Power crossover switch

[James H Phelan]

Team

The question is how to switch from battery power to bench power and back without having to shutdown and reboot.

Google hasn't been much help as I may not be phrasing my search right.

My main power switch is a mushroom head emergency stop switch.

The battery connects to the NC side.

My thinking is I could connect the bench power to the NO side and have it switch over when I bop the button. The trick is to keep the power going during the crossover. Claude agrees I need a moderate size capacitor to provide power during the crossover but I don't fully trust or understand how the circuit should be structured, the size of the capacitor, or any additional components for a 14-16V 4s LiPo system. "I'm a doctor, Jim, not an Electrical Engineer."

I know at least several of you know all about this sort of thing.

Thanks

James

-- 
James H Phelan
"Nihil est sine ratione cur potius sit quam non sit"
Leibniz

[Chris Albertson]

This is a solved problem; lots of systems do this in many industries.  The key is that you do not think of it as a “crossover”.    What you have is a power bus that the robot runs from.  Then you have one, two, four, or more sources of power that can be connected or removed from the bus.  Either manually or many times using MOSFETs or contactors under computer control.

As for the big red E-stop switch, there is one and only one correct design: NEVER run power through that switch.  If you do, then it is not the fail-proof kill switch that you want.   The correct way is to use a contactor to supply power.  The contactor is held closed by a small control current; when that current stops, the contactor opens.    This contactor is between the load (the robot’s motors) and the power bus.     The E-stop switch ONLY has the control voltage going through it.  So when E-stop opens, the contactor opens or if the control wires are cut off or damaged, the contactor opens.     This design allows for any number of E-stop switches to be wired in series, and pulling any of them opens the contactor.       You do NOT want high current in the E-stop switch because if so, the switch can fail by arc welding the contacts.  Your E-stop system has to be designed so that any one-in-a-million failure mode results in a stopped machine.  Fire is a common failure mode; in big machines, self-collision is a bad thing too because a powerful robot arm can rip apart the robot controller.    It should always result in a stopped machine.        If you’re using heavy power wire to the switch, a common failure is a short, and then the E-stop does not shut the robot down.         I’ve been in an industrial background and mostly powerful machine tools, but the industry solved this nearly 100 years ago, and everyone does it the same way.

Adding and removing power sources to a bus is easier with DC power.    When AC is used, you have to adjust the phase and connect on a zero crossing.   With DC, it is easier.   Just roughly match voltage and make sure all your sources can handle a back feed.    The always connect BEFORE disconnect, and you never have a gap in power.      If the source can be backed, use diode shunts or something.

About capacitors,   yes.  But those go in their power supplies.    On a modern robot, those supplies are likely DC/DC converters that use some high-frequency switching design.  They will already have capacitors.   If there are long power leads place the caps on ther "load end”.  But this is mostly a rule for PCB design.   

Next, or rather first, comes fuses.     A fuse is there to protect the wire.  A fuse has a rating less than the capacity of the wire.   Best practice is to place fuses physically close to the battery and even inside the battery compartment.       As the power is divided, say with multiple DC/DC converters, the output of each supply is fused.   NEVER use a small wire to “tap” off a big wire, unless the fuse is sized for that small wire.     

There are professional societies for Marine, Aircraft, Autos, and even residential wiring, and they all say about the same as the above.  In some cases, the rules are codified into law, in others, it is just industry best practice.

That said, for a small hobby project, nothing bad is going to happen. But if you get into power supplies that can source 100+ amps, then it does matter.      I made a design mistake on a robot experiment here at home and had an 8-AWG cable in my hand when the cable vaporized.   #8 is “big”, at least for personal-sized robots.   Not just got hot but a section of the cable instantly went missing, likely the solid cable changed to gas copper.  It made a loud noise, enough that people came down from upstairs to see what happened.      It was a stupid design error. I thought I had overcurrent protection, but the design was too complex to visually inspect; the eror went unnoticed.

This is one reason they say NEVER route power through the E-stop.  When they fail, you have some melted plastic and bits of metal, and they do unpredictable things.  But if you have the wires going to the E-stop on 0.25A fuses, they can’t do much harm, and if the 0.25A fuse does blow, the machine stops.   Another common failure is when a wire heats and the insulation melts, and you have two wires in contact.   Now current flows in unplanned ways that you can not predict.    The machine needs to stop when unpredictable things happen.    It is very easy to design a machine that works when it works.  It ia very hard to design one that remains safe even if something unexpected happens.  So industry has a set of best practice.

