96-97 Specific Purge Valve Mod Question
 

Hey guys, tomorrow I'm doing the purge valve mod to help out with some A/C issues and I noticed the wiring harness is different.

ADMIN EDIT: REMOVED OLD IMAGE

I was talking to Joe and he said that 2X is actually 4T on my unit but were unsure where 1Q would be on my mspnp. Best guess is that it would be the green/black wire on 1K. Before I place the jumper I wanted to see if anyone has done this on the 96-97 unit and can confirm 1K.

Thanks

listen to Joe. that diagram is old and has been replaced.

http://www.boostedmiata.com/gallery2...ne_harness.jpg

and the fans are incorrect on the above. lol

Thanks man, I pulled that from an old thread. Appreciate the help guys.

This morning I went outside and did the purge falve mod. This is the one that activates the unused charcoal canister's purge valve to give a little idle bump when the a/c turns on. So I placed my jumper between 1K and 4T and it worked! Everytime the a/c compressor turned on the valve would open and voila, no stall.

Until it started stalling. It stalled with the a/c on, it stalled when I turned the a/c off, it just refused to run. I turned the car off and pulled the jumper out but no dice.

What was happening is that the a/c compressor was not shutting off, at all. I finally noticed green a/c oil dripping out so I shut the car off and disconnected the two large a/c relays at the front on the passenger side.

So anyone got any guesses as to why the a/c refuses to shut off? I'm assuming the MSPNP is no longer closing the connection. When I press the switch in the cabin the little light will turn on and off but it doesn't seem to make any difference whatsoever. I checked the MS itself and no smells of ozone or apparent damage on any connectors or solder points. I also checked the harness but all seems to be well. The wires I jumped both appear to be firmly seated in their connectors. However if there were a wire that would cause this which would it be?

At the moment the car runs, just no A/C, if I plug the relays back up the whole problem comes back. However with the GA summer looming I'd really like to find a solution.

I suppose it's possible that this problem isn't MS related although Occam would say it would be odd for it to happen as I was modifying something to do with the a/c. I just don't see how a jumper could cause this.

Ok so I did more diagnosis but I'm still stuck. It's my understanding that the ECU sends a signal out through the Lt/G wire I jumpered. That signal goes through the pressure switch, the temperature switch, the button on the dash, and then out to the relay that turns on the compressor clutch.

I checked the relay and it seemed to be operating as normal. Also, the button on the HVAC does nothing, so it should be safe to assume that's fried. However with the a/c constantly on the pressure switch never seems to kill the signal either. So it would seem that, at least, the pressure switch and HVAC switch are dead and stuck open.

Does that seem odd to anyone else? Usually when a switch breaks it breaks in the closed position. I understand the pressure switch is normally closed but the dash button is mechanical, it wouldn't seem like one or two extra volts would somehow just make it a solid connection.

I'm really stumped on this one. Is there another way the compressor clutch can activate? Can the ECU override the switches? From the wiring diagram it doesn't seem possible but it's certainly happening.

When the LG/B wire is pulled to ground, the ECU sends a ground out on the BLU/BLK wire to energize the A/C relay and A/C clutch. What you did was jumper another potential ground path into the LG/B wire. The function of the YEL/RED wire is to provide a ground to close the purge valve solenoid. To be safe, a diode should have been used so that a ground on the LG/B wire would be seen by the YEL/RED wire, but not vice versa.

In addition, whenever you add an inductor (such as a coil) to a silicon circuit, you need a protection diode to dissipate flyback voltages that happen when the inductor discharges. Your stock wiring for the LG/B wire only has switches . . . no coils. Therefore, I would presume it is an unprotected circuit. By jumpering into the YEL/RED wire, you've added a coil to the circuit (the coil in the purge valve solenoid).

Time to get out the multi-meter. Is the LG/B wire at ground? If so, where is the ground coming from? I suspect internal frying. You've likely also done some mechanical damage if you're leaking R-134 oil.

The mechanical damage I expected. That doesn't worry me too much as a compressor can always be replaced, it's the ECU stuff that bothers me.

So what you're saying, just to make sure I understand correctly. Is that the LG/B wire is likely grounded which causes the BLU/BLK wire to send the signal for the compressor to activate?

