Hospital robot suffering noise issues?

[Flyback]

Hello

We are bidding to start work on a hospital robot…. “Hospitabot”. It’s a wheeled robot which goes round the hospital and can pick up parcels and then carry them and then deliver them to other parts of the hospital.
It also has radio receivers on it to give it specific commands when needed.

We don’t yet know the full spec, but we know enough to have grave concerns. Please confirm or refute our concerns?.......................

I am told the power is that given by its 24V, 20Ah battery.

All of its electric motors are Brushed Motors. We are also told that there is a noise problem with its comms system. Can you confirm that Brushed DC motors are a poor choice here, and a swap to BLDCs is in order? …. because all the electronics and comms and motors are packed tightly together, and as is surely known to all, brush noise is very aggressive.

There are offtheshelf SMPS’s on the robot, eg the JTH1524S15 for example….(and others of the same family)

JTH1524S15 24VIN 15VOUT, 15W DCDC module
https://www.xppower.com/portals/0/pdfs/SF_JTH.pdf

They have told us that the power supplies are causing noise problems with the comms systems.

Would you agree that these type of DCDC modules are optimised for Thermal performance and Efficiency…not for EMC? Do you agree that they probably switch the FETs on superfast, resulting in super high dv/dt and resultant noise issues?

Also, maybe the underside of these DCDC modules is not metal shielded? , and so maybe the underside can output noise which can couple to nearby electronics and cables?

In fact, would you agree that when electronics is packed tightly together like in this “Hospitabot”, it is often better to NOT use offtheshelf SMPS modules? , ……but instead, to use custom designed SMPS’s, with relatively high value FET gate resistors so as to damp the V(ds) switching transitions (lessen the dv/dt). This will reduce noise issues.

They are suspicious that the Hospitabot’s cable looms have not been routed correctly, and not been made correctly, and are picking up noise.

Also, the above JTH1524S15 power module has a recommended input filter as in the attached. If Vin = 24V gets hot plugged into this module, then, its input will ring up to 48V, and that will damage the module. Would you agree with this? (PDF schem and LTspice sim attached of input filter).

Also, regarding the comms system noise, I am presuming that where possible, converting to differential signalling methods will reduce noise? Also, use of common mode chokes in signal lines.

Also, we are thinking that using common mode chokes at the output of the battery will be good. Also, common mode chokes at the input to all DCDC modules?

The one noise combatting situation that comes up repeatedly in similar things, is the “grounding” of the circuit ground to the chassis. ….Where should this be done, and how many times.? Eg a grounding wire connects chassis to circuit ground, but where best to connect it? Also, can it be beneficial to connect chassis to circuit ground with more than one wire in more than one place? Of course, in the hospitabot, there is no earth ground, so we assume that from a common mode filtration point of view, the chassis is effectively to be treated as “earth”. Also, would you agree that all metal chassis parts must be electrically connected together? (for purpose of noise combatting).

Also, the Y capacitors, that connect chassis to circuit ground. Would you agree that the place where they actually connect can impact on noise? Also, over-use of Y capacitors from chassis to circuit ground can actually make common mode noise worse?

 

[alec_t]

I can't answer any of your questions, but have one for you. Is use of optical fibre for the comms an option to avoid/reduce interference from cable pick-up?

 

[Flyback]

Thanks yes, but it needs to be "on a lead" then...so not often a bonus

 

[KeepItSimpleStupid]

Brushed motors will be a good source of EMI. That should go.

Grounding needs to be thought out at the beginning. Fixing it later can get to be expensive. A reminder that EMI is reduced by twisting and RFI is reduced by shielding. Ground the shield at one end only.

The next big issue is to divide the "commons" into classes and don;t depend on the frame. + and - of motor distribution should originate at two points, one for + and one for -. Minus of the battery can be connected to the frame after you have short circuit protection. Oneor many breakers and.or a fuseable link. The ability to be able to kill motor power quickly is a good thing. I'm, sort of thinking of a visible safety switch.

The classes are mainly current and noise classes and of course a reference class. You keep them separate and tie them together at one place.

it's the best you can do because every wire to the same point is impractical. Sensors, like an automobile should be referenced to mid supply. That means whatever it is at the time. What's commonly called ratiometric.

CAN bus?

They are my "BIG" comments. Ripple is somewhat high, but that was the norm. Lower ripple is possible now.

300 kHz is a low switching frequency, but that doesn't tell the full story.

In a car the major noise sources are:
The ignition system. We went from distributer, points, wires and plugs to a coil on each spark plug. Wasted spark is a step down/

The alternator. There are -200V and +50V spikes.

How will you handle someone connecting the battery backwards?

Eventually the auto industry went to two different size terminals.

The user is stupid.

Aside:
My cousin said do you want to take this 18 wheeled tractor trailer loaded with paper around the parking lot. I was like 18 YO at the time.
It was scary, but probably one of the best things that ever happened to me. That short trip told me to give them a break.

You had to double-clutch everything. Clutch to get out of gear and clutch to get in. That was new. I could drive a stick.

What freaked me out was when I went to apply the brake, it "looked like" an automobile accelerator.

There were other instances of that like a farm tractor having three brake pedals. One for each wheel and one for both. The throttle is on the column and the starter switch looked like the old style high beam switch that was on the floor.

I was probably 17 at the time and had to drive a pick-up with the shifting on the column. Step 1: Where are the gears?

I taught one person how to drive a stick once too. Iinitially concentrated with downhill only. Then flat and finally hills and then steep hills.

I have no idea why I'm saying this except that bot will be unfamiliar to everybody. Emergency stop should be visible.

 

[Flyback]

Thanks, we are wondering if an earthing strap may solve the noise issues....(as in like the attached)

 

[KeepItSimpleStupid]

I don't think so. people might ask why your robot has a tail.

I think they were designed to help prevent shocks when sliding on the seats when getting into the car.

 

[gophert]

Old HP wide format plotters had copper wire mesh straps connecting each of the three PCBs to the chassis and each motor housing to the chassis.
The original HP plotters used brushed motors with encoders (NOT STEPPERS). There was good grounding, back EMF control and capacitive decoupling. Also, power cables to each motor were carefully twisted and terminated with large, tight-fitting connectors. Those plotters were bulletproof electrically and I never heard of a sheilding/crosstalk issue. The logic speed may have been much slower than your bot but they are a good benchmark for assembly techniques.

 

[Nigel Goodwin]

I don't think so. people might ask why your robot has a tail.

I think they were designed to help prevent shocks when sliding on the seats when getting into the car.

And they usually don't

The issue is usually NOT the car becoming charged, it's normally the occupent becoming charged due to the interaction between seat and clothing. You get a shock as you get out of the car as you discharge to earth via the car bodywork.