So, in short (1) don’t do crossover; you add and remove source and loads from a power bus.  Use fuses; E-Stop has to work even after bizarre corner cases happen.     We generally don’t build industrial robots and can lift a car chassis, but with these new 3-phase robot motors used in humanoid robots, that can cause big fires and injury if you do an ad-hoc design.  I’ve seen lithium fires and other problems firsthand; plan for it; it will happen.

--
You received this message because you are subscribed to the Google Groups "HomeBrew Robotics Club" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hbrobotics+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/hbrobotics/e937686e-3739-415a-b1df-701dfaad4671%40hal-pc.org.

[Michael Wimble]

Pretty good advice. I will following on in a few hours with a short paper about Titania’s new power system, including a schematic. It has what Chris said and more, because it has to handle on order of 100 amps of current. 

On Aug 30, 2026, at 11:51 AM, Chris Albertson <alberts...@gmail.com> wrote:

This is a solved problem; lots of systems do this in many industries.  The key is that you do not think of it as a “crossover”.    What you have is a power bus that the robot runs from.  Then you have one, two, four, or more sources of power that can be connected or removed from the bus.  Either manually or many times using MOSFETs or contactors under computer control.

To view this discussion visit https://groups.google.com/d/msgid/hbrobotics/44BE6365-E6EE-4502-A190-79DA17E6689F%40gmail.com.

[Sergei Grichine]

A small robot can use a couple 50V 30A Schottky diodes - here's what I do:

The robot's power bus is energized when either power source is connected.

Best Regards, 

To view this discussion visit https://groups.google.com/d/msgid/hbrobotics/5CDFC65A-CCB8-4666-B58A-63E0559B0935%40gmail.com.

[James H Phelan]

Team,

Much appreciate y'all's expertise!

Stormy is not Michael's industrial strength 36-42V Titania. It's only a 15V 3.3 Kg floorbot not to say that safety first should be ignored.

Putting bench power at 16V to the bus while the battery is down to 13V would cause the system to try to charge the battery without the benefit of the balance charger which, I've been told, is a very bad idea. Otherwise a simple parallel Y connector would do the trick during the swap (which I did in the past without incident but have stopped). Would the diodes in Sergei's schematic prevent this? Would a STPS10L25D be sufficient? But the battery is supposed to absorb back EMF so that wouldn't work.

https://www.digikey.com/en/products/detail/STPS10L25D/497-2738-5-ND/603763?itemSeq=281160145

I don't need Sergei's yellow 'to charger' connection as charging is done off the robot by swapping batteries.

James H Phelan
"Nihil est sine ratione cur potius sit quam non sit"
Leibniz

--

[Chris Albertson]

….Putting bench power at 16V to the bus while the battery is down to 13V would cause the system to try to charge the battery 

I think this is why the 50A Schottky diodes were suggested, to address that very problem.

[James H Phelan]

But what about the battery being a sink for back EMF?

James H Phelan
"Nihil est sine ratione cur potius sit quam non sit"
Leibniz

On 8/30/2026 6:05 PM, Chris Albertson wrote:

….Putting bench power at 16V to the bus while the battery is down to 13V would cause the system to try to charge the battery 

I think this is why the 50A Schottky diodes were suggested, to address that very problem.

--
You received this message because you are subscribed to the Google Groups "HomeBrew Robotics Club" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hbrobotics+...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/hbrobotics/DFC718D9-B19F-4DCE-92E2-AF7DE1C0EA66%40gmail.com.

[Marco Walther]

On 8/30/26 16:07, 'James H Phelan' via HomeBrew Robotics Club wrote:
> But what about the battery being a sink for back EMF?

I usually divide the 'powering' problem into two parts: The computer
side and the motors.

The motors don't have a problem with short breaks in the power supply as
long as you don't switch power sources during high speed driving;-) So,
I normally don't handle those for the un-interrupted power case.

All the back-feeding prevention/charging/... would still need to be
handled there.