Would it most likely be grounding within the unit itself or somewhere on the car? If it's somewhere on the car than it would likely be the A/C switch in the HVAC right as the other two switches don't connect directly to a ground? If it's somewhere inside the ECU would you expect it to be something that could be fixed or just a fried unit (like perhaps internally to the chip)?

Thanks for your response btw

I wouldn't have expected the jumper he added would kill anything, but it sounds like the a/c relay is stuck open. pull the connector out of it and replace.

My guess is that something inside the ECU has failed. Likely a diode or transistor (i.e., a silicon PN junction somewhere). Quick test, pull the connector and see if the pin for the LG/B wire is at ground. If everything is OK, it shouldn't be.

the compressor is activated through the a/c relay...something is holding it open. the ECU provides a ground to that circuit.


pull the wiper fuse and replace and see what happens.

Hornet: Which connector do you mean? Does the car need to be running for this test? I was just wondering because I know I can't unplug that section of the harness while the car is running. Excuse my ignorance but what is a PN Junction? I've been trying to learn this stuff recently but I havent gotten to that yet.

Hopefully it's just a diode or transistor as I have a ton of those laying around from other projects. I'm still trying to figure out exactly what I need to do to locate the faulty piece. Do I check for ground and then just work my way from the pin until I find a point that is no longer grounded?

Brain: I assume you mean the relay under the hood next to the passenger side headlight? Right now I have it unplugged so I can drive the car. I checked it with my multimeter and it was closed although I have no solid way to test with the car running. I'm currently searching around to see if anyone has one locally I can try. Does the car need to be running for the ECU to hold the ground to the relay open or can it just be on?

I drove the car for about 20 minutes today with the relay unplugged and it ran fine. But is there any danger of the unit possibly overheating if it is holding that pin open like that?

BTW, I was talking to Joe about this earlier and I think, at least till a proper fix is discovered, there might need to be a warning in the "Great A/C FAQ" thread. I know people still reference that thread regularly.

The wiring diagrams I have show that the AC Relay is normally open. It is pulled closed by a ground on the ECU BLU/BLK wire. So, what you suspect is that the relay is stuck closed (melted contacts)?

yeah that.

I meant the ECU connector. Car not running. If you have a failed component, it won't matter whether the ECU is powered or not. However, Brain knows the guts of these things better than just about everyone else, and he seems convinced that this jumper should not have caused damage.

You just stated that "you checked [the AC relay] with my multimeter and it was closed . . . ." That's a failure. The AC relay should be normally open. Heck of a coincidence but a nice easy fix if that is all it is.

You should get a wiring diagram for your car BTW.

The only wiring diagram I have is the Hayne's manual.

I actually I just pulled the relay and I misspoke. Completely disconnected from the car the switch inside the relay is open. So the MS must be powering the relay.

So next step is to open the MS on Thursday and find the faulty connector. Any recommendations on how I could narrow down the search and isolate the fried part? Diagnosing circuit boards is still foreign to me.

Bummer.

So, start at the AC relay socket and, with power on and AC controls off (car does not need to be running), confirm that a voltage and a ground are present to operate the coil. If the ground is present, move on to the ECU. ECU tests will be done with power off since we have to pull the connector.

On the ECU, check if the ECU pin for the LG/B wire is at ground. If it is, there is an issue inside the ECU. If not, then do the same thing on the wiring side. Ohm out the normal LG/B wiring path and your jumper separately. Start writing the stuff down so you know where you are.

If it is internal to the ECU, then it will get a bit tricky. I don't believe that schematics for the MSPNP daughter board are available. Anyway, don't open the MS until you know that is where the problem is. Hopefully you'll find something in the wiring.

follow the output you jumped back into the MS. They use a fairly simple relay circuit to activate the A/C. Most likely a red zener diode, a p2n2222 transitor, and a 1K resistor.

Matt at DIY seems to think it might be the switch on the dash, which would be ideal. I have tomorrow off so I'm going to dive into this and at least find out whether it's in the MS or in the switches. I'll let you guys know what I find.

I assume to test a diode I simply see if it's allowing current through in both directions?

yes. or none in any.

Well it's confirmed, it's something in the ECU. The pin for the LG/B wire was groused. So I opened the ECU to start looking for the faulty component but the connections where the pin connects to the circuit board are not visible.

