announce that all 248 BMUs are now TESTed Id-ed and CALibrated.
Here's what that looks like:



To everyone who has helped us, in any way shape or form, with this long-running BMS development project, I just want to say a heartfelt:

[Nevilleh]

Good job! Now to fit them to the cells.......

[weber]

Long time no post. But rest assured there has been progress on the MX-ϟ. These photos are from a many weeks back. Conformally coating the battery monitoring units. The coloured dots are just keeping the lacquer out of the piezo speakers.

[weber]

Here are some photos from today's effort with Coulomb and Newton. They show the MX-ϟ's eight battery boxes, containing all 228 40 Ah cells. They are all clamped up and have Battery Monitoring Units (BMUs) on them.

The BMUs and high current links are held onto the cells with M6 stainless bolts with a belleville washer and a flat washer. The belleville replaces the usual spring washer. The force required to bottom out the conical belleville washer is 3 to 4 times that of the split spring washer. The lowest online price I could find for the bellevilles was from Blackwoods. Even when ordering 500 they were nearly 30 cents each, so I hope they're worth it.

The box of 60 cells on the trolley on the floor have been fully charged and top balanced. The set on the round table (54 + 28) are being charged at present.





The charge is being controlled by a modified Tritium Driver Controls Unit (DCU). We have added four Industrial Fibre Optic (IFO) ports to it (2 in, 2 out) and modified the DCU software. One pair of IFOs connect to the BMS, the other to the Elcon/TC Charger.

The BMS sends regular status bytes to the DCU. These contain a "stress level" between 0 and 15 and an "all-in-bypass" bit. The stress level is that of the most stressed cell in the pack. Stress can be due to voltage or temperature or even the voltage drop across the links between cells. But in normal charging it will depend only on voltage. One stress level for every 10 mV above 3.500 V.

Only stress levels of 8 or more are considered "dis-stress". And only levels of 11 or more are considered "alarming distress" which means the BMU turns on it red light and beeps. Stress 7 or less can be called "eustress".

A lot of time in the past few weeks has been spent developing and debugging the new charging software (which doesn't photograph well).

To control the charger, we've implemented a version of a suggestion by Tritium_James, to run a PID loop on the highest cell's voltage, with a set-point just above the voltage at which bypass turns on, and terminating when the lowest cell's voltage reaches the bypass level. However we don't run the loop on max voltage per se, but the 4 bit stress level, and we don't terminate on min voltage per se, but the single all-in-bypass bit.

Controlling stress, rather than voltage, will automatically cause the charger to back off the current if any cell or BMU gets hot, or if there's a high resistance joint between cells.

Bypass turns on at stress 6 (> 3.56 V). The set-point is at stress 7 and so has a dead-band from 3.57 to 3.58 V.

The same PID controller (with different parameters) will be used to automatically back off the drive or regen current from the motor controller (Tritium WaveSculptor 200) when any cell is distressed.

[PlanB]

The series connected battery pack always reminds me of that old proverb about a chain only being as strong as it's weakest link. Is that your main motivation for modifying the standard Tritium kit, to better identify suspect cells in the pack?
I've always thought there is a great opportunity for data logging with EV packs so a failing cell can be flagged for replacement before it actually stands you up on some busy highway?

[weber]

The series connected battery pack always reminds me of that old proverb about a chain only being as strong as it's weakest link.

Very true.

Is that your main motivation for modifying the standard Tritium kit, to better identify suspect cells in the pack?

No. The only reason we have to modify the standard Tritium DCU is to add optic fibre interfaces. Tritium interface with their BMS via CAN bus, while we use optic fibre.

I've always thought there is a great opportunity for data logging with EV packs so a failing cell can be flagged for replacement before it actually stands you up on some busy highway?

Yes. Definitely. Tritium's BMS and ours are equally amenable to such data logging. But we haven't got that far with our software yet.

[weber]

There was a teenage Halloween party at my place last night. Little did they know what was lurking in the darkened lab next door.

[PlanB]

Is it sentient? The ripple pattern in the flashing reminds me of some patters I've seen on this

[weber]

We decided we needed a better interface between our netbook and optic fibre for debugging purposes. The current one (on the left in the photo below) required batteries and was unwieldy due to being connected with skinny wires that keep breaking off. You can see the new model on the right. Designed by coulomb and weber and built by Mark Aylott. Schematic below.

To eliminate the need for batteries, we modified the Novus USB to RS422 interface box internally to repurpose one of its termination-resistor terminals. We put its isolated 5V supply out on that instead, with a 22R resistor in series for short circuit protection, and connected its GND terminal directly to its internal isolated 0V instead of via a 120R resistor as it was.

