Testing MOSFETs_U2&U4 without using an O-scope

Inside the Roomba and Scooba and more, Cool mods, Repair and Upgrades - including the all new iRobot Create Kit. Let's void that warranty baby!

Testing MOSFETs_U2&U4 without using an O-scope

Postby Gordon » July 28th, 2010, 2:11 pm

A SIMPLE U2, U4 TEST METHOD FOR R2 (4XXX, 400) ROOMBAS

In past years, I, and others, have said services of an oscilloscope is needed to determine whether power MOSFET transistors, U2 or U4, have suffered failure, and to learn which it is. Time whizzed by, a few things were learned and then supported by testing, and I can now report there is a simple (sans O-scope) means to check for duff FETs in a 4XXX-Roomba's charging controller circuit! Is there anyone left out there with that sort of interest? If so, then, read on.

Every time you put any R2 or R3 Roomba into its charging mode its charging FETs experience a sequence of test pulses in the first half minute. If the system is satisfied with those test data, and with other conditionals (such as battery temperature and voltage) charging is then enabled. A manual rendition of that testing constitutes the following process and requires only one DMM, and is not concerned about battery conditions -- in fact, the battery must not be connected to Roomba.

This is an active test which must be performed on a partially disassembled 4XXX-Roomba in order to access specific contact points on the main electronics assembly ("PWB-asm.", hereafter). The main thrust of the test is to determine failure of U2, U4, or both.

--- EQUIPMENT ---

Test equipment is an elemental group:

* One DMM is required, but two may provide more convenience.
(One can be connected in current mode to measure current conducted through U2+U4 (when it occurs), while the second can be used to measure voltage any voltage of interest).

* A test load-resistor stands in for the battery.
(Its resistance value should be selected to cause only a large fraction of normal high-rate charging current (I've used a 30 ohm, 25W power resistor, which reduces current to ~0.74A)).

* Two needle-point probe tips are required.

* A small assortment of clip leads, including a couple with hook-type MiniGrabbers (rather than alligator clips) will be found necessary.

* Roomba's Charging PSU, or a bench-PS providing 20 to 22Vdc at about 0.5 to one ampere, is required.

--- THE SETUP ---

With battery out and set aside, Roomba must be disassembled to permit its PWB-asm to be lifted out of its chassis slot and arranged to make connecting clips and probing points on its aft face as easy as possible (peripherals may remain connected to the board, or left disconnected but the power charging path must remain connected). J7 (to the battery connector) must have its cable mated. Same with J15 or J25 (whichever one will be connected to Roomba's PSU (Fast Charger, or Std. Charger PSU). J15, the side/rear charging jack will be easiest to use.

Test-leads placed underneath the chassis must clip onto battery connector 'pins', (+) & (-), "blades", actually, then taped to secure, and their leads brought out to clip to the ammeter (in series with the test load-resistor) and other end of the load resistor. Alternatively, a DVM may be clipped across the test load-resistor, and used to assess current, if any. I think that covers the physical setting up.

--- THE PROCESS ---

Test-1 Check if both FETs have shorted:
With meters powered ON, and set to proper function and scale (use 20A, FS, setting on ammeter), the load resistor connected across Roomba's battery-connector, and charging PSU plugged into the the side charging jack, connect the PSU to a mains outlet.

Dc-current "I_load" should remain nearly at zero amps (or delta "V_load" should be nearly zero volts). Test-1 shows that both FETs have not failed as in-line conducting shorts (i.e., as switches, one or both are OPEN).

Roomba has been placed in charging mode, and it may complain because it senses no thermistor or battery voltage, but, that will not matter, since for this process you are in control, not Roomba's MCU!

If any appreciable current results, you should jump past the next few tests, and continue at Test-5.

Test-2 Verify that U4 is OPEN:
This test requires the operator to electrically bypass FET U2 by shorting together its source and drain pins, (U2S<==>U2D), while the test setup is powered. If the other FET, U4, is erroneously conducting, the expected test current (I_load = 0.74A, in my trials) will be seen on the ammeter. Remove the jumper cable when done.

Conduction means U4 is conducting when it should not, hence, U4 needs to be replaced.

Test-3 Verify that U2 is OPEN:
This test is identical to Test-2, but U4S is to be jumpered to U4D, U4S<==>U4D. Remove the jumper cable when done, and de-power the setup.

If conduction occurs, it is FET U2 that should be replaced.

If both FETs seem to not be shorted, the next test can be done to verify their turn-ON & conduction capability.

Test-4 Switch FETs ON together:
If neither FET is shown failed by tests #2 or #3, switch both FETs ON simultaneously by pulling their gates low. Do that by re-powering the setup and follow by shorting Q47's collector to GND. Q47 is a SOT-23 SMD located along the very bottom edge of the PWB-asm. and directly below U4. (The heat-sink tab on U5, the TO-220 cased +5V regulator IC, provides an easy GND point to clip onto.).

If operable, both FETs are expected to turn ON by that single action. However, if no current flows through the test load-resistance, it is probable that one, or both FETs are faulty. There is a method to discriminate which it might be, but the steps are out of scope to this process. (If one FET is found failed, they should both be replaced).

