RPM lags MAP by 0.4 seconds

[Jim_Blackwood]

Sounds worth trying. Of course I still have that noisy TPS signal to deal with, but at idle just for a test that could possibly be unplugged?

Jim

[jsmcortina]

Sounds worth trying. Of course I still have that noisy TPS signal to deal with, but at idle just for a test that could possibly be unplugged?

Jim

You can't run alpha-N without a TPS !

On one car I was helping tune it had a bad idle oscillation. In the idle portion I adjusted the VE table numbers to counteract MAP changes to give a near constant pulsewidth i.e. values go down as MAP goes up.
PW = Deadtime + (ReqFuel * MAP * VE * others)

Also set your timing to avoid huge changes in the idle oscillation area.

James

[Jim_Blackwood]

You can't run alpha-N without a TPS ! James

Yes of course. But if the throttle does not move, just for the purposes of the test surely a resistor pair would do. However, I do have another TPS I can try plugging in for the test. I think it will also be noisy but I can try it.

I just got back home, but once I've re-mapped the VE again I'll be ready for more testing. I'll do my best to take your suggestions into account, just as soon as I've got it running again.

Jim

[jamies]

'data' out of the ecu is only as good as the data going into it.
resolve your noise issues and setup the TPS first else you will probably find yourself going around in circles.

[Jim_Blackwood]

Yes, you are quite right.
So I'll ask yet again: Is the GM TPS any less noisy than the Ford TPS? Not trying to start a ford vs gm war here but I've had two noisy Ford TPS sensors on the car now, both brand new, and I'm a little tired of it. The noise goes away when I unplug the sensor.

Jim

[turbo conversion]

I have used both sensors with out any issues.

I have had noise if the TPS wires are to close to noise source.

Also if the TPS ground is grounded to any ground source other than the sensor ground.

David

[Jim_Blackwood]

Two posts here, first the TPS then the test of the new VE map in the next post.

In TPS1 below you'll see the typical noise spikes I'm getting on the TPS trace between 6.9 and 8.3 seconds.
The second file is a longer datalog and if you expand it out so you have about 50-80 seconds onscreen you can see the fuzz that the trace has become. Although the major spikes in TPS1 do coincide with the rpm spike I do not think there is any true correlation based on the longer datalog.

First question: Are the spikes small enough that they can be removed with filtering? If so what would be suitable settings? As it is they do cause accel enrichment if I have it turned on.

2nd question: The wires do run in close proximity to the injector wires. Is this likely to be the cause? If so, should they be separated out all the way back to the controller? (I do think they are a twisted trio but unshielded.

Jim

[Jim_Blackwood]

Files attached.
I was able to scale the last VE map by X .4 and get in the ballpark on idle and fast idle at 10.9 req. fuel and 5.45 ms., 4 squirts alternating. My VE at idle is about 28 which seems reasonable.

My timing map is not flat in the affected area (SO obvious from the datalog) so I will address that next. Then the MAP as suggested.

Bill, I am still trying to understand what "multiply MAP" does, are you saying that it scales the map for a user-friendly display, or is there some other purpose?

Jim

[jsmcortina]

I am still trying to understand what "multiply MAP" does

It is a key part of the Speed-Density maths.

The fuel equation is in the manual: http://www.msextra.com/doc/pdf/html/Meg ... 4-104.html

James

[Jim_Blackwood]

The result of flat timing was apparently to reduce the excursions slightly, to a 900rpm-1600rpm oscillation. Not a lot to show for it, but some maybe.

Thanks for the link James. I'm no mathematician but I do get that most of the terms are multipliers, meaning I guess that for the most part they reduce the pulsewidth. So I guess with "Multiply MAP" turned off it just removes *MAP from the equation, right? Leaving just the VE table values for RPM,MAP for the MAP influence. I get that is a table value, and a multiplier, but not rpm factored by MAP. For the moment I won't mess with that, but do you think turning multiply MAP off could have a benefit? Seems it would reduce the influence of MAP without eliminating it entirely.

If I do end up trying Alpha-N, that means creating a new table? Not a big deal.

I'm off to try modifying the VE map based on the MAP trace. More in a bit.

Jim

[turbo conversion]

TPS wiring close to injector wires could be an issue.

Also is the sensor grounding correct, to the MS sensor ground not block or head or firewall etc?

David

[Jim_Blackwood]

I think it is, I went through and redid the sensor grounds, but I'll check. I think I have some twisted pair with shield that might be usable.

Jim

[Jim_Blackwood]

Well, the resolution of the VE map bins isn't as fine as I'd like for trying to tune based on the MAP excursions during oscillations so that limits me unless I want to go back and redefine the bins. Which I could do, if it makes sense to try it.

