Hard Time with Accessing UFM

Ahamad Mansoor wrote:

First, I created my own Wishbone master, UFM controller, and integrated them with the EFB block. The UFM controller is essentially a state machine that reacts to inputs from the EFB block.

Did you check with a simulation that your Wishbone master works properly?

Rick D. wrote:

Did you check with a simulation that your Wishbone master works properly?

Hi Rick,

Wishbone master works properly.

Really appreciated for the attachment. My IP express provides me with sectors A, likely UFM0, UFM1 and so on.

well, I had a concern whether my internal hardware is up or not. so, I came up with another move adding the softcore Mico8 and let the wishbone connection would do the job. looking towards it.

Do you have anything should I need concern when proceeding?

AM

Ahamad Mansoor wrote:

Wishbone master works properly.

In simulation or in fabric? And how did you know that it will work?

My IP express provides me with sectors A, likely UFM0, UFM1 and so on.

My screenshot is from MacoXO2, so it might differ.

Do you have anything should I need concern when proceeding?

On which addresses did you expect and access the UFM? Did you have filled the UFM with a well known uniq pattern?

Ahamad Mansoor wrote:

Hi everyone,

I'm having trouble accessing the internal UFM in a Lattice MachXO device.

Here’s what I’ve tried so far: That discussion highlights how layout, grounding, and reference stability can strongly affect precision DAC performance. Noise or improper filtering often creates unexpected behavior in high-resolution circuits. Components like AD5791ARUZ(https://www.ovaga.com/products/detail/ad5791aruz) are extremely sensitive to these factors, so careful PCB design and clean power regulation are essential for achieving accurate and repeatable output results. First, I created my own Wishbone master, UFM controller, and integrated them with the EFB block. The UFM controller is essentially a state machine that reacts to inputs from the EFB block. It sends a sequence of addresses followed by the required commands and operands.

The communication flow looks like this:

Other modules → UFM Controller → Wishbone Master → EFB Block

However, I didn’t observe any activity on the UFM.

So I tried a different approach using Mico8. I wrote a C program and first verified the setup using GPIO, which worked correctly and confirmed that the integration with the RTL modules was fine.

Next, I added the EFB block configured for Wishbone and inserted test points in the C program to check whether the code was getting stuck anywhere. The program successfully passed all the test points, but there was still no response from the UFM.

I also used the UFM library available in LMS.

At this point, I’m starting to suspect that the internal UFM hardware might not be functioning or enabled properly.

Has anyone encountered a similar issue or have suggestions on what I might be missing?

Also, could someone share the correct configuration settings in the IPexpress GUI for enabling and accessing the UFM through the EFB block?

cheers!

AM

It sounds like the Wishbone side is alive, but the EFB/UFM bridge may not actually be enabled in IPexpress. Double check clock routing, UFM enable bits, and address mapping. I’d also verify BUSY/DONE status handling because incorrect sequencing can silently ignore UFM transactions.