- Joined
- Dec 25, 2002
- Location
- mont, AL
This may belong in showroom, but it's definitely got a performance aspect to it. I kinda teased it out a little on the facebook group (mostly because I posted in between meetings and work, partly to draw interest and discussion along the way), so anyone over there already knows how this went down. I'll do the TL;DR up front this time, and show other things as the post develops, as well as provide more information here than FB (most of it can be inferred, but the nature of social media is a quick splash and move on, whereas you can get a little more verbose in a forum setting-which I prefer, personally)
Sam brought his chumpcar/champcar/whatever over weds to do tuning with a new turbo setup and shake down the new clutch setup prior to an upcoming race. the new turbo did well (As was expected, it's not huge, mostly an upgrade on the turbine side, not looking for huge power here), although some overzealous porting on the wastegate led it tappering boost as rpms went up rather than holding. at some point that will be addressed via a new turbine housing, but for the purposes of this, it only bears mention as an artifact of the setup and not so much a direct contributor to the data. boost was set to wastegate pressure (no shens, from the compressor cover to the actuator), roughly 9.5-10psi on the hit, tapering 1-2psi by redline, or so it was with the factory late 940 intercooler fitted on the car.
Once he got done tuning, he wanted to test out the stereotypical ebay intercooler, the old tried and supposedly true 24x12x3 unit that everyone runs because they're awesome and cheap. I will preface everything here and say that I and Sam both had expectations of a roughly 10ish hp increase, maybe a little more. To say that we were surprised by the results is a little bit of an understatement.
We went on to test the garret core intercooler from the gold car (because why not, and readily available), and I'll break that down with the image here in a second:
Pink run: Stock intercooler avg of 3 pulls back to back (little bit of a spread, iirc ~2hp up/down, I can light up the dyno computer later and get all 3)
Dark blue: Ebay intercooler avg of 3 pulls back to back
Orange run: Garret intercooler. That's the high value of 3 pulls back to back, the other two pulls were 193.4 hp, I figured yall wouldn't mind the extra .2hp. I think torque fluctuated a couple ft/lbs across them.
to be fair to the stock intercooler, it did have the roughest time of all three, as it was stuck in where it belongs, and the other two were sitting on top of the core support, like so:
The datalog data bore out some interesting things as well, namely that there was a consistently lower pressure start to finish with the ebay intercooler, indicating a healthy restriction pretty much from start to finish (as can be seen in the overall shape of the graphs, where it starts lower and ends lower. This created a minor controversy on the internets, I'll explain that towards the bottom, but it has to do with the dyno setup)
ebay intercooler:
garret intercooler:
you can see the taper still exists on the garret setup, but it's not quite as bad as the ebay intercooler.
As this was kind of a spur of the moment test, not much thought was given to gathering additional data (eg pressure differential across the various cores, pre and post air temps, etc.). Of course, we didn't expect the results to be what they turned out to be either.
So, the setup:
Car is a b230f with a custom turbo.. I think it's like a gt2871 or 75 or something compressor wheel and a stage3 turbine wheel, .63 a/r back housing, 3 inch exhaust. Microsquirt, manual transmission, RSI Stage3 cam. Pretty basic, not unlike the normal tbricks "turbo intercooler chips exhaust" kind of thing. No magic, no tricks, no excessive cool downs to benefit one over the other (I mean, we expected different results hah)
the dyno was setup with a fixed load amount (I don't recall if it was 5 or 10%, sam can speak to that, but it didn't change), and had a 3 second hold-to-start value, which basically gives you 3 seconds to build boost before the run starts.. so all of the runs benefitted from that, and one of the things it did create was a gap from the beginning as boost was already building. Without having the datalogs in front of me I can't say what the ecu was seeing for boost at the beginning of the runs, only that it started in boost.
This is why the runs don't all start at the same point and diverge from there.. on a more 'normal' run where you hit the go button and mash the gas, you would expect and see that the instant output at the beginning would be more or less the same for all 3.
One thing I noticed across the stock pulls was the early 'bump' in torque from the stock intercooler, and I'd like to see the datalogs but I suspect it spikes a little and drops right away before settling down to a gradual falling boost curve, and it's possible that at that pressure level the stock intercooler is starting to choke a little, but without the pre-intercooler pressures to go along with the manifold pressure it's speculation on my part and certainly not proven fact.
