560 is possesed

[skeeter]

Hi all. I have a 560 that started doing something rather odd. At times it will make the sound that the battery is being installed then act like it is totally dead. An hour or two later and sometimes much shorter or longer times it will make the sound again and it will work again.

It will do the same while on the docking station, running or even just sitting powered off. For example I can walk up to it a random time it may or may not turn on when I push start. If it does it may run until the batter is low and it docks or it will make the battery inserted noise and just stop dead. I can push it over to the docking station and it will not charge but then after a while it will make the sound again and start charging.

There does not seem to be any rhyme or reason. I have pulled the board and checked the contacts as well as looked for cold solder joints. I don't visually see anything. Wondering if anyone else has seen this and has fixed it without replacing the main board.

Thanks,

Skeeter

[vic7767]

I suspect that your 560 is in the early stage of complete shutdown which is what many 560 owners have been reporting in this forum. So far a solution has not been discovered.

[skeeter]

I am guessing by complete shutdown you mean a catastrophic board or board component failure. I do see many posts where the 560 simply will not power up but have not seen something similar to this where it is random and intermittent, but who knows it may just be another variation.

[Gordon]

I am guessing by complete shutdown you mean a catastrophic board or board component failure.

A single component failing open, or value-drifting far from nominal can stop these 500-series main_PCAs from working. The culprit does that by preventing operation of the +5VREG buck-converter section. If the board lacks the five-volts power-form, the MCU does not boot, so pressing any button does no more than tire the fingers.

I do see many posts where the 560 simply will not power up but have not seen something similar to this where it is random and intermittent, but who knows ...

There is a power resistor on your board that is performing marginally. Its value may be acceptable when cool, but move out of bounds as it dissipates more power. I can't recall reading about that on / off action. My personal experience was a step function -- the test PCA was functional up to one of its auto-time-out shut downs, then only a few minutes later I pressed the Clean switch and there was no response. (I must admit that I spent no time trying to coax it back to working, I merely set about seeing what its trouble could be).

I was able to simulate your report (not the variable part -- yet I could) by disconnecting the power resistor, then apply battery power (no LEDs lit, or sound emitted); then connect the power resistor (to get the buck-converter SMPS going and boot the MCU. When that finished, the MCU sounded the battery-connect musical notes; and I pressed Clean to shift into the 'ready-to-clean' state. Clean-SW LEDs lit, This was done on a free standing 510 PCA, to which only the battery and speaker had been connected.

I have been drafting a trouble-shooting thread for this dead-Roomba topic, but it is days from what I was thinking the OP should contain. If you would tell me what your interest is, so far as digging in to repair your main_PCA (requires SMD resistor R&R), or if you are only interested in knowing whether the fault lies on the main_PCA, and replacing it will get your Roomba going, it would help me figure what I need to tell you -- without prattling on, and on.

[TechGuy]

... If the board lacks the five-volts power-form, the MCU does not boot, so pressing any button does no more than tire the fingers.

As always, you can pin point the problem area. I may have working in software area for too long and forgot the basic hardware troubleshooting skill. Thanks Gordon, I will check my dead 5xx Roomba for 5Vdc when I have time.

...There is a power resistor on your board that is performing marginally. Its value may be acceptable when cool, but move out of bounds as it dissipates more power...

I did a search and find your post on the buck converter. Which power resistor that you are referring to from the SMPS_schemat_3? Please post a larger image of MPS_schemat_3 when you have time. My old eyes cannot see very well on small print. lol

[mig_akira]

Just out of curiosity, all the 5xx series have this problem? Will most of them eventually die of this?
How old is your 560, skeeter?

Thanks!

[Gordon]

..., I will check my dead 5xx Roomba for 5Vdc when I have time.

HINT: If you are working at the robot's top assembly level, remove its Top Cover so you can access the Open Interface MiniDIN jack, then measure TXD voltage between sockets #4 and #6//#7. A fixed (following completion of MCU-boot) +5Vdc should represent quiescent TXD.

...I did a search and find your post on the ...buck converter... Which power resistor that you are referring to from the SMPS_schemat_3?

Look at left, where a serial list shows the current path from VBAT to D47.

Please post a larger image of MPS_schemat_3 when you have time. My old eyes cannot see very well on small print. lol

The quickest thing for you to do is:

a) Click the image on the rr page.
b) Copy that enlarged image to your clipboard.
c) Paste that copy into your favorite photo editor, and save the copy.
d) Use the 'magnify', or 'zoom' capability of that photo editor to enlarge any zone you wish.

