Roomba 520 flashing orange (power disconnected ..)

[bolero1000]

Hi,

Tried to search for a similar behaviour (also on the service manual), no joy ..

It's the 2nd time that my roomba 520 is flashing orange (like when is charging), but power is NOT connected ...

Don't remember the exact steps last time.
This time I simply unplugged the roomba from the charger (I don't use the base), the Clean button stayed green/on (normally is going off once unplugged), no response pushing Clean (more than 10 sec., it should go off), did a reset (spot + dock), reset music and Clean flashed Orange for 1-2 min.
Then Clean went off and roomba starts to work normally (I could switch it on & now it's cleaning ...).
Well, almost normally ... after 2 or 3 minutes (for this room it works normally > 25 minutes) it stops, no beeps/error messages, Clean button off, I could restart it again (for 2 or 3 minutes).

Last time, to reset it I had to unplug the battery (but perhaps I simply didn't wait enough time ...)

Anyone has an idea of what does it means this flashing orange ??

This evening I'll try the selftests.

Thanks & Ciaoooooooo,
Luciano

[vic7767]

It may be time to measure the battery voltage after a full charge and then if the voltage is between 16 and 17 volts perform a load test on it. How old is this battery ?

[bolero1000]

Hi Vic,

It may be time to measure the battery voltage after a full charge and then if the voltage is between 16 and 17 volts perform a load test on it. How old is this battery ?

Battery should be new (~ 1 month), it's a replacement unit I received for a "broken" roomba green. It was (& hopefully still is) working for 1 hour +

I don't have a multimeter (& BTW I don't know how to use it ..).

Update on the tests, this morning at the 3rd try roomba worked for at least 15 minutes (then I had to go to work), trying again tomorrow.

Tried the selftest as per service manual, not so clear to me how to run them & read the BUILT-IN-TEST CHART.

Tried first the AUTO-ADVANCE MODE, quite confusing to me, seems to stop (Clean RED) to test 14 ...so tried the MANUAL-ADVANCE MODE instead.
I was able to run all test successfully, but I'm not 100% sure I did the tests correctly or read correctly the results .. Clean green means passed (or not) ??

Test from 0 to 10 works as I was expecting, Clean start red I do the action in Tester Action column and Clean become green.

Test 11 probably still OK, spot + dock green, clean green, but +check robot RED + dirt detect (3 blink ?)

My "interpretation" problem start with test 12, Tester Action says : Make sure wheels are
in the air and unobstructed. Wait
until DOCK is on solid. Then briefly stall
left wheel until SPOT is on solid.

Test start with DOCK green and Clean green, stalling the wheel SPOT is on, freeing the wheel SPOT is going off.
Is it correct ? I was expecting SPOT stays on ... and that Clean become green at this point ...

Test 14 Clean + spot + Dock green, Check Robot LED off (ok or not ?)

Other "confusing" things are:
- if Clean is red (i.e. test not yet passed) and you go back (Dock) to a test that passed, Clean stays red
- Test 20 Clean and dock were green BEFORE I plugged the charger ..

Thanks & Ciaoooooooo,
Luciano

[bolero1000]

Hi,

Just a fast update. Roomba "behave" normally this morning, once charger unplugged Clean went off. No flashing orange.
It's more than 1 hour that's working, Clean still green, so I suppose battery should be OK ...

Ciaooooooooo,
Luciano

[Gordon]

... I simply unplugged the roomba from the charger (I don't use the base), ...

Luciano, please state whether you elect to not use a Home Base, or not use it because the 520 has no Home Base and no HB's electrical contacts on the robot.

BTW, I'm glad to see your 520 has corrected itself, but be sure to say something if it reverts to its error state.

[bolero1000]

Hi Gordon,

Luciano, please state whether you elect to not use a Home Base, or not use it because the 520 has no Home Base and no HB's electrical contacts on the robot.

I simply don't use the base, but I've it & the 520 (at least the European one) have the contacts.

BTW, I'm glad to see your 520 has corrected itself, but be sure to say something if it reverts to its error state.