For the compute side, I use something like
https://www.digikey.com/en/products/detail/smc-diode-solutions/MBR1545CT/6022120
, with > 5A at 3S or 12V, that was enough for my compute needs;-) So, I
can keep the SBC etc powered for the day and when I want to really test
the driving, I can plug in the battery, disconnect the power supply and
drive off.

https://ardupilot.org/copter/docs/common-powering-the-pixhawk.html#advanced-power-supply-configuration

Just an idea;-)

-- Marco

>> hbrobotics/DFC718D9-B19F-4DCE-92E2-AF7DE1C0EA66%40gmail.com <https://
>> groups.google.com/d/msgid/hbrobotics/DFC718D9-B19F-4DCE-92E2-
>> AF7DE1C0EA66%40gmail.com?utm_medium=email&utm_source=footer>.

>
> --
> You received this message because you are subscribed to the Google
> Groups "HomeBrew Robotics Club" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to hbrobotics+...@googlegroups.com

> <mailto:hbrobotics+...@googlegroups.com>.

> To view this discussion visit https://groups.google.com/d/msgid/

> hbrobotics/49c6fe6c-da5d-49d3-aee2-160f7d95b391%40hal-pc.org <https://
> groups.google.com/d/msgid/hbrobotics/49c6fe6c-da5d-49d3-
> aee2-160f7d95b391%40hal-pc.org?utm_medium=email&utm_source=footer>.

[Chris Albertson]

On Aug 30, 2026, at 4:07 PM, 'James H Phelan' via HomeBrew Robotics Club <hbrob...@googlegroups.com> wrote:

But what about the battery being a sink for back EMF?

To answer that, we need details, like does the robot have high inertia, how much energy is needed to bring it to a stop?    Are the drive gears even backdrivable?   

How does the H-bridge work?  Are you using a motor controller chip or separate MOSFET transistors?   If a chip, likely it already has overcurrent protection and diodes for reverse voltage.

If the robot motor is small and there is so little inertia that it stops as soon as you reduce power, then there is very little energy being created and TVS across the leads will be enoughor even overkill.    If the robot were heavy like a forklift or a Tesla Model Y, then you need more complex engineering.   A human-sized robot is someplace between the limits.

[James H Phelan]

does the robot have high inertia   It's 3.3Kg

Are the drive gears even backdrivable?   Yes

motor controller chip or separate MOSFET transistors  RoboClaw motor controller

already has overcurrent protection  No. Recommended to use.

TVS across the leads   I have a shunt regulator across the leads

James H Phelan
"Nihil est sine ratione cur potius sit quam non sit"
Leibniz

--
You received this message because you are subscribed to the Google Groups "HomeBrew Robotics Club" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hbrobotics+...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/hbrobotics/FFAA9D81-0EA2-4E86-8E0F-A5355D00E690%40gmail.com.

[Chris Albertson]

On Aug 31, 2026, at 6:08 AM, 'James H Phelan' via HomeBrew Robotics Club <hbrob...@googlegroups.com> wrote:

motor controller chip or separate MOSFET transistors  RoboClaw motor controller

OK, you have outsourced the entire design of the motor controller and the software that drives it.   The docs say to never connect the RoboClaw driver to a switching power supply, unless you use their “vclamp" product.  This will protect the power supply from overvolts and allows you to connect the braking resistor, and most importantly it works with their ecosystem and only costs $20.

Roboclaw is kind of a closed system but it saves a robot builder from a ton of work.

So, there are two options: Connect Roboclaw to the motor, then power it via Vclamp and then Vclamp connects to the power bus and then connect battery and/ or switching power to the power bus.  The downside is that power does not flow back into the battery when braking but rather heat is resistor that is connected to the Vclamp.

Or you use the Vclamp only if the switching power supply is connected and rig it to it is switched off the bus when the power supply is switched off the bus.

I think once you are using Roboclaw product you need to be either “all in” or “all out”.

I did use Roboclaw on one robot I made.  I found their customer support is very good. You could draw a detailed schematic using both battery and bench supply, the diodes and Vclamp and switches and ask them “is this OK?” and I bet they will tell you.   The lifetime warranty is good too. I had to use it.  They swapped out a blown cap.   But on my next robot I used a $6 h-bridge that did the same function as the $100 Roboclaw.   

