KTM's Claims Honda Exceeded The Moto3 Rev Limit: Substance or Specter?

Were Honda exceeding the Moto3 rev limit in 2015? This is the accusation made by KTM Sports Director Pit Beirer in a story on the German-language website Speedweek. Beirer alleges that KTM came across the issue when talking to riders who switched from Honda to KTM this year, who were complaining of how abruptly the KTM hit the rev limiter. Beirer further claims that KTM were able to look at the data of the Honda Moto3 machine held by a former Honda mechanic. In that data, he alleges, the Honda ran flat out to the 13,500 RPM rev limit, then gradually tailed off to 13,600 RPM.

These claims, if they are true, would be a massive breach of the Moto3 regulations. Though Beirer does not mention Danny Kent by name, the insinuation was that this may have been a factor in a Moto3 title which ended up being decided by just six points.

We spoke to Peter Bom, crew chief to Danny Kent both this year and last, during his successful Moto3 championship campaign, and a key factor in the Englishman's title. Bom denied the allegations, and explained that the claims can only be based on Beirer misinterpreting the facts. The difference between the Honda and KTM Moto3 rev limiter strategies was marked, Honda having invested a large amount of time and money in optimizing both gear change and rev limiter strategies, making the bike as smooth as possible and as easy to ride.

A primer on rev limiters

To understand the issue at stake, some background is needed. In every race series with a rev limiter, factories expend a lot of energy in optimizing the point at which the rev limiter cuts in. It is not a matter of the bike being under full power at (in the case of Moto3) 13,500 RPM, and all power being cut as soon as it hits 13,501 RPM. If that were to happen, the consequences could be very dire. If a rider hit the rev limit while still leaned over in a corner, they could easily be thrown from the bike and injured.

What happens when an engine hits the rev limit is that the ECU detects that the engine is at the rev limit, and decides to cut power. The speed at which it can make that decision depends on the processing power of the ECU. The spec Dell'Orto ECU is a very basic unit, with limited power, and so cannot react quickly enough. The result is that for a split second, the revs exceed the limit, before being cut by the ECU. As the engine speed drops below the rev limit, the ECU is slow again to return the sparks to the ignition, causing it to drop below a little way below the limit before ignition returns. This causes the engine to sputter as it hits the limiter, the ECU alternating between cutting power and returning it again.

KTM vs Honda

This process appears to be at the core of Beirer's accusation. KTM has made a conscious decision to allow full power all the way up to the rev limiter, making it a very hard transition. As the rev limiter is at or around peak power of the engine, this allows the rider to extract the maximum performance from the bike. The downside is that it is much harder for the riders to manage the transition around the rev limiter. The bike is unsettled, and requires more input from the riders.

Honda have chosen a different approach, putting in a vast amount of work aimed at smoothing the transition where the rev limiter cuts in. Their engineers have worked hard at optimizing engine management strategies to make it less abrupt, while retaining as much power as possible, yet giving the riders the feeling they need to shift up.

Feeling the switch

The accusations by Beirer had stung Peter Bom, and prompted him to go through his data, to see what could have triggered the claims by KTM. Before having Danny Kent under his wing in 2015, Bom was crew chief to German rider Luca Gruenwald in 2014, the Kiefer team using KTMs that year.

"It's true that the young riders switching from Honda to KTM are complaining that the KTM rev limiter is a lot more abrupt," Bom said. "That's just a difference in the strategies the two bikes use. Honda worked really hard on the rev limit strategy, putting a lot of work into getting it right. The rev limiter and gear changes were really strong for Honda, because of the work Honda did on using the spec Dell'Orto ECU. The gear changes on the Honda almost feel like a seamless gearbox. That's just all down to getting the strategies right."

Had Honda been cheating? "I really didn't understand the nature of the accusations," Bom said, "so I went back and looked through my data, comparing it to Luca Gruenwald in 2014. It's a bit difficult to make the comparison, Danny was fighting for the championship, Luca was much further down the field. And the rev limit was changed this year, so the Honda was only allowed to rev to 13,500, the 2014 KTM was still allowed to rev to 14,000 RPM."

