i'll post specs and more pictures soon, probably two-three weeks. but basically it's powered by a 5.7L chevy LS 'crate motor' sitting longitudinally and driving a transaxle. RWD, 400-600hp, 1,600 lbs. target price around $40K for a roller (no engine or gearbox). it'll be sold as a kit or complete roller so if you don't like the LS then pretty much anything would fit in there.
to clarify the 95% thing - what i meant is that dp1 is 95% cheaper than ferrari FXX (i.e. 5% of $2.5M, give or take
these programs are just tools, whichever one enables you to be most effective and works with your way of thinking is the right one to use.
creating one's own car is a very frustrating but also a very rewarding process. i don't know much about your background but if you have an interest in doing that i would encourage you to go for it. it'll likely take years (see www.kimini.com - the guy took 10 years to complete his design) but just the learning experience is well worth it. whatever tools you use is secondary.
it's a very different way of thinking. you need to think more conceptually in sw, i.e. in terms of solid shapes you create. everything starts with a sketch and the key thing is that a sketch basically defines RELATIONSHIPS and SHAPES rather than 'hard data' - it can all be tweaked at any time and all subsequent steps rebuilt with new data. rather than exact dimensions think 'concentric', 'tangent', 'parallel', 'vertical', etc... dimensions are changeable and secondary.
the real trick to using a parametric system is to think of building objects in terms of parametric blocks that relate to each other. i keep hearing autodesk guys call it 'history' but that is fundamentally wrong. it is not 'history' because you can go back and change it at will and that is the essential difference from 'deterministic' programs. i think of the steps used to create something as a 'structure' of the object.
ideally, for a 90 degree FOV your eyes need to be half of screen width (actual width, not diagonal) away from the screen. so for a 40" wide panel like i have, the eyes should be 20" from the screen at 90 deg. that's not quite practical with the present setup (as a background project i'm thinking of how to put something together that makes it work). i'm about 30" away and as a compromise run about an 80 deg FOV which is not perfect but pretty darn good. i like that i have to physically turn my head to look through corners!
with large monitors dropping in price every day there might come a time when i just get a 60" wide panel and all will be good. until then one has to make do
the electric stuff is a separate endeavor, i like to think of it as my 'carbon credits' that way i don't feel guilty about wantonly burning up dinosaur juice on the track, i offset it with doing good elsewhere. there are, of course, plans for greener powerplants for the racecars too but that's a ways down the road. some fun things though
the main issue is that on the prototype the tires are road-legal R compounds which by definition are made for much heavier road cars. i WAS able to heat them up on a kart track where the turns never stop but on a full-size track there are too many straights also, the prototype only has a stock 180 hp 4-cyl hayabusa motor. my engine builder tells me he should be able to get close to 500hp from the 3.0L version of the V8. good to have a few spare horses just in case the weight comes in a pound or two over target.
lesson learned though - this thing has to run on slicks. i don't anticipate any low-heat problems once i put proper racing tires on it. we'll see.
there are lots of factors that affect suspension setup. generally, stiff rebound helps control body pitch/roll at the expense of mechanical grip. ideally the purpose of the shocks is to keep the force at the tyre/pavement interface as constant as possible, both at high and low frequencies. in practice the shocks also have to control the body motion to limit undesirable geometry changes, unless the spring rate is high enough relative to vehicle weight that the body essentially doesn't move. this is how race cars end up with spring rates that may seem ridiculous in terms of traditional 'natural frequency' approach which is comfort-based rather than performance-based.
other factors are things like unsprung weight and sprung/unsprung ratio, geometry specifics, tyre particulars, ad nauseum.
the dp1 has unusually long halfshafts due to wide track and long suspension arms, so they tend to act a little like torsion bars
at 1.5g braking the fronts are indeed twisted 20 deg but it's only 54% yield strength. each shaft is seeing 440 lb-ft of torque at that point. like everything in engineering it's a tradeoff - the shafts are loaded but the suspension arms are not so lots of things can be lighter. and since the shafts see the same load under acceleration they already have to be this strong anyway and yes that means they twist under acceleration too, though it's the rears that do most (72%) of the work in that case.
as long as the tyres stay in contact with the pavement it's actually a bit of an anti-lock feature and works quite well in real life. simulating this accurately can be quite a trick of course.