Good product, kind os the Cadillac solution though

[Michael Wimble]

BTW, before the vclamp came out, I used a Pololu shunt protector device. It worked a treat for years, but every time there was a serious over-voltage event, it essentially fried the protective device. I now use the vclamp. My new robot, Titania, is going to brushless motors and the very versatile ODrive S1 motor controller has the projection circuit built in. All you need as a bank of resistors to absorb the power. I also don’t know if every battery BMS for lithium chemistry batteries does the “right thing” when presented with the back power. I seem to recall mention of some BMSes responding by cutting off the power, which would produce a glitch, but I haven’t tried to verify that. Well, here’s what the Iron Brain has to say:

No, standard lithium-polymer (LiPo) Battery Management Systems (BMS) do not handle brushed motor power surges well.

They frequently disconnect the battery, which causes severe system glitches.

Why BMS Circuits Fail with Brushed Motors

• Over-current tripping: Brushed motors pull massive inrush currents when starting or stalling. A BMS views this sharp surge as a short circuit and instantly cuts power.
• Regenerative braking spikes: When a brushed motor slows down or spins freely, it acts as a generator. It pushes high voltage back into the system.
• Over-voltage disconnection: Standard switching power supplies cannot absorb this returned energy, causing the voltage rail to spike. If this spike reaches the BMS, its over-voltage protection triggers and disconnects the battery to prevent overcharging.

So, the AI said the BMS disconnecting will result in even higher over voltage. Their solution? Add a shunt regulator—which the Pololu device is as is the vclamp. Or add capacitors. Which causes the whole inrush problem which is why I’m adding the precharge solution to my new power controller.

--
You received this message because you are subscribed to the Google Groups "HomeBrew Robotics Club" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hbrobotics+...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/hbrobotics/392B10A9-84E5-4F6E-B29B-16D080D1AA98%40gmail.com.

[James H Phelan]

"The docs say to never connect the RoboClaw driver to a switching power supply"

It is not. It is connected to bus power.

James H Phelan
"Nihil est sine ratione cur potius sit quam non sit"
Leibniz

Good product, kind os the Cadillac solution though --

You received this message because you are subscribed to the Google Groups "HomeBrew Robotics Club" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hbrobotics+...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/hbrobotics/392B10A9-84E5-4F6E-B29B-16D080D1AA98%40gmail.com.

[Michael Wimble]

Well, for what it’s worth, that’s pretty much how I have always powered the roboclaw. I needed isolation. 

On Aug 31, 2026, at 12:26 PM, 'James H Phelan' via HomeBrew Robotics Club <hbrob...@googlegroups.com> wrote:



To view this discussion visit https://groups.google.com/d/msgid/hbrobotics/609a0bd7-8b64-483e-a3aa-0c6614d950e6%40hal-pc.org.

[James H Phelan]

Team,

This has been a fascinating journey into power management. Thanks!

But...

I don't have a huge robot.

I don't have big motors.

I don't use a switching power supply.

I don't have a Battery Management System on my Zeee 9000 LiPo batteries.

BasicMicro states a pre-charge resistor is not needed for the RoboClaw 2x7A.

Pololu states: "Note that the VClamp requires a controller to activate it. It does not contain logic to detect a voltage spike on its own, so it must be used with a RoboClaw or MCP motor controller. For a similar lower-power transient suppressing solution see our shunt regulators [Which I have]. Like the VClamp, they dissipate excess motor energy through a resistor, but they feature a set or potentiometer adjustable over-voltage cutoff, so they do not need any input signals or controller."

Back EMF = Vapp - (I x R) meaning it will be less than the battery voltage. So my 25V 10A power diode is fine.

SO... How do I swap power without rebooting and without things going Snap, Crackle, or Pop?

Do I just plug in the new battery or bench power then unplug the tired battery?

Is that safe?

James

James H Phelan
"Nihil est sine ratione cur potius sit quam non sit"
Leibniz

--

You received this message because you are subscribed to the Google Groups "HomeBrew Robotics Club" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hbrobotics+...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/hbrobotics/e937686e-3739-415a-b1df-701dfaad4671%40hal-pc.org.