[ Edit Coulomb: the +5 V turns out to me more like +6.2 V. ]

[weber]

Here's the stripboard layout for the optic-fibre sniffer/injector, whose schematic and photo appear in the previous message in this thread. The "Sniffer/Injector" sounds like it has a serious drug problem, but in fact it is designed to be inserted in the optic fibre input and output of a system under test, such as the string of battery monitoring units (BMUs) in our BMS, and sniff its output (display it on a computer screen) at the same time as letting us inject commands into its input (by typing on the computer keyboard or sending files).

We can either wait until the master unit (the modified Tritium DCU) is not sending anything to the BMUs, or we can deliberately type at the same time, to test its error tolerance. The previous version could either sniff, or have sole control of the input and output, it could not do both at the same time, and it could not inject.



[ Edit Oct 2013: Changes to above image in red -- modified for optional wired sniff RX; remove jumper and use upper two pins for wired RX ]

[weber]

Here's my attempt at a minimal modification to a 12V 30A 20kHz PWM speed controller used for quieting the power steering pump. The purpose of the modification is to add current limiting at 20 to 30A for when the steering hits the stops.

The idea is to use the MOSFET itself as the current sensing element. The pair of 1N914 diodes, one to the drain and one to the gate, and the associated pullup resistor and filter capacitor, are intended to produce a signal proportional to the average current through the MOSFET, plus an offset due to the diode drops.

I note that the LM393 dual comparators have open collector outputs.

I'd be pleased if anyone could comment on whether they see any problems with the proposed implementation, or have any suggestions to improve it, or indeed whether they think it has any chance of working at all.



[Edit: For an update, see viewtopic.php?t=980&p=40580#p40580]

[Johny]

You are relying heavily on the TC4427 getting nearly as close to VSS as when the MOSFET is on. I think this won't be the case and the MOSFET low voltage when on will be swamped by the condition when the TC4427 is on. Of course, I may be totally wrong. Given the level of modification I think I'd just make a veroboard PWM using a TL494 (my favourite PWM IC) which has easily implementable current limit.
It depends on whether you want to learn something or just get it working quickly. I have a feeling I know the answer.
Interesting PWM generator - I hadn't seen that before - does it work?
There will be thermal drift due to no compensation of the diode voltage drops - it may not matter in the whole scheme of things but worth measuring.

[weber]

You are relying heavily on the TC4427 getting nearly as close to VSS as when the MOSFET is on. I think this won't be the case and the MOSFET low voltage when on will be swamped by the condition when the TC4427 is on. Of course, I may be totally wrong.

Good point. Just looked at the datasheet and it guarantees the TC4427 output will be max 25 mV (presumably at zero current) and will have a resistance of less than 12 ohms at 25°C. And with a 3k9 pullup that's 3 mA it has to draw (after it has discharged the MOSFET gate). So that's 3*12 + 25 = 61 mV. MOSFET at 5 mR pulling 20 A will only be 100 mV. So it's looking pretty dodgy. It certainly isn't going to limit the current into a short-circuit (when the PWM duty cycle would have to go close to zero). I guess I should increase the pullup resistor by a factor of 10 to at least get the TC4427 output voltage down to 30 mV.

Or ... What if leave the pullup as it is and put say 22R in series with the diode that's sensing the drain voltage? That should add about the same offset to it. And then there's the -2 mV/°C temperature coefficient of the diodes. All pretty crude.

Given the level of modification I think I'd just make a veroboard PWM using a TL494 (my favourite PWM IC) which has easily implementable current limit.

Maybe so. But it would need a shunt to measure the current.

It depends on whether you want to learn something or just get it working quickly. I have a feeling I know the answer.

Yeah. I just want to get it working quickly. But I can't know in advance whether that means modifying what I've got or starting again from scratch.

I'd rather just buy one, but as you may have seen in the For Sale/Wanted section, I can't find one that runs at an inaudible 20 kHz.

Interesting PWM generator - I hadn't seen that before - does it work?

Yes. Learned that one off Ross Pink. And of course it's even simpler (omit the 3 resistors) if you use a logic schmidt inverter instead of a comparator. Don't use an op-amp for this as some will generate RF bursts as they pass through zero and fry your MOSFET or its driver.

[Edit: Added op-amp warning]

[Johny]

Remove the IN914 diodes and use a lower power JFET. Gate connect to output of 4427, drain to drain of MOSFET and source to -ve input of comparator.
Viola - sample and hold. Maybe a very high value resistor from -ve comparator input to VSS - to be sure!

