PER testing and setup, in need of help please!!!

Hi I am trying to do PEr testing (please see attached jpeg of setup) for 802.15.4 and since it is fairly new and that we do not have the money to purchse a test machine, I am trying to set up a test for PER. We have code where one device is a "transmitter"(defined as the deivce that sends out packets, it also waits to receive an acknowledge from the "Receiver device") and the other is the "Receiver" device(it transmits only an acknowledge back to the "transmitter deivce" when all packets have been received). I have the device that is the receiver in a metal box and have an attenuator that vaires from 0 to 110dB. Is this a valid way to connect the tx port of the "tranmitter" to the attenuator and then to the Receiver. The two devices are being tested conducted adn are about 5 feet apart with the "Receive" device in a metal closure and "tranmitter" device in open air . I did not put any attenuation between the "receiver's" transmit port (to send acknowledge) to the "transmitter" receiver port because it only sends an acknoledge adn not used for calculating PER, or is this my mistake and should I put attenuation here as well The "Receiver device sends an acknowledge after all packets are received back to the "transmitter" device Of course the reciver has code where it spits out the number of missed packets out of how many I decide to send. I have the attenuator set to 110dB and I still get at most 9 packets out of 1000 that are not received which puzzles me and I know that there is osmething wrong but not sure how to set it up propoerly. The specification for 802.15.4 for receive sensitivity is -85dBm with less than 1% PER. Now the bandwidth of the signal is 2MHz so the theoretical noise floor if I did my calculations correctly would be approximately -111dBm.

Are you sure a single metal shield box is enough for the isolation between the radiated path and the conducted path?