So what's the takeaway here? Do all ebay intercoolers suck? Can I get a second opinion? Why are you cheating the data and fudging the results?
-For the moment, I would say the takeaway is you get what you pay for to a certain extent. Not all ebay intercoolers are the same, not all are bar and plate. If you can get your hands on one to look at before you buy, I would look for one that has larger passages through it rather than smaller, and I would probably look at a tube-and-fin over a bar and plate. You can also go bigger (if it's available), a thicker and taller core will give you more of a chance of not choking.
This is what you're looking at in the average ebay bar/plate (and has been the case for at least 15 years or so):
vs the garret core:
a more direct comparison. the solid bars are where the air flows through the cores
-there are probably better ebay intercoolers floating around, but some of the more name-brand budget options get pretty close and you can buy a known product at that point. Treadstone cores are pretty good, I have one in the blue 242 as well as one in the v8 car, they seem to fight a bit above their weight class. Precision makes cores now (or did..), and garret as well.. garret last time I checked was kind of the gold standard, and certainly on the higher end of the price range.
-you can get as many as you want if you want to do the leg work
We did this experiment quite a while ago (-04 or 05, I forget when exactly) on a buddy's turbo honda.. he had the same ebay intercooler and was turning the wick up and noticed that he just wasn't making nearly the power that was expected for the boost level, and got talked into an intercooler upgrade. The specific numbers are probably lost to time now, but the car went from roughly 450-480hp to like 600 or 620hp, ran out of fuel and fuel injector at the same time. He ended up turning the boost down quite a bit to get the car to a more moderate 500 or 550hp.
-no data changing or results fudging. I'm not in the business of selling intercoolers, I don't plan to get into that business, and honestly we expected a different result. If data were being fudged, we would've turned the boost up and lied about it or something I guess. I may do some more tests with the black wagon, as it has the early style oem intercooler and is otherwise virtually stock except for the boost controller. I guess I'll turn off learning on LH so I can get consistent results from the ecu, that's easy enough though.
Reviewed the runs again while I was in the shop yesterday and realized I'd sandbagged a little (or.. something. whatever). I put up all three runs for each intercooler individually, and then put everything on one graph. The trends are still very clear. The ebay intercooler had an average of 170.5ish, and the stock intercooler had an average of roughly 179 to 180.
All 9 runs for ALL THE GLORY:
Of course I managed to do a real bangup job getting the numbers in the shot :-/ Top three are stock, middle three are ebay, bottom three are garret. If you work right to left you can see the three distinct trends.
stock intercooler runs:
numbers are 180.2, 180.6, 178.9
ebay intercooler runs:
garret intercooler runs:
You can see on one of those runs sam was either a little early getting on the go button, or late getting on the gas after hitting the go button because the torque at the start of sampling was 25-30ft/lbs lower than the other two runs.
That's actually something interesting to talk about as well, because that's the basis of argument for one of the tools on facebook "the runs didn't start from the same spot".. it is almost impossible to start at exactly the same spot, human error and all that, and at the end of the day unless you just skip half of the powerband, it doesn't really change much.
On that run (the green line), it essentially got a slow start, but this only delayed peak boost by what looks like about 100 rpms over the orange line that started higher than the other two, and around 50 from the pink one that started at a lower rpm (And thus a lower boost amount). They all look to have hit peak boost around the 3500-3600 rpm mark, and after that point all of the runs were about as close to identical as you can realistically hope to get.. normally if I'm doing checks to see if a tune is going to back itself up (i.e. I've done a decent job with it) you wanna see a tight grouping, but normally the spread is still a bit more than that.. more along the lines of 1-2hp or so at that power level (it doesn't really get to be much larger of a gap at higher hp levels either, but on something north of 500 hp that's turbocharged, I'd like to see 3 runs back to back that are more or less within 5hp of each other). Turbo cars generally have a little more variability in them than NA cars or blower cars (which behave essentially like big NA cars). Anyway, I'm off track. enjoy the datas, there will likely be more to come.
Last edited: