need advice: boost overcorrect / under shoot with log graph

I got my ALH tuned by Kerma a few weeks ago. Its my first mod, and overall I'm very happy with it. Charlie made me two tunes and my stock tune switchable in VCDS (once my file is loaded on to ECU, I can use the auto, manual, 4-mo soft coding in VCDS to switch).

But I wanted some feedback about my 03 and 011 logs @ WOT. I think they look good overall, no overboosting at all. But I do get a dip (about 200 mbar) after my initial spike even though there is no overboosting. Kerma/ Charlie's recommendation is to lengthen the actuator rod if I want to even it out more.

The first two graphs are from the same tune, and the second two are from the more aggressive tune. Note: For all my logs, the duty cycle never gets above the low 60s (at the initial spike) and is usually in the mid 50s in the 3000-4000rpm range.

What I have done:

• tested all vacuum lines (28 into N75, 2-22 in VCDS n75 output test, confirmed actuator starts moving at 3.5/4 and stops ~18ish)
• sprayed lube on actuator rod, exercised vnt mechanism a bit, couldn't find anything sticking or catching
• checked and found snow screen was already gone
• cleaned MAF, discovered it was the original failure prone Bosch model (not revision C)--it was a used replacement my mechanic had on hand
• intake should be clean as it was done about 10k ago, along with air filter as well

Should I:

• leave it alone, forget about it and enjoy driving it? [It is difficult for me to do VCDS logs except late at night here in Los Angeles].
• lengthen the actuator rod? (Kerma recommends I try this, if I do anything)
• replace my MAF with rev. C just in case?
• something else to check or tweak?

Tune 1, first run (dips to 1978 mbar when specified is 2193).


Tune 1, second run (dips to 1989 mbar when specified is 2193)


Tune 2 (more power), first run (spikes to 2264, dips to 1978, specified 2193)


Tune 2 (more power), second run probably in 4th gear (dips to 1938, specified 2193)

You need to shorten the rod, not lengthen it. Not much though, do half a turn shorter and re-log. And when you log data, only log one group at a time. Resolution is better that way.

Kevinski4 said:

You need to shorten the rod, not lengthen it. Not much though, do half a turn shorter and re-log. And when you log data, only log one group at a time. Resolution is better that way.

Click to expand...

Thanks Kevinski. Can you explain a little more? Shortening increases boost response, so I would expect a bigger initial spike, right? That is fine as long as it quickly settles to specified, but would it potentially undershoot more?

EDIT: Also, would you bother or leave it as is? How much better am I likely to get it?

You have mail

The response on the graph will change each time you make a run. It depends on a host of factors including the load on the engine (like uphill/downhill) initial engine/road speed, final engine/road speed, initial pedal position, final pedal position, how quickly you slap the pedal to the floor, how sticky the vnt is, you get the idea. The same exact test conditions will never be repeated twice. Anyone who tell you differently is either feeding you a line of bull or simply is repeating what they've been told.

There will always be some degree of over/undershoot because that is the nature of this particular feedback governor system. The question is: does it matter?

Lots of variables that you can't control like you do in a laboratory or controlling a valve in a plant with relatively fixed conditions. But the system is meant to be adaptable on the fly and compensates. No two "logs" runs will be 100% the same, and the system is designed around this. It manages the boost within a specified tolerance band, constantly comparing and compensating and predicting and remembering.

If you dampen it too much, responsiveness suffers. If you make it too responsive, then it "rings" or oscillates overboost/underboost. Both conditions will exist to some degree and there have to be design decisions made to adjust the components of the various feedback gains in order to achieve the correct balance. The Physical properties such as spring tension on the actuator and the VNT travel and stickiness come into play here too, along with combustion properties.

But in the end, two things are important here.

1) turbo protection, important not to overspeed it or induce compressor stall/surge. That's where the forum obsession with boost spikes comes in. There were some really poorly put together tunes in the "olden days" that did not handle things such as injectors well. (and one could argue there still are)

2) Driveability. Are any minor variations/deviations intrusive on the driving experience? Logs are in the virtual world,but what matters in the real world is what you feel in the seat of your pants.

I design the tunes so that the driver can't tell if there is a small overboost/underboost, it simply does not matter. It does not translate into SOP. What matters is the boost is controlled enough to meet those two conditions: safety and drivability.

Keep in mind: We don't drive at WOT all the time, or on the dyno. In the real world driving is done at part throttle, even when driving hard.

