Tuner Studio Reads 1 volt higher than actual battery voltage

[elaw]

That would depend on the DC power supply. Some have a pretty "clean" output, others do not.

Is there any chance you can power it from a battery of some sort? In the car, or on the bench - even a small 9-volt battery as used in radios and such will power a Megasquirt for a while. A battery is as close to a noiseless power source as you'll find... if you see spikes when running (the MS, not the engine) on battery there's definitely a problem.

[billr]

Unfortunately, intermittent problems are very difficult and tedious to track down. I can only offer my sympathies there. As I recall, this is a PNP unit, right? Can you open it up and inspect the soldering and such in there? Point is, if the soldering is OK, it is very unlikely that any of the components are "intermittent". Modern solid-state components rarely flicker "on-and-off" like a broken incandescent lamp filament being shaken. An intermittent problem is most likely, by far, in your wiring or charging system.

Are there any analog inputs on a PNP MS that could be used to read and log the raw 12V? Can any of the electronics experts here suggest a simple circuit that would use the raw 12V as an input to provide a simulated ECT/IAT resistance that could be logged?

[lagos]

I did some more investigating and looked at my circuit board. Seems I found a burned out transistor. This is right next to the R15 resistor on the DIYPNP.
Does anyone know what this transistor is use for and if it could be the cause of my issues? I consulted with Matt Cramer, and he said this was part of the wled circuit, but I just wanted to confirm this. The blue wires next to the transistor are for my AC, which has never worked right ( probably because of this transistor).

Here is a picture of the carnage.

[Matt Cramer]

The transistor itself isn't part of WLED (I had been looking at the trace) - the transistor is part of the nitrous input circuit.

[lagos]

The transistor itself isn't part of WLED (I had been looking at the trace) - the transistor is part of the nitrous input circuit.

Could a bad transistor on one of the inputs cause the voltage rise issue that I have been dealing with?
I am using this input for my AC system. The in wire is going to my ACT pin and the out wire is going to AC, on my factory harness. ACT should be the amp signal and AC should be for magnetic switch. Why would the transistor burn out like that with this setup and how can I prevent it from happening again?

I de-soldered the bad transistor and cleaned up the area with alcohol. Happy to report that the transistor was the only victim and the circuit board is still in good shape.

[billr]

You really, really need to log what the "raw 12V" input to MS is doing! Can the O2 sensor input be used to do that, with appropriate voltage-divider resistors to bring it down into the 0-5V range? Speak up, electronics experts, this OP needs to determine for sure which direction to be searching; into the MS or out into the car wiring/charging.

[lagos]

You really, really need to log what the "raw 12V" input to MS is doing! Can the O2 sensor input be used to do that, with appropriate voltage-divider resistors to bring it down into the 0-5V range? Speak up, electronics experts, this OP needs to determine for sure which direction to be searching; into the MS or out into the car wiring/charging.

I ended up probing the 12v input voltage and ground with my multimeter. The input voltage is reading a steady 14.2v at all times.
I also have it setup now so that I can easily probe this 12v line while on the road , if the voltage rise issue pops up again with the transistor removed.

Matt mentioned that there could be something on the 5v vref that could be pulling the voltage down and messing with the signal. The only odd thing in this regard would be with my coolant temp and Mat readings. On a cold engine, both of these two sensors never reach true equilibrium. The coolant temp sensor always reads 10F higher than the MAT sensor. But that seems more like a resistor bias issue, than anything else.

[billr]

Do you have the multimeter probes (both + and -) connected to the DB37 connector on the MS mainboard itself? They need to be there to eliminate all other sources of error when comparing the meter to what TS displays.

[lagos]

There was also some odd behavior that I noticed in my gair readings. I had the car fully warmed up and pulled over to get gas. After leaving the gas station, I noticed that my afr was very rich. Looking over at the laptop, my Gair was reading 106 at 59F ( drove like this for about a mile so the car was fully warmed up after the restart). I pulled over and power cycled the ecu. Now gair was correctly reading 102 at 62F. What could have cause this? If you look at my two data points, it seems as the calculation for linear interpolation was being done wrong and corrected itself after the restart

Here is where my Mat settings were

air correct issue.JPG

Here is the log I started when I pulled over to the side of the road.

GairProblem.msl

Seems like a similar problem to the voltage issue that is resolve by a power cycle to the ecu.

[lagos]

Do you have the multimeter probes (both + and -) connected to the DB37 connector on the MS mainboard itself? They need to be there to eliminate all other sources of error when comparing the meter to what TS displays.

I probed the 12v and ground directly on the circuit board that come in from the jumper wires. For my "on the road" testing, I setup a tee tap on the power wires form the factory harness to be able to check the in car voltage going into the ecu, to rule out any alternator issues. This is a DIYPNP so there is no db37 connector.