If you don't have a favorite image editor, try out a free-editor called IrfanView. It is a bit geeky, but can accomplish many editing tasks.

[Gordon]

Just out of curiosity, all the 5xx series have this problem?

They all make use of the same basic printed circuit board (AFAIK). The basic 510 has many component sites which are not needed, so are left empty on its main_PCA. OTOH, the more expensive 580's PCA will have nearly all of those empty sites occupied. The common feature between those two PCAs is: both make use of the same string of components (five items when battery powered, or three when charging PSU powered) to route power into the +5VREG SMPS. In that regard, all 5XXs are susceptible to any item in that string acting in a manner that starves the SMPS of operating power. While it is possible to point to other single-point failure candidates in that string, the resistor seems, IMHO, to have the greatest potential for failure.

Will most of them eventually die of this?...

We have essentially no evidence that makes it fair to say anything about that. But, I think the problem can be related to attempting to reduce mfg costs by increasing procurement lot sizes and allowing component mfrs to pay less attention to the quality of their products. IOW, if a procurement contract was let for a huge quantity of these specific power resistors, thus lowering the total cost, and giving enough stock to build tens of thousands of 5XX-robots, while also relaxing the resistor's reliability requirements (thus pushing cost even lower) one can visualize a recipe for building latent defects into a huge number of robots.

[skeeter]

Not sure how old the unit is as I have several that were purchased over the past 4 years. It is no older than 4 years and no newer than 2 years. I would like to see if I can repair it. If there is something I can do myself testing wise without an oscilloscope and can be repaired with a standard Weller soldering pen without magnification I am all for it.

[Gordon]

... I would like to see if I can repair it. If there is something I can do myself testing wise without an oscilloscope and can be repaired with a standard Weller soldering pen without magnification I am all for it.

Magnification: These days, I find it a must when working on these iRobot PCAs, but 40 years ago I might have been able to resolve the 0.025-inch high component markers w/o any visual aids.
Soldering Pen: If it is the product I visualize, it might work to R&R R138, but not be able to R&R through-hole mounted diodes. When you order replacement components, plan to add some solder removal aids, such as small width SODER-WIK, or a quality solder sucker.
O-scope: An O-scope should not be needed to detect a faulty R138, a DMM will do that.

I will get you started along the trouble-shooting path, but I want to be brief while doing so. I prefer to not spend time creating instructions tailored for this reply, and subsequent, when I need to do similar work for the dead-Roomba thread which I hope to post in the next couple days.

I think the path you might follow is to:

1) Verify lack of +5VREG on your PCA.
2) If there is no +5V power-form, go inside Roomba to the board level and measure voltage between R138's pads while Roomba is powered into the STBY state, then at the 'ready-to-clean' state (those are my terms, not iRobot's).
3) If the largest dV measured in (2) is less than 0.1V we might have to look for a different villain; but, if a much greater dV was measured, you know you have the faulty component.
4) Assuming R138 appears to be the faulty component, you may actively verify that to be the case by by-passing R138 with a similar valued resistor. When by-passed, the +5VREG SMPS should start up, then its 5-volt power will boot the MCU, and the MCU will sound the battery connect song. Then, if you press the Clean switch, its green LEDs, two each will light.

Now, I expect you will require elaboration of at least three of those list elements.

Test-1: Follow the hint that I gave to TechGuy in an earlier reply. IOW, you can verify whether the five-volts power form exists by measuring it at the Open Interface (SCI) jack sockets.

Test-2: You need to know where R138 is located on the PCB; you might want to know what I mean by "STBY state"; and/or 'ready-to-clean' state. I'll define the two states (and similar terms I use) first, then tackle the component location.

