Honda Power Unit Hardware & Software

[alexx_88]

Modern ECUs use FPGAs and co-processors to do the time-critical events (injection, ignition for a normal car, probably much much more for an F1 car).

So it's not really the raw speed of the processor that impresses, but how adept it is at triggering the timing events accurately and with very little lag. Trying to do that on standard single-core MCU is a painstakingly hard job. Speaking from experience here

[SameSame]

Yes that was a subjective question, I guess I was looking for a performance metric that could be compared with the order of magnitude of modern personal computing; which you kindly gave

And relating to your efficiency comment, I can imagine these days with limits imposed on flops and time; efficiency becomes a place where you could essentially score CFD time over your competitors.

Modern ECUs use FPGAs and co-processors to do the time-critical events (injection, ignition for a normal car, probably much much more for an F1 car).

So it's not really the raw speed of the processor that impresses, but how adept it is at triggering the timing events accurately and with very little lag. Trying to do that on standard single-core MCU is a painstakingly hard job. Speaking from experience here

I am in the process of using a single core (low clock speed) processor to sample at precise frequencies and it is painful as you mention

The FPGAs I guess would be great for all the low latency parallel timing operations (similar in function in terms of parallel computing to a GPU for software?)?

To get back on topic, maybe because Honda has been out of it for so long they have lost some finesse on the mapping side. But to me it seems as a direct confirmation of the poor correlation between dyno and track. If that correlation was good, their mapping would be close to ideal from the get go.

[SameSame]

Yes that was a subjective question, I guess I was looking for a performance metric that could be compared with the order of magnitude of modern personal computing; which you kindly gave

And relating to your efficiency comment, I can imagine these days with limits imposed on flops and time; efficiency becomes a place where you could essentially score CFD time over your competitors.

Modern ECUs use FPGAs and co-processors to do the time-critical events (injection, ignition for a normal car, probably much much more for an F1 car).

So it's not really the raw speed of the processor that impresses, but how adept it is at triggering the timing events accurately and with very little lag. Trying to do that on standard single-core MCU is a painstakingly hard job. Speaking from experience here

I am in the process of using a single core (low clock speed) processor to sample at precise frequencies and it is painful as you mention

The FPGAs I guess would be great for all the low latency parallel timing operations (similar in function in terms of parallel computing to a GPU for software?)? Edit: The realisation of ladder logic has just hit me and why it is so useful, besides the practical side of it, thank you!

To get back on topic, maybe because Honda has been out of it for so long they have lost some finesse on the mapping side. But to me it seems as a direct confirmation of the poor correlation between dyno and track. If that correlation was good, their mapping would be close to ideal from the get go.

[PlatinumZealot]

How much horsepower do you guys reckon Honda is at now?

[godlameroso]

Impossible to say, they're 20kph slower than Mercedes right before the braking point for turn 3 @ RBR so the gap can be anywhere between 60 to 80kW. 60 if the McLaren has more downforce and drag, 80 if the McLaren is carrying less downforce and drag. My guess is somewhere in between or ~70kW.

This is assuming Q3 mode going all out.

http://en.mclarenf-1.com/index.php?page ... 20Hamilton

According to this the McLaren was on average about 1.7 seconds slower than the Mercedes in the race hardly scientific, but shows more or less the pace gap, as it's similar for Bottas.

The McLaren had pretty much constant pace however, so at least it seems from this one race that they didn't have to do too much fuel saving.

Either that, or the power is probably pretty similar but the Honda PU slowly de-rates the MGU-K towards the end of straight, whereas the Mercedes doesn't.

[GhostF1]

- The development Honda was able to do in first two seasons is similar to the development any other manufacturer would do in... 6 months? due to the severe restrictions wich didn´t allow to solve problems as they show. Traditionally when some manufacturer had some faulty design it was solved in no time, as there were no restrictions about allocations, tokens or costs, so they were able to bring a completely new ICE next GP. For example they could develop two or three different routes for same ICE, and if first was a fail, they could bring the second or third one in next race and solve the problem instantly. But that´s no longer possible because of both tokens and allocations, so Honda was forced to keep their 2015 faulty design for two full seasons, as not even in 2015-16 winter they were allowed to change their architecture due to the limited tokens. What I´m trying to say here is Honda 2017 PU whould have been their 2016 PU, or even some spec 3 or 4 from their 2015 PU, with rules similar to past eras, but they are what they are now so they couldn´t introduce a new PU until 2017.