[Chris Albertson]

On Aug 31, 2026, at 1:07 PM, Michael Wimble <mwi...@gmail.com> wrote:

Well, for what it’s worth, that’s pretty much how I have always powered the roboclaw. I needed isolation. 

On Aug 31, 2026, at 12:26 PM, 'James H Phelan' via HomeBrew Robotics Club <hbrob...@googlegroups.com> wrote:



"The docs say to never connect the RoboClaw driver to a switching power supply"

It is not. It is connected to bus power.

But what powers the bus?     I thought the topic bneing discussed here was a “cross over” that allows either a battery of a switching power supply (or both) to power the bus.

But in any case if using Roboclaw, they have excelent applaction notes.    Basically they tell you to use some kind of device like Vclamp if using anything but a raw battery and even then use Vclamp if the back EMP generate too high volts for the battery.

I learned this when I blew a capasitor on the roboclaw board.   They fixed it for free under their lifetime warenty but then I looked into why it blew up.  Voltage clamps were needed in my case. either DIY or Vclamp.

The App note also suggests shorting the on/off switch with a “recharge resister” of about 1K to limit inrush current.

I will.  likey never again use a brushed DC motor.   3-phase BLCD are more complex but hugly more controlable and powerfull.

[James H Phelan]

"But what powers the bus?" Either the battery, the bench supply, or a fresh battery.

James H Phelan
"Nihil est sine ratione cur potius sit quam non sit"
Leibniz

--
You received this message because you are subscribed to the Google Groups "HomeBrew Robotics Club" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hbrobotics+...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/hbrobotics/B211E71A-CC58-48F3-B769-6B8F9DF59958%40gmail.com.

[James H Phelan]

Michael,

If I read you and Sergei's schematic right, y'all have Battery Management Systems IN the battery so you only need sufficient outside power to charge.  My Zeee 9000s require an EXTERNAL balance charger and individual cell sensor wire plug. I can't connect 2 batteries into it at the same time.

Google / Gemini says I need an "Ideal Diode Dual Power OR-ing Controller" but none of them (Pololu, Adafruit) can handle the current.

OR have the bench power EXACTLY match (to 0.1V) the tired battery voltage then swap. Increase the voltage to the fresh battery and swap again.

OR I can do the dual diode thing, sacrifice back EMF path to the battery, and rely on only the shunt regulator.

James

James H Phelan
"Nihil est sine ratione cur potius sit quam non sit"
Leibniz

On 8/31/2026 4:17 PM, Michael Wimble wrote:

On Aug 31, 2026, at 1:58 PM, 'James H Phelan' via HomeBrew Robotics Club <hbrob...@googlegroups.com> wrote:

Team,

This has been a fascinating journey into power management. Thanks!

But...

I don't have a huge robot.

I don't have big motors.

I don't use a switching power supply.

I don't have a Battery Management System on my Zeee 9000 LiPo batteries.

BasicMicro states a pre-charge resistor is not needed for the RoboClaw 2x7A.

Pololu states: "Note that the VClamp requires a controller to activate it. It does not contain logic to detect a voltage spike on its own, so it must be used with a RoboClaw or MCP motor controller. For a similar lower-power transient suppressing solution see our shunt regulators [Which I have]. Like the VClamp, they dissipate excess motor energy through a resistor, but they feature a set or potentiometer adjustable over-voltage cutoff, so they do not need any input signals or controller."

Back EMF = Vapp - (I x R) meaning it will be less than the battery voltage. So my 25V 10A power diode is fine.

SO... How do I swap power without rebooting and without things going Snap, Crackle, or Pop?

Do I just plug in the new battery or bench power then unplug the tired battery?

Is that safe?

