Mesa 7i92 Read Error

I've been getting very intermittent read errors from my Mesa 7i92. By intermittent, I mead less than once an hour. For years, I was running LInuxCNC on a Intel NUC running a not particularly fast Celeron, and I put up with the errors. I figured that the processor was just not up to it. I recently upgraded to a Lenovo ThinkCentre Tiny running an Intel i5 at 2.9GHz. However, I'm still getting the errors. The exact error is:


Of course, this error puts LinuxCNC offline and the lathe loses all its homes. Often, just restarting LinuxCNC won't get things going again, and I have to power down the 7i92 completely and power it back up to get it working.

So, any advice? Is there some sort of software / firmware / hardware upgrade I should try? Is there some setting in the INI / HAL that I should try? Is it possible that my 7i92 is defective?

My system is all up to date, so I'm running the latest versions of Debian, the RT kernel, and LinuxCNC. But I don't think this is the problem since these read errors have been persistent through multiple versions of all the software and different computers.

You really should never get these errors
They mean the the host PC has lost all contact with
the 7I92 even after multiple re-trys (like you pulled the Ethernet cable)

It does sound like a bad 7I92, power supply issue, or cable issue

Peter-

Thanks for the quick reply. It helps a lot to know that I should not get the error, and that it only occurs after multiple retries.

I've been down the cable route before based on an earlier comment (probably by you). It's now a shielded internet cable. I've tried other cables also. I don't think it's the cable.

I also don't think it's the power supply. I have a transformer-type (non-switching) 24VDC power supply, with the 5VDC supply through a buck. That should be fairly reliable.

I do have two Ethernet connections - the onboard connection is set for the 7i92, and a USB Ethernet adapter connects to my home network. I've suspected this for various reasons, but unplugging the home network usb does not seem to eliminate the problem.

So that leaves me with the 7i92. Have you seen this sort of failure before? I'm thinking that replacing it is my next logical step.

I have not seen this type of failure except in initial testing where things like one of the through-hole transformer pins was unsoldered. leading to very weak
Ethernet signals. Sending the card back for repair/replacement is probably the thing to do (ask for a RMA # from Mesa) You might make sure the 5V is OK before this though, a 7I92 resetting because of power issue will act like this.

I did some testing of the power supply. I tried using a bench power supply, a different external power supp0ly, with and without a filter cap on the 5V input and with a 220 uF cap on the input in case I was getting ripple. I looked at the 5V signal with an oscilloscope. I even wondered for a while if having the enclosure open vs sealed had an effect, even though it has fan ventilation.

At first, I thought I was on to something. The card ran over 11 hours without a problem using the bench power supply. However, in a subsequent test, it only ran about 2 minutes before getting the error with the same setup.

Before I pursue the RMA, I've got a question. I don't have a Mesa brand breakout board. I don't know exactly how the error code is generated, or if downstream devices affect it, but is there a possibility that it's a result of the breakout boards glitching?

Thanks for the help. I can't even guess how much time I've put in to this problem over the last couple of years. Since a single test can run an entire day, it's very time consuming to try out different things!

It could be an intermittent problem with the 7I92 or something external
If you are using a standard breakout board is the 7I92 jumpered to _not_ supply 5V to the breakout?

The read error means you have lost communication with the 7I92 (even after multiple retry attempts) , the most common causes of this kind of behaviour are power related

another possibility is real time issues with the host (power management left enabled etc)

I had this effect also while the jumper has been not correct
and i did a ELKO in the 5V supply 470uF

that fixed it for me

I checked the jumpers. I think they're all OK:


I checked, and power management was disabled in the BIOS. This is also the second computer I've used for control, and I had power management disabled in the other one also.

I guess I'm at the point where I'll pull the board.

Looks OK.