Troubleshooters: Get a handle on fixing R3's Whl-Mtr Circuit

[Gordon]

I think we have seen numerous reports describing Roomba 500 (a.k.a., "R3") drive-wheel failures. Some have been due to busted tachometer magnets. Some reports talked about broken drop switches, and there may have been some in which the motor no longer worked.

This OP deals with none of those failures. The failure modes this OP should help with are ailments where the motor passes a stand-alone power check (the ubiquitous "9V battery" applied to spin the motor check) but the motor won't respond while being powered by its own robot.

Some members have pulled their robot out of those doldrums by replacing one or more of the motor-driver's H-bridge power transistors, however, a subset of those transistor replacers have not met with success, and, were then left to ponder what other devices in that wheel-motor driver circuit might have failed.

Lacking schematic diagrams for R3 H-bridge circuits, the only help I could offer was to point at R2's (4XXX Roombas) wheel-motor driver schematic(s) while saying something like: "R3's circuit ought to be somewhat the same..., blah, blah.".

Member latzka's plight prompted me, this month, to journey around R3's main_PCA copper tracks to visit each of the semiconductors and passives that form a wheel-motor driver circuit. Now it is clear to me that what I said in the past about this circuit was rubbish!

Once one goes outside of the H-bridge core (i.e., the four power xstrs that we envision as forming vertical lines of a capital "H", the electric motor and its pigtail wires that form the horizontal bar between the uprights, and four power diodes (used to snub voltage generated every time motor power is cut off)), thus, nine components in all, the 5XX, and 6XX wheel-motor driver circuits differ greatly when compared to the design used in the Roomba Discovery series.

See for yourself. Compare the following image with either schematic2 or schematic5 (find by searching this board).

{THIS IS A DRAFT VERSION.}

Here are a few other things to think about that drawing:

1) It may not be 100% correct (even after discounting that none of four input / control signals are precisely known).
2) Via deduction I asserted what those signal functions might be. Instrumented (O-scope) measurements may be needed to confirm. Three of them appear to have no direct connection to the MCU, marker U28, yet the MCU must source those signals.
3) I have not performed rigorous checking accuracy (but did discover, and correct a couple errors that were revealed while I drafted tables of logic states (tables attached later)). Relative to "accuracy", I have to point out that I purposely omitted showing four components (two are certainly insignificant, and two are poor candidates towards disabling this circuit). Here they are:
a) Two each zero-ohms jumper 'resistors' that carry motor current to/from sockets of card-edge receptacle J16.
b) Two each ceramic capacitors that suppress conducted motor EMI, which also board mount near J16. Schematically, these caps parallel diodes D52 & D53.{edited to change "D25" to "D52".}
4) 'On paper' the circuit looks quite reasonable, quite functional, but bear in mind that what you see is all based on continuity checks, resistance measurements, visual identities, and latzka's live data. I did no live-board measurements to support this project.

Here is another copy of Right Wheel circuit, but symbolically powered up and running its wheel in a direction corresponding to forward travel of the robot:

Without member latzka's xstr base measurements, that he reported over here...: viewtopic.php?p=111779&sid=a3e6fa68892e57f24d5b0cbaea9fd557#p111779
..., I doubt I would have been able to assign FWD (or REV) motor-current flow.

One major difference between R2 & R3 motor driver circuits is: increased complexity in R3's. I can indicate that difference by device and component counts. In the following counts I will not tally the four (bridge) power xstrs and their four power diode voltage snubbers in the following table. IOW, numbers are for the outliers beyond the H-bridge:

Code: Select all

Roomba version__| R2 | R3 |
----------------+---------+
Active devices__|..3.|..8.|
Passive devices_|..9.|.22.|
----------------+----+----+
Component count_|.12.|.27.|
----------------+----+----+

From a EE's viewpoint the R3 motor driver circuit may be seen as better behaved than was R2's. But it should also be obvious that reliability of that performance is suspect when the peripheral parts' count is more than double that in the R2 circuit. When system performance is important the main tactic used to assure reliable performance is to build with certified-reliability components. I don't mean aerospace high-rel level components should be used in floor-sweepers, but only suggesting that iRobot might have to avoid buying electronic components at "minimum-bid" cost level.