I think so. If you can, do the diode thing that Sergei has. I power my robot using the very same charger that the battery uses. it’s just:

charger+ ——— + ——— fuse ——— power switch

                       battery

charger- —— —————————————————————

So I unplug the charger, turn on the power switch and it runs off from the charger. I just unplug the charger and it runs off from the battery, no glitching. I can plug the charger back in any time. If I turn off the power switch, the charger can still charge the battery. If you want to swap batteries, well, plug in the charger, unplug the battery, plug in a new battery. As log as the battery has enough power in it when you plug in the new battery, it shouldn’t glitch the system. I don’t use diodes as my bench supply is the charger which has enough power to run my robot if the motors aren’t actually spinning. There isn’t the ability for wrong polarity or voltages that are unexpectedly out of range, so no diodes are needed. The shunt regulator is always on the RoboClaw so back EMF from the motors can’t hurt anything. If you are going to just do software work while the battery charges, no need to unplug the battery. Put a simple voltage/current monitor inline, especially an I2C device, and you can monitor when your battery is low and use that signal to initiate a docking maneuver. Here are the devices I was using:

I have since moved on to this device because it gives me 85V instead of 36V ability. There’s more detail I can add about current/voltage sensors if you find you need it. Mostly my main power is more than 10 or 20 amps on Titania so I’ll have to remove the shunt resistor on the board and use an external shunt. And I have code to read the sensors, but you can probably do well just using the library code.

[Marco Walther]

On 8/31/26 16:00, 'James H Phelan' via HomeBrew Robotics Club wrote:
> Michael,
>
> If I read you and Sergei's schematic right, y'all have Battery
> Management Systems IN the battery so you only need sufficient outside
> power to charge.  My Zeee 9000s require an EXTERNAL balance charger and
> individual cell sensor wire plug. I can't connect 2 batteries into it at
> the same time.
>
> Google / Gemini says I need an "Ideal Diode Dual Power OR-ing
> Controller" but none of them (Pololu, Adafruit) can handle the current.
>
> OR have the bench power EXACTLY match (to 0.1V) the tired battery
> voltage then swap. Increase the voltage to the fresh battery and swap again.
>
> OR I can do the dual diode thing, sacrifice back EMF path to the
> battery, and rely on only the shunt regulator.

How much do you expect to get back here? Somewhere you said, your robot
is small with small motors?! Yes, you might see some voltage spikes, but
you won't really see much energy/charging unless you try to let the
robot roll down a hill and use the motors as dynamos;-)

BTW, your bench supply `might` be the switching kind!

-- Marco

>
> James
>
> James H Phelan
> "Nihil est sine ratione cur potius sit quam non sit"
> Leibniz
>
> On 8/31/2026 4:17 PM, Michael Wimble wrote:
>>> On Aug 31, 2026, at 1:58 PM, 'James H Phelan' via HomeBrew Robotics
>>> Club <hbrob...@googlegroups.com> wrote:
>>>
>>> Team,
>>>
>>> This has been a fascinating journey into power management. Thanks!
>>>
>>> But...
>>>
>>> I don't have a huge robot.
>>>
>>> I don't have big motors.
>>>
>>> I don't use a switching power supply.
>>>
>>> I don't have a Battery Management System on my Zeee 9000 LiPo batteries.
>>>
>>> BasicMicro states a pre-charge resistor is not needed for the
>>> RoboClaw 2x7A.
>>>
>>> Pololu states: "Note that the VClamp requires a controller to
>>> activate it. It does not contain logic to detect a voltage spike on
>>> its own, so it must be used with a RoboClaw or MCP motor controller.

>>> *For a similar lower-power transient suppressing solution see our
>>> shunt regulators* *[Which I have]*. Like the VClamp, they dissipate

>>> excess motor energy through a resistor, but they feature a set or
>>> potentiometer adjustable over-voltage cutoff, so they do not need any
>>> input signals or controller."
>>>
>>> Back EMF = Vapp - (I x R) meaning it will be less than the battery
>>> voltage. So my 25V 10A power diode is fine.
>>>

>>> *SO... How do I swap power without rebooting and without things going
>>> Snap, Crackle, or Pop?*
>>>
>>> /Do I just plug in the new battery or bench power then unplug the
>>> tired battery?/
>>>
>>> /Is that safe?/