i've been playing with my LFS setup and after thinking about it for a while realized i had to sit A LOT closer to the screen... the result is quite amazing actually (there is a slightly more detailed writeup on my site http://www.dpcars.net/etc.htm)
yep - see http://www.dpcars.net/atom.htm for the full story on 'project bikini' (just click on page 1 upper right for the beginning). it's taking a lot longer than hoped for but so it goes. i'm awaiting arrival of a wideband O2 sensor to facilitate mapping of the ecu. too bad the rains have started already...
just a perspective: being someone who regularly takes both bikes and cars to the track i can say that riding a bike is a very PHYSICAL experience. that means the body positioning, body weight transfer, the feeling of dragging knee along the pavemet are what it's all about. i would think that this experience would be very diffcult to simulate. none of the arcade bike games even come close. driving a car on the other hand is a 'sitting down' exercise that is pretty well suited to simulation. of course the g forces and 'seat of the pants' feedback are all missing but i've found with a decent ffb steering wheel the experience is quite satisfying. and no, motion simulators don't help - i've tried a bunch, from simple to very fancy, and i've yet to experience one that helps the simulation rather than distracts from it by providing confusing/unrealistic feedback. no feedback is better than bad feedback.
my point? i think that the physics of bikes can be simulated well (although it's a subject complex enough that many of the professional bike race schools don't agree on things as fundamental as how a bike is steered!). but the experience just doesn't lend itself well to simulation, at least not with any controller hardware i've ever seen.
- motion ratio: 1. put a jack under one wheel and raise it 1" 2. take a zip tie and put it on the shock shaft up against shock body. 3. lower the wheel exacly 1". 4. measure distance between zip tie and shock body. that distance is your motion ratio (i.e. how much the shock travels for 1" of wheel travel).
- unsprung mass: (this only works as long as you're using rodends in the suspension and not rubber bushes and the rodends are reasonably free and are not binding up). 1. put the car up on jackstands 2. put corner scales under each wheel (or just one front and one rear) 3. disconnect the shock/spring and a/r bar link completely. the scale will read the approximate unsprung weight for each corner. depending on how tight the rodends are i'd add 5-10% for friction. it may also help to raise the wheel off the scale and put it back down a couple times and take averages.
- brake torque is not difficult if you can measure your deceleration (in g) and know your tire diameter.
the things that are hard:
- CG height. you won't get that without having a spreadsheet or CAD model with every component and its accurate weight and location. measuring from roll angle is very error prone as it requires knowing your exact roll stiffness front and rear, the exact cornering g, the exact longitudinal position of the cg (which is the easiest of the bunch) and to some degree chassis torsional stiffness although on a race car that's pretty minor. you might as well just eyeball it or use 13" for a formula car just because.
- CD, aero centers of pressure, etc. you need a rolling road wind tunnel or a good and accurate CFD model with sophisticated software and someone who REALLY knows what they're doing. eyeballing it is absolutely pointless. even good data acq with motion and force transducers can only get you in the ballpark, depending on how sophisticated your filtering is and how carefully you run the tests. best you can do is guess whether there's downforce or lift and whether center of pressure is forward or aft of CG by how the car feels at high speed vs low speed. even coastdown tests would need to be performed very carefully and the results are not at all obvious since you have to filter out rolling resistance and drivetrain drag. it's far easier to 'reverse engineer' the CD if you know your horsepower and frontal area by looking at your actual top speed and rpm and then calculating what the CD must be to achieve that. you can get to within 5% of the actual number that way if your top speed is high enough where drag is the overwhelming factor, say 150 mph or so. at lower speeds you need to account for mechanical factors more.
perhaps a default setup file per vehicle that is automatically loaded when vehicle is loaded might be handy? not a big deal, would just make it easier and more intuitive. right now if you go from a 4-speed car to a 7-speed car and just try to run the sim without redoing setup or reloading a setup file the results are a bit odd
would there be any harm in setting the minimum at 0 instead of 0.5? if someone actually leaves a gear as 0 it can be skipped when the sim is run. just an idea for another way to solve the problem.
overall, great work! let me know when i can put a link to it on my site.
looks good (hopefully dp1 will be back soon - is there a way to model the center sprag? if not a 30/70 torque split works ok)
one minor thing - how come the picture of the bmw f1 car says 'renault' on it?