Where the difference is

The most accurate comparison Bom could find was at Mugello. Both Kent in 2015 and Gruenwald in 2014 had been fighting in a large group during the race, though for different positions. Both had been leading the group at points, in the slipstream at others, and both had been flat out in sixth gear at the end of the straight just as they crest the hill and start to go downhill. "If I look at the data, I can see that when Danny's engine was on the rev limiter, the revs were bouncing around by about 50 RPM. Luca's KTM was bouncing around the limiter a lot more, maybe 100 RPM or so." Perhaps, Bom said, there was something in the way that the data which KTM had seen which led them to interpret it as having gone over the rev limit.

The engine speed data needs to be treated with some caution in terms of accuracy, Bom explained. "Dell'Orto (who make the spec ECU) define the sampling rate for each of the channels in the ECU. I don't know exactly what the sampling rate is for the engine speed, but it feels way too low. 200Hz would be normal, 500Hz would be ideal, but asks a lot from an ECU." The Dell'Orto spec ECU does not have the processing power to spare on that kind intense workout. As a result, the ECU isn't sampling every single revolution when the engine is at the rev limit. At 13,500 RPM, a Moto3 engine is spinning 225 times per second, and if Bom is correct, and the engine speed is being sampled a much less than 200Hz, then the ECU will be having to average out the signal to calculate the engine speed.

Bom denied outright that the Leopard Moto3 team had exceeded the rev limit last year. "100 RPM is not going to make that much of a difference," Bom said. "The rev limit is fixed by Dell'Orto before they hand us the ECU, so we couldn't break it if we wanted to." Bom also said that his data had been checked several times by IRTA, and been found to be within limits. Everything they had done had been perfectly legal.

Investigations continue

MotoGP's Technical Director Danny Aldridge confirmed this. "At every event, the data was taken from the first 3 places, plus 1 or 2 random riders after every qualify and Race," Aldridge told MotoMatters.com. "For example Danny Kent data was taken direct from his machine 20 times in 2015, the most of any rider." Aldridge acknowledged that Pit Beirer had contacted Director of Technology Corrado Cecchinelli and himself a few weeks ago, and IRTA and Dorna were investigating the claims. "Of course we take any alleged cheating very seriously, especially if it comes from a manufacturer. Since the report we have informed Dell'Orto, who in turn are checking the data from 2015."

The process will take some time, Aldridge said. "We are just asking for time to make sure we carry out this investigation correctly," he added. IRTA had not received the information from KTM which the article on Speedweek had alleged they had supplied, Aldridge confirmed. "I am sure you can understand that this is making everything take a lot longer," he said.

Were Honda really exceeding the rev limit by 100 RPM last year, as KTM claim? Was Danny Kent's championship really obtained through fraudulent means? Peter Bom is adamant that it wasn't. The title was earned as a result of the hard work by Danny Kent and his team, and Honda had put in the work to extract the maximum potential from the spec Dell'Orto ECU. One of the reasons the Honda Moto3 bike was so expensive to lease was because of the work HRC had done on the ECU, and making it work for the bike.

The affair is now in the hands of Dorna and IRTA, with Corrado Cecchinelli and Danny Aldridge leading the investigation. They have informed Dell'Orto, who are going through their 2015 data looking for anomalies. There is no timescale of when results are expected, but such serious allegations need to be checked thoroughly and carefully, so it will take some time.


Gathering the background information for long articles such as these is an expensive and time-consuming operation. If you enjoyed this article, please consider supporting MotoMatters.com. You can help by either taking out a subscription, buying the beautiful MotoMatters.com 2016 racing calendar, by making a donation, or by contributing via our GoFundMe page.

Back to top

Comments

Peter Bom told me that 200Hz would be adequate, 500Hz would be ideal.