Not sure what to do next I remembered Brain saying that a 1K resistor would likely be used on the switch so I checked and both 1k's on the daughterboard and they were fine. So then I figured I would just check for continuity between the pin and the resistors to see what it a part of the circuit. Well, almost every resistor on the daughterboard is connected to the pin, most are between 600 and 800 ohms. So that didn't really narrow anything down for me. Brain also mentioned a 2N2222 transistor, there are 4 of them on the daughterboard next to the harness but they were the same as the resistors, all connected to the pin with varying amounts of resistance.


I really need a bloody diagram for this thing but I know DIY won't provide it.

I *think* you hosed U1. The 2N2222 is likely fine, unless you did something like hook your coil to the A/C command input...

That's exactly what he did. The coil of the purge valve solenoid was jumpered into the A/C command input with no protection against flyback voltage.

Yeah, he used 1K and not 1G, which would have had been protected.

So I should start by locating the faulty transistor? There's only 4 on the board so it shouldn't be too difficult.

Yes, check all the 2N2222's on the MSPNP daughter board.

Readup on "flyback voltage" or "protection diode" (Wikipedia works fine). In the MS1 Extra Hardware Manual, there are a lot of examples of flyback voltage protection diodes. See, for example, the circuit for "Outputs 1, 2, 3 and 4." The 1N4002 in that circuit is there to dissipate flyback voltage. To be safe, you need something like this going across the coil of your purge valve solenoid. Note that 1N's are pretty much interchangeable, with the larger numbers able to resist larger reverse currents. I typically use 1N4004's, as they are readily available in packages of two at Radio Shack.

So, I went back and looked at the thread that started this whole thing:

https://www.miataturbo.net/showthrea...ht=purge+valve

I think there is some information missing from this. In fact, there seems to have been at least one other person on that thread who experienced Dutton's problem (although his failure mode was AC clutch not engaging instead of stuck engaged). I'm getting ready to do this myself to give a little more kick when I engage A/C. I really don't feel like replacing my compressor or burning up something in my MSPNP.

First of all, a little bit of wiring definition because the ECU pinouts have changed over the years:

Wire ** Function ** (1.6 ECU Pin/1.8 ECU Pin)
Y/R ** Ground for Purge Valve ** (2X/4T)
LG/B ** AC Command Input ** (1Q/1K)
L/B ** Ground for AC Relay ** (1J/1G)
W/R ** Power for ECU and Purge Valve ** (1B/4B)

Joe's design jumpers Y/R into LG/B without providing for flyback voltage protection. Brain has suggested that Y/R should be jumpered into L/B, which makes more sense to me. What I am after is activating the purge valve when the A/C clutch engages, and L/B is closer to the AC clutch than LG/B. In addition, L/B has some flyback protection because it is designed to be connected to a coil whereas LG/B is only designed to be connected to switches.

However, I don't think the existing flyback protection on L/B is necessarily sufficient, because the purge valve solenoid can be powered in run and start, whereas the AC clutch can only be powered in run. You might potentially get flyback voltage damage during start in this case.

So, I would propose the following implementation and would appreciate feedback.

First, use a diode jumper to ground the Purge Valve Solenoid when the AC Clutch is engaged (note: diode used to prevent a ground on Y/R from interfering with L/B function):

Y/R ----->|----- L/B

Second, a diode jumper to dissipate flyback voltage on Y/R and protect L/B:

Y/R ----->|----- W/R

I think this will be safe. Thoughts?

Ok, so I may be getting another MS in a trade with a guy building a track car. If that falls through I'm contacting DIY about repair, and if that falls through I have a plan.

So the LG/B wire and the Blu/Blck wires that go to the ECU are just grounds correct? Other than connecting the two all the ECU is doing is turning on the seconday fan.

So what if I just wired those two directly together? When the dash switch is turned on it would ground the A/C relay directly which would turn the system on. If the pressure or temperature switch get out of range it would turn the compressor off. Since the Blu/Black wire only grounds the relay I can't imagine there's a ton of amps running through it so it should just ground through the switches, thus bypassing the ECU completely.

Now, what am I missing? There's no way it's this simple.

