I'm current using a TiWi01-R2, with the WL1271 module connected to the AM/DM37x Evaluation Module from TI/Mistral. I'm using the TiWi01-R2 to connect to an Apple iPad2 via bluetooth/rfcomm. Once the connection is made, I'm seeing approximately 750 kilobits/second between the LSR device and the iPad when they are at small distances, using large packet size (64kb). Smaller packet sizes reduce throughput, and larger packet sizes are not supported by the Apple hardware. Given that all the hardware involved claims to support BT2.1 + EDR, my expectation is that I should see much closer to the theoretical value of 2.1 megabits/second. Is the TiWi01-R2 capable of achieving more than 750kb/sec when connected to Apple hardware?As a comparison, I created an rfcomm connection between the TiWi01-R2 device and a USB Bluetooth 2.0EDR device on a desktop linux machine. The bluetooth specification claims that the practical maximum throughput for a bluetooth 2.0 EDR link is about 2.1 megabits. My setup yielded about 1.4 megabits. My sender and receiver were right next to one another. I know that the USB dongle is capable of ~2 megabits when connected to another linux desktop with another USB BT module.In conclusion, my two questions:1) Why is the device yielding such a low value for throughput when connected to an iPad2 (approximately 1/3 of practical spec bandwidth)2) Why is the device faster but still much slower than practical maximum when connected to a desktop PC? (approximately 2/3 of the practical spec bandwidth)My goal here is to determine whether the TiWi01-R2 module is capable of more (meaning I should continue trying to find where my slowdown is) or if this is the practical maximum for this part.
Hi David,
In order to analyze it we would need to have a BT sniffer to see what happens on the air (retries, frame rate and so on) however normally that kind of equipment is expensive and hard to get.
Can you send me those exact commands that you used on the EVM for the throughput measurement?
Regards,
Eyal
Unfortunately, I don't think that I can share the code that I'm using to generate the bandwidth measurement, due to the Apple NDA in force. I could share the code for a similar test I ran between the dev board and a PC. That connection ran faster (1.4mbps), but still nowhere near the advertised 2.1mbps.
I am using the BT_Demo.sh script to pair and connect the BT devices, then custom code to create a connection and push data for measurement. Should I paste that code into a post here?
We have special HW that we used to evaluate the 127x chips and were able to pass 2.1Mbps one direction, but i am not sure how is the SW you are using is working but probably us used L2CAP layer or above that has overhead as far as traffic information (not all bits are data) and in that case 1.4Mbps is very good number.