For MFI Gurus. . .and friends of Warren . . .

[304065]

Well, as some of you may have read I was shooting flames out the tailpipe of my 1971 911E at the Club Race a few weeks back. I knew what was causing this-- lack of fuel cutoff on overrun, which is accomplished by the MFI "speed switch."

I knew about the function, troubleshooting and common failure modes of this switch due to reading Warren's posts (RIP buddy) from way back. In one particular thread, he said:

Apparently all of these speed switches are destined to fail because of corrosion induced by a lack of sealing from the elements and the original adhesive compound used to hold the long-legged capacitors to the board and in some cases -- adjacent transistors! While contact has induced rust on every transistor in my unit, the adhesive qualities are gone and the compound had hardened to the consistency of brittle hard chocolate! Handling to trace the circuit resulted in two leads of the largest transistor fracturing due to rust underlying the gold-plated iron/steel leads!

http://forums.pelicanparts.com/showthread.php?t=8074&highlight=corrosion+gold-plated

So I started working on the circuit this weekend, beginning with the speed switch. As soon as I removed it from the electical console, I held it to my ear, shook it and heard the transistor rattling around inside. Just made me smile. The loose transistor says: (Siemens logo) B C Y 5 8 VIII Z L (or 7 Z, but the Germans usually used a bar thorugh the Z and the 7, so don't I'm not sure)

So here's what I've got. . .



You can see the transistor that was just rattling around inside the case. Here is an extreme closeup.



Finally, here is my board. I'll check it against the schematic to verify which one it is. Evidently there are multiple revisions of this circuit out there.



Anyway, all the information and advice I need is right here on Pelican. I'll keep you posted of my efforts to locate a new transistor, desolder the old one and put the new one on. We'll see if it works!

Warren, you helped me find the problem, now you're going to help me fix it! Thanks from all of us for your help over the years.

[FenderBender]

Hi John, Thanks to Warren (RIP) and this board I was able to repair mine also. I cheated and found a strand of wire approximatly the same size and soldered a new leg on it. It's been working for approxitmatly 500 miles so far.

[304065]

Well I desoldered the broken leads from D1 last night. Turning to Warren's historical work on the subject I see that this is going to be a tricky one. . .

It turns out that the emitter-base junction of one of the smaller BCY58-1X [Should be IX -JC] transistors is being used as a Zener diode to regulate the supply Voltage to 8.4 Volts for the timing circuits. Variations in that regulated Volatage, and the actual value of C4 require selection of the values for R5 and R6, which are unique to each unit, and replacement of D1 would require that new values be found by testing!

So I will probably have to change the R5 and R6 resistors also to get the voltage to equal 8.4.
Of course everybody says the Siemens transistors are NLA, so it looks like a 2N2222 will have to be used-- these are available in the original TO-18 package.

But it turns out I can use a Zener instead:

Yes, it is usually a simple repair ... just a replacement of the 8.2 Volt Zener diode D1, or the transistors T1 - T4. In your case, as in many of the Speed Switches, D1 -- the Zener Diode function is being performed by the emitter-base junction of a transistor, but a replacement would be a regular 400 mW or 1 Watt Zener diode. The cathode band of the replacement diode will go into the hole stenciled 'E' at the D1 location.

[Grady Clay]

John,

Why not just get an 8.2 V zener diode? Those are readily available.

It is instructive that the electronic components were so expensive that they could spend the labor to custom fit the least expensive part. Not so today.

Good thread.

Best,
Grady

[Jim Richards]

Great thread! My speed switch might need some love.

[declang]

I also had a transistor rattling around inside the case.
I bought some 8.2V/1.3W zener diodes (BZX85C8V2) and some 2N2222A transistors.

Never got around to fitting them yet !

[304065]

Just to get it all in one place

For the following tests, I suggest a DMM and a TTL/CMOS logic probe such as the one Radio Shack sells for $19.95. All tests should be done with the logic probe set to the CMOS position, and the power leads for the logic probe connected to the power leads #2 and #3 of the Speed Switch.

When connected and open on a bench for testing, check the following with engine running just above 1600 rpm:

1. Low resistance ground from Terminal #3 -- Brown wire -- to chassis ground ... less than 0.5 Ohms!

2. Steady +14 Volt power (11.0 Volts to 14.2 Volts) between Terminal #2 and #3 on the circuit board.

3. Regulated +7.5 Volt to +9.0 Volt bus Voltage at 'north' end of large resistor R12 -- nominal 8 Volts.

4. Pulse activity at base lead of transistor T1 -- silkscreened 'B' on component side of board. This can be checked with a DMM with Hz input function, oscilloscope, or CMOS logic probe.

5. Pulse activity at collector lead of T1 ... silkscreened 'C' on component side of board.

6. Pulse activity at the base lead of transistor T2.

7. Pulse activity at collector lead of T2.

8. Steady high logic level at base lead of transistor T4.

9. Logic low with CMOS logic probe at collector lead of transistor T4.

10. +12 Volt to +14 Volt output at output terminal #1.

