Nexus-H7 FD. Fast grblHAL controller for professional SMT placement.

430 views
Skip to first unread message

Alex

unread,
Apr 3, 2026, 1:56:09 PMApr 3
to OpenPnP

Nexus-H7 FD. High-Performance Motion Controller.

The Nexus-H7 FD is an advanced motion controller based on the STM32H723 (550 MHz), designed specifically for professional SMT placement machines (OpenPnP) and high-speed CNC systems running grblHAL.

 Key Features

* 0.8 MHz Step/Dir: Stable pulse generation on the X/Y axes.
* Dual Isolated CAN FD: Two independent buses for distributed control (X/Y Servo + Tower/Feeders).
* Isolated RS-485: Dedicated interface for smart feeders (Nexus-Feeder) via Modbus/Custom protocol. Industrial Grade: Full galvanic isolation of power drivers (UCC21520), CAN and RS-485 interfaces.

Alex

unread,
Apr 12, 2026, 3:10:13 AMApr 12
to OpenPnP
Nexus-H7_FD.jpg
NEXUS_TOWER .jpg
пятница, 3 апреля 2026 г. в 20:56:09 UTC+3, Alex:

guo feng

unread,
Apr 13, 2026, 11:28:32 PMApr 13
to OpenPnP
good job

Alex

unread,
May 2, 2026, 12:38:00 AMMay 2
to OpenPnP
Prepared the Nexus-Drive-FD board.Drive FD.png

вторник, 14 апреля 2026 г. в 06:28:32 UTC+3, gfen...@gmail.com:

Alex

unread,
May 16, 2026, 12:21:37 AMMay 16
to OpenPnP
Added the Sense board to the project – an intelligent sensor hub.   https://github.com/AcaA-sys/Nexus-Sense-FDSense.png

суббота, 2 мая 2026 г. в 07:38:00 UTC+3, Alex:

guo feng

unread,
May 19, 2026, 11:54:37 PMMay 19
to OpenPnP
Hello, is this project currently open source? I haven't seen the stencil creation files. If possible, I'd like to create my own stencil for verification.
tks!

Alex

unread,
May 20, 2026, 3:29:46 AMMay 20
to OpenPnP
Hello ,
To clear things up: the Nexus hardware ecosystem is a custom development, and we do not provide ready-made fabrication Gerber or stencil files for replication.
Furthermore, using Step/Dir control (even at 0.8 MHz via grblHAL) completely defeats the purpose of the Nexus-H7 FD architecture for an SMT installer. Standard pulse/direction signals cannot leverage the industrial-grade dynamics required for professional pick-and-place performance.
The entire Nexus project is designed around a distributed architecture:
  1. Master Controller (Nexus-H7 FD) offloads real-time physics and communicates via high-speed CAN FD.
  2. Axis Control must be handled by intelligent FOC servo controllers (like Nexus-Drive-FD) combined with Sin/Cos encoders.
  3. Peripheral Nodes (valves, vacuums, steppers) are managed via dedicated slave nodes like Nexus-Tower-G4 and Nexus-Sense-FD.
If you want to achieve professional-grade placement, you should look into the complete ecosystem rather than trying to adapt a single board for legacy Step/Dir servo drives. You can review the project topology and public documentation here: github.com/AcaA-sys.
Best regards,
Alex

среда, 20 мая 2026 г. в 06:54:37 UTC+3, gfen...@gmail.com:

simpl...@tuta.io

unread,
May 20, 2026, 7:46:31 AMMay 20
to Openpnp, OpenPnP
"Standard pulse/direction signals cannot leverage the industrial-grade dynamics required for professional pick-and-place performance"

I think that's nonsense.

The use of pulse train signals (Step/Dir) is still the gold standard on a great many industrial machines in the lower price range. My self-developed motion controllers (also closed source) deliver clock rates of several MHz across 18 axes simultaneously, although, of course, only 7-segment ramps are used for full jerk control. 

But I wouldn't dare to constantly promote untested, unfinished closed-source hardware or software in an open-source forum.

Best regards,
SM

dc42

unread,
Jun 10, 2026, 4:12:00 AMJun 10
to OpenPnP
The problem with performing high speed motion using step-dir commands is that the controller has no idea what is coming next. If the interval between a pulse and the previous one is smaller than the previous interval, this could mean that the drive needs to accelerate, or just that the controller doesn't time its step pulses very accurately and there is some jitter in the interval even at a steady speed. For open loop controllers, this is probably immaterial. It matters more for closed loop controllers, because they typically use velocity and acceleration feedforward to improve performance, and knowing how the acceleration is changing allows them to optimise the feedforward.
Reply all
Reply to author
Forward
0 new messages