Ferrari Power Unit Hardware & Software

[Craigy]

Your reading sensor data, and processing it, it's not on/off single bit digital data, it's potentially more complex than recognising 'cats'.

Its surely not. Every sensor is a single dimension only.

A system as the one we are discussing might only want to know when the car is standing still, cornering, accelerating beyond a threshold, braking from above a threshold speed, one or two safety parameters; altogether only a handful of dimensions. Turning switches on or of in an F1 car should be pretty straightforward with the main challenges being integrating the machine learning system with the design/function of the car.

Recognising cats on the other hand, requires juggling millions of dimensions.

Talking about turning switches on or of in an F1 car to change functionality ... It could be anything from simple off/on switch actions, to loading custom software on the go or taking circuits offline to reprogramme them. The AI software could also be hidden like a virus to pop up when required. All these techniques can both change functionality of an engine and mask the method of change without requiring massive computational power.

There's no custom user code that can be installed for execution into the standard controller unit.

It's a shitty Microsoft/Williams Unit using a ancient dual IBM cell cores and a TPU timing processing unit, couple hundred Mhz per core. Two SoC one for powertrain and one for chassis.
The TPU manual is 200+ page document.
It's really handicapping the TJI usage as it's designed before HAJI let alone TJI in shared fuel injector, and prechamber scavenge injector/valve.

The standard F1 ECU is called the TAG320, and it's from McLaren, not Williams or Microsoft. The CPUs are PowerPC's but not from IBM - they come down the Motorola side, and are currently from NXP.

I really do wish people would look stuff up before posting...

[Zynerji]

Im surprised they haven't upgraded those since 2008...

A raspberry pi with an fpga would have more power than those specs...

[Phil]

Anyone else thought about why Ferrari have already released the Spec3? Given we are barely past the half way point of the season, i do wonder why they havent held back for a perhaps more substantial upgrade later around Monza or after?

[GrandAxe]

Your reading sensor data, and processing it, it's not on/off single bit digital data, it's potentially more complex than recognising 'cats'.

Its surely not. Every sensor is a single dimension only.

A system as the one we are discussing might only want to know when the car is standing still, cornering, accelerating beyond a threshold, braking from above a threshold speed, one or two safety parameters; altogether only a handful of dimensions. Turning switches on or of in an F1 car should be pretty straightforward with the main challenges being integrating the machine learning system with the design/function of the car.

Recognising cats on the other hand, requires juggling millions of dimensions.

Talking about turning switches on or of in an F1 car to change functionality ... It could be anything from simple off/on switch actions, to loading custom software on the go or taking circuits offline to reprogramme them. The AI software could also be hidden like a virus to pop up when required. All these techniques can both change functionality of an engine and mask the method of change without requiring massive computational power.

There's no custom user code that can be installed for execution into the standard controller unit.

It's a shitty Microsoft/Williams Unit using a ancient dual IBM cell cores and a TPU timing processing unit, couple hundred Mhz per core. Two SoC one for powertrain and one for chassis.
The TPU manual is 200+ page document.
It's really handicapping the TJI usage as it's designed before HAJI let alone TJI in shared fuel injector, and prechamber scavenge injector/valve.

That's the ECU (Craigy offered some corrections above). However, as well as the batteries, the teams manufacture the controllers for the MGU-K and MGU-H. I believe they can run any algorithms they can in these controllers.

I very much doubt that the power boost comes from any mode of engine firing which is an inflexible area compared to the systems that generate electric power.

[wuzak]

Anyone else thought about why Ferrari have already released the Spec3? Given we are barely past the half way point of the season, i do wonder why they havent held back for a perhaps more substantial upgrade later around Monza or after?

Spa is the next race and Monza is the race after that.

The two races where they would get the biggest benefit from a new PU.

[wuzak]

The crankcase of a formula engine is ran in a partial vacuum, that design system has been used for ages.
The “H” doesn’t deliver anything to the ICE. The only delivery allowed directly to the ICE is by the “K”.

The H cannot be directly connected to the crankshaft. That is correct.

But neither can the ES.

The K cannot provide mechanical power to the crankshaft if does not receive electrical power from somewhere.

Luckily, the rules allow two systems which could provide electrical power to the K.

One is the ES.
The other is the MGUH.

One has restrictions on the direct path (2MJ in/4MJ out).
The other has no such restrictions.

[saviour stivala]

Although it have nothing to do with what I was saying, the other have a 120KW limit out to the crankshaft regardless of what is thrown at it unlimited or not.

[wuzak]

Although it have nothing to do with what I was saying, the other have a 120KW limit out to the crankshaft regardless of what is thrown at it unlimited or not.

Yes, but you have been consistent at claiming the MGUK is limited to 120kW for 33s.

[saviour stivala]

Although it have nothing to do with what I was saying, the other have a 120KW limit out to the crankshaft regardless of what is thrown at it unlimited or not.

Yes, but you have been consistent at claiming the MGUK is limited to 120kW for 33s.

although once again it has nothing to do with I was saying in the first post, Isn’t that correct as per rules/regulations?