Having a greater number of failure points built into this R3 circuit may help explain wheel-drive failures that can't be corrected by trial replacement of H-bridge power transistors. There are many opportunities for failure between the H-bridge and the MCU control source. Failure of any one gate in U6, or a diode in dual diode packages D11 or D12, can cause a migraine to a troubleshooter working without a schematic diagram.

Thus, with that in mind I have created and attached two tables of test points that a troubleshooter might use to narrow down the faulty area in this circuit. These are Tables of Bias States, one with motor running forward, and the other with the motor rotation reversed.

FWD-TableOfBiasStates.pdf

(20.16 KiB) Downloaded 52 times

REV-TableOfBiasStates.pdf

(20.15 KiB) Downloaded 46 times

Readers may wonder whether a Left Wheel Motor Driver schematic diagram has been prepared. No. But, I have seen (and spot checked left vs. right PCB constructions) enough to claim there is adequate similarity between the sides that anyone working at this detail level ought to be able to draft a schematic of the Left Wheel Motor Driver by using as a guide the right-side schematic and components' positions. By and large, the line drawings will be identical and only the component markers will differ.

[vic7767]

WoW Gordon, great effort. I wondered what you were up to lately.

[TechGuy]

Thanks Gordon.

I found a couple of 500 series Roomba with the value of the R138 1-ohm resistor changed to over 10 times higher than 1-ohm. This affects the Buck converter output +5REG.

I am wondering whether the zero-ohm resistors on the wheel circuit may change affecting the current delivered to the wheel.

[Gordon]

Thanks Gordon.

Hope the OP helps you fix some things.

I found a couple of 500 series Roomba with the value of the R138 1-ohm resistor changed to over 10 times higher than 1-ohm. This affects the Buck converter output +5REG.

While googling (using sea-$: "re SMT zero ohms resistors how are they made?") to find answers for your "wondering" about the 0-ohm resistors shifting value I came across this page: https://en.wikipedia.org/wiki/Resistor which, about 4/5ths way down the page -- under "Failure Modes", talks about the start up stress put on a resistor, such as R138, by a SMPS like the buck convertor.

I am wondering whether the zero-ohm resistors on the wheel circuit may change affecting the current delivered to the wheel.

If you go to this page: http://www.analog.com/static/imported-f ... stors.html and scroll down about half way to the topic "Rectangular surface mount film resistors. / Thick film", you can read about just how poor the stability of thick-film carbon-comp resistors can be. Writers at this site: http://www.learningaboutelectronics.com ... -resistors place the resistance of zero-ohm resistors around 0.004 ohms. We can't argue against value drifting, so lets take a severe case in which drift amounts to factor 100X, giving a value of 0.4 ohms. The question then is: How might that higher value effect wheel drive?

Going to the 500 Series Service Manual: It tells us about normal and excessive BiTs limits:
left / right drive current 75 to 150 mA wheel motor
left / right stall current 700 to 850 mA wheel motor {I rearranged current values to make sense.}

At the momentary high limit of stall current the voltage drop across our 0.4 ohms jumper resistor would be: dV = 0.4 * 0.85 = 0.34 volts; which is a small fraction of battery voltage impressed across the series arrangement of two power xstrs and the motor. I can't see that as being disruptive. Operating in the "drive current" range would be even less concern.

[Gordon]

WoW Gordon, great effort. I wondered what you were up to lately.

vic, thanks for the note of appreciation. I was hoping to snare one of the 'failed wheel-drive' posters to test my tables and report any errors, but looks like I'll have to troll the archives to ferret out those buggers!

My best resource, "latzka", side stepped his R3's fault by R&R-ing his ailing robot's mobo, so he is out of the picture for the moment.

[vic7767]

I run into that fault once in a while, the next one I get I'll ship you the mobo.

[Gordon]

I run into that fault once in a while, the next one I get I'll ship you the mobo.

Thanks for the enticement, vic. When that time arrives we can work the details.

I think before messing with a faulty main_PCA I should first take the lid off my 535 and check out the two tables using its wheel-drivers (drivers that function OK). In addition to verifying tabled logic statements I want to determine how badly nodes that carry PWM signals into the circuit tend to misconstrue "status" when a dc-voltmeter is the only available troubleshooting instrument.

I have no way of foretelling when I can steal time away from doing: what must be done; what I should do; or, what 'honey-do comes along, in order to insert what I wish to do into available hours. Therefore I have great difficulty scheduling any of my "wish-to-do" tasks!