Convert battery voltage to AC

Thread Starter

homerdodd

Joined Feb 10, 2018
69
I have a pesky transmission light on and it drives me nuts! There is nothing wrong with the way my car drives or shifts, but the light glares at me. The reason is because of a permanent magnet type speed sensor, which generates an AC voltage of 5VAC and higher to the PCM. As long as the cars computer (PCM) sees this 5VAC, then the light stays off. The sensor itself is buried in the transmission, approx $2,000 in labor. Some guys have designed OP amps to simulate 5 VAC into the computed, and thus fooling it, causing the light to go off (what I want !). I'm thinking that if I can get help designing a small DC to AC transformer, and send 5-10 VAC to the computer, my problem will be solved. Can someone help me out with a transformer, or inverter circuit, whatever is the most compact and simple to build. Whatever it takes to get 5-10 VAC is all I need. Thanks !
 

LesJones

Joined Jan 8, 2017
4,174
You say in your first post that the car is working but the light annoys you. Post #3 seems to contradict your first post.
Are you sure you only need a fixed frequency ? I would expect the frequency of the signal is what it uses to make the decision to change gear.

Les.
 
Last edited:

Alec_t

Joined Sep 17, 2013
14,280
If the transmission shift controller is operating with no obvious problem then it must be receiving pulses of sufficient amplitude. Presumably it is reporting a less-than-intended amplitude as an error. Nevertheless, those pulses will have a repetition frequency related to shaft speed. If you make a DC-to-AC convertor the frequency will be unrelated to shaft speed, so isn't that likely to throw another error signal or confuse the shift controller? I would have though an amplifier would be more effective than a DC-to-AC converter.
 

Thread Starter

homerdodd

Joined Feb 10, 2018
69
If the transmission shift controller is operating with no obvious problem then it must be receiving pulses of sufficient amplitude. Presumably it is reporting a less-than-intended amplitude as an error. Nevertheless, those pulses will have a repetition frequency related to shaft speed. If you make a DC-to-AC convertor the frequency will be unrelated to shaft speed, so isn't that likely to throw another error signal or confuse the shift controller? I would have though an amplifier would be more effective than a DC-to-AC converter.

Thank you. No, it won't throw another code, as long as the PCM see's the parameter voltage of 5-9 volts, even if it is fixed. The permanent magnet two-wire sensor (when working correctly) generates 5-9 VAC. I road tested my car with a DMM hooked up, and the most it generated was 1.5 volts, which is way below the parameter, turning on the "Check Engine" lamp. This is a common problem on these transmissions. Several guys said they bypassed the sensor with an OP Amp, but I can no longer locate them to get their schematic, as the posts were from and older forum. The main reason is because the car will fail a visual state inspection with the light on, even though it has nothing to do with emissions. My car passed the tailpipe emissions test with flying colors, so I am not polluting. Anyway I can get 5-9 VAC to the sensor harness is what I need; it is really that simple. If I could amplify the 1.5 VAC coming out of my defective sensor, that would work also.
 

Thread Starter

homerdodd

Joined Feb 10, 2018
69
You say in your first post that the car is working but the light annoys you. Post #3 seems to contradict your first post.
Are you sure you only need a fixed frequency ? I would expect the frequency of the signal is what it uses to make the decision to change gear.

Les.
I simply have to get the light off to pass a visual inspection. The trans shifts and operates with no problem.
 

Alec_t

Joined Sep 17, 2013
14,280
Here's a simple amplifier you could try, at your own risk. It boosts the input signal from the sensor by a factor of ten or so. Observe the correct polarity of the capacitors.
SensorAmp.PNG

BTW, a DMM won't give you accurate voltage readings if the sensor signal is not a reasonably pure sine-wave.
 

Thread Starter

homerdodd

Joined Feb 10, 2018
69
Here's a simple amplifier you could try, at your own risk. It boosts the input signal from the sensor by a factor of ten or so. Observe the correct polarity of the capacitors.
View attachment 145637

BTW, a DMM won't give you accurate voltage readings if the sensor signal is not a reasonably pure sine-wave.

Thank you much for the schematic. Would the output voltage be 25 VAC or is it DC? The computer needs the AC signal to operate correctly.
 

ebp

Joined Feb 8, 2018
2,332
I think, but don't know for sure, that you could probably simply use a unipolar squarish wave. The receiving circuit probably simply ignores the negative part of the waveform and uses the positive part. The big unknown is just what the computer tries to derive from the signal. It seems unlikely to me that it looks for anything other than period, so as long as you feed it a signal for which it can determine the start and end, it'll probably be fine. It sounds like it doesn't even measure period, but just looks for a regular poking.

If you need AC relative to "ground" and the specific waveform isn't important, the simplest thing to do is generate a square wave at the appropriate frequency and AC couple it - use a capacitor-resistor circuit.
Connect a capacitor to the output of the square wave generator and a resistor from the free end of the cap to ground. You need to select the R & C appropriately to get a suitable waveform. You can easily fiddle with the circuit on the bench or using any of the free circuit simulation tools. The rectangular waveform probably doesn't really need to be square, so something like a 555 running at car battery voltage would probably work, though this shifts the relative amplitudes of the positive & negative. You can clip the excursions with zeners if necessary. The circuit which receives the signal from the pickup will probably load the signal (its own R to ground), so additional testing might be required.

You might buffer and AC couple the capacitor voltage from a 555 astable. A bipolar emitter follower would be OK for buffering, since the minimum voltage on the cap is 1/3 of the 555's supply voltage, unless you fiddle with the thresholds.

All manner of exotic solutions are possible, but I doubt their necessity - e.g. unipolar square wave driving a comparator or over-driven op-amp that has a negative supply rail from a charge pump; use a charge pump to produce negative rail & an RS-232 transmitter driven with a square wave to produce relatively low-impedance true AC output; all sorts of circuits using discrete transistors as drivers
 

ebp

Joined Feb 8, 2018
2,332
An aside on the sensor:
If there is a permanent magnet in the sensor, it is likely a "variable reluctance" pickup. They are common for detecting gear teeth or other ferromagnetic bits passing by. You can find details on the web.

Depending on the physical arrangement, the output waveform may spend a significant time at zero: e.g. with a typical gear, the tooth crest is considerably less than half the tooth pitch; if it is set up to detect a key in a keyway or a boss on a flywheel, it will give one more-or-less sinusoidal pulse per rev as the key or boss passes, so trying to measure with voltmeter isn't instructive unless you know the expected waveform.

VR sensors receivers sometimes use differential amps at the input. A diff amp, depending on design (sigh, always that blasted caveat) doesn't care about where the signal is relative to "ground" as long as it falls within the limits that the amp can handle. If could easily interpret a signal swinging from -10 V to -9 V exactly the same way as one swinging from -0.5 V to +0.5 V or from +99 V to +100 V.
 

Alec_t

Joined Sep 17, 2013
14,280
Would the output voltage be 25 VAC or is it DC?
About 6V RMS maximum. AC because of the output capacitor. Since there is no DC path to ground from the output, the PCM might not be happy with the signal if it is doing any preliminary checks for the presence of the sensor coil.
 

Thread Starter

homerdodd

Joined Feb 10, 2018
69
Here's a simple amplifier you could try, at your own risk. It boosts the input signal from the sensor by a factor of ten or so. Observe the correct polarity of the capacitors.
View attachment 145637

BTW, a DMM won't give you accurate voltage readings if the sensor signal is not a reasonably pure sine-wave.

Thanks again, but another question: if I am looking at the diagram correctly, my sensor remains in the circuit and the battery voltage is the "driving force" to amplify the output? I noticed the 25V amount listed, how was this calculated? Sorry for all the questions, I am only a novice at basic electronics. This gives me hope, and will probably give the computer the signal it needs.
 

Thread Starter

homerdodd

Joined Feb 10, 2018
69
An aside on the sensor:
If there is a permanent magnet in the sensor, it is likely a "variable reluctance" pickup. They are common for detecting gear teeth or other ferromagnetic bits passing by. You can find details on the web.

Depending on the physical arrangement, the output waveform may spend a significant time at zero: e.g. with a typical gear, the tooth crest is considerably less than half the tooth pitch; if it is set up to detect a key in a keyway or a boss on a flywheel, it will give one more-or-less sinusoidal pulse per rev as the key or boss passes, so trying to measure with voltmeter isn't instructive unless you know the expected waveform.

VR sensors receivers sometimes use differential amps at the input. A diff amp, depending on design (sigh, always that blasted caveat) doesn't care about where the signal is relative to "ground" as long as it falls within the limits that the amp can handle. If could easily interpret a signal swinging from -10 V to -9 V exactly the same way as one swinging from -0.5 V to +0.5 V or from +99 V to +100 V.

It is a two wire permanent magnet sensor. The Cadillac schematic indicates one output as "low" and one output as "high". I have attached picture from the service manual, that I slightly modified for wiring color coding. As I said before, it is common for these sensors to wear out and not send the minimum 5VAC (mine is only sending 1.5 - 2.0, way below spec). Very labor intensive to remove the transmission and replace it. Whatever I can do so send 5-9 VAC to the computer is all I need, even if if means disconnecting the sensor completely from the wiring harness and inserting some other supply of AC voltage. Again, thanks for your time and assistance.
 

Attachments

Alec_t

Joined Sep 17, 2013
14,280
if I am looking at the diagram correctly, my sensor remains in the circuit and the battery voltage is the "driving force" to amplify the output?
Yes. My circuit assumes the Blue/Wht (low side) line is grounded. Is that the case? If not, you may need another electrolytic capacitor, to link that wire indirectly to ground.
I noticed the 25V amount listed
That is the suggested Working Voltage (WV) rating of the capacitors, i.e. a voltage well above the battery voltage. A WV value >25V would also be ok.
 

ebp

Joined Feb 8, 2018
2,332
I can't really tell what sort of sensor it is from the pic. It is likely a variable reluctance type since they will sense any ferromagnetic (like steel) "thing" that passes close enough. I have no idea why one would "wear out" unless the magnet in it loses strength. It could be just an inductive pickup, but this would require magnets on the moving thing, which is usually a lot more expensive to do.

Here's a data sheet for a signal processor for a VR sensor:
http://www.ti.com/lit/ds/symlink/lm1815.pdf

Page 8 has a couple of interesting things:
- an oscillogram that shows why a meter might read "too low" - note what looks like individual full sine waves separated by considerable time at zero. Could you borrow a battery powered oscilloscope (and a co-pilot) to look at the signal to confirm the amplitude really is too low? Even the most basic low-bandwidth scope would do the job. There are some multimeters that have rudimentary scope functions and would be quite adequate.
- a description of the input signal clamp: this particular device would work perfectly well with a signal swing between 0 volts and some positive voltage - it lops off the part that is negative with respect to ground. In general, bad things can happen if any pin on an integrated circuit that operates from a single positive supply is pulled more than a few hundred millivolts negative with respect to the circuit's ground. This circuit uses an elaborate scheme to prevent that.

Are you sure the computer requires a signal that swings negative or are you assuming that because that's what normally comes out of the sensor?
 

ebp

Joined Feb 8, 2018
2,332
Found something on the web that suggests that the sensors are VR type and the magnet can attract metal particles over time and the accumulation can interfere. This does sound quite plausible. The sensor may be OK, but you still have to pull it out to clean it up, which is no help if you have to pull the transmission to get at it. It's like those things in modern cars where it takes two minutes to fixed the fault - once you've taken out the front seats and pulled off the dashboard. I went without a car for about half of last year because I couldn't get up the ambition to tackle the repair. When I got around to it, it was quite easy & wound up costing me less than it cost to fix my moderately-priced bicycle after I hit a pothole.
 

Thread Starter

homerdodd

Joined Feb 10, 2018
69
I can't really tell what sort of sensor it is from the pic. It is likely a variable reluctance type since they will sense any ferromagnetic (like steel) "thing" that passes close enough. I have no idea why one would "wear out" unless the magnet in it loses strength. It could be just an inductive pickup, but this would require magnets on the moving thing, which is usually a lot more expensive to do.

Here's a data sheet for a signal processor for a VR sensor:
http://www.ti.com/lit/ds/symlink/lm1815.pdf

Page 8 has a couple of interesting things:
- an oscillogram that shows why a meter might read "too low" - note what looks like individual full sine waves separated by considerable time at zero. Could you borrow a battery powered oscilloscope (and a co-pilot) to look at the signal to confirm the amplitude really is too low? Even the most basic low-bandwidth scope would do the job. There are some multimeters that have rudimentary scope functions and would be quite adequate.
- a description of the input signal clamp: this particular device would work perfectly well with a signal swing between 0 volts and some positive voltage - it lops off the part that is negative with respect to ground. In general, bad things can happen if any pin on an integrated circuit that operates from a single positive supply is pulled more than a few hundred millivolts negative with respect to the circuit's ground. This circuit uses an elaborate scheme to prevent that.

Are you sure the computer requires a signal that swings negative or are you assuming that because that's what normally comes out of the sensor?

Great info, thanks !
 

Thread Starter

homerdodd

Joined Feb 10, 2018
69
I can't really tell what sort of sensor it is from the pic. It is likely a variable reluctance type since they will sense any ferromagnetic (like steel) "thing" that passes close enough. I have no idea why one would "wear out" unless the magnet in it loses strength. It could be just an inductive pickup, but this would require magnets on the moving thing, which is usually a lot more expensive to do.

Here's a data sheet for a signal processor for a VR sensor:
http://www.ti.com/lit/ds/symlink/lm1815.pdf

Page 8 has a couple of interesting things:
- an oscillogram that shows why a meter might read "too low" - note what looks like individual full sine waves separated by considerable time at zero. Could you borrow a battery powered oscilloscope (and a co-pilot) to look at the signal to confirm the amplitude really is too low? Even the most basic low-bandwidth scope would do the job. There are some multimeters that have rudimentary scope functions and would be quite adequate.
- a description of the input signal clamp: this particular device would work perfectly well with a signal swing between 0 volts and some positive voltage - it lops off the part that is negative with respect to ground. In general, bad things can happen if any pin on an integrated circuit that operates from a single positive supply is pulled more than a few hundred millivolts negative with respect to the circuit's ground. This circuit uses an elaborate scheme to prevent that.

Are you sure the computer requires a signal that swings negative or are you assuming that because that's what normally comes out of the sensor?
I vaguely remember one of the Cadillac guys saying it was a square wave AC signal, if that makes sense...?
 

Thread Starter

homerdodd

Joined Feb 10, 2018
69
Yes. My circuit assumes the Blue/Wht (low side) line is grounded. Is that the case? If not, you may need another electrolytic capacitor, to link that wire indirectly to ground.

That is the suggested Working Voltage (WV) rating of the capacitors, i.e. a voltage well above the battery voltage. A WV value >25V would also be ok.

I'm going to give it a try............thanks much !
 
Top