a) When Roomba has no power source connected to it, I say the robot is in its "OFF Mode". If a battery is connected to the robot, its MCU boots and ends up in what I call the "Standby State", abbreviated as "STBY". No UIP LEDs are lit, but the robot is standing by for some user input (e.g., button push, Remote-Control code, connection of charging PSU, or scheduled start).
b) If the user briefly depresses the Clean button, status changes from STBY to 'ready-to-clean' state, and the MCU turns ON the Clean button's two green LEDs (and starts an inactivity time-out monitor).
c) If the applied power source is Roomba's Charging PSU (whether a battery had been powering the robot, or not), the MCU will detect that alternate power status and launch its Charging Mode routine.
d) Component "addresses": On the 5XX PCA I superimpose a Cartesian Coordinate System with its origin centered on the Clean-switch dome. Its x-axis runs // to the main-brush spindle axes, and plus-x occupies the right half plane. The plus-y axis points along Roomba's forward travel direction. Each component that I study, and make note of is tagged with its coordinates of position in millimeter units. For example resistor R138's address is "R138(-16,100)" to body center (+/- a couple mm). In addition, it is handy to know whether a component is mounted to the upper face of the PCB, or to the bottom face. One end of R138 connects to diode D47, which is mounted under the PCB. When I need to know (in the future, and w/o looking at the PCA) the component is bottom side mounted I will prefix its address with a lower case "b", this way: "bD47(-3,109)". Note that coordinates of bottom face components are viewed X-ray fashion.

Test-3: (no elaboration).
Test-4: If you have an assortment of resistors to select a test-bypass resistance, choose one in the range one to 50 ohms and 1/4W or more. If you have no assortment of resistors, your only option might be to briefly short-circuit R138 while performing this trial. I you are concerned about doing that, we can talk about it.

Hopefully the above info will get you going.

[skeeter]

Gordon, I have a DMM, solder wick and a good soldering removal tool. I will go through these later today. Thanks for the guidance.

-Skeet

[TechGuy]

...when I need to do similar work for the dead-Roomba thread which I hope to post in the next couple days.

Both of my dead Roombas have 5Vdc at the SCI port.
I am looking forward to read your dead-Roomba information...

[Gordon]

...Both of my dead Roombas have 5Vdc at the SCI port.

I have some ideas about this condition, but you could be of great help in reducing the extent of my guessing what may be wrong by replying with descriptions of what you can see, and or hear those dead Roombas do.
Let us say, you remove and replace the battery. Does the battery connect song get sounded?

Next, starting with the robot in STBY state, you then depress and release the Clean button. Do the green LEDs under the Clean button light up?

If yes on that one, what happens if you again depress the Clean button?

If you then plug in a powered charging-PSU, to the side jack, is there any LED action under the Clean button?
I reckon you have monitored the OI-jack for charging-data output, and got nothing(?).

You may guess, I am trying to judge whether those dead Roomba MCUs are doing anything. The MCU itself is a single-point failure, but we expect it to have greater Hi-Rel than two- and three-terminal components.

You might get a clue about MCU-life by setting up your scope to measure TXD on the OI (SCI) jack as the MCU boots. a) Remove the battery. b) Set up your scope probe on the two jack sockets. c) Push the battery into contact and hold it for a few seconds to capture a brief 'hello world', 'I'm listening" sort of transmission. Looks like this:

The violet trace is the TXD signal. Once that ~ 11ms of binary data (high density bit pattern) goes quiet TXD remains HI at +5V (unless, I presume, RXD says something).
That may tell you that the MCU completed its boot and has started 'housekeeping' tasks.

If you don't see that TXD output you might measure the clock-xtal's (X1) signal with the scope. On the 510_PCA I measured a tight 8.0Mhz, clean sinusoid having quite a low VPP amplitude (it seems too low, but works). Remember, a couple members have had to replace X1 in their 5XX robot.

Add any other info that you can acquire from those two bad boys.

I am looking forward to read your dead-Roomba information...

I worked on it most of today, and reached the point where I will detail the R138 fault-finding trail. I had not thought about writing in the OP my early guesses about what may be wrong IF the 5V-SMPS is working OK!

[TechGuy]

Gordon, I have an old Heartkit o-scope which does not have storage function. For a brief, non-repeating wave form, my scope is useless.

But I was able to capture text coming out of SCI port white it was booting up.

=====================================
Working Roomba 560 booting information
=====================================

bl-start
STR730
bootloader id: #x4709564D ADB97FFF
bootloader info rev: #xF000
bootloader rev: #x0001
2007-05-14-1715-L
Roomba by iRobot!
str730
2007-08-08-1111-L
battery-current-zero 259

2007-08-08-1111-L
ADB9 7FFF
assembly: 3.5
revision: 0
flash version: 8
flash info crc passed: 1

battery-current-zero 260

Please note the information on the last line - "battery-current-zero 260" For the non-working Roombas, both indicated "battery-current-zero 0".

battery-current-zero 0
saving bbox vars
languages: english (0)
bbox vars saved!

