Debug RS232 command latency in GBL2660T

Hello,

We have reason to suspect there is something weird happening with command latency in GBL2660T, using two channels for controlling separate motors. 

My question relates to the processing and latency time in the GBL2660T and how it works. 

How many bytes/commands is the controller capable of processing at once? 

For example, if I want to update the command to both channels, can I send one big concatenated command like this: [!][G][ ][1][ ][1][0][0][\r][!][G][ ][2][ ][1][0][0][\r] ? 
Will all of those commands/bytes be processed without latency? Or is there a problem with sending one right after the other? Will there be an issue? 

We are sending these commands approximately every 4 ms (250 Hz). 

We are running the GBL2660T controller in open-loop mode, but using an external microcontroller to do closed loop control of our system. In this case, we need extremely low latency to have good performance. 

We have another system that is using custom BLDC drivers that we made, and there is basically no latency (i.e. we have direct PWM control of the FETs). In this system, the closed loop is working great. In the larger system with Roboteq GBL2660T, the system behaves as if there is significant (on the order of 10-20 ms) latency. I'm trying to debug what might be the cause of this latency, and if there are any RS232 strategies that are better than others for controlling two motors simultaneously. 

Please advise with any comments or questions to help figure this out. 

Thanks.