[turbof1]

No. The 33.33s was a simplification/error made by the press early in 2014. 33.33s of 120kw is 4MJ. Only, we already established they are able to output more than 4MJ across a lap (although more of a Q lap than anything else).

You are correct to state the peak power output on the crankshaft is limited. However, how long you are allowed to deploy the 120kw is unlimited. Take a hypothetical lap where you are 50s on throttle. If you are able to, you are effectively allowed to deploy 120kw those whole 50s.

[saviour stivala]

No. The 33.33s was a simplification/error made by the press early in 2014. 33.33s of 120kw is 4MJ. Only, we already established they are able to output more than 4MJ across a lap (although more of a Q lap than anything else).

You are correct to state the peak power output on the crankshaft is limited. However, how long you are allowed to deploy the 120kw is unlimited. Take a hypothetical lap where you are 50s on throttle. If you are able to, you are effectively allowed to deploy 120kw those whole 50s.

in my opinion anybody claiming that they are able to output more than 4MJ per lap is out of order. I fully understand your explanation and reasoning and thanks for your troubles, but I don't agree. and also what we are talking about has nothing to do with qualifying modes. this just to eliminate that from the discussion.

[sosic2121]

No. The 33.33s was a simplification/error made by the press early in 2014. 33.33s of 120kw is 4MJ. Only, we already established they are able to output more than 4MJ across a lap (although more of a Q lap than anything else).

You are correct to state the peak power output on the crankshaft is limited. However, how long you are allowed to deploy the 120kw is unlimited. Take a hypothetical lap where you are 50s on throttle. If you are able to, you are effectively allowed to deploy 120kw those whole 50s.

in my opinion anybody claiming that they are able to output more than 4MJ per lap is out of order. I fully understand your explanation and reasoning and thanks for your troubles, but I don't agree. and also what we are talking about has nothing to do with qualifying modes. this just to eliminate that from the discussion.

I assume you haven't seen the video where Cowell says they can deploy all the time in Canada, but they can't in Baku.
He also said that during Q they can deploy whole lap.

[saviour stivala]

No. The 33.33s was a simplification/error made by the press early in 2014. 33.33s of 120kw is 4MJ. Only, we already established they are able to output more than 4MJ across a lap (although more of a Q lap than anything else).

You are correct to state the peak power output on the crankshaft is limited. However, how long you are allowed to deploy the 120kw is unlimited. Take a hypothetical lap where you are 50s on throttle. If you are able to, you are effectively allowed to deploy 120kw those whole 50s.

in my opinion anybody claiming that they are able to output more than 4MJ per lap is out of order. I fully understand your explanation and reasoning and thanks for your troubles, but I don't agree. and also what we are talking about has nothing to do with qualifying modes. this just to eliminate that from the discussion.

I assume you haven't seen the video where Cowell says they can deploy all the time in Canada, but they can't in Baku.
He also said that during Q they can deploy whole lap.

i have no doubt that mercedes can "deploy all the time" (are able too) around a lap, of course senza braking and lifting, but that i have already explianed how its done numbers deployed and all. and by the way, the same Andy Cowel you calling on is the one i was lambasted for qouting on here. it is now emerging that when it suits some it is OK to qoute Andy Cowel but not when it does not.

[sosic2121]

in my opinion anybody claiming that they are able to output more than 4MJ per lap is out of order. I fully understand your explanation and reasoning and thanks for your troubles, but I don't agree. and also what we are talking about has nothing to do with qualifying modes. this just to eliminate that from the discussion.

I assume you haven't seen the video where Cowell says they can deploy all the time in Canada, but they can't in Baku.
He also said that during Q they can deploy whole lap.

i have no doubt that mercedes can "deploy all the time" (are able too) around a lap, of course senza braking and lifting, but that i have already explianed how its done numbers deployed and all. and by the way, the same Andy Cowel you calling on is the one i was lambasted for qouting on here. it is now emerging that when it suits some it is OK to qoute Andy Cowel but not when it does not.

But he is talking exact part of the Baku circuit where they ran out of energy and engine "derated".
https://m.youtube.com/watch?v=DCZsXQ98FCM&t=2s

[henry]

Although it have nothing to do with what I was saying, the other have a 120KW limit out to the crankshaft regardless of what is thrown at it unlimited or not.

Since you enjoy pedantry, what do you make of this part of the regulations:

5.2.2 Energy flows, power and ES state of charge limits are defined in the energy flow diagram shown in Appendix 3 of these regulations.

When the car is on the track a lap will be measured on each successive crossing of the timing line, however, when entering the pits the lap will end, and the next one will begin, at the start of the pit lane (as defined in the F1 Sporting Regulations).

Electrical DC measurements will be used to verify that the energy and power requirements are being respected.

A fixed efficiency correction of 0.95 will be used to monitor the maximum MGU-K power.

The part in bold tells you where they measure the power. It’s not after the MGU-K it’s before it.

The part in italics tells you the maximum power at the measurement point is 126kW. And that means 4MJ will only last 31.7 seconds.

The diagram shows the 120kW on the output of the MGU-K as a design goal, not the solution that is actually implemented.

Now take another look at the diagram and think of how that goal to solution scheme Helps in understanding what is really being done by the engine manufacturers.