I think bbox stands for Roomba blackbox in nvram. Roomba saves the problem data in nvram for diagnostic purpose. It is a shame that iRobot does not service their robot after their engineers put in data to locate the problem area quickly. It would be helpful if we can get our hand on the blackbox program.

I don't want to hijack this thread. I will create two separate threads to discuss the problem with my Roombas.

[Gordon]

Gordon, I have an old Heartkit o-scope which does not have storage function. ...

If you were to ignore the desire to capture any displayed data and instead just watch the CRT as you connect the battery, then seeing the trace rise to +5V, followed by a short bit-packet flash would be satisfactory.

Of course, doing so is now moot, since you have recorded a degree of proper MCU activity piped out of the OI-jack!

... I was able to capture text coming out of SCI port white it was booting up.

What was powering that boot? Battery? Or, Charging-PSU?

...Please note the information on the last line - "battery-current-zero 260" For the non-working Roombas, both indicated "battery-current-zero 0".

If you do that same test on a functional 5XX, what does that battery-current record display?

battery-current-zero 0
saving bbox vars
languages: english (0)
bbox vars saved!

So, what happened next? Did Roomba simply abort, and not even light its Status-LED RED?

What about answers to my other questions?

I don't want to hijack this thread. I will create two separate threads to discuss the problem with my Roombas.

Well, that will be fine (however, I would not think we are high-jacking since we are staying on topic).

At this point: It appears the MCU is able to boot, but as it begins to sample its environment it finds fault with system current and quietly gives up (in an unknown manner). Obviously, current is not zero, since tens of (minus, if battery powered) mA are operating the MCU, its power-supply, and probably the remote-control IRM. I would take that as a clue to verify: 1) whether analog current data is being properly developed and 2) do those data correctly reach an MCU port?

[vic7767]

I have a working 510 PCAB and when a 15 volt power supply is connected it will indicate .07 amp at 15 vdc at powerup it will turn on the two Clean LEDs and display the folowing SCI data:

bl-start

STR730

bootloader id: #x470B5659 AD0BEFFF
bootloader info rev: #xF000
bootloader rev: #x0001

2007-05-14-1715-L
Roomba by iRobot!
str730
2007-08-27-1705-L
battery-current-zero 259

2007-08-27-1705-L
AD0B EFFF
assembly: 3.0
revision: 0
flash version: 2
flash info crc passed: 1
battery-current-zero 259


Then when the Create cabe is connected to a 535 that has no display and no power at the SCI the power supply indicates 0 amps at 15 vdc.

[matofill]

What is one interesting clue myself exactly once rescued a half-dead plate 530 points.

I did the baking motherboards, because usually you may destroy the bad joints.

What does not work anyway even more difficult to destroy ............. link:

http://www.overclockers.com/forums/show ... p?t=606658

[TechGuy]

Of course, doing so is now moot, since you have recorded a degree of proper MCU activity piped out of the OI-jack!

... I was able to capture text coming out of SCI port white it was booting up.

What was powering that boot? Battery? Or, Charging-PSU?

I was using battery without power supply on all of my tests.

Sorry, I was interrupted earlier before I had a chance to answer the other questions you have in new thread. Since you don't think I am hijacking, I will put them on this thread.

================================
Good Roomba 560 booting normally
================================

bl-start
STR730
bootloader id: #x4709564D ADB97FFF
bootloader info rev: #xF000
bootloader rev: #x0001
2007-05-14-1715-L
Roomba by iRobot!
str730
2007-08-08-1111-L
battery-current-zero 259

2007-08-08-1111-L
ADB9 7FFF
assembly: 3.5
revision: 0
flash version: 8
flash info crc passed: 1

battery-current-zero 260 <---- It detected current.

==================
Roomba 531 Problem
==================
I got this Roomba one year ago. I forgot what was the problem until this morning after I put them together again.

