AM3358: EtherCAT Master PLC

Hello everyone,
I'm Luca Campedelli electronic technician at the textile company Vibemac, see https://www.vibemac.com/, we produce sewing machines for jeans and we export all over the world. In the last few years we have made custom electronic boards, similar to small PLCs, for motor and I / O control. The current card drives two motors in step / dir from 0 KHz to 100 KHz and has an incremental encoder input (A, B, Z). The two motors can be independently driven, they can follow a master encoder, work in cam, in electronic gearing, etc ... A typical application in the field of sewing is cam movement following the master encoder of the sewing needle. The slave motor remains stationary while the needle remains in the fabric, when the needle comes out of the fabric the slave motor performs a movement that corresponds to the length of the sewing stitch, this happens very quickly, at 5000 rpm a stitch is performed in 12ms, the useful time to move and about 120 degrees, that is, in 4ms the slave motor must accelerate and stop before the next point, all following the needle axis which varies the speed from 0 to 5000 rpm following the analog signal of the machine pedal. Usually, in order to be able to move well with the fabric, the cam is instantaneously anticipated with a factor related to the speed of the needle motor. This example to give an idea of the speed required.
The new electronic board should have at least double the performance, 4 step / dir motor outputs and two absolute encoder inputs, but, given the diffusion of fast protocols, IEEE 1588, we are considering implementing the EtherCat master in our board, for this reason we thought of using Sitara processors. We need help to define the best hardware for us, such as demo boards to buy and help in software development, perhaps with already tested libraries, etc. .. Unfortunately there are only three of us in the electronic office and we have very short time to create the prototype, your products are very complete and powerful, but we don't have much experience in using them. We would love to use them, but to do it we need, if possible, a lot of your help. This new card with EtherCat should manage 14/16 axes simultaneously with a cycle time of 125us, the Trio Flex-6 Nano reference product. We have performed EtherCat tests with the black beaglebone, AM3358, Xenomai and Code Composer Studio, but we are very far from the required performance, with the resources alone our forces the development of the new card would be very long. In addition, the new CPU must have ModBus TCP, manage I / O modules and remote connection. It is also very useful to give the customer the opportunity to update the software easily and without the use of the PC and network connections, in the previous electronic board it can be updated with a simple file on USB drive, very easy and fast.
The expected consumption is about 1000 pieces per year, which recommended solutions? What demo boards are needed? What software?

Il Customer support center TI mi ha consigliato la famiglia ARM574X per soddisfare le nostre esigenze, the product is certainly very fast and more than adequate for fast industrial control, but the AM574X are expensive for us, we thought of the AM335X or AM437X family, are they enough? Very important for us is the software development time to implement the EtherCat bus and all the management part of the step / dir motors. To drive the four motors by impulse (Step / Dir) we thought to use the Pulse Train Output (PTO) function, can it work? Can you control the four engines at the same time? Are there working examples for using EtherCat? Libraries? Are there any other fees to be incurred such as Acontis EtherCAT Master Stack (www.acontis.com/.../ecmaster.html) If it can help, we can pay for development courses, but we have to be quick. You can give us suggestions of how to move, what to buy, etc ... sorry, but we don't know the product, thanks.

I'm sorry for my English,
Thanks.