This is the normal ending point. De-power the meter(s) and system.

-----------------------------------------------------------------

Test-5 Checking the FETs' driver:
If Test-1 shows both FETs to be in their ON state, it may be due to fault conditions with Q47 & Q45 (driver xstrs). (Refer to Test-4 for details regarding Q47's location and where to find GND). Use your DVM to measure from GND to either Q47's collector, or to the gate of U4. Do this while the PWB-asm. is PSU powered.

Since there is no chargeable battery connected to Roomba, Roomba's MCU will not have commanded Q45 & Q47 ON by raising their base voltages. Therefore, expect the measured collector voltage, V_Q47c, to be in excess of 10V. If you find it to be less than 10V, you can also measure voltages at Q45 & Q47 base terminals. If V_Q45b and V_Q47b are (essentially) non-zero it will be necessary to research the cause of both Q45 & Q47 being commanded ON at the same time. Such action is beyond the scope of this process.

This is the abnormal ending point. De-power the meter(s) and system.

That's about all there is to the process. Now, isn't that easy?

======================= END ======================
Gordon
Robot Master
 
Posts: 4304
Joined: April 6th, 2005, 2:02 am
Location: Santa Ynez, CA USA

Re: Testing MOSFETs_U2&U4 without using an O-scope

Postby kevinlongisland » July 26th, 2012, 9:26 pm

Gordon, thanks for posting this. Great resource. I have a roomba that's not charging and this is a great bit of info to test u2/u4. Thanks!
kevinlongisland
Robot Groupie
 
Posts: 56
Joined: November 10th, 2011, 8:41 am

Re: Testing MOSFETs_U2&U4 without using an O-scope

Postby Gordon » July 26th, 2012, 11:19 pm

Welcome you are, Kevin. Good luck with testing.
Gordon
Robot Master
 
Posts: 4304
Joined: April 6th, 2005, 2:02 am
Location: Santa Ynez, CA USA

Re: Testing MOSFETs_U2&U4 without using an O-scope

Postby Dindo » July 11th, 2013, 7:39 am

Gordon wrote:A SIMPLE U2, U4 TEST METHOD FOR R2 (4XXX, 400) ROOMBAS

In past years, I, and others, have said services of an oscilloscope is needed to determine whether power MOSFET transistors, U2 or U4, have suffered failure, and to learn which it is. Time whizzed by, a few things were learned and then supported by testing, and I can now report there is a simple (sans O-scope) means to check for duff FETs in a 4XXX-Roomba's charging controller circuit! Is there anyone left out there with that sort of interest? If so, then, read on.

Every time you put any R2 or R3 Roomba into its charging mode its charging FETs experience a sequence of test pulses in the first half minute. If the system is satisfied with those test data, and with other conditionals (such as battery temperature and voltage) charging is then enabled. A manual rendition of that testing constitutes the following process and requires only one DMM, and is not concerned about battery conditions -- in fact, the battery must not be connected to Roomba.

This is an active test which must be performed on a partially disassembled 4XXX-Roomba in order to access specific contact points on the main electronics assembly ("PWB-asm.", hereafter). The main thrust of the test is to determine failure of U2, U4, or both.

--- EQUIPMENT ---

Test equipment is an elemental group:

* One DMM is required, but two may provide more convenience.
(One can be connected in current mode to measure current conducted through U2+U4 (when it occurs), while the second can be used to measure voltage any voltage of interest).

* A test load-resistor stands in for the battery.
(Its resistance value should be selected to cause only a large fraction of normal high-rate charging current (I've used a 30 ohm, 25W power resistor, which reduces current to ~0.74A)).

* Two needle-point probe tips are required.

* A small assortment of clip leads, including a couple with hook-type MiniGrabbers (rather than alligator clips) will be found necessary.

* Roomba's Charging PSU, or a bench-PS providing 20 to 22Vdc at about 0.5 to one ampere, is required.

--- THE SETUP ---

With battery out and set aside, Roomba must be disassembled to permit its PWB-asm to be lifted out of its chassis slot and arranged to make connecting clips and probing points on its aft face as easy as possible (peripherals may remain connected to the board, or left disconnected but the power charging path must remain connected). J7 (to the battery connector) must have its cable mated. Same with J15 or J25 (whichever one will be connected to Roomba's PSU (Fast Charger, or Std. Charger PSU). J15, the side/rear charging jack will be easiest to use.

Test-leads placed underneath the chassis must clip onto battery connector 'pins', (+) & (-), "blades", actually, then taped to secure, and their leads brought out to clip to the ammeter (in series with the test load-resistor) and other end of the load resistor. Alternatively, a DVM may be clipped across the test load-resistor, and used to assess current, if any. I think that covers the physical setting up.

--- THE PROCESS ---

Test-1 Check if both FETs have shorted:
With meters powered ON, and set to proper function and scale (use 20A, FS, setting on ammeter), the load resistor connected across Roomba's battery-connector, and charging PSU plugged into the the side charging jack, connect the PSU to a mains outlet.