I found a way to "replicate" it (perhaps...), if I unplug the charger during the charging (flashing orange), the roomba continue to flash orange also unplugged for 30sec-1min .. perhaps it's done on purpouse (stays in charge "control" mode) to handle possible power problems (spikes or other) ??

Perhaps it could means that for an unknown reason my roomba decided that was in "charging mode" ?? Just a guess ..

I'll let you know if it'll happen again.

Thanks & Ciaoooooooooooo,
Luciano

P.S.: not really related ... someone could clarify me, for the selftest in MANUAL-ADVANCE MODE, how do you understand if a test passed or not ?? I thought it was when Clean become green but not so sure now ...

[Gordon]

Hi Gordon,

Luciano, please state whether you elect to not use a Home Base, or not use it because the 520 has no Home Base and no HB's electrical contacts on the robot.

I simply don't use the base, but I've it & the 520 (at least the European one) have the contacts.

Thanks for that. Perhaps the "520" is the USA-530 bundled with a 240VAC-mains capable charging-PSU!

BTW, I'm glad to see your 520 has corrected itself, but be sure to say something if it reverts to its error state.

I found a way to "replicate" it (perhaps...), if I unplug the charger during the charging (flashing orange), the roomba continue to flash orange also unplugged for 30sec-1min .. perhaps it's done on purpouse (stays in charge "control" mode) to handle possible power problems (spikes or other) ??

No, it is suffering a failure. After reading your OP, I gave this a lot of thought, and did some tracing of the charging current path on a loose 510-PWB, but could not get enough of the circuit diagram in front of me to be certain about my thoughts.

As a test, you might try setting up the Home Base for charging and let the 520 charge from that source. See how well that works.

In any event, you might begin looking around for a 520/530 main board to buy -- assuming you can get someone to R&R main boards.

... ... someone could clarify me, for the selftest in MANUAL-ADVANCE MODE, how do you understand if a test passed or not ?? I thought it was when Clean become green but not so sure now ...

I can only parrot what the manual says, and it is this:

MANUAL-ADVANCE MODE
This is designed to be used for debugging particular sensors or actuators. In this mode, the robot
DOES NOT decide PASS/FAIL for each test number.

[vic7767]

The only way to determine a Pass/Fail in manual mode is to monitor the test results via the SCI port.

[bolero1000]

Hi,

I found a way to "replicate" it (perhaps...), if I unplug the charger during the charging (flashing orange), the roomba continue to flash orange also unplugged for 30sec-1min .. perhaps it's done on purpouse (stays in charge "control" mode) to handle possible power problems (spikes or other) ??

No, it is suffering a failure.

Do you mean that you tried to unplug the charger while roomba is charging (flashing orange) & it not flashing orange (unplugged, for 30sec-1min) ??

As a test, you might try setting up the Home Base for charging and let the 520 charge from that source. See how well that works.

I could try, problem is that this "behaviour" is random (as usual ... ) so it'll be difficult to understand if charging from base "works". Moreover also charging without base "works" (i.e. at least last night & right now & the first 20-25 days of this roomba's life) ...

Just to avoid misunderstanding for "works" I mean:
- roomba charge battery (clean flash orange, then clean fixed gree)
- when I unplug the roomba clean goes off
- when I press on clean it switch on (clean on), pressing again clean music & roomba starts

The only way to determine a Pass/Fail in manual mode is to monitor the test results via the SCI port.

Well, at this point I've a major understanding problem ...so, if no SCI port connection run the selftest is useless ??

Just to make an example, the simple test 2 for bumpers, Tester Action "Press and release left bumper. Press and release right bumper.", if the Spot & Dock LED goes on (& the Clean goes green) could I says that test's passed or not ?

Thanks & Ciaooooooooo,
Luciano

[vic7767]

Just to make an example, the simple test 2 for bumpers, Tester Action "Press and release left bumper. Press and release right bumper.", if the Spot & Dock LED goes on (& the Clean goes green) could I says that test's passed or not ?

Yes, when the documentation mentions a visual way to determine Pass/Fail then use the results that are mentioned. What I attempted to explain is that on the tests that have no Pass/Fail indicator you could use the output of the test in ASCII via the SCI port.

