Extra large h-bridge

@dubbadan I believe standard design tolerances are 500%. If the spec peak is 30A, put in a 150A

Hi @dubbadan,

   Sorry to hear your experience with the Cytron ended badly. Although my sermon probably (as is usual for me) sounded all doom and gloom, I was hoping I was being over cautious, and you would strike lucky.

---------

I am not clear as to the actual type and cause of your high current spikes ... not because I can't think of a reason, but more a case of there being more than 1 possibility. Whilst it is plausible the exact value you measured may not be precisely accurate, I suspect it is a reasonable indication that you are dealing with substantial currents.

Have a look at the traces, or maybe try to remember them if you don't have a record. Think of the timing of the spikes, and try to figure out what would cause each major peak to start and stop at that rate.  Also, that sort of information can be useful when choosing devices to handle it. Semiconductor devices can often withstand a higher pulse current than the continuous current, providing they do not get too hot.

In this case, I am not really talking about the outside transistor case temperature, although that is a factor, but more importantly the temperature of the 'sliver' of semiconductor (probably silicon) that is controlling the current. The local heating can be so fast, there isn't time to conduct the heat away from this area, to the device case and so on. The rule of thumb often being that a temperature rise of the silicon (in milliseconds or less) of more than 60 degC is likely to crack the device internally.

Also, of course, if the spikes are repetitive and close together, then the cumulative effect can also rapidly overheat the device.

Sorry, I can't give you an easy diagnosis method, but the more information you can get, the better you will gradually begin to understand the issues, and maybe find ways of mitigating them.

------

The other problem when currents through coils, (motors have several coils), are being switched, is that high voltages can be generated ... often much higher voltages than the power supply is providing. These voltages can also destroy unsuspecting components in their path that are not suitably rated. Again, sorry, but I do not have simple answers. It is common to add protective components, but with the current peaks you mention, these may need to be substantial parts to absorb and direct the energy safely. Care is needed with the physical arrangements of the wiring to direct the energy away from the more 'delicate' components, etc.

Good luck with your investigations. I fear this may not be your last unfortunate experience, but with each one try to learn a bit more, and you may well succeed in the end. I suspect most of the best power electronic engineers can recall a number of 'experiments' that didn't go as well as they hoped! Ironically, it is often the failures that taught them the best lessons.

Best wishes, good luck and stay safe, Dave

@davee @dubbadan OR, just replace the marginal 13A controller with 25A or 30Amp, that will probably handle it.

Hi Ron @zander,

   OR, just replace the marginal 13A controller with 25A or 30Amp, that will probably handle it.

Ron, you might be right ... to me that suggestion reminds me of betting at a casino....

First Google hit was thepihut.com/products/30a-7-35v-smartdrive-2-channel-motor-driver

this is a two channel version for nearly £80.

The advert includes "This is the SmartDrive Duo-30 motor driver which is capable of controlling BIG motors - up to 35V, 30A (Peak 80A ~1 second)!"

Note the 'peak' rating of about 1 second refers to a single peak, with 'lots of time', possibly minutes, before the next peak, so that the heat can be dissipated. This is a 'taste', but far from a full specification, of the information I was referring to above, to find a reliable solution.

Also, I haven't checked again, but when I wrote my earlier sermon on this topic, I had concerns about the control side interface ... not necessarily a show stopper, but could require care when directly connecting to GPIO pins of an Arduino or similar.

---------

I have never been to a casino ... and have no intention of doing so ... but obviously others think differently ... presumably some win?

Best wishes and good luck, whatever decisions you make, Dave

@davee I was looking at a Cyton unit with the same specs; they are a well-thought-of brand. I think the cost was more like $35 USD.

I am not sure what your concerns are with the control side; this is a typical logic level controlling much bigger power; it's what MCUs do.

Regarding casinos, all the computer conventions I attended as either a guest, speaker, or vendor-invited guest were where Casinos existed. I think I had one night where I lost the $100 that I allocated for gambling; all the other nights, I won about $300, which paid for our dinner and stage show. 

