RomRaider

I bought a used set of Perrin 816's a good while back, and recently sent them off for cleaning and flow testing. There wasn't massive change before vs. after, but I'm still happy I decided to do it.

I chose Witchhunter because they only charge $17 per injector, and the guy was very helpful over the phone. I asked about latency tests (which he can do for another $20 per) but he said with modded stockers they should be virtually identical to the stock values. He went ahead and tested a couple (without charging me the fee - he did it as much for himself as me) and verified his assumptions.

So without further ado, here are the results. The scanner did a horrible job at picking up the grey areas, so I had to go back in with Paint afterward (what a PITA).



Now with all that on the table, what do YOU recommend I use for the scalar and latency values? Thanks for any input.

I'm using ~770cc with the 270090/x (or something like that?) scalar. I use the stock latency values.

I'm pretty happy with these injectors. It looks like Perrin isn't doing this anymore, though. It's gone off their website.

The scalar should be very close to 820. It is very easy to check.

Before installing them, just do a WOT log from 2000 rpm to 4500 rpm. Check AFR with WBO2 at about 4000 rpm. After installing them, do another WOT log from 2000 rpm to 4500 rpm. AFR should be the same at about 4000 rpm. Assuming your load is in the 3+ g/sec range, latency will have very little effect.

For latency, do a similar check. Before installing injectors, find an empty parking lot and drive VERY slowly at constant throttle. Log AFR #1, MAF(g/s), throttle, MAP, IPW direct, AFR correction #1, and AFR correction #2. Install injectors and then do a similar log. Compare trims. I have found this to be more accurate than checking stationary idle trims.

I will be surprised if these have stock latencies... but the proof is in the pudding. The electric latency may be the same but I think there is mechanical latency necessary to increase flow from 0->520 vs 0->820. I'ld like to see your results.

Thanks for the replies fellas.

I assume once the modded-stockers cat got out of the bag, Perrin stopped making money on them and moved on.

I am planning to install these before I make any other changes (bigger turbo on the way), so I should be able to have good before & after comparisons.

I have a set of modded stock injectors installed in my car right now flow tested by Witch Hunter 822/816/816/814. I left the latencies stock in a map that I flashed to the car a week ago and had massive problems at idle. Car was leaning out and adding +25 A/F correction. Once I adjusted the latency I got the car back on the road.

Thanks for your post. Can you share what scalar you're using and what latencies you ended up with?

I've also seen these numbers in a nabisco thread:
http://forums.nasioc.com/forums/showthr ... ?t=1436277

Sorry I have been without the internet for a while. I will post up tomorrow what I am working with right now. They are by no means perfect values but it is a good starting place.

Here's what I have plugged in for right now.



I am attaching a log of idle with the learning from the past couple of weeks. If anyone has any input on what I should bump the values to I am all ears.

Do you have a similar log of your data before installing the injectors?

If you log IPW direct, you can calculate how much you need to change your latency. Right now, I am guessing that you need to lower your latency. Maybe -0.1 ms?

However, it is just a guess. I have found that stationary idle trims can vary by themself. That's why I use my spreadsheet (or you can use Kascade's build 028) to calculate latency or you can use very slow speed driving.

However, you need a reference point of what your data looked like BEFORE installing injectors.

I do not have a log of similar data. I was using Pinks and had a protune with the AP so I never had any reason to log it. I am going to look into using the test release and try to tune it better. Thanks for the advice.

You were extremely close! I used the test release rev033 and came up with -.12 ms yesterday.