[lagos]

I looked at my log again. Seems the higher fueling was caused by barrow correction being added in, and not any interpolation issue, so at least thats reassuring. Any idea on how to get barrow correction to stop giving me false readings for restarts without power cycle?

[billr]

By "factory harness" do you mean flying-leads coming out of the DIYPNP? If so, and that voltage doesn't closely match TS, then I'm convinced the problem is in the MS! I thought the "factory harness" was the car's OEM harness. I like the theory that the 5V is being affected and shifting the reference for the ADC.

[lagos]

By "factory harness" do you mean flying-leads coming out of the DIYPNP? If so, and that voltage doesn't closely match TS, then I'm convinced the problem is in the MS! I thought the "factory harness" was the car's OEM harness. I like the theory that the 5V is being affected and shifting the reference for the ADC.

Currently both the probed voltage and what tunerstudio shows are matching perfectly. The problem is intermittent and has yet to return, but I have a good rig set up so that I can monitor both the supplied voltage and what tunerstudio is reporting for the next time it pops up.

[lagos]

Okay...so this issue happened again..

Here are two logs, but show a votltage of about 14.8 (supplied voltage is 14.1-14.2v). In both logs I drove locally and not on the highway, but I have a feeling if I got on the highway, I would see the voltage rise like id did before. In both logs you can see me power cycle the car. Right after the power cycle the voltage goes to 14.2 and exactly matches the supplied voltage. I decided to disconnect the tps sensor and measure the reference voltage, and I was seeing 4.5v on the 5v line with the engine running, while tuner studio was reporting 14.7v.

https://www.dropbox.com/s/u171fhccxup7p ... .19.31.msl

https://www.dropbox.com/s/rm7kkfb8u5d96 ... .18.02.msl

Here is a video of the difference in voltage readings.
https://www.dropbox.com/s/3jex4vpqjgg6j ... .22.25.mp4

Current battery calibration is set at 0 and 29.5.

[billr]

" I decided to disconnect the tps sensor and measure the reference voltage, and I was seeing 4.5v on the 5v line with the engine running, while tuner studio was reporting 14.7v."

So... you disconnected the TPS and measured the 5V as 4.5, while TS showed 14.7 for battery voltage, what was the raw 12V input at that time?

[lagos]

" I decided to disconnect the tps sensor and measure the reference voltage, and I was seeing 4.5v on the 5v line with the engine running, while tuner studio was reporting 14.7v."

So... you disconnected the TPS and measured the 5V as 4.5, while TS showed 14.7 for battery voltage, what was the raw 12V input at that time?

14.1-14.2

[lagos]

So maybe this is a sign into what is going on... When I had the ecu on the bench with the dc power supply, my 5v line was reading 4.9v and Tunerstudio matched my multimeter at 13.2v.

Now on the car its reading 4.5v at the senor and 14.7v in Tunerstudio (14.3 actual voltage).

So since the difference between the two readings is 4.9-4.5 = 0.4, then the actual voltage and the error is 14.3 + 0.4 = 14.7 (the incorrect voltage).

So is it possible that this is just a battery calibration issue? Since the battery calibration is really a slope between two points, if that was wrong, couldn't that cause me to see much higher reading with just a small drop of the 5v line?

What is the official procedure for setting up the battery calibration, besides just lowering the high number? The documentation on this is almost none existent.

[slow_hemi6]

You need at least 2, preferably 3 or more measured points of, accurate input voltage (actual) Vs indicated voltage. Wider spreads would be better. Then you can use your existing ADC factor to work out what ADC derived from indicated, corresponds to actual. If you have 29.5V corresponding to 1024ADC then the factor would be 0.0288v/ADC. Then do a plot of ADC vs actual voltage to find your 0ADC voltage and your 1024ADC voltage to enter into the calibration. But you have to have a stable and repeatably constant cpu voltage, or you will be wasting your time. Also if after plotting several ADC vs actual V points you don't get something resembling a linear result you will have an issue that can't be well dealt with by calibration.

[lagos]

You need at least 2, preferably 3 or more measured points of, accurate input voltage (actual) Vs indicated voltage. Wider spreads would be better. Then you can use your existing ADC factor to work out what ADC derived from indicated, corresponds to actual. If you have 29.5V corresponding to 1024ADC then the factor would be 0.0288v/ADC. Then do a plot of ADC vs actual voltage to find your 0ADC voltage and your 1024ADC voltage to enter into the calibration. But you have to have a stable and repeatably constant cpu voltage, or you will be wasting your time. Also if after plotting several ADC vs actual V points you don't get something resembling a linear result you will have an issue that can't be well dealt with by calibration.

What about a more experimental way of finding 0ADC? I originally assumed that 0ADC was just my zero reading, and that I could basically measure the voltage at the battery and then measure the voltage at the ecu, and the difference would be my zero reading. Would that work?

[slow_hemi6]

No. Do it how I described.