>>>
>>
>> I think so. If you can, do the diode thing that Sergei has. I power my
>> robot using the very same charger that the battery uses. it’s just:
>>
>> charger+ ——— + ——— fuse ——— power switch
>>                        battery
>> charger- —— —————————————————————
>>
>> So I unplug the charger, turn on the power switch and it runs off from
>> the charger. I just unplug the charger and it runs off from the
>> battery, no glitching. I can plug the charger back in any time. If I
>> turn off the power switch, the charger can still charge the battery.
>> If you want to swap batteries, well, plug in the charger, unplug the
>> battery, plug in a new battery. As log as the battery has enough power
>> in it when you plug in the new battery, it shouldn’t glitch the
>> system. I don’t use diodes as my bench supply is the charger which has
>> enough power to run my robot if the motors aren’t actually spinning.
>> There isn’t the ability for wrong polarity or voltages that are
>> unexpectedly out of range, so no diodes are needed. The shunt
>> regulator is always on the RoboClaw so back EMF from the motors can’t
>> hurt anything. If you are going to just do software work while the
>> battery charges, no need to unplug the battery. Put a simple voltage/
>> current monitor inline, especially an I2C device, and you can monitor
>> when your battery is low and use that signal to initiate a docking
>> maneuver. Here are the devices I was using:

>> 61xB2OPzEKL.jpg_BO30,255,255,255_UF750,750_SR1910,1000,0,C_QL100_.jpeg
>> INA226 Voltage Current Monitor 0-36V 20A Tester I2C IIC Power

>> Monitoring Sensor Module <https://www.amazon.com/dp/B0DRNG1VKM>
>> amazon.com <https://www.amazon.com/dp/B0DRNG1VKM>
>>
>> <https://www.amazon.com/dp/B0DRNG1VKM>

>>
>> I have since moved on to this device because it gives me 85V instead
>> of 36V ability. There’s more detail I can add about current/voltage
>> sensors if you find you need it. Mostly my main power is more than 10
>> or 20 amps on Titania so I’ll have to remove the shunt resistor on the
>> board and use an external shunt. And I have code to read the sensors,
>> but you can probably do well just using the library code.
>>
>>
>>

> --
> You received this message because you are subscribed to the Google
> Groups "HomeBrew Robotics Club" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to hbrobotics+...@googlegroups.com

> <mailto:hbrobotics+...@googlegroups.com>.

> To view this discussion visit https://groups.google.com/d/msgid/

> hbrobotics/f8193752-354f-488e-ab5f-cb8dfbcc4751%40hal-pc.org <https://
> groups.google.com/d/msgid/hbrobotics/f8193752-354f-488e-ab5f-
> cb8dfbcc4751%40hal-pc.org?utm_medium=email&utm_source=footer>.

[Chris Albertson]

On Aug 31, 2026, at 6:28 PM, Marco Walther <marc...@gmail.com> wrote:

On 8/31/26 16:00, 'James H Phelan' via HomeBrew Robotics Club wrote:

Michael,
If I read you and Sergei's schematic right, y'all have Battery Management Systems IN the battery so you only need sufficient outside power to charge.  My Zeee 9000s require an EXTERNAL balance charger and individual cell sensor wire plug. I can't connect 2 batteries into it at the same time.
Google / Gemini says I need an "Ideal Diode Dual Power OR-ing Controller" but none of them (Pololu, Adafruit) can handle the current.
OR have the bench power EXACTLY match (to 0.1V) the tired battery voltage then swap. Increase the voltage to the fresh battery and swap again.
OR I can do the dual diode thing, sacrifice back EMF path to the battery, and rely on only the shunt regulator.

If you really need a part, Digikey or Mouser will have exactly what you need.  I just ch. ecked Mouser has 176,521 different kinds of diodes in stock.

Put for small motors, it hardly matters.  A 3-ton Tesla needs a liquid heat exchanger and a radiator with a fan to dump the heat; my much lighter Prius gets by with a big resistor, but every robot I’ve built, I just keep the bare, unmanaged battery in the circuit.  For a small motor, anything you do will likely work.  And if you place a fuse in the circuit, nothing else will break but the fuse.

To unsubscribe from this group and stop receiving emails from it, send an email to hbrobotics+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/hbrobotics/e59f581a-180c-4392-a918-d9ae8db23358%40gmail.com.

[Sergei Grichine]

I use these:

https://www.amazon.com/dp/B0CKSNMQBN

Best Regards, 
-- Sergei
   

To view this discussion visit https://groups.google.com/d/msgid/hbrobotics/CDC6BA5D-FD39-4C52-A76C-4BB023A94290%40gmail.com.