Edit: Actually I'd just use a wire shunt on the source of the MOSFET and integrate that into the -ve of the comparator.

[klaasdc]

I had a private request for dimensions of the adapter plate for the gearbox bellhousing of the Mazda MX-5, also known as the Mazda Miata (USA) and the Eunos Roadster (Japan). I thought I might as well post it here too, in case someone else can benefit. Apparently it may suit some other Mazdas as well, such as the Mazda 121.

See
viewtopic.php?t=980&p=26473#p26473
for photos and other info.

The obvious origin or reference point is midway between the two locating dowels (the two holes that are almost on a horizontal line through the shaft centre). The dowel centres are 161.0 mm either side of the origin and the centre of the shaft is 2.0 mm below the origin. The dowels are 19.0 mm OD. These are the only critical measurements.

The six lower holes are all on a circle of radius 161.0 mm.

The two lowest are at angles of 22.6 degrees either side of vertical (61.9 mm either side of the origin and 148.6 mm below it) apparently forming 5:12:13 Pythagorean triangles.

The other two without dowels are not symmetrical and are 56.1 and 47.9 degrees off the vertical (one is 133.7 mm to the side and 89.7 mm below the origin and the other is 119.5 mm to the side and 107.9 mm below the origin).

The two top bolts are on a radius of 204.1 mm and at angles of 23.0 degrees either side of the vertical (79.8 mm to the side and 187.9 mm above the origin).

If anyone ever get this into some kind of CAD file, I'd appreciate a copy. You'd also need to describe the outline as a series of lines and arcs as well. We haven't done that. Pascal just traced around the bellhousing for that.

[Edit: Made URL into a live link]

Edit2: Note that the two bellhousing bolt circles I give radii for above, are centered on the aforementioned origin, _not_ centered on the shaft. However the bolt circle for the electric motor flange _is_ centered on the shaft, 2.0 mm below the bellhousing origin.

Hello weber,
First of all I wanted to congratulate you on the work you guys have been doing on the car. It's very interesting to read!
I stumbled upon this forum in search for a good bellhousing template. We're doing an electric conversion of a '94 MX-5 ourselves and had a hard time to get an accurate measurement of the transmission bolt holes. I even tried scanning the gasket that sits in between, but it seems not accurate enough (due to the gasket not being flat against the flat bed scanner).
So a big thank you for making your data available!

I made a drawing in Autodesk Inventor so I could print a paper version first.



All the files and pictures can be found at:

mx5-bp-boltholes.pdf
mx5-bp-boltholes.ipt
inventor.PNG
mx5-bp-gasket-scan.png

[weber]

Hi KlaasDC! Thanks for the kind words, and thanks so much for the files and drawings. It is wonderful to be able to cooperate in this way. Can you provide a link to more information about your conversion?

Also, if it's not too much trouble, could you please change the note in your file to, "Original measurements by "pascal" and "weber" at:". Thanks.

[klaasdc]

Hi KlaasDC! Thanks for the kind words, and thanks so much for the files and drawings. It is wonderful to be able to cooperate in this way. Can you provide a link to more information about your conversion?

Also, if it's not too much trouble, could you please change the note in your file to, "Original measurements by "pascal" and "weber" at:". Thanks.

Hi,
I changed the files as you requested

We're not far enough in our project yet to have a dedicated web page or album Our initial goal is to have the wheels rotate as soon as possible (so no battery racks etc), hopefully somewhere at the end of this year.
The motor will consist of 2 Motenergy ME0913 BLDC motors coupled together on their axles, each with its own controller (96V 400A from KellyControls). The whole clutch and transmisson will be kept at first to keep things simple. For the batteries, we will be using 2 strings of 32 ThunderSky 100Ah batteries, with an Electromotus BMS.
Eventually, we'd also like to remove the power steering system and convert it to a manual rack. In the ideal situation, we find someone that wants to trade it for a "real" manual rack.

I noticed that you basically made a complete new clutch plate/flywheel out of aluminium. We were thinking of turning down the original one to lighten it but maybe your idea is a better option. Did you have to balance the disc?

[weber]

I noticed that you basically made a complete new clutch plate/flywheel out of aluminium. We were thinking of turning down the original one to lighten it but maybe your idea is a better option. Did you have to balance the disc?