Roomba 531 Symptom:
1. It does not sing the battery installed song.
2. The clean button lights up green briefly and then flashing red. It said "Please charge your battery" when it begin flashing red (I think). But the battery is fully charged and is the same one that I used to test the good 560 above. I think the current sensing circuit is faulty on this one.
3. Pushing clean, spot or dock buttons do nothing. It continued to flashing red.
4. When I put in the battery without connecting the SCI cable, it does not have the green/red clean light and voice described in 2.

bl-start
STR730
bootloader id: #x4716524D E6ECFFFF
bootloader info rev: #xF000
bootloader rev: #x0001
2007-05-14-1715-L
bl-start
STR730
bootloader id: #x4716524D E6ECFFFF
bootloader info rev: #xF000
bootloader rev: #x0001
2007-05-14-1715-L
bl-start
STR730
bootloader id: #x4716524D E6ECFFFF
bootloader info rev: #xF000
bootloader rev: #x0001
2007-05-14-1715-L
bl-start
STR730
bootloader id: #x4716524D E6ECFFFF
bootloader info rev: #xF000
bootloader rev: #x0001
2007-05-14-1715-L
Roomba by iRobot!
str730
2010-06-08-1628-L
battery-current-zero 0
battery-current-zero-not-in-specs

2010-06-08-1628-L
r3_robot/tags/release-3.1.1:4159 CLEAN

bootloader id: 4716 524D E6EC FFFF
assembly: 3.3
revision: 0
flash version: 10
flash info crc passed: 1

battery-current-zero 0 <--- It did not detected any current.
saving bbox vars
languages: english (0)
bbox vars saved!

When I charge it from the side, the current is flowing backward and clean light is flashing yellow. I think that it is the charging MOSFETs problem. I am going to locate and check the MOSFETs. Replace if necessary and report back.

Without connecting the SCI cable this morning, I will not get the "Please charge Roomba" message. It was dead and does not do anything. Thanks for asking me to measure the TXD output waveform. Otherwise, I am not thinking of connecting the SCI to see what is coming out of SCI port.

SCI cable is a great debugging tools. I am finding more ways to use the cable!

bat: min 3 sec 10 mV 2588 mA -164 tenths-deg-C 835 mAH 0 state 2
bat: min 3 sec 11 mV 2588 mA -164 tenths-deg-C 835 mAH 0 state 2
bat: min 3 sec 12 mV 2588 mA -164 tenths-deg-C 835 mAH 0 state 2
bat: min 3 sec 13 mV 2588 mA -164 tenths-deg-C 835 mAH 0 state 2
bat: min 3 sec 14 mV 2588 mA -164 tenths-deg-C 835 mAH 0 state 2
bat: min 3 sec 15 mV 2588 mA -164 tenths-deg-C 835 mAH 0 state 2

==========================
Roomba 540 - Water Damaged
==========================
- The owner told me that the left wheel started to turn during cleaning this Roomba with water. He left the wheel going until it stopped.
- I found residue on the main board.

Roomba 531 Symptom:
1. I put in a good battery. The left wheel continues to rotate. It does not sing the battery installed song.
2. I changed one power transistor. The left wheel rotates a few seconds and stop. It does not sing the battery installed song.
3. It does nothing when pressing clean button. It resets when I hold spot and dock together. But it does not sing the system reset song. The left wheel rotates for a few seconds and stop.

bl-start
STR730
bootloader id: #x470D5F60 B01D9FFF
bootloader info rev: #xF000
bootloader rev: #x0001
2007-05-14-1715-L
Roomba by iRobot!
str730
2008-01-14-1241-L
battery-current-zero 0
battery-current-zero-not-in-specs

2008-01-14-1241-L
bootloader id: 470D 5F60 B01D 9FFF
assembly: 3.4
revision: 0
flash version: 8
flash info crc passed: 1

battery-current-zero 0 <---- It does not detect current
saving bbox vars
bbox vars saved!

This water damaged 540 has other damages. Although the SCI messages indicate the charging FET is okay, but the voltage and current are both zero during charge.

bl-start <---- after spot/dock reset with power supply plugged in at the side
STR730
bootloader id: #x470D5F60 B01D9FFF
bootloader info rev: #xF000
bootloader rev: #x0001
2007-05-14-1715-L
Roomba by iRobot!
str730
2008-01-14-1241-L
battery-current-zero 0
battery-current-zero-not-in-specs