Total votes: 5

Beautiful geek out you two!

Betting Honda can re-map their bike and dodge this, too late to nail them.

Total votes: 8

Honda are unable to retroactively remap anything. Dell'Orto hold the keys to the data and control the ECU. If Honda had been cheating and the revs were really going that high then it would have been logged and discovered during the season. Also the Honda teams ECU's, wiring loom and data were checked over 50 times during the 2015 season (20+ for Danny Kent alone) by both race control and Dell'Orto.

For MotoGP level Moto3 (as opposed to national level Moto3 which can have different rules) the Dell'Orto DoPe Moto3 spec ECU combined with the homologated software is a very basic package with a low power ARM processing unit. Its sampling rate for RPM is only about 200hz (I have seen sightly different rates in different software revisions). This means the maximum duration that a Moto3 engine can go over 13,500 rpm for, without detection is 0.005 seconds.

The only way too cheat this and for it not to show in the data is a secondary ECU box. The secondary ECU would have to sit between the crank sensor (RPM sensor) and modify the signal data to report an RPM speed that is slower than the real RPM. If the secondary ECU reports an RPM value of 0.7% slower than reality then the main Dell'Orto DoPe ECU would think the engine was turning at 13,500 RPM when it was really turning at 13,600 RPM. While this isn't actually very difficult to achieve, it's almost impossible to hide a second ECU in the wiring loom of a Moto3 bike and for race control to not discover it. Dorna are pretty wise to the secondary ECU trick, they have just spent 3 years defining MotoGP regulations which do allow a secondary ECU for the MotoGP class. The manufacturer specific and unregulated "inertial platform" used in MotoGP is exactly this, a secondary ECU (and sensors) that works in conjunction with the spec Marelli ECU and software.

Total votes: 24

Maybe I'm not understanding how the Dell-Orto CPU works but I would think that something as crucial as crank speed and position sensing would occur on a hardware interrupt and not a scanned channel. The rpm is (should be?) more of a system state variable that is reported by the cpu than a simple channel scanned by the datalogger.

And as mtiberio pointed out, sampling a 225Hz signal at 200Hz is simply not sufficient for accuracy.

Chris

Total votes: 3

For RPM measurement there is just a simple analog output inductive sensor that connects to pin 45 of the ECU. This isn't a crank position sensor as such as it only outputs frequency (RPM). The Dell'Orto DoPe ECU then samples this, via an ADC, at an unspecified rate (about 200hz from what ive seen and from others I have spoken too).

As Peter Bom said (he also thinks the sampling rate is about 200hz), 200hz is a quite low and in an ideal world you would want the sampling rate to be >= (Max RPM / 60) * 2. That's 450hz for 13,500 RPM. However 200hz is enough for it to work without major problems and it helps to keep the hardware to be kept as cheap as possible which was Dorna's primary concern when the ECU was specified.

It all comes down to cost. The 2015 spec Dell'Orto DoPe 2.0 ECU, Dashboard, firmware and software package costs about €5K for a team to buy. 90%+ of this is to cover the cost of Dell'Orto's development, The actual hardware costs when you look at the bill of materials is easily under €500 and probably closer to €400 (not including assembly, testing, etc). When looking at just the ECU hardware (not the dash and ancillaries) then the bill of materials cost drops below €200, even for small volume manufacturing.

Total votes: 11

One inductive sensor and $1 in IC components is all that is needed for a good high speed crank position and rpm sensor. Also, it is not sensing crank rpm, it is counting teeth on the 30 tooth skip tooth timing wheel so the actual signal is 30x crank speed. The Nyquist rate for that signal is 13.5kHz. The dell'orto docs listed show the crank input _sensor_ bandwidth as 20kHz. That is plenty fast to accurately scan the timing wheel. The REVLIMIT functionality in the manual seems to refer to pit lane and TC functionality, not the max rev limit. That would definitely not be in the user software.

