Amazon UWA6 - Ports 801 & 803

At Bastian Solutions, we continually strive to provide superb customer service at the highest standard, and not only reactive but proactive support. Transparency is a key component in helping achieve this objective and with that in mind, we are providing a full root cause analysis for the disruptions in service that began on Sunday December 8th, 2025 around 3:00 am CT described below.

Summary:

Sunday, December 8th

2:59 AM CT
Site reached out regarding Multiport 801, specifically Port19 experiencing consistent [233] M1_REFERENCE_TOO_FAR_FROM_POS errors. Site confirmed that all available troubleshooting was complete including multiple part swaps (motors and index sensors). Confirmed with site that I would pull logs and forward them to AutoStore for next steps, with follow up expected in the morning. 

3:24 AM CT
Site (Schylar) confirmed the following steps completed:

Absolutely, Luke and I (Schylar) have swapped out fork senso on front cart, we have checked levels of both rear and front cart belts, tried replacing front cart motor again, didn't fix issue, put old one back as it was new last week, replaced rear cart motor, it had same issue as front cart motor last week, encoder cable in motor was out of place, and that didn't fix issue, we also replaced slider motor, tested it and didn't fix issue, going to be putting old slider motor back in. Checked belt tensions, checked belts for any damage, checked all cables/wires and encoders and found no damage and made sure they were all secure in their spots. 

3:28 AM CT

Provided image of belt level check to site.

7:37 AM CT

Site (Erik) followed up regarding the Port:

1. Port 19 which had its front motor replaced last week, when it was discovered the encoder terminal block on the circuit board was ripped out of the motor, did not have the flange installed on the motor pulley to prevent belt walking that exists on Chilled Fusion Ports only. See attached image. 
2. It was also mentioned/discovered by RME this morning that the rear cart motor also had the encoder cable ripped out of it causing similar damage and needing replaced. 
I (Erik) have just completed troubleshooting with RME and showing them the technical information in console again that can be used to determine the issue. After checking various cart positions I saw that the rail index sensor never changed from “disconnected” status unlike that of Port 20 its neighbor. 
We are performing that index sensor change now.  

7:46 AM CT

Index sensor swap completed and states indicated correctly according to technical info.

8:36 AM CT

Motor swap resulted in [212] MOTOR_ENCODER_NOT_OK leading to thoughts of a bad encoder cable. Additional errors became present during component swaps leading to the discovery of a bent pin on the encoder/power cable. 

2:06 PM CT

AutoStore provided feedback based on events from 12/1 and 12/8 / 12/8:

For Dec 1:
From the 233 errors, it is clear that either encoder is malfunctioning or belt is climbing on pully. SENSOR_CART_HARD_CONNECTION was found during these errors. SENSOR_CART_HARD_CONNECTION means that carts are physically connected. While cart gap (estimated from m1 and m2 encoder positions) is increasing up to almost 10cm. (Either encoder or belt.)
Front cart bumps into Rear cart when driving backwards. In this particular scenario rear cart is standing still and Front cart moves towards it. And when we compare cart gap and fork sensor reading again, we can see that physical impact happens before encoders say impact should have happened, again either encoder or belt climb.

Dec 7/8:
Again 233 errors look similar with encoder estimated gap drifting from physical gap reported by fork sensor. Either way, it really looks like an encoder issue, but if they have swapped both front cart and rear cart motor, and all following cables, then the next step is the driver box.

3:18 PM CT

Bridge call shared with Bastian from Amazon to continue troubleshooting.

Began preparing to get new parts ordered and getting quotes + lead times.

4:46 PM CT

Alex, with Bastian, brought up to speed to take over. Next steps were to swap the Driver Box due to suspected damage from damaged encoder/power cable.

This swap was performed and firmware was attempted to be updated due to [1104] INVALID_DRIVER_UNIT_VERSION error. Programming through console was unsuccessful and ST-Link was partially successful. 

Tuesday, December 9th

12:39 AM CT

Call was disbanded after unsuccessful attempts to program the Driver Box, call would resume at 8:00 AM CT.

7:57 AM CT

Chime call information shared, and discovered that power supply for Driver Box was non-functional. Replaced the power supply and attempted to connect to the Driver Box to confirm configuration through the web-portal and an ops laptop. 

10:55 AM CT

Was able to get the Driver Box to communicate through console using the old SD card and updated Port CPUs through the console. Began getting the same [233] M1_REFERENCE_TOO_FAR_FROM_POS error again. 

11:33 AM CT

Swapped slider motor as a sanity check from Port 20 however this did not change anything. Carts are able to initialize, but when brough close together the error would happen (front cart running into rear cart, or rear cart running into front cart).