[Gordon]

...Do you mean that you tried to unplug the charger while roomba is charging (flashing orange) & it not flashing orange (unplugged, for 30sec-1min) ??

NO. I did no active testing, only analysis by examining a spare circuit board.

Your 520 appears to remain in charging mode even after removal of charging power. We don't know with 100% certainty that it is remaining in charging mode, but based on your description it seems to be doing that. If so, the main board has a charging system fault, and it is not likely that you will be able to fix it -- other than by replacing the (alleged) failed board.

[bolero1000]

Hi,

Roomba worked fine also this morning (was plugged directly last night), did a mission with no problems.

Just FYI I tried to "replicate" also using the home base:
- roomba on the HB, clean flashing orange
- I unplugged the power cord (from the power socket or directly from the HB, it's the same)
- clean still flash orange for 45 secs (this time I measured it ..)

NO. I did no active testing, only analysis by examining a spare circuit board.

It'll be interesting, at least for me, if someone could try on his 5xx to unplug while's charging & confirm if the Clean goes off (or continue to flash orange) ...

If so, the main board has a charging system fault, and it is not likely that you will be able to fix it -- other than by replacing the (alleged) failed board.

I don't think I mentionned it, but I still have the warranty, problem's that right now I don't know what to tell to CS .. I risk that they ship it back saying "no problem found" if problem isn't continuos and/or find a way to replicate it ..
Not sure the european CS (not Irobot) have the tools/expertise to read the BIT results, I've the impression are more based on the "parts swap" model ...

Thanks & Ciaooooooooooo,
Luciano

[Gordon]

...It'll be interesting, at least for me, if someone could try on his 5xx to unplug while's charging & confirm if the Clean goes off (or continue to flash orange) ...

I ran my 535 on carpet for about half an hour, and then set it to direct charging. After waiting about 10 minutes, I unplugged the dc power and noticed that the Clean-button LEDs went dark within a second or two, and stayed dark until I resumed charging. About 30 minutes later, I repeated the charging interrupt, and the LEDs again went dark very quickly.

[bolero1000]

Hi Gordon,

...It'll be interesting, at least for me, if someone could try on his 5xx to unplug while's charging & confirm if the Clean goes off (or continue to flash orange) ...

I ran my 535 on carpet for about half an hour, and then set it to direct charging. After waiting about 10 minutes, I unplugged the dc power and noticed that the Clean-button LEDs went dark within a second or two, and stayed dark until I resumed charging. About 30 minutes later, I repeated the charging interrupt, and the LEDs again went dark very quickly.

Thanks a lot for the confirmation, at least I've something sure to tell to the CS.

Thanks & Ciaooooooooooo,
Luciano

[Gordon]

The subject line of this thread alludes to the possibility that, once charging power has been disconnected, current from the charged battery is able to flow backwards along (partially // to) the charging current path to then keep the system (specifically five-volts regulated power and the MCU) up and running in charging mode!

The 4XXX Roomba circuit disallows that occurrence because there is a blocking diode (D31) between the battery's V_pos terminal and the final charging FET (U4). However, a quick look at the 500 Roomba charging path shows no such blocking diode between the final charging FET (Q18) and the (VBAT) battery terminal -- the connection is direct.

Once the externally supplied charging power is no longer impressed on the 1st FET's (that would be Q17) input pin the battery potential existing at the output of Q18 is then able to forward bias each FET's parasitic diode to then feed power back to Q17's source-pin and to the on-board SMPS that develops +5VREG voltage. That backward flow can happen so fast that the MCU is kept powered, and in the same mode, hence it keeps pulsing the LEDs under the Clean-button. But, nothing happens, save for the small current drawn to run the SMPS, the MCU, and the IRM-sensor.

However, I must point out the fallacy in this line of thinking: If that's all there were to it, ALL 500s would behave this way, but, they don't. The one feature that is easy to verify is the direct connection of battery to Q18's drain terminal, i.e., no blocking diode. A thorough reverse-engineering of the 500's charging-control circuit should reveal why this self-powering does not occur in all 500s; and, it should also suggest a component failure mode that supports what Luciano is seeing his 520 do.