We made a new flywheel, but the clutch plate (= friction disk) and pressure plate are standard. The reason we did this was because we wanted to use a taper-lock hub so it was removable, rather than shrink fit a hub to the motor shaft. And the inner bolt circle that used to mount the flywheel to the end of the ICE crankshaft was too small and interfered with the taper-lock bush needed for our 38 mm dia keyed shaft and expected torque (360 Nm).

So we designed a flywheel with an integral taperlock hub and had it made. It was not made from aluminium alloy, but from SAE-grade 1045 carbon steel. This was required for the strength of the taper-lock hub. See the discussion here: viewtopic.php?title=weber-and-coulombs- ... 142#p30142

We have not yet had it balanced and are limiting the rpm, but we expect to have it (and the motor rotor) balanced before allowing it to go to max design rpm (6000).

With your 22.23 mm dia shaft you may be able to use, and turn down, the original flywheel, or use an aluminium alloy one.

[klaasdc]

We made a new flywheel, but the clutch plate (= friction disk) and pressure plate are standard. The reason we did this was because we wanted to use a taper-lock hub so it was removable, rather than shrink fit a hub to the motor shaft. And the inner bolt circle that used to mount the flywheel to the end of the ICE crankshaft was too small and interfered with the taper-lock bush needed for our 38 mm dia keyed shaft and expected torque (360 Nm).

So we designed a flywheel with an integral taperlock hub and had it made. It was not made from aluminium alloy, but from SAE-grade 1045 carbon steel. This was required for the strength of the taper-lock hub. See the discussion here: viewtopic.php?title=weber-and-coulombs- ... 142#p30142

We have not yet had it balanced and are limiting the rpm, but we expect to have it (and the motor rotor) balanced before allowing it to go to max design rpm (6000).

With your 22.23 mm dia shaft you may be able to use, and turn down, the original flywheel, or use an aluminium alloy one.

Aha, that's clear! Very very neat, but outside what we can fabricate here But the torque for your motor is much higher than ours (90-100 Nm), so we'll probably be safe using a simple adapter:

You can find some pictures of our car in this album: https://picasaweb.google.com/1033574858 ... directlink


Did you consider using a damper or anything in between the gearbox and motor? An engineer friend recommended us to do this, so that alignment of the two axles is not too critical. I'm not decided about this.

[weber]

Aha, that's clear! Very very neat, but outside what we can fabricate here But the torque for your motor is much higher than ours (90-100 Nm), so we'll probably be safe using a simple adapter:

Yes, I agree that should be fine.

You can find some pictures of our car in this album: https://picasaweb.google.com/1033574858 ... directlink

Oh! I so wanted a blue one. Beautiful.

Some cosmetic tips, just in case you don't already know:
1. It's very common to see the back of the pop-up headlights being lower than the bonnet, but this is only because we lean on them when we work under the bonnet and they bend down. You will feel the bumps on the side where they are creased, 50 to 60 mm from the rear. It is possible to gently and carefully bend them back to almost level with the bonnet, but you might as well wait until you're finished converting before you bother.
2. The rust in the rear sills is usually because the plastic tubes that drain the soft-top have blocked up and overflowed, with leaf-litter then possibly blocking the drain slot in the seam at the bottom of the sill itself. Ensure that these root causes are fixed before you bother repairing the symptoms.

Did you consider using a damper or anything in between the gearbox and motor? An engineer friend recommended us to do this, so that alignment of the two axles is not too critical. I'm not decided about this.

Sounds like he's talking about a flexible coupling. This is not necessary if you are keeping the clutch and gearbox. The spline on the gearbox input shaft allows for this. It has a small amount of free play inside the gearbox and a small amount inside the clutch-plate. But you probably still need to get the motor shaft within 0.2 mm (preferably 0.1 mm) of the ideal position (i.e. midway between the two dowels and 2.0 mm below them).

But you will certainly need a flexible coupling between your two motors. See the excellent example here, and read my comments at the end.
http://zeva.com.au/Projects/RX7/?section=drivetrain

[klaasdc]

Oh! I so wanted a blue one. Beautiful.

Thanks! I'm told it was a limited edition for the '94 model year. It'll be a very limited edition after the conversion

Some cosmetic tips, just in case you don't already know:
1. It's very common to see the back of the pop-up headlights being lower than the bonnet, but this is only because we lean on them when we work under the bonnet and they bend down. You will feel the bumps on the side where they are creased, 50 to 60 mm from the rear. It is possible to gently and carefully bend them back to almost level with the bonnet, but you might as well wait until you're finished converting before you bother.
2. The rust in the rear sills is usually because the plastic tubes that drain the soft-top have blocked up and overflowed, with leaf-litter then possibly blocking the drain slot in the seam at the bottom of the sill itself. Ensure that these root causes are fixed before you bother repairing the symptoms.