Dc-current "I_load" should remain nearly at zero amps (or delta "V_load" should be nearly zero volts). Test-1 shows that both FETs have not failed as in-line conducting shorts (i.e., as switches, one or both are OPEN).

Roomba has been placed in charging mode, and it may complain because it senses no thermistor or battery voltage, but, that will not matter, since for this process you are in control, not Roomba's MCU!

If any appreciable current results, you should jump past the next few tests, and continue at Test-5.

Test-2 Verify that U4 is OPEN:
This test requires the operator to electrically bypass FET U2 by shorting together its source and drain pins, (U2S<==>U2D), while the test setup is powered. If the other FET, U4, is erroneously conducting, the expected test current (I_load = 0.74A, in my trials) will be seen on the ammeter. Remove the jumper cable when done.

Conduction means U4 is conducting when it should not, hence, U4 needs to be replaced.

Test-3 Verify that U2 is OPEN:
This test is identical to Test-2, but U4S is to be jumpered to U4D, U4S<==>U4D. Remove the jumper cable when done, and de-power the setup.

If conduction occurs, it is FET U2 that should be replaced.

If both FETs seem to not be shorted, the next test can be done to verify their turn-ON & conduction capability.

Test-4 Switch FETs ON together:
If neither FET is shown failed by tests #2 or #3, switch both FETs ON simultaneously by pulling their gates low. Do that by re-powering the setup and follow by shorting Q47's collector to GND. Q47 is a SOT-23 SMD located along the very bottom edge of the PWB-asm. and directly below U4. (The heat-sink tab on U5, the TO-220 cased +5V regulator IC, provides an easy GND point to clip onto.).

If operable, both FETs are expected to turn ON by that single action. However, if no current flows through the test load-resistance, it is probable that one, or both FETs are faulty. There is a method to discriminate which it might be, but the steps are out of scope to this process. (If one FET is found failed, they should both be replaced).

This is the normal ending point. De-power the meter(s) and system.

-----------------------------------------------------------------

Test-5 Checking the FETs' driver:
If Test-1 shows both FETs to be in their ON state, it may be due to fault conditions with Q47 & Q45 (driver xstrs). (Refer to Test-4 for details regarding Q47's location and where to find GND). Use your DVM to measure from GND to either Q47's collector, or to the gate of U4. Do this while the PWB-asm. is PSU powered.

Since there is no chargeable battery connected to Roomba, Roomba's MCU will not have commanded Q45 & Q47 ON by raising their base voltages. Therefore, expect the measured collector voltage, V_Q47c, to be in excess of 10V. If you find it to be less than 10V, you can also measure voltages at Q45 & Q47 base terminals. If V_Q45b and V_Q47b are (essentially) non-zero it will be necessary to research the cause of both Q45 & Q47 being commanded ON at the same time. Such action is beyond the scope of this process.

This is the abnormal ending point. De-power the meter(s) and system.

That's about all there is to the process. Now, isn't that easy?

======================= END ======================


Hello Gordon,

Is this test applicable to Roomba 550?
Mine does not charge and flags the Error 3. I have tried Vic's advise to pre-charge using the PSU directly for 15 minutes, then try to charge via the roomba, I got 17V but error 3 is still prompting.
Dindo
 
Posts: 5
Joined: July 9th, 2013, 7:45 am

Re: Testing MOSFETs_U2&U4 without using an O-scope

Postby Gordon » July 11th, 2013, 11:05 am

Dindo wrote:...Is this test applicable to Roomba 550?
Yes, the electrical aspects are almost identical. But, performing the tests on 5XX series mobos is made difficult by the lack of handy clip-on test-connection points on its 'PWB assembly'. Also, a troubleshooter would have to translate the test procedure from Roomba-4XXX board & components nomenclature to Roomba-5XX nomenclature before proceeding.
Mine does not charge and flags the Error 3. I have tried Vic's advise to pre-charge using the PSU directly for 15 minutes, then try to charge via the roomba, I got 17V but error 3 is still prompting.
I have read your posts and agree with the battery replacement assessment given by others. You should be more concerned about doing a measured load test on your abused battery. Load testing a suspect battery is the quickest way to move ahead, since the test can reveal that a battery has weak cells. Weak cells hold little charge, but can develop near normal voltage when fresh charged. So the summation of weak cells' voltages with strong cells' voltages is what you measure at the battery's terminals; and, if the value is near normal you may be fooled into thinking you have a good battery.
-------------------------------------------------------------
BTW, please refrain from copying an entire post into your message.
Gordon
Robot Master
 
Posts: 4304
Joined: April 6th, 2005, 2:02 am
Location: Santa Ynez, CA USA

Re: Testing MOSFETs_U2&U4 without using an O-scope

Postby Dindo » July 11th, 2013, 5:19 pm

Thanks Gordon.
I will try the load test and let you know he result.
Dindo
 
Posts: 5
Joined: July 9th, 2013, 7:45 am


Return to Robotic Hacking

Who is online

Users browsing this forum: anjosva, jdong and 344 guests

cron