Mercedes Power Unit Hardware & Software

[Hoffman900]

I have said it here multiple times, but taking the derivative of velocity to get acceleration would be more insightful when comparing PU’s and deployment.

It is a far more accurate a tool than many would believe if you can get accurate data.
(having a little whinge) I can remember using and suggesting this in earlier times and got criticised by a few knobs but it was over the question "What is the power of the ICE" and nobody knew. At least it gave an indication

I just worked on a project where I used a Savitzky–Golay filter to get smoothed velocity and acceleration data from positional data with some noise in it. It worked exceptionally well!

I’m not sure what the camshaft measuring software uses, but this is an issue when measuring camshaft data. You really need a linear encoder and measurements of multiple points / degree to get anything super useful for Velocity / Acceleration / Jerk plots. Hand measured plots at one degree increments is useless.

A quick and dirty way use to use a polynomial smoothing function in excel. It helps get a sense of the shape of the curve, but useless in terms of absolute values.

[dans79]

It is a far more accurate a tool than many would believe if you can get accurate data.
(having a little whinge) I can remember using and suggesting this in earlier times and got criticised by a few knobs but it was over the question "What is the power of the ICE" and nobody knew. At least it gave an indication

I just worked on a project where I used a Savitzky–Golay filter to get smoothed velocity and acceleration data from positional data with some noise in it. It worked exceptionally well!

I’m not sure what the camshaft measuring software uses, but this is an issue when measuring camshaft data. You really need a linear encoder and measurements of multiple points / degree to get anything super useful. Velocity / Acceleration / Jerk plots. Hand measured plots at one degree increments is useless.

A quick and dirty way use to use a polynomial smoothing function in excel. It helps get a sense of the shape of the curve, but useless in terms of absolute values.

I needed the data "real time", so I went Savitzky–Golay filter at the right side of the window.

If you just want a higher sampling rate, a windowed polynomial fit, that you then sub sample should give very good results. You would need to spend a little time determining the best window, and fit order, but you should get pretty good results.

[johnny comelately]

It is a far more accurate a tool than many would believe if you can get accurate data.
(having a little whinge) I can remember using and suggesting this in earlier times and got criticised by a few knobs but it was over the question "What is the power of the ICE" and nobody knew. At least it gave an indication

I just worked on a project where I used a Savitzky–Golay filter to get smoothed velocity and acceleration data from positional data with some noise in it. It worked exceptionally well!

I’m not sure what the camshaft measuring software uses, but this is an issue when measuring camshaft data. You really need a linear encoder and measurements of multiple points / degree to get anything super useful for Velocity / Acceleration / Jerk plots. Hand measured plots at one degree increments is useless.

A quick and dirty way use to use a polynomial smoothing function in excel. It helps get a sense of the shape of the curve, but useless in terms of absolute values.

The last cam profile I did was using the lathe and a timing wheel and dial indicator and now you tell me it was useless I came to the same conclusion because it was for a Harley

[Hoffman900]

I just worked on a project where I used a Savitzky–Golay filter to get smoothed velocity and acceleration data from positional data with some noise in it. It worked exceptionally well!

I’m not sure what the camshaft measuring software uses, but this is an issue when measuring camshaft data. You really need a linear encoder and measurements of multiple points / degree to get anything super useful for Velocity / Acceleration / Jerk plots. Hand measured plots at one degree increments is useless.

A quick and dirty way use to use a polynomial smoothing function in excel. It helps get a sense of the shape of the curve, but useless in terms of absolute values.

The last cam profile I did was using the lathe and a timing wheel and dial indicator and now you tell me it was useless I came to the same conclusion because it was for a Harley

I mean, it’s fine for displacement and comparing cams that way, but we have had issues with grinders when running a cam over multiple lobes and you can’t see that without measuring it on a camshaft analyzer or better yet an Adcole machine.

[ispano6]

A quote from Hamilton during the US GP race weekend.

"Oh my God, man, jeez,” he said. “I came here after Montreal and I drove last year’s car, and ah, it was so good, and I remember getting out, just beaming from ear to ear. It was so smooth and I had good downforce, good power.

“This year we lost a bit of power with the biofuels and then the car is so stiff, and now coming and driving this car…

I thought Mercedes had made up the power loss from biofuels? Is this something they can't remedy until the next engine formula?

Of course they can't (no one can) and why do you think all power loss has been made up? Since day one of 2022 all evidence has pointed to this not being the case and even in principle E10 has less energy density as opposed to regular fuels. Initial estimates when E10 was mandated in the rules put power loss at 5% at the least without mitigations. 5% is not something you can magically make up, this goes for all manufacturers as all of them are down on HP.

I remember a senior engine guy from honda saying in pre-season he has doubts over a rival engine manufacturer claiming they've regained all power lost from E10 and I believe he was very much right.

Hold on a sec, I'm talking about NET power compared to last year. Just because all manufacturers stood to lose 5% power from E10 doesn't necessarily mean they are down 5% on power from last year. Gains could have been made in other areas of the PU, such as hybrid electric power or changes to the turbo. Asaki-san didn't say he "doubts that rivals regained all power lost", he simply said what he read of other manufacturers seemed to "downplay" the difficulty in regaining the power loss due to E10. From my understanding of what Horner said is that their partners Honda and Mobil had done a good job to make their power unit competitive, likely from the 2021 PU architecture(which didn't actually realize full potential during 2021) and new E10 fuel blend. Marko said with just the E10 he expected a 20HP loss, and that Honda/Asaki were working to ensure the power loss as minimal as possible. Ferrari were reported to have recouped the losses, and Alpine was also said to have an even more powerful PU in 2022 compared to 2021. Based on Hamilton's comments, it seemed that the 2022 Mercedes PU is down on power compared to the 2021.

And looking at the regulations again it does seem that Mercedes is locked in with all component specifications including fuel. So what I was asking is there still an avenue such as modifications for (or in the guise of) reliability or ancillaries that they can employ to overcome the power deficit from last year for 2023-25.

[gruntguru]

5% power loss is very pessimistic. The energy content of E10 is down 3.8% on petrol/gasoline but it has some redeeming qualities too - mostly octane number.

[Zynerji]

It is a far more accurate a tool than many would believe if you can get accurate data.
(having a little whinge) I can remember using and suggesting this in earlier times and got criticised by a few knobs but it was over the question "What is the power of the ICE" and nobody knew. At least it gave an indication

I just worked on a project where I used a Savitzky–Golay filter to get smoothed velocity and acceleration data from positional data with some noise in it. It worked exceptionally well!

I’m not sure what the camshaft measuring software uses, but this is an issue when measuring camshaft data. You really need a linear encoder and measurements of multiple points / degree to get anything super useful for Velocity / Acceleration / Jerk plots. Hand measured plots at one degree increments is useless.

A quick and dirty way use to use a polynomial smoothing function in excel. It helps get a sense of the shape of the curve, but useless in terms of absolute values.

Wouldn't a rotary encoder on both ends allow for a twist offset?

[Hoffman900]

I just worked on a project where I used a Savitzky–Golay filter to get smoothed velocity and acceleration data from positional data with some noise in it. It worked exceptionally well!

I’m not sure what the camshaft measuring software uses, but this is an issue when measuring camshaft data. You really need a linear encoder and measurements of multiple points / degree to get anything super useful for Velocity / Acceleration / Jerk plots. Hand measured plots at one degree increments is useless.

A quick and dirty way use to use a polynomial smoothing function in excel. It helps get a sense of the shape of the curve, but useless in terms of absolute values.

Wouldn't a rotary encoder on both ends allow for a twist offset?

Huh?

Grunt, you may want to revisit this:

[Mansell89]

Forgive the non-technical question….

Is there anyone who thinks Honda have a locked in performance advantage over Mercedes in this engine freeze?

Lewis Hamilton made a reference to “100% deployment” over a lap for Honda which Mercedes don’t have - obviously he is likely dramatising a little in his comments but everything about the Honda looks incredible this year.

It cannot only be drag surely?

Are there any reliability grounds for Mercedes to exploit an up tick in performance? And see engine makers able to still ask fuel providers to keep working on improved fuels? I wondered if this was one way of extracting performance in a freeze?

[djones]

Forgive the non-technical question….

Is there anyone who thinks Honda have a locked in performance advantage over Mercedes in this engine freeze?

Lewis Hamilton made a reference to “100% deployment” over a lap for Honda which Mercedes don’t have - obviously he is likely dramatising a little in his comments but everything about the Honda looks incredible this year.

It cannot only be drag surely?

Are there any reliability grounds for Mercedes to exploit an up tick in performance? And see engine makers able to still ask fuel providers to keep working on improved fuels? I wondered if this was one way of extracting performance in a freeze?

I’m sure there are mild differences and the Honda and Mercedes engines are better/worse at various things.

But for overall performance I’d bet the Mercedes’ is still best. The poor speed is drag and I think that shows when you start to compare other teams using these engines.

[organic]

Forgive the non-technical question….

Is there anyone who thinks Honda have a locked in performance advantage over Mercedes in this engine freeze?

Lewis Hamilton made a reference to “100% deployment” over a lap for Honda which Mercedes don’t have - obviously he is likely dramatising a little in his comments but everything about the Honda looks incredible this year.

It cannot only be drag surely?

Are there any reliability grounds for Mercedes to exploit an up tick in performance? And see engine makers able to still ask fuel providers to keep working on improved fuels? I wondered if this was one way of extracting performance in a freeze?

I’m sure there are mild differences and the Honda and Mercedes engines are better/worse at various things.