No I meant ignition coil 25k+ volts. My money is on U1 needs to be replaced, and likely that's it. LM324N. Come by and I will give you one.

I would also check the car's a/c switch to make sure it still goes to ground when depressed.

It pretty much is that simple.

Here's the gist, the LG/B wire is an INPUT. The stock ECU uses this input for many things. You already mentioned the fan. It also kicks up the idle speed internally and it grounds the L/B OUTPUT to the AC relay.

In our case, the MSPNP connects the LG/B INPUT to the L/B OUTPUT and takes care of the fan. There is no idle speed compensation. We're trying to assist that with the purge valve, but you got burned because you connected the purge valve solenoid to an INPUT instead of an OUTPUT.

If you do what you suggest, you still have to handle the fan, but otherwise it should work (you can also wire the purge valve into that connection without worry). It's a bit more wire surgery than I want to do on my car, however. I'd like to leave the wiring and connectors as intact as possible and add my jumpers and circuits inside the MS. That's just me though.

I think you would be surprised at how much flyback voltage a simple solenoid coil can generate. Not much power, but a heck of a lot of voltage.

My fans are wired in parallel so that's not an issue for me. This isn't by any means my #1 choice. First I'm going to check U1 and the dash switch like Ben suggested. Also I'm hoping this trade goes through and it will fix a lot of problems for a few people (my unit would go to a track car).

I'm aware. But I also have access to the schematic. ;) U1 would take the hit.

Thanks for the offer on the chip Ben. I live so far south I'll probably just have to order one though (the only day I have off is Sunday and you guys are closed).

Is there anyway to test the chip before I pull and replace to confirm?

Ahhhhh . . . . Access to schematics, eh? Sneaky.

BTW, as long as the schematics are in front of you . . . care to weigh in on my suggested use of diodes to make this safe? In particular, is the isolation diode between Y/R and L/B needed?

I figure you owe me since you were so coy.

Whoo hoo Ben it looks like you're spot on. So I found another user who had this exact same problem (only they have a 95) and it was the same op-amp you were saying. Hopefully (fingers crosses) all I have to do is locate this chip on my daughterboard and replace.

Of course my compressor is probably still fried but at least that's something I know I can fix. Besides, my a/c had a slow leak anyway so it was probably time to replace all the o-rings.

I plan to order the chip tomorrow (maybe I can send DIY a few bucks so shipping will be fast) unless someone knows a local place to buy them. It looks like radioshack sells an LM324 but is that the same as a LM324N?

probably not. the DIY one I assume is an 8 pin, not 14pin.

Glad you found your problem. So, you definitely had a flyback voltage issue. You're probably OK with the RS part, but if I were you, I would get the exact op-amp from Ben. Unless you did the design or have access to design data, it can be hard to tell why an exact variant of a chip has been used.

Also, since you're talking to Joe about this, can you run the diode jumpers I posted above by him? I want to do this, but I don't want to repeat your experience. I dabble in electronics, but Joe is a EE (I think), and I'd feel a lot better if a trained person would weigh in. Thanks for creating a thread about this. You've certainly helped me.

To protect from flyback when the solenoid closes? Probably couldn't hurt to put a 1N4001 between Y/R and +12V.

Drew, send an email to websales@diyautotune.com with your shipping info, and I'll send it to you.

Thanks Ben, just got the confirmation.

I've passed the idea on to Joe for consideration. Until someone confirms a jumper design that works I'm just going to wire a small switch into the purge valve and operate it manually. I'll let someone else play Guinea Pig this time ;)

Well, my A/C is operating normally. I got the new chip in today (oddly enough it was a LM2902N instead of the LM324N in the board). I did my first ever PCB circuitry, I was so scared of damaging the board it took me an hour and a half roll of solder braid to get the old chip out. But I just fired up the car and the compressor comes on and off with the dash button as it should.

There isn't much freon or oil in the system so the air wasn't much colder than ambient but any mechanical damage will be much easier to diagnose.

Thanks for all the help guys. Hope this thread can serve as a warning to anyone in the future.

Also Hornet, let me know how your jumper+diode works. I don't want to abandon this idea completely. For the time being I'm just going to wire the purge's ground to a switch in the cabin but I would like a more automated system if you figure it out.

Planning to try it this coming weekend.