Thanks for the tips. After the "mechanical" parts are done, the car will be stripped (again) and repainted (in the original color). I believe those sills will have to be cut out, and the previous owners' bills indicate that some welding has already been done in the rocker panels.
Also, because the car was lowered, the frame rails underneath the car are flattened in the middle (speed bumps or incorrectly lifted?...). It is visible on one of the pictures of the underside. Maybe we can knock them back out from inside the car. Considering the thickness of the metal there and the fact that they are closer to the middle, they are likely not super-critical for the structure.

Did you consider using a damper or anything in between the gearbox and motor? An engineer friend recommended us to do this, so that alignment of the two axles is not too critical. I'm not decided about this.

Sounds like he's talking about a flexible coupling. This is not necessary if you are keeping the clutch and gearbox. The spline on the gearbox input shaft allows for this. It has a small amount of free play inside the gearbox and a small amount inside the clutch-plate. But you probably still need to get the motor shaft within 0.2 mm (preferably 0.1 mm) of the ideal position (i.e. midway between the two dowels and 2.0 mm below them).

But you will certainly need a flexible coupling between your two motors. See the excellent example here, and read my comments at the end.
http://zeva.com.au/Projects/RX7/?section=drivetrain

Very interesting. I never really gave it much thought before, but indeed you can only constrain an axle on two points... So I'll be looking at such a spider coupling then as well

[klaasdc]

...viewtopic.php?title=weber-and-coulombs- ... 142#p30142

We have not yet had it balanced and are limiting the rpm, but we expect to have it (and the motor rotor) balanced before allowing it to go to max design rpm (6000).

With your 22.23 mm dia shaft you may be able to use, and turn down, the original flywheel, or use an aluminium alloy one.

At viewtopic.php?title=weber-and-coulombs- ... 121#p30121 you mentioned that the critical distance between clutch and engine is 38 mm. Do you have an idea if this also applies on a 1.8?

I just noticed that the flywheel and pressure plate are quite different between 1.8 and 1.6. The 1.8 flywheel has "stubs" to mount the thinner pressure plate, while the 1.6 one is completely flat.
Unfortunately, our engine is already sold so I can't measure it anymore .

[weber]

At viewtopic.php?title=weber-and-coulombs- ... 121#p30121 you mentioned that the critical distance between clutch and engine is 38 mm. Do you have an idea if this also applies on a 1.8?

Logically, it must be the same, because it is as much a part of the design of the gearbox, as the engine, and these engines are interchangeable without having to alter the gearbox.

I just noticed that the flywheel and pressure plate are quite different between 1.8 and 1.6. The 1.8 flywheel has "stubs" to mount the thinner pressure plate, while the 1.6 one is completely flat.
Unfortunately, our engine is already sold so I can't measure it anymore .

Do you mean the 3 dowel pins that accurately locate the pressure plate assembly on the flywheel? If so, what makes you think the 1.6 flywheel doesn't have them? Ours is a 1.6 and it has the 3 dowel pins. You can see them in a photo in the post you link above.

[klaasdc]

Logically, it must be the same, because it is as much a part of the design of the gearbox, as the engine, and these engines are interchangeable without having to alter the gearbox.

Well at the bottom of this page http://www.topdownmx5.com/clutch.htm there is a side by side picture of the two flywheels.
From what I understand, the flywheel and clutch etc is only interchangeable between the 1.6 and 1.8 as a unit.

Do you mean the 3 dowel pins that accurately locate the pressure plate assembly on the flywheel? If so, what makes you think the 1.6 flywheel doesn't have them? Ours is a 1.6 and it has the 3 dowel pins. You can see them in a photo in the post you link above.

No, the dowel pins are the same, but due to the "stubs" they are positioned deeper into the bellhousing. But the pressure plate is thinner, so it could be possible that the "critical distance" ends up being the same. That's what I hope anyway

[weber]

I see what you mean. But the distance we are talking about is from the face of the adapter plate that the bellhousing bears on, to the face of the flywheel that the clutch plate bears on. This will be the same (or close enough).

We've been calling this the "critical distance" but it isn't really all that critical since there's plenty of spline on the gearbox input shaft (GIS). We settled for a 0.5 mm error. A slightly more critical distance is to the bearing in the centre of the flywheel that the GIS pushes into. But you have a flywheel and so can measure the distance from that to the face that the clutch plate bears on.