2008-01-14-1241-L
bootloader id: 470D 5F60 B01D 9FFF
assembly: 3.4
revision: 0
flash version: 8
flash info crc passed: 1

battery-current-zero 0
saving bbox vars

processor-sleep
charger-wakeup
slept for 0 minutes 1 seconds

2008-01-14-1241-L
bootloader id: 470D 5F60 B01D 9FFF
assembly: 3.4
revision: 0
flash version: 8
flash info crc passed: 1

do-charging-checking-fets @ minutes 0
bat: min 0 sec 13 mV 0 mA 0 tenths-deg-C 993 mAH 0
Charging FET test passed.
do-charging-wait-initial @ minutes 0
bat: min 0 sec 14 mV 0 mA 0 tenths-deg-C 993 mAH 0
bat: min 0 sec 15 mV 0 mA 0 tenths-deg-C 993 mAH 0
bat: min 0 sec 16 mV 0 mA 0 tenths-deg-C 993 mAH 0
bat: min 0 sec 17 mV 0 mA 0 tenths-deg-C 993 mAH 0
bat: min 0 sec 18 mV 0 mA 0 tenths-deg-C 993 mAH 0
bbox vars saved!
bat: min 0 sec 19 mV 0 mA 0 tenths-deg-C 993 mAH 0
bat: min 0 sec 20 mV 0 mA 0 tenths-deg-C 993 mAH 0
bat: min 0 sec 21 mV 0 mA 0 tenths-deg-C 993 mAH 0
bat: min 0 sec 22 mV 0 mA 0 tenths-deg-C 993 mAH 0
bat: min 0 sec 23 mV 0 mA 0 tenths-deg-C 993 mAH 0
bat: min 0 sec 24 mV 0 mA 0 tenths-deg-C 993 mAH 0
bat: min 0 sec 25 mV 0 mA 0 tenths-deg-C 993 mAH 0
bat: min 0 sec 26 mV 0 mA 0 tenths-deg-C 993 mAH 0
bat: min 0 sec 27 mV 0 mA 0 tenths-deg-C 993 mAH 0
bat: min 0 sec 28 mV 0 mA 0 tenths-deg-C 993 mAH 0
.
.
.
bat: min 75 sec 29 mV 0 mA 0 tenths-deg-C 993 mAH 0 <--- After 75 minutes, still 0V, 0A and 0mAH
bat: min 75 sec 30 mV 0 mA 0 tenths-deg-C 993 mAH 0
bat: min 75 sec 31 mV 0 mA 0 tenths-deg-C 993 mAH 0

[TechGuy]

What is one interesting clue myself exactly once rescued a half-dead plate 530 points.

I did the baking motherboards, because usually you may destroy the bad joints.

What does not work anyway even more difficult to destroy ............. link:

http://www.overclockers.com/forums/show ... p?t=606658

Interesting idea to re-flow the solder joints. Special care to set the temperature and amount of heating time in order to protect the plastic parts, SCI port and connectors, on the 5xx main board. These parts survive the manufacturing heat and they will survive the re-flow heat... I hope!

[Gordon]

I have a working 510 PCAB and when a 15 volt power supply is connected it will indicate .07 amp at 15 vdc at powerup it will turn on the two Clean LEDs and display the folowing SCI data:...
battery-current-zero 259 ...

Thanks for that info vic. I gather that it is now the belief that a "259" or "260" at the end of that "battery-current-zero" record means battery current is not zero! Are you able to convert that number to an actual current level?

Now, regarding your power up process on that 510, I did the same thing with my 560 test_PCA: 15Vdc bench-PSU clipped onto the PCA's spring contacts powered up the board, its connect-song played, and I (DMM in series) monitored about 5 mA of current (along with periodic current burps) while in the STBY state. After a short closure of the Clean switch, its GRN LEDs lit up and current was around 90 mA (with periodic higher pulses) in the 'ready-to-clean' state.

I had a hunch that playing of the connect song might be battery-voltage sensitive, so I repeated the above process after reducing PSU voltage to 14Vdc. The connect song did play, and other comments remain about the same.

Again, I repeated the above process after reducing PSU voltage down to 13Vdc. This time the connect song did NOT play. Other comments remain about the same, except I noticed input current for the 'ready-to-clean' state was then around 104mA.

I suspect that if I had dropped input voltage another volt, to 12Vdc, I might have heard the "Please charge Roomba's battery" message.

Then when the Create cabe is connected to a 535 that has no display and no power at the SCI the power supply indicates 0 amps at 15 vdc.

Don't you suppose that 535_PCA's MCU has no Vcc voltage?