[mfortuna]

I wonder if the battery voltage, divded down goes into the MCU via an ADC (analog to digital) input. Maybe the battery voltage is higher than expected until it bleeds down a bit, making the MCU think the charger is attached.

[bolero1000]

Hi

Just an update, local CS exchanged my 520 (that was already an exchange of a 4xx/green) with a 531.

Ciaoooooooooo,
Luciano

[rangatanga]

One of my Roombas 560 starts to behave very strange - light pulses amber although powersupply or homebase are not connected. And more often- after reser it runs for a while and then suddenly dead - no beep, no message, and again - green for five minutes an then pulses amber, etc. I did a diagnostic test and everything has passed, Found similar isuues on the forum, but no solution. It seems like some part on the MB gets warm and then fails. Any suggestion where to find the culprit? What voltages to measure to identify the malfunctionig part?
Thanks
Michal

another rekmark - when MB is connected to regulated PS it works within range 12-14 Volts. When the voltage exceeds roughly 14,5 Volts, the charging light starts to blink. Thats why it blinks with battery connected since charged battery voltage is more then 15 Volts.

[Gordon]

rangatanga, I am going to recommend that your post be moved to tag onto the end of bolero1000's same topic thread at:
viewtopic.php?p=79949#p79949

One of my Roombas 560 starts to behave very strange - ... .

Basically the same as bolero1000 experienced.

Any suggestion where to find the culprit?

None have been posted, as yet. I have been studying the fault as time permits. I think the self-powering occurs as I described in the referent thread, but have no clue how the circuit may be designed to overcome such a built in fault, or why all 500-Roomba owners don't experience the problem.

What voltages to measure to identify the malfunctionig part?

I can suggest measurements that will confirm how battery power flows in a reverse direction through the charging FETs to then power the robot, but resulting data will only confirm that flow, rather than lead to a correction.

another rekmark - when MB is connected to regulated PS it works within range 12-14 Volts. When the voltage exceeds roughly 14,5 Volts, the charging light starts to blink. Thats why it blinks with battery connected since charged battery voltage is more then 15 Volts.

Michal, I think that may become useful data as we learn more about this problem, but can't see what to make of it at this time.

If you wish to confirm reverse current flow through the parasitic diodes (='bulk-diode' = 'body-diode') int the series charging FETs, then locate them, Q17 & Q18 at Cartesian coordinates (-27,76) & (-14,76), millimeter units, where the origin is centered on the Clean SW button, and +x points to Roomba's right, & +y = forward travel.

Get the robot into the amber pulsing mode, then measure from terminal "VBAT" to Q17S. I expect to see a positive voltage drop of about 1.5Vdc, which confirms conventional current is flowing through the FET's parasitic diodes.

You can go one step further by measuring from VBAT to U3-6 (which includes one additional diode (D47, bottom face) drop and the dV across a one ohm resistor). Pin-6 of U3 is its I_sense input, and U3 is the SMPS controller MC34063, 8-pin SMD_DIP at (15,115). U3 has to be operational to provide power to the robot's MCU.

There is a lot more to learn about this situation.

[Addendum]
I just noticed that I have more of the U3 pin-6 node fleshed out. Here it is:
bD47[1,1]K=pth=>U3[2,2]6(15,115){MC34063,I_sens}==>U3[2,3]7{Vcc}==>U3[2,4]8{drive_Q2c}==>
==>C280[1,1](+)(20,115){220uF,35V}==>R78[2,1](19,24*){204,SMD}==>R111[2,1](19,117){R270,1510}==>
==>D2[1,1]K(15,96){WJ,SMD}==>R158[1,1](11,96){103,SMD}==>Q7[2,3]S(8,103){ZXMP6A17,Pchan}==>
==>via(14,96)=b=>TP#129==>via(7,2)=u=>Q14[2,3]e(9,93){2T,SMD}==>D10[2,3]K2(5,93){A7,SOT23}.
I have visually verified connections to that node, but can't be 100% certain about accuracy unless R111 is first dismounted -- which I don't wish to do on a functional PWB!

[Gordon]

Bump -- to announce addition of the Addendum.