another small item - units preference does not propagate to vehicle tools when creating a vehicle. if i set units to imperial and save preferences, vehicle data is still in metric. yes i know i can use the units converter but would be more convenient if the same units applied throughout. not a big deal of course.
also, in gearing setup the text boxes don't seem to accept overdrive (0.xx) selection - the whole thing goes haywire. using sliders works fine though. when switching vehicles gearing setup seems to remain from the previous vehicle and does not appear to switch to vehicle defaults. overall the gearing setup functionality seems a bit unstable (i'm running on vista btw, if that has any bearing on it - a horrible OS that came with my machine and i'm thiiiis close to actually buying a copy of XP and reloading the whole thing).
haven't played around too much with other functions but what i did try seems to work well.
it has same wheelbase as UFR. some of the perceived 'nervousness' of the prototype may be due to alignment and/or aero weirdness, it's too soon to tell. changing front toe from in to out definitey helped. and maybe i just need to get used to it. the fastest i've gone in it so far is 126 mph. i've spun it twice, both at low speed and both under braking. i did manage to catch a number of slides so it is doable. because the driver sits fairly far back in the car it is very easy to tell what the back end is doing. however i would imagine without the seat-of-pants feedback slides/spins would be much harder to catch in LFS because it all happens FAST. if anyone here has ever driven a 125 shifter kart, the dp1 drives VERY similar to that.
so far based on the prototype testing and whatever acceleration sims were done, the v8 dp1 should actually out-accelerate BF1 up to around 80-90 mph simply due to better traction. at over 100 BF1 would pull away definitely. the main advantage of dp1 is the ability to put power down out of corners, it is quite evident even in the prototype. to give an LFS example, if you mash the throttle on medium or low speed corner exit in FO8 or BF1 the back end will want to take the lead. do the same in UFR and the front end will run wide. the dp1 is between the two (for obvious reasons), it just pulls cleanly out. also, mechanical grip should be pretty much unmatched due to very wide track, lots of rubber on the road, very low unsprung weight (small wheels/tires, inboard brakes and shocks). so on tighter tracks dp1 may well be the fastest car out there.
where the car is at a bit of a disadvantage is on long straights and high-speed corners, it tends to be a bit 'nervous' and if a spin does develop it's very difficult to catch. the solution is to always be ready to apply more power, it stabilizes the car. so on longer tracks with high-speed corners and long straights the power and downforce of BF1 would make it faster. possibly the FO8 as well, but that's a close call. well, that's all assuming dp1 ever makes it into LFS
anyway, this is all just theories at this point and everyone is entitled to their own theory
they do, although IMO audio remains something that could be improved... it seems the major challenge is finding actual car sounds? i don't know much about doing sound files but if you let me know exactly what is needed, i could try sending my friend's DAT deck along for a ride in a few cars over the summer. it's a bit unpredictable as to who is running what and when but generally cars are here and i do have varying access to them, though some may take a while to get. if this would help just email me direct with details and wish list...
been playing with the W patch on my newly built system - finally can run it at 1920x1080 and it looks great! getting over 90 fps at that resolution at LOD 1.0 when i don't limit it. whether my car will ever make it into the game or not i don't know (would be great, obviously) but i like the improvements in rendering. as i said before, i'm sure the devs can manage their priorities
btw, after a couple RL trackdays i got a whole lot worse at LFS - not that i ever got good at it to begin with but it is more clear to me now where the differences in skillsets lie. LFS is definitely useful and continues to impress. a friend of mine tried it yesterday and immediately asked for the link to download. he is a hardcore car guy and was involved as a sort of informal 'consultant' in development of both PS and Xbox games a couple years ago (one of his cars was recorded on a dyno by microsoft for the game sound) and he said it was the fist time he saw a game that was reasonably close to driving, as opposed to just being a game.
both a/x and hillclimb would be great competition venues for the car once it's done (h/c especially, what with power/weight and traction), but as development tools they are far from ideal. you get very little tracktime per event and only see the course a couple times so a huge portion of performance is determined by driver's ability to quickly figure out the course and navigate it. track testing on the other hand is repeatable and provides a lot of seat time. the track becomes a non-factor because it's a given and so the testing can focus on how the car behaves. it is also easy to experiment with different lines/strategies from one lap to the next and immediately see the results in datalogs.