That was true only for in-season development in 2015 and 2016. So, while I agree that they couldn't have shown spectacular progress in-season (because of the tokens), nothing was stopping them from developing an amazing PU in Sakura and bringing it to the first test of next year. From season to season there were enough tokens to completely change the architecture, which they've done.

I can´t find it now, but I can remind as if it was yesterday Honda people claiming their 2016 PU was only a compromise because they couldn´t change all they need due to the tokens restriction, because they would have need to change everything in the PU, components, layout... and there was not enough tokens for such a dramatic change. They changed a lot of things, true, but not all they needed to change, or in other words, they were stuck with 2015 PU concept and were not allowed to change it until 2017.

I remember it perfectly because to me that was one of the most dissapointing statements I´ve ever read in F1. It was a confirmation 2016 season would be a disaster again even before the winter tests.


btw, it was pretty similar to 2017 season in the aspect Hasegawa said their PU design was risky and it would probably bring unintended reliability problems, but then when those reliability problems showed people started to bash Honda as if they don´t know what they´re doing.... when they warn everyone about that before the season start

I recall what Andres is saying as well.
2016 PU they did whatever they could with the architecture they had currently, knowing it was flawed. 2015's main issue was an undersized compressor, the start of 2015 the idea was clever with their axial flow compressor but it couldn't achieve high enough boost pressures, it's advantage was spool time, which was irrelevant anyway with the MGU-H being able to spin it up. Notice the 2016 PU was much taller, they had to raise it's positioning in the V to allow them to increase the compressor to a reasonable size which compromised the CoG and the packaging. And I recall Honda saying they cannot make all the changes they want due to the token system restricting them as they would have to make extreme changes.

This years PU, FINALLY all new, is still radical, now with the compressor and turbine pushed outwards of the engines footprint, the MGU-H is causing issues now though, and that's likely due to the increased distance (larger connecting shaft), this was chatted about frequently for Merc and when Hamilton had a string of problems with his. Honda also said they had managed to make the engine smaller and lighter (opposite the rest of the fields trend). Go back a lonngggg way in this thread and you'll see a lot of discussion when Honda said it was a smaller engine this year, and speculation on the idea of them utilising a split conrod design, so two rods and pistons utilising the same space on the crank etc was riot.

Couple this with Hasegawa's comments in Canada when Alonso's actual internal combustion engine failed for the first time, he mentioned "it was the first time we had seen an actual issue with the internal combustion engine, so it was quite surprising and likely an anomaly, but so far we are glad with this design's longevity"

That comment pretty much confirmed for me they have something unique going on with the block, crank, rods etc, mixed with the unusual half cylinder low rpm and idle operation.

I'll say it again, I did in Feb this year. There'll be a shining light towards the end of the season but the real potential won't be realised until next years engine.

[alexx_88]

This years PU, FINALLY all new, is still radical, now with the compressor and turbine pushed outwards of the engines footprint, the MGU-H is causing issues now though, and that's likely due to the increased distance (larger connecting shaft), this was chatted about frequently for Merc and when Hamilton had a string of problems with his. Honda also said they had managed to make the engine smaller and lighter (opposite the rest of the fields trend). Go back a lonngggg way in this thread and you'll see a lot of discussion when Honda said it was a smaller engine this year, and speculation on the idea of them utilising a split conrod design, so two rods and pistons utilising the same space on the crank etc was riot.

[...]

I'll say it again, I did in Feb this year. There'll be a shining light towards the end of the season but the real potential won't be realised until next years engine.

I never questioned what you and Andres said in the first paragraph I quoted, but it also means they are essentially 1-2 years behind the top-runners. Also, they've been developing the engine for over 4 years now and they aren't even close to where Mercedes was after 4 years of development.

As I said, it's an iterative process and you can't burn the steps in acquiring the knowledge of building today's Mercedes PU without:

A. Spending the same amount of man-hours as they did developing it. And this is assuming the optimistic scenario where their engineers are as many and as good as the ones in Brixworth and that their dynos and other equipment is just as good. And, from what people reported, neither is true.

