Page 2 of 2

Re: [FAQ] OBD Benchmarks

Posted: Mon Jun 16, 2014 12:11 pm
by Doug123_2
Harry, I'm a little confused about your comparison between BT and WiFi dongles. I get that Bluetooth devices are slower than WiFi, but do you mean that WiFi dongles require less energy/are easier to handle than BT, or the other way around?

Doug.

Re: [FAQ] OBD Benchmarks

Posted: Mon Jun 16, 2014 4:59 pm
by Harry
Not easy to answer. In general, BT is slower but requires less energy. However, I saw different statements on the speed side too (OBD Solutions e.g. rates both their BT and wifi dongle to operate at the same speed). The only dongle I tested with both BT and wifi is the Kiwi(1). The difference in update rates is obvious, the reason for this is not: tests where conducted on different platforms / devices (Android and iOS) and the internals of the Kiwi BT and Wifi may differ too. It seems the testing sample is not yet big enough to derive advantages from individual components (wireless technology, manufacturer, OS). For most scenarios, BT will be the solution that is more convenient.

- Harry

Re: [FAQ] OBD Benchmarks

Posted: Thu Jul 03, 2014 10:25 am
by Harry
I ran some additional OBD benchmarks yesterday and updated the master sheet... (see first post in this thread).

So what's new? I have added GoPoint BT1A support to the Android version (will get available soon) and ran benchmarks for both iOS and Android using the BT1A version. I cannot see differences in speed on the iOS side although the BT1A seems to use a later Bluetooth standard than the BT1. The Android version is somewhat slower than the iOS version - not sure about the reason. As I have seen this effect for the OBDLink MX too, I assume the current Android stream handling is less effective.

As testing the dongles using the (pretty fast) ECUsim2000 simulator removes most performance constraints from real life car busses, I have added some benchmarks for the 911 GTS (Porsche 997/2) too. Comparing the (impressive) OBDLink MX wifi results for the ECUSim with that of the 911, it becomes clear that ECUSim results need to be considered a kind if "best case" a dongle can achieve. Effective update rate rates measured by LapTimer are always the result from the complete chain of car bus, OBD gateway, OBD dongle, connection technology, OS stream handling, and LapTimer processing. The MX is obviously not the bottleneck, it is the 911's bus. Anyway, best update rate I have seen for it so far (for this car): 11 to 12 Hz

- Harry

Re: [FAQ] OBD Benchmarks

Posted: Tue Jul 15, 2014 3:22 pm
by Harry
Just added the Bluetooth versions from the OBDLink series: MX and LX Bluetooth. Please note both dongles cannot be used with LapTimer's iOS line of apps as they are not Mfi certified. For the Android side they showed a pretty consistent picture and held an update rate between 12 and 14 Hz for hours.

- Harald

Re: [FAQ] OBD Benchmarks

Posted: Tue Jul 15, 2014 7:28 pm
by foo_fighter
Now that you have an LX/MX...will it be easier to debug dropouts and noisy RPM readings(especially with Porsche cars?) Both seems to show great update rates...but don't seem that way in practice so it could be car specific?

Re: [FAQ] OBD Benchmarks

Posted: Tue Jul 15, 2014 7:37 pm
by Harry
It seems most of the dropouts are due to BT / wifi polling conflicts. The next update will come with some further optimizations here - especially because any BT / wifi discovery activities are stopped while driving with a BT dongle connected already. This is a challenge only apps like LapTimer face. As an example, OBD tools can stop searching for alternatives once they have found their OBD dongle. LapTimer can's as there is probably a GPS or a cam to connect around.

Furthermore, please check the small footnotes in the benchmark sheet. Any tests ran against the simulator (ECUSim2000) will reveal bottlenecks from BT and wifi connections and the dongle itself. In reality, the main bottleneck is the car's bus. This is the reason I have added some samples gathered for a 997/2 Porsche. The MX Wifi for example delivers more than 20 Hz while using the simulator, but delivers 12 Hz when connected to the 997 (which comes with a reasonable fast bus).

- Harry

Re: [FAQ] OBD Benchmarks

Posted: Mon May 04, 2015 5:27 pm
by carramrod
I was checking out posts in the Automatic forums and they are being coy about support for external programs. Do you have any benchmarks from that device yet? Or do you have to wait till they announce external support for that? hah

Re: [FAQ] OBD Benchmarks

Posted: Mon May 04, 2015 5:42 pm
by Harry
I have no relevant benchmarks so far. The new adapter is still in development, so it is too early.

- Harry