The oscillation describes an oval on the VE map. Looking at the datalog, MAP is coming down around 1500 rpm between 34 and 27 MAP, and it's going up between 900 and 1200 rpm at about 33-37 MAP. That's not too hard to track in real time on the VE map so I lowered the VE map points in the 15-40 area below 1200 rpm and left it raised it in the 1800 and 2600 area. Got down to about 18-19 in the low area and sloping upwards to 26/27 on the 1800 line.

That got me down to about 1200 rpm before oscillation set in. Not really a significant improvement from my starting point last week, so maybe it's time to take the MAP out of the equation and see if it can be made to idle slower without it.

For right now though I think I'd better get the rest of the map sorted out as the car can't be driven as-is.

Jim

[billr]

Generally, turning "multiply MAP" off does not help, quite the opposite. With multiply off, the VE table values will have to be "faked" to account for density change, it will no longer be a true VE (Volumetric Efficiency) table and values in it will no longer follow much common sense. That's not to say it can't work (turning multiply off), but your tune will be a lot harder to understand or work with. VE values with multiply off will have more drastic swings, you may run out of range with the values that are allowed (0-255?). Got it? There are several good reasons why TS advises that "99% of installs should use multiply on".

[Jim_Blackwood]

I do get it, and it does make sense.
So far the single most consistent factor through all of these efforts though is the persistent behavior of the engine below 1200 rpm. Really, nothing we have done has changed that except to slightly change the amplitude of oscillations and the centerline. It's gone from 400-1800 rpm down to the latest 900-1600 rpm, still beginning at around 1200-1300 rpm. You might say that's an improvement and it sort of is, but off idle tip-in has suffered quite a lot so on average, really not any better, and the steady speed idle has not changed more than MAYBE 100 rpm at the most. Tip-in can be tuned, I know but the point is there really has been no gain.

Oddly enough my last datalog showed almost no noise on the TPS line. More testing there is needed obviously but if I can get a reasonably clean TPS signal I think I might be ready to try Alpha-N.

Jim

[billr]

"Really, nothing we have done has changed that"

I have to be brutal here... What has not been done is probably crippling this effort. Without flowing the injectors for rate and DT, and then using a realistic VE table, you may "wander in the weeds" forever.

[jamies]

do you have access to an oscilloscope?

[Jim_Blackwood]

I do. I have an older Tektronics, and also a pocket digital storage scope. Can't claim to be currently completely competent with either of them but with a little review I've been getting by. What do you have in mind?

Jim

[Jim_Blackwood]

Some good news at least. No more noise on the TPS line.

Now it isn't exactly in accordance with noise suppression guidelines but here's what worked and it might help somebody else. Twisted pair fully braid shielded teflon insulated cable, so very high performance wire. Probably overkill but better than inadequate. Vref and TPS on the pair, and here's the deviation: Shield connected to the TPS ground and on the other end terminated to one of the controller ground leads. No other ground on the shield. (Adding a ground clip lead from the controller end of the shield to the external sensor ground introduced large periodic spikes of 0.3 to 3.2 seconds duration.) Guidelines call for the shield to be connected at only one end to avoid ground loops and use of the shield as an antenna or radiating element, but in this particular case it works, and the signal line is dead flat with no detectable noise. The board itself is grounded to the battery via a dedicated line for sensor grounds as well as power grounds.

I'm still seeing noise on the CLT line and AFR. Those are less critical at this time but CLT will be taken care of, and I have a new O2 sensor for install later. I also will recalibrate the baro once I know exactly that I'm doing it right.

Bill, I greatly respect your experience and opinion so don't take this as confrontational. Just seeking more light.

Req fuel is a multiplier. Pretty straightforward too it seems. Therefore the value just moves the end result up or down. If it is set wrong the VE map scale will be off. But I have a question: If you already know the acceptable range of VE values then can't you use the installed MS-EFI system as an injector flow bench? Set your map and work backwards from there? Adjust the req fuel until the engine runs well with those values, copy them down, and then work the injector flow number to give that req fuel? At the end you should have injector flow rates that match what you bought.

As for the DT, it's a simple adder in the range of 0.5 to 2.0, and most are around 1.0 from my understanding. being off a couple tenths isn't a killer. All it'll really do is throw off your extreme low idle calibration a small amount, meaning the VE map will be a smidge high or low there, and since you'd really prefer to be around 5-6ms at idle, a couple tenths is less than the resolution of the VE map. Is this not so? What I'm asking is this: Wouldn't the individual cells be able to accommodate the difference?

In the end all it is under any condition is more or less fuel.

I'm ready to try Alpha-N.

Jim

[racingmini_mtl]

The board itself is grounded to the battery via a dedicated line for sensor grounds as well as power grounds.

That's not how the sensor ground works. You do not want it to be connected to an external ground. You only connect the sensors to the sensor ground and that connects them to the board reference ground. An external connection to a ground defeats the purpose of using the board reference ground and can easily introduce noise.

Jean