NB alternators & how to not blow stuff up
 

I got tired of seeing run away alternator induced damage and decided to dig into the root of the problem. Run away voltage is a problem I have seen all variants of NB with multiple ECUs, but seeing as how the 20V+ can damage my ECU -- and then possibly ruin a race weekend for one my customers if they didn't have a backup ECU -- I wanted to figure it out. And honestly it was fun to get back into a lab with a scope and do some actual engineering / R&D instead of management.
This lead to the discovery of a few things and caused us to tweak our ECU hardware for Miatas to harden against high operating voltages.

However I have also figured out the root of the problem. The Miata alternator is poorly grounded. At higher RPM, when poorly grounded, the field transistor internal to the alternator can latch on. When latched, the alternator enters the charge state regardless of charge command. If you have experienced a run away alternator that resolved itself after a power cycle, this is what happened. Extended operation with the transistor latched will eventually fail the transistor, and the alternator is then not controllable.

My suggestion to cure is to grind some paint off the alternator case, and run a proper ground strap between the alternator and block. 10 minutes worht of work and a couple bucks in wire might save your next race weekend.
Frankly, Miata grounding in general sucks so a ground strap between block and battery is also a good thing. Otherwise, you are grounding to the battery through PPF and its hardware.

Interesting find; thanks for sharing!

The recommended installation for Innovate WBO2 sensor issues on NA/NBs is to ground it directly to the battery.
I'm using the power windows wiring to power+GND mine and have never had issues but I'm guessing that the direct block-to-battery ground connection should achieve the same thing? (please correct me if I'm wrong)

Im trying to visualize where to attach this ground strap to on the alternator and my mind isnt wrapping around it...

Are you using one of the case bolts to put a ring terminal on then running to the block? Or using a small self tapper on the case itself somewhere?

Has anyone had this happen to them on an MS3? I've never seen this issue on an MS3 so far.

A friend of mine had something like this happen with his Lucky Dog car with MS3, yes.

--Ian

Also seen this happen on an MS3 car. Alternator jumped to 22v for an extended period and fried the ECU, had to be sent back to DIY for repair. Car moved to an NA alternator to prevent future occurrences.

Thanks for the tip! On my alternator there's a threaded hole for a small metric screw on the back.

https://cimg5.ibsrv.net/gimg/www.mia...4f82d42cee.jpg

Yep, the NB2 alternator that came installed on my spare MSM engine also has that threaded hole. It's M6 x 0.8mm thread pitch. I measured 14mm long (threaded section) from the tape I wrapped around the tap to the tip but that might be slightly too long. With an eyelet underneath the bolt head 14mm might be ok but it also might require sanding/grinding the last few threads. 12mm would be about perfect length. I can't tell from the picture, hopefully this can be added without removing the intake manifold brace. If not it's only three bolts (14mm IIRC) to pull the brace. Many choose to leave the brace off.

Never had a problem with an NA alternator. Why not just back date to one with a built in voltage regulator that doesn't have the voltage control issues the NB2's sometimes do?

We nuked a half dozen or more NB alternators on an NA, with stock ground wiring. Nuked the ecu eventually. Switched to NA alternators, similar problem. Wanted to under drive it, but custom pulley cost was a little high when we were considering an engine swap anyways. Last year of the vvt we ran a fresh air duct to the alternator, never had a problem again.

For a street car there's a theoretical minor drivability improvement because the ECU can know when it's going to dial up the alternator load and adjust the idle/etc controls appropriately. I have no idea if a megasquirt actually does this or not though.

For a race car I agree it's simpler to just go with a 94-97 alternator, and I believe that's what my friend wound up doing. A lot of low-buck racers don't want to spend the money replacing a working alternator and do the rewiring though.

--Ian

If it truly is a grounding issue, which I can believe, the NA alternator would suffer from regulation issues as well.

that said, I suspect it is the grounds via the ppf and driver side braid that are the issue, see below. The alternator has multiple solid metal connection points to the motor.

NB alternators let you regulate to 14V regardless of engine speed, which is a big advantage at idle, especially stock idle.

I have always been a fan and user of the westfield style external HW regulator circuit as opposed to the internal MS processor solution.

the stock miata grounding solution is atrocious. You see people do crazy stuff like cutting signal grounds in an attempt to reduce noise (this is dumb! If unwanted currents are returning to source through signal ground you have an issue somewhere else). I have been running my own much larger braided straps (one to the head, one to the block) for years and it solved all the gremlin issues.

Where are you grounding your straps to the head and block? Using one of the stock grounding points in the engine bay, or running something all the way back to the battery?

During "Run" all currents are sourced via the alternator, not the battery. These currents circulate essentially in the engine bay and do not return to the battery. IOW the common "running a ground to the battery" concept is flawed. Furthermore, it is impedance that we care about, which is a function of cross sectional area at DC and surface area for high frequency. No dedicated wire back to the battery will ever match the cross sectional area or the surface area of the main chassis. Where the issue lies with chassis grounds is:

the chassis is painted and paint is non-conductive, so connection points should have no paint and corrosion prevention

while aluminum is a decent conductor, oxides of aluminum are frequently insulators, and begin forming nearly instantaneously in the presence of oxygen. All aluminium connection points should have oxide removed immediately prior to connection with some form of a sealing solution (like a sealing washer)

System grounds are not chassis ground, but are frequently treated as such. They may be common with chassis ground, but are separated by the impedance of the wire per unit length.

here you can see I have run a much larger braid to the head and to the block (the block connection is through a steel bracket). These are all located in a way that allows visual inspection if needed.

I should note that the block connection is directly to the alternator through the tensioner.

Hope this helps


https://cimg1.ibsrv.net/gimg/www.mia...a0ae7c3340.jpg
Chassis
https://cimg3.ibsrv.net/gimg/www.mia...399f3e3e75.jpg
Block
https://cimg4.ibsrv.net/gimg/www.mia...54248560dc.jpg
Head

Thanks for the pictures and write-up. I see what you did, I'll have to run mine mirror-image since I have an NB (kind of weird how the alternator is on the opposite side from yours).

yeah, some idiot went and installed a supercharger where the alternator should be on my car.
/jack

I hate when that happens! At least I know I'm not losing it.

I'm adding pictures of how I ran my ground wires in case a) it helps someone else, and b) I screwed something up someone else can tell me about. Ground to alt: https://live.staticflickr.com/65535/...0b054ef7_c.jpg

Ground to chassis: https://live.staticflickr.com/65535/...42d7b931_c.jpg

Ground to head: https://live.staticflickr.com/65535/...7c22c2c2_c.jpg

Ground to chassis: https://live.staticflickr.com/65535/...8218a16e_c.jpg