With even the cheapest cpu what is usually done with a n-2 tooth timing wheel style crank sensor is that this sensor input line is scanned continuously. When a tooth goes by it is scanned and processed to get a zero crossing point which triggers an interrupt channel to tell the cpu to start counting clock cycles until the next tooth comes along. Even cheap CPUs are in MHz, far above 200Hz. The clock cycle count between teeth gives an accurate time duration which combined with a fixed angular travel of one tooth gives a rpm value accurate to ppm. That rpm value is what the cpu uses to interpolate from the spark and fuel lookup tables. The skip tooth tells the cpu where TDC is since a doubling of the timing count over one tooth cannot physically happen (think halving your engine rpm in less than 1/10 of a revolution). The cpu counts clock cycles (again in the MHz range) from the skip tooth tdc indicator to whatever the lookup table values for the injector and spark signals are and triggers them appropriately.

All of this engine management is happening at the cpu speed of MHz. Then every couple of hundrd Hz it reports the current engine rpm to the datalogger. The datalogger should have no influence on the max rpm limit of the engine.

Chris

Total votes: 6

The only interrupt that comes into play with the RPM strategy is the REV_LIMITER IRQ which, when active, sends a signal to hard inhibit the sparks when the sampling rate gets above 225hz.

Dell'Orto have some pretty decent documentation available for the DoPe ECU that you can download. For those interested in how it works and the type of strategies available for Moto3 teams you can learn a lot from the DoPe 2.0 System Manual.

System Manual PDF:
http://www.dellorto-pe.com/app/download/5663300411/DoPe+Team+Engine+Cont...

Electrical Spec:
http://www.dellorto-pe.com/app/download/8263950585/Dope+Eelctrical+specs...

Total votes: 9

A couple things intersting here where design diffences in motor could have an influence.

First we are talking about revs and sampling rate limitations, perhaps Honda have chosen to use an option in the rules to use a Hall Effect sensor on the Cam, which rotates at half the revs as the crank and can then give twice the resolution for same sampling limitation compared to working off the mandatory inductive crank shaft sensor alone. It might not sound like much but each manufacture must setup a configuration file for their motor and if you can begin to soften your fuel maps a rotation before that next spark than it could make a decernable diffences in the feel when approaching the limiter.

Second if we are just talking about a hundred rpm then this could be consequence of the tire driving the motor when one considers the diameter change in the profile of the tires from crown to edge. Wrapping a string around the crown of a tire and doing the same about an inch in from the edge will give you a 10% differnce in circumference. So if you have any kinks on a straight and lean over the bike, like that Muggello straight mentioned in the article, then the change in diameter will drive up the RPM easly by a hundred RPM or more. In theory the motor revs could go up 10% or 1,350revs more without any added power or sparks firing just by the rear tire change in rolling diameter. In reality a few hundred rpm which momentarily exceed 13,500 and tapper off as bike slows by natural causes like wind and rolling restance should be expected.

If KTM can not see this in their data, but Honda can, than maybe they should consider choosing to sample from the slower spining parts.

Here's an except from the rules:
"Timing Option 1 Crankshaft Pickup only
Crankshaft timing pattern is “n-2” type, where “n” can be between 12 and 30. For optimum performance it is recommended that the first tooth after the missing teeth corresponds to TDC (top dead centre)
Timing Option 2 Crankshaft and Camshaft Pickups
Crankshaft timing wheel has between 12 and 30 teeth, and the camshaft timing pattern is one single tooth"

Total votes: 10

From the downloadable specs at Dellorto; the ECU reads the crackspeedsensor at 1 KHz, the readout provided on the CAN-bus for datalogging is maximised at 100Hz and quote "The proper functioning of the CAN line is not guaranteed"

from specs document: "The maximum sampling frequency is 100Hz. If you set a frequency greater than 100Hz, then this frequency will be saturated to 100Hz."

Total votes: 4