[304065]

[304065]

Hmm. . . turns out Warren was EXACTLY right, the base-emitter junction of the transistor is used as a voltage regulator. If you look closely at the BACK side of the circuit board (as far as I know, this has never been photographed) you can see that the Collector of D1 is just out there on its own in an island of solder, not connected to anything, but holding the transistor in place on the board!

I also highlighed the trace that takes +14v power, this was helpful in doing continuity testing to the various devices connected to the power bus. So far, everything OK, but I haven't tested the individual components yet.

I actually bought an inexpensive signal generator for this project, unfortunately I will have to make my own tachometer tester, as I can't replicate the waveform that drives the speed switch just yet. This is easy enough, but I need to try some different components-- driving the coil of a reed relay with the signal generator results in an output frequency that is a multiple of the input! Must be the wiper bouncing.

[Flieger]

Subscribing.

I found a core that rattles inside the can. I want to replace my missing RPM transducer and do it myself to $ave.

I should be trying this on the weekend.

[304065]

Well, the TO-18 packaged 2N2222 transistors are pretty cheap, what is expensive is all the test equipment-- function generator, oscilloscope or logic probe, power supply, etc. If saving money is the goal, it is far cheaper to send it somewhere to have it rebuilt by a professional.

That is not my goal, my goal is to learn discrete components troubleshooting, which is becoming a lost art! And it's kind of fun. I was able to solder the replacement transistor to my board last night, using one of those "cold heat" soldering irons (worked extremely well) and trim the leads-- only to find that Warren was exactly right again (there is a pattern here)-- the voltage on the emitter side of D1 was about 7.9 volts, not the 8.2 volts called for in his schematic! This means that the resistor values are affecting the system voltage, which of course will alter the rate at which the capacitor is charged, altering the timing.

Once I get a decent test signal figured out I'll see exactly how this affects the on/off points for the relay. The relay clicks loudly, alarmingly so.

[declang]

I haven't looked at the circuit in any great detail, so forgive me if this is silly..... but would using an 8.2V zener diode instead of the transistor BE still mean tweaking the two resistor values ?

I remember recently finding a thread with a test circuit that Warren had come up with, but I can't find it now.

[304065]

Declan,

Welcome back, I saw you were looking for a replacement for the Siemens transistors in the other thread.

You are probably right, insofar as the resistors affect how fast C4 charges. Even variability within the tolerance range of the resistors will change the charging voltage. Makes me wonder whether I could substitute a potentiometer to fine tune the charge time.

I would be very interested in your analysis of the circuit, I'm trained as a lawyer, not as an EE

[304065]

Here is Warren's test schematic. You flip the switch at the upper right to select which pullup resistor you want-- 33 ohm for the early CDI, 100 for the later box. Of course his tester is fundamentally driven off the 60Hz mains current here in the USA-- doesn't Ireland use 240V 50Hz? The output values would be different.



So I tried replicating this with a reed relay with the coil driven by my function generator-- driving it with a square wave, I got an output frequency that was a multiple of the input-- must be the relay coil bouncing or something-- and nothing like a square wave for the output.

You know what makes a good square wave? A computer fan's tachometer output. I think I'll try one of those with a pullup resistor. The five watt rating is critical-- I tried it with a .5 Watt 33 ohm resistor and it started SMOKING after a few seconds of operation. . . Doh!

[declang]

John,
Yes, with my 50Hz mains I'd get 50Hz or 100Hz, so 1000 or 2000 rpm. Should still be good enough for testing the speed switch though

Taking 14V and 33 ohms, gives I=V/R = 14/33 = 0.424 Amps
Not surprised your 0.5W resistor got hot, even though its not 100% flowing.

There must be some (very) simple circuits out there for square wave switching at these low frequencies.

The computer fan idea is a good one.

I'll follow your progress with interest as I'll be testing my speed switch, eventually.

[304065]

Declan,

The easiest square wave generator would be a 555. 555s are controlled by two resistors and a capacitor-- the most basic layout would be R2 of 1M, R1 of 700K (giving you a 63% duty cycle like a 6-cylinder Porsche distributor) and C1 of 0.01uF (for 1037 RPM output) or 0.005uF (for double that, 2074 RPM). You would switch the timing capacitors to vary the frequency while maintaining the same duty cycle.

This is worth a try, the parts are probably under five bucks.

[Flieger]

So it appears that all the small transistors can be replaced with 2N2222A models but what about the big transistor? Does anyone know an equivalent part number?

[Flieger]

Is there only 1 transistor that can be replaced with a Zener diode?

The schematic shoes a couple different values for resitors. Should any resistor values be changed or was the schematic printed wrong and revised by hand?

[mjshira]

very interesting stuff, one has to have the patience of Job to tackle this one. good luck.

[Flieger]

I have a transistor (T4) that is Siemens BSX45-10, not a -25 as shown in the diagram. Can this also be replaced by a 2N2102 transistor?

Does the dash number indicate a power capacity or current value? I know you said the 5 Watt rating was important for the resistor.