Charlie, I look forward to trying out the tweaked tune you sent me to see what difference it makes. I mostly agree with your take on this.

I ran lots of logs under varying conditions, different sections of highway, etc. and the over correction was consistent. My question basically was 1) whether it matters, 2) is there anything I'm missing that could help.

You mentioned "combustion properties." I'm curious. Given that I have a high blend of biodiesel in the tank now, would that make any small differences in VCDS logs vs. D2 or a new high-cetane renewable diesel we have in CA?

SFHGolfTDI said:

Thanks Kevinski. Can you explain a little more? Shortening increases boost response, so I would expect a bigger initial spike, right? That is fine as long as it quickly settles to specified, but would it potentially undershoot more?

EDIT: Also, would you bother or leave it as is? How much better am I likely to get it?

Click to expand...

Correct, you'll get a bigger initial spike (right now you technically don't have a spike since it never really gets above requested) then it will settle much closer to requested and stay there. A perfectly set up VNT actuator on a clean turbo with a tune will spike to about ~20psi and then immediately drop to and hold ~18psi. Depending how aggressive the tuner is with the boost control P term. If you shorten it too far, say make it 3 full turns shorter, then yes, you'll get a massive spike, followed by a bad undershoot.

It won't hurt anything to run it as is. You'll just be running a little less power than it's capable of after you stab the throttle until it corrects itself. I'd shorten it a fuzz myself, unless the updated tune does the trick for you.

So I did a little more testing with the vacuum pump today to confirm a few things that I missed.

• With a vacuum gauge hooked up to the out port of the N75 at idle it puts out 10psi from that port. Normal? Within spec? Duty cycle is around 47% at idle. I don't know if idle numbers really mean much, but I've read in some places that you should expect to see at least 18 at idle (should be at full boost at idle?). Last I checked, I had full vacuum of 28 psi at the hose to the vac port on the N75. So I don't think I have vacuum problem before the N75. But why only 10 out at idle? And does it matter if it seems to cycle well from the full boost / no boost positions?
• My N75 output test in VCDS goes between 986-996 (vacuum on/off), which shows about 2/21 on my vacuum gauge.
• I hooked up the vacuum pump to the actuator at the top of the hose to check for leaking / bleed down. I pumped it up to 23psi. It took about 2 minutes to go down to 22 psi. Is this within spec Vacuum hoses are new silicone (~10k miles). Connections seem tight. My tool isn't the best, but only lost a fraction of that amount when I held my finger to it and did the same test. I know since it is a dynamic system, the ECU should constantly adjust the vacuum, so I wouldn't imagine this would be a big deal since it holds vacuum pretty well.

Bump for update. Still haven't had a chance to try out the updated tune. Hopefully tonight. Above are a few more numbers from more vacuum tests I did on Friday.

With all due respect, you're being too obsessive with numbers. All your numbers and tests are good. Your N75 is doing its job correctly (as told by the tuned ECU), and you have good vacuum.

If your actuator starts moving at around 4" Hg (with Mityvac), and hits the max at 18" Hg (sometimes can go up to 20, and even 22 if REALLY needed), then it's set up fine.

Enabled said:

With all due respect, you're being too obsessive with numbers.

Click to expand...

Haha. Sounds good. Helpful reminder.

What were the logs before the tune? How does the car feel when the boost is slightly below request? Mine felt pretty bad. I actually wound up getting the VW tsb on actuator replacement/adjustment and followed that to a T. Runs fine after that. I think it was starting to open at 3 and fully open at 18. Long time ago and my memory might not be exact. I actually think my actuator is starting to leak down a little bit again. ��

Here are graphs on the car stock, pre-tune. You'll notice there is no spike and turbo seems a little laggy? (in line with the actuator being too long?):


And here is my older style MAF:


Performance-wise I can't say that I really feel too much of what is shown in the graph at WOT. But it is hard to say since with the new tunes everything feels awesome in comparison.

Also, updates on a few things. Charlie's updated tunes were a definite improvement, but the dip is still there so it is obviously hardware setup. I went ahead and put a new rev. C MAF in just because I've been suspicious of my old one (which was a used replacement a guru had on hand). I haven't had a chance to log to see if the MAF made any difference. After that I'll either leave it alone or shorten the actuator rod a hair.

One more question about actuator adjustment: adjusting from above, I would turn the locking nut to the right/driver side, and also turn the adjusting sleeve to the right/driver side to shorten? Correct? That is what I can tell from looking at pictures of actuators.