2:07 PM CT

Pieced back together Port 20 to begin testing components before install into Port 19. Components operated as expected. The dump files were pulled from both SD cards and sent to AutoStore for review.

4:39 PM CT

Call disbanded and will reconvene on 12/10 pending analysis of the DUMP files by AutoStore. 

10:11 PM CT

Multiport 803, specifically Port 16, began experiencing the same error following a PM.

"RIGHT after they swapped a belt (rode up on rear pulley as improperly tracked) and front cart motor (burned out due to extra resistance from belt riding up) which is the EXACT thing done to Port 19 on 801 before that was down for 2 days"

10:52 PM CT

Amazon RME has requested Bastian send someone on-site due to the “systemic issues” being seen to aid in resolving the issues they are having with what they identify as “systemic issues”:
1. Belts walking and shredding on the rear pulleys (which is not occurring at UNY4 and I believe the root cause to be that the motors are not being correctly rotated when belt is tensioned and that the tension is being changed during non-intrusive weekly PM's meant for cleaning and visual inspection only). 

2. Fusion Port issues with the motors failing and needing swapped out (most of these are similarly related to the cart belts). 

Requesting Bastian quote the cost for a technician/knowledgeable individual to come out and look at the Fusion Ports. 

Wednesday, December 10th

8:08 AM CT

Analysis from AutoStore development provided indicating issues with front motor/front motor encoder due to pulses being "lost" or falsely generated pulses on the rear motor/rear motor encoder. Concerns called out over potential damage as seen with other encoder cables.

8:37 AM CT

Chime call information shared with Bastian and we forwarded to AutoStore.

2:13 PM CT

With all troubleshooting efforts exhausted over multiple calls, Amazon and Bastian determined next steps are for an on-site visit to deep dive into the Port. Quote shared and travel booked for arrival on site the next day.

Thursday, December 11th

12:00 PM PT

Erik (Amazon) and I arrived on site and began visual inspection of the Ports shortly after. This included checking all cable connections, belts, and motor installation. 

12:46 PM PT

Upon powering up Multiport 803 and performing cart moves identified that the front cart on Port 16 was moving further than the front cart on Port 15. This led to the identification of an incorrectly installed front cart motor drive belt. This was clear when compared to belts on Port 15. Identified the same to be true of Port 19. 

This was also identified by AutoStore overnight as a potential cause based on images/videos shared and a height difference identified between belts compared to the retention bracket. 

1:53 PM PT

RME completed the reinstall of the belt on Port 16 and Erik and I confirmed operation of the port shortly after. We turned the Port back over to operations and no additional issues were identified.

2:42 PM PT

RME completed the reinstall of the belt on Port 19. Site also had to piece back together Port 20 as parts were used from this Port for validation.

Once powered back on and handed over to ops [304] SLIDER_POSITION error began showing up on heavier bins within Port 20. It was found that the carrier for the slider motor was never tightened, so the belt was loose. Erik tensioned the belt and tightened the motor carrier. The Port operated as expected starting around 4:10 PM CT.

4:49 PM PT

Erik (Amazon) and I determined the issue resolved with both ports operating as expected and the error no longer occurring.

Root Cause:

Fusion Port 801

During PMs of Fusion Port 801, site identified that the retention flange was not re-installed on the front cart motor installed on 12/1 as RC for the same error. This flange missing resulted in the belt walking. When the belt was re-installed for the front cart motor it was installed backwards resulting in slipping which caused the [233] M1_REFERENCE_TOO_FAR_FROM_POS error. 

Fusion Port 803

Site found that the belt was riding up on the rear pully causing additional resistance on the front cart motor, requiring replacement. The belt was also replaced at this time as it was found damaged and installed backwards resulting in slipping which caused the [233] M1_REFERENCE_TOO_FAR_FROM_POS error.

Resolution:

RME re-installed the belt on both ports with the groves in the correct direction.

Preventative Measures:

AutoStore to update the Service Manual to include a visual check of belt grove orientation post install. Target completion of January 2026.

Amazon RME to reiterate current service manual procedure when attaching the new belt to the old belt with the zip tie:

"Make sure the belt teeth are pointing in the same direction."

Amazon RME will also cease unnecessary belt tensioning during regular PMs. Per AutoStore belts should only be re-tensioned once they fall below the threshold for re tensioning:

The threshold value for Front Cart belt re tensioning is 25 Hz

The threshold value for Rear Cart belt re tensioning is 15 Hz.

Bastian to retain reference images from the occurrence for training purposes.

Our commitment to you:

Bastian Solutions understands the impact of the disruption that occurred and affected operations for your organization. It is our primary objective in providing our clients with superb customer service and we assure you we're taking the required preventative measures to prevent reoccurrence.