But for overall performance I’d bet the Mercedes’ is still best. The poor speed is drag and I think that shows when you start to compare other teams using these engines.

Lewis wouldn't be talking publicly about poor engine performance if the Merc was still the best overall package. He tends to repeat stuff in interviews that he's been told/heard in briefings.

[Big Tea]

Forgive the non-technical question….

Is there anyone who thinks Honda have a locked in performance advantage over Mercedes in this engine freeze?

Lewis Hamilton made a reference to “100% deployment” over a lap for Honda which Mercedes don’t have - obviously he is likely dramatising a little in his comments but everything about the Honda looks incredible this year.

It cannot only be drag surely?

Are there any reliability grounds for Mercedes to exploit an up tick in performance? And see engine makers able to still ask fuel providers to keep working on improved fuels? I wondered if this was one way of extracting performance in a freeze?

I’m sure there are mild differences and the Honda and Mercedes engines are better/worse at various things.

But for overall performance I’d bet the Mercedes’ is still best. The poor speed is drag and I think that shows when you start to compare other teams using these engines.

Lewis wouldn't be talking publicly about poor engine performance if the Merc was still the best overall package. He tends to repeat stuff in interviews that he's been told/heard in briefings.

Things did not fall well for Merc, they were too confident.

They undoubtedly had an engine that was the top of the heap, and 'allowed' the engine regs aimed at equalisation to be adopted, but under estimated just how much ground Honda had made up.

Ferrari seemingly the head of the field was seen to be 'not quite kosher', so we must assume Merc thought they would still be at least equal with anything fielded before the new regs.

Honda seem to have played a bit of a blinder here (as is their job) and initially gone for power over reliability, and promptly sorted out the reliability without having to lose any power, or in all probability, increasing power.

Merc look to have been too complacent with the engine and agreed to things they maybe should not have done so readily.

Having said that, even in the Red Bull I wonder how thing would now stand without the Max Factor? At least less definite, possibly even not champions, looking to be repeat champions.

[saviour stivala]

When the FIA froze the FI power unit design and development from 2022 they did so because they were sure that out of the four power unit manufacturers nobody had a significant advantage.

[PlatinumZealot]

It is a far more accurate a tool than many would believe if you can get accurate data.
(having a little whinge) I can remember using and suggesting this in earlier times and got criticised by a few knobs but it was over the question "What is the power of the ICE" and nobody knew. At least it gave an indication

I just worked on a project where I used a Savitzky–Golay filter to get smoothed velocity and acceleration data from positional data with some noise in it. It worked exceptionally well!

I’m not sure what the camshaft measuring software uses, but this is an issue when measuring camshaft data. You really need a linear encoder and measurements of multiple points / degree to get anything super useful for Velocity / Acceleration / Jerk plots. Hand measured plots at one degree increments is useless.

A quick and dirty way use to use a polynomial smoothing function in excel. It helps get a sense of the shape of the curve, but useless in terms of absolute values.

Very interesting.
The question is.. are the camshaft lobes designed on polynomial curves (whether multiple joined poly-curves) or trial and error?

[Hoffman900]

I just worked on a project where I used a Savitzky–Golay filter to get smoothed velocity and acceleration data from positional data with some noise in it. It worked exceptionally well!

I’m not sure what the camshaft measuring software uses, but this is an issue when measuring camshaft data. You really need a linear encoder and measurements of multiple points / degree to get anything super useful for Velocity / Acceleration / Jerk plots. Hand measured plots at one degree increments is useless.

A quick and dirty way use to use a polynomial smoothing function in excel. It helps get a sense of the shape of the curve, but useless in terms of absolute values.

Very interesting.
The question is.. are the camshaft lobes designed on polynomial curves (whether multiple joined poly-curves) or trial and error?

Use to be polynomials, now it's B-Splines, polynomials, and a few other curve methods etc. I think most of the commercial cam lobe designers are still using Blair's program: http://www.profblairandassociates.com/G ... esign.html , some are using housemade (Jones), other's Lotus, and others Andrews Products.

Harold Brookshire (General Kinetics, Reed Cams, Comp Cams (first designer), Ultradyne, Lunati, Arrington (NASCAR)) back in the 1970s and 1980s use to use a time share computer and brute force his way with polynomials. A few old cam designers use to joke that Cosworth didn't know how to design cams, they were just a base circle and a nose radius with a straight line. I've never looked at them on a Camshaft Analyzer to know, but a bunch of people made careers or became notable fixing their cams.

A strong suit of a graphical design is you can control the acceleration curve at the valve and it integrates back through the valvetrain to a lobe shape.

As for constraints, it's "it depends". It use to be trail and error but now Spintrons allow designers to get very close and then a firing engine gets them that last little bit, usually on the exhaust side with a pushrod engine due to the amount of deflection.