B. Directly employing people from Mercedes HPP to get up to speed on what the Germans are doing and move from there.

Nothing that I've seen so far suggests to me that, by using 'A', they can ever catch up under the current engine regs. Both Mercedes and Ferrari seem to continue getting excellent results from PU development. The fact that Honda went into such a bad direction with the PU initially suggests to me that their resource pool is inferior to that of the other manufacturers, a statement backed by all the reports we've heard from inside their camp. Simpler engine dynos, smaller budget and the perspective that the F1 participation is more of a training tool for their young engineers rather than a chance to showcase their engineering prowess.

I'd love to be as optimistic as you guys are, but my bet is that they'll never catch up under the current regs, unless the engine development is frozen for the other manufacturers or they start changing their outset and heavily recruit from the companies that got it right in the first place. And, even if they do that, I think they have no chance of being better than the 4th team before 2019 - and that's assuming they go on a massive hiring spree RIGHT NOW.

[Andres125sx]

Also, they've been developing the engine for over 4 years now and they aren't even close to where Mercedes was after 4 years of development.

Really? What facts are you using to make this sort of comparison?

[alexx_88]

Also, they've been developing the engine for over 4 years now and they aren't even close to where Mercedes was after 4 years of development.

Really? What facts are you using to make this sort of comparison?

Mercedes started development of the new PU in 2012. 4 years of development would put us in the 2016 season as a comparison point. Even Hasegawa admits they haven't matched Mercedes' PU from last year in terms of performance, not to mention reliability which is very far even from the 2014 numbers. And the bigger problems is that they don't just need to match the others' development pace, but surpass it, otherwise they'll always be at least one year behind.

I really want Mclaren to do well, I want Alonso to be able to fight for titles, but that's not a good enough reason to be overly optimistic about things. What have you guys seen that makes you so optimistic about this partnership's future success?

[alexa]

Also, they've been developing the engine for over 4 years now and they aren't even close to where Mercedes was after 4 years of development.

Really? What facts are you using to make this sort of comparison?

For start 2014 Merc PU was much more reliable-hence no penalties and yet they didn't have that opportunity to use races and develop engine using that experience as Honda did over the last 2,5 years.If Honda is at the moment at 2016 power level they had,and we all know that in 2016 they could match 2015 Ferrari engine,it's pretty clear that this project is going nowhere.

[Craigy]

Also, they've been developing the engine for over 4 years now and they aren't even close to where Mercedes was after 4 years of development.

Really? What facts are you using to make this sort of comparison?

Mercedes started development of the new PU in 2012.

According to Luca Montezemelo, Mercedes started on their PU in 2007.
http://www.grandprix247.com/2017/02/01/ ... ince-2007/

When did Honda start again? 2013?

[alexx_88]

According to Luca Montezemelo, Mercedes started on their PU in 2007.
http://www.grandprix247.com/2017/02/01/ ... ince-2007/

When did Honda start again? 2013?

Andy Cowell, head of HPP said they've started in the second half of 2011. I'd rather believe him than what others have heard through the grapevine.

And, even presuming that was true, Ferrari surely didn't start in 2007. So just compare where they were in 2016 vs where Honda are now.

“From the new regulations coming out to the first race, it was ‘let’s all look at that on a Gantt chart, and let’s work out what are the key events’,” Cowell remembered.

“Christmas appears a couple of times in there – this kind of milestones help to tie your work-life in the life you have always known –, track testing before the first race appears in there.

“[And then] we look at it to say ‘How many full V6 iterations can we do in that period of time?’ It is the harmony between simulation and experimentation that enables rapid iterations to be done. So you need to get your simulation and your experimentation correlated well and then you need to be in a situation where you can do as many of those iterations as possible before the first race, before the point you have to commit.

Now compare the approach outlined by Cowell with what was described from the Honda camp.

[alexa]

Really? What facts are you using to make this sort of comparison?

Mercedes started development of the new PU in 2012.

According to Luca Montezemelo, Mercedes started on their PU in 2007.
http://www.grandprix247.com/2017/02/01/ ... ince-2007/

When did Honda start again? 2013?

May 2013.

[Jef Patat]

Didn't see that one coming.
"Sauber Honda engine deal off, claim sources"
https://www.motorsport.com/f1/news/saub ... ff-929575/

[dren]

Didn't see that one coming.
"Sauber Honda engine deal off, claim sources"
https://www.motorsport.com/f1/news/saub ... ff-929575/

I understand the team owners want the team to be competitive, but they have to have operating funds flowing in from somewhere.