Hi Ron @zander,

 You obviously have a lucky streak ... I wish you the best if you ever return to a casino. I have never exceeded a couple of pounds at a local charity raffle ... but engineer salaries in the US are a lot more generous than on my side of the pond, so I would never have dreamt of £50-£100 (depending on exchange rate at the time)

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

 I do not query the quality of the Cytron brand ... I have never used them, but they seem to get a good press, so I presume they probably do what they claim. The PiHut advert referred to robots and showed motors taking up to 2.1 Amps as possible motors ... a long way from a wheelchair carrying a possibly heavy adult.

I fear the load currents and/or back emf voltages may exceed the device's specification, as I don't have a proper view of either the requirements imposed by the motors or the capabilities of the Cytron devices.

I included the UK reference price ... I do not claim it was the best price possible ... and it includes 20% VAT ... but for UK, Pi_Hut usually looks 'fair' for R-Pi related stuff, so I doubt if I could buy it much cheaper, unless maybe it was on AliExpress or similar.

My concerns on the control side relate to the possibility of transients from the motor side being accidentally coupled into the input. The simple wiring in the PiHut advert showed an RC receiver, which is probably being powered from a small Li-ion battery and electrically isolated from the main power wiring.

It is possible this could be duplicated in this case, but other options might include trying to take power from the main battery, and unexpected problems may arise. I have previously looked (on the web) at the 'box' sized drivers for stepper motors, and even though they were lower power, they all had opto-isolated inputs to minimise problems. Only the matchbox sized drivers found in home 3D printers (and Inq's robots) directly couple to the inputs.

To be clear, I am not sure that Cytron does not have opto--isolation ... I just failed to find it mentioned when I looked. By contrast, the stepper motor driver information always showed it.

Since I can't see the whole machine, etc. this is a concern, but not a definite will or won't work diagnosis.

Best wishes, Dave

@davee, To tell the truth, I think this design path will fail. For beach/loose sand, a tracked vehicle is needed, and that will involve gears, so the spikes may be mostly eliminated. Something tells me, however, that the OP will keep trying his current failing path.

BTW, I am not an engineer, my education level is technical high school plus one year of tech at a level one below college.

All my computer education was done at IBM, roughly 3 months out of every year for about 15 years.

.

Hi Ron @zander &  @dubbadan,

I personally very much hope this project can be made to work, as I think it is a worthwhile aim, but I also appreciate it is not an easy one.

I have no experience about wheels versus tracks on beach type surfaces, in terms of traction, etc. so I will leave that to others.

However, in terms of current transients, and possibly voltage spikes, it is important to try to determine not only their magnitude in Amps or Volts, but also time.

With brushed motors, there will inevitably be sparking at the commutator, and this will result in short duration transients, possibly of around 1 millisecond time duration, as contact is made and broken, as the motor spins.

Alternately, if the mechanical load on the motors varies greatly, perhaps because the wheels encounter rocks in the path, then this will temporarily reduce the speed of the motor, possibly to the point of almost stalling. Under such conditions, the back emf created when the motor is spinning freely will be much reduced, and the current draw will correspondingly greatly increase. I would expect such current transients to take place over a slower timescale, possibly a second or more.

In practice, you may well be able to find pulses of both of these types at the same time ... and the waveform is likely to look rather chaotic, so you will also need some imagination to try to work out the causes of the patterns, and make an overall interpretation. Remember, I am typing this sitting on a couch, with only a speculative imagination ... I am probably not realising lots of factors, so please take my comments as encouragements as to the type of things that might be happening. It is only when someone tries to it in the 'real world' that many 'obvious' points will become visible.

And after you have made some measurements and analysed the results, can you begin to form a view about the specification of motor driver required. I doubt if your first measurements will tell you everything you need to know, so still be prepared for further trials and problems, but hopefully you can converge on a path to success. That is basically how research and engineering proceeds when venturing into a new 'unknown' area.

I wish you all the best ... please try not to be too disappointed when it doesn't go as well as you hoped, Dave

@davee @dubbadan Tracks are vastly superior in beach conditions, less so in pavement situations. I also think it is far less likely that stalling will happen; however, hard maneuvering may result in some stalling due to the nature of skid steering. If turns are kept to more gentle maneuvers, I think it can be managed by clever software-controlled motor control. As far as the commutator arcing, why not switch to brushless motors? I have no idea if that is possible and what the pros and cons are, but I do know they are very popular today.