Why is the vesc edu $127 usd?

1

I’m just looking at the specs and stuff, and frankly the boards we are throwing around on the forum here have equal or higher specs on the important stuff, apparently. I don’t see any expensive components on there or anything.

I think it warrants explanation, if we are going to barrel ahead thinking we can get similar specs for $20 or less. Seriously, there are clearly a lot of applications that can’t work with that kind of price point, and the lower levels need to have some options. Are we missing something here? I noticed it has EMI filters on the phases, which I’ve never seen mentioned as important on this forum. But I don’t see anything else on there that warrants the price, by a long way. And that’s on sale, it’s usually like $191 usd. That’s kind of nuts.

I feel bad thinking we could under-price this guy which might be inconvenient for him, but that’s a rather impractical price point for a lot of applications and we really need to do better. Also he has a rather bizzarre and apparently inefficient approach that may be responsible for the high price point.

To be honest I wouldn’t really trust it, though, I would get an odrive for that price. I looked at the code for the sensorless drive and it doesn’t look very solid, and there are a lot of people describing serious issues with basic stuff on the forums. Mind you to be fair the sensorless observer for the odrive looks like it does not work with PM type motors, which are very common. So that’s pretty bad too…

2

link: VESC SIX Education - with IMU

the odrive is $100 usd, too, it’s pretty much what I want if it had the right processor etc, but $100 is way way too much.

Just kind of frustrating because this stuff would basically be done already if they just took a nice logical and efficient approach the first time around. If we can do it, so could they have already done it.

3

The margins are very low. Cost of doing business in US is very high. You have two choices. Make a lot of money on a single board or sell thousands of boards per month. How many people would buy this type of niche product? For example, ODrive’s owner (Oskar Weigl) lives in SF Bay Area (quite literally he is my neighbor). Cost of living here is so high I clearly see why he charges $200+ for his boards.

I can sell my designs (make them for $10 and sell for $20) but unless I make 10,000 sales a month, it’s not even worth me getting my ass off my chair. Can you sell 120,000 boards a year, year after year, dealing with so many customers, renting space, paying taxes, etc. here in San Francisco, California? Forget it.

What I can do is work out of my garage, make 100 boards a month at $10, sell them at $200 and deal with a few bitchy customers. That’s 3 to 4 boards per day, very manageable by a single person.

VESC is made by Benjamin Vedder in Sweden. I’m sure his situation is no different. EU has a rather poor reputation as far as cost of doing business, too.

Cheers,
Valentine

2 Likes
4

So no fundamental reason we can’t get boards made for $20 each with the specs we are hoping for if we get them made straight from jlcpcb? That’s the way it seems to me. I think we are already close, people are already ordering boards with the comparable current capacity and so on, I’m aware there is more to the picture by far, but none of it strikes me as expensive.

If so, it’s a good argument for the reality that we should proceed further with the board designs, towards a board we can mostly get behind. Sucks to be Vedder if someone comes along and makes his edu board obsolete, but honestly that’s clearly too expensive. The really hard part seems like the firmware and wizards etc, which he has worked hard on, but he could do like blheli and go closed source. I wouldn’t use it, but… Some people are trying to build the xESC which is compatible with the firmware and software tools he made, but much cheaper and smaller, but no sensorless capability so I can’t use that. Also they haven’t gotten very far…

Edit:
The fact that it’s a long road to make a motor driver seems to me to be a strong argument in favor of the modular, plodding approach I am a fan of. Neither the one person/company does everything, nor the everyone for themselves trying to roll their own. Rather, a really good flagship board, which gets well refined over time and gets and increasingly good code base built up around it, modular all the way down, even within the firmware and any user interface tools. And because it is modular, this stuff with people trying to manufacture and re-ship stuff when they live in high cost of living areas is largely sidestepped. If I have $200 to spend on boards, I just click click order straight from jlcpcb, and I get 10 perfectly good well designed boards instead of 1, which let’s face it I’m sure an odrive or vesc can get fried too so better to have more than one. Eventually with a well enough built code base, people can pick a program and load it on an they are ready to go relatively fast. Eventually you wouldn’t even need to know much arduino. I can re-flash my 3d printer with Marlin pretty easily, helps to know some arduino but you don’t really need it. Mixing and matching is great, but you need to have building blocks that are well enough along. We can’t make a good lego car out of plastic nurdles. They have to be blocks, at least, right. What we have right now is complete overpriced toy cars on one hand and a pile of nurdles on the other.

There may seem to be a situation where you are willing to buy an odrive or vesc and just roll with it, but honestly I don’t think either of them would do what I need, for more reason than one. First is noise, second is I don’t think the odrive or maybe even vesc is capable of driving my motor, having looked at the forums and code. I had problems with the texas instruments FOC driver, it was fundamentally incapable of driving many motors even within the specified size range, just regular 3 phase motors. They just hadn’t refined the firmware well enough. And it had other problems like acceleration and reliability issues, too. Pretty good chance the odrive and vesc are going to have some issues, if texas instruments does. So their business model doesn’t work for me as a customer on basically any level at all.

1 Like
5

So, if you think it is so easy to make and sell such boards cheap, then just do it. However, I would predict, that while you may be able to manufacture some really good boards at around 25$ in small quantities, you will fail on selling them profitably, unless your per hour salary is close to nothing and you have an excellent chat bot handling all the support stuff :wink:.

6

I think there are several legit questions here. " how do I make some money?“,“how do I do something useful and still get paid ok?” and " how do we get stuff done/improve life?”

The first two are good questions which I also ask at other times. But here I’m only concerned with the last one, basically.

Nobody in an expensive country needs to sell the boards. We just order them straight from jlcpcb. They get paid, problem solved.

7

I do feel like we’ve had this discussion several times before :wink:

Let me try to find an analogy:

The difference between making your own driver for personal use, like we do here on SimpleFOC, and what VESC, ODrive etc are doing, is like the difference between cooking pasta for yourself, or running an Italian restaurant.

While the Coca Cola you pay $4.50 for may only cost the restaurant only 15c in terms of the actual liquid inside it, and noodles are super-cheap in large volume, its all the other things that need to be done that make up the cost of that coke or that plate of spaghetti:

  • rent the restaurant space, and outfit it with a dining area and a kitchen
  • comply with food safety, fire code and all the other regulations, and prove it
  • hire waiters and a cook
  • design your menus, price and source your raw materials, negotiate with suppliers
  • design and print marketing materials, run ad campaigns and build your reputation
  • factor in occasional losses: food stocks that aren’t used in time, disgruntled customers you reimburse
  • cost of doing business: Mastercard, bank accounts, lawyer, tax advisor
  • and so on, the list is probably endless…

Somewhere, at the bottom of the list is this:

  • food, raw material costs

It will be among the smaller cost items on a well managed restaurant.

And so it is with electronics. The cost of the components, even for a very good generalist type of board would be fairly low compared to its commercial sale price. If you choose cheap stuff, $20 may even do it, although I would aim more for $40-$50 assembled cost for a nice board with reliable components, convenient plugs and different features on there to make it fully functional.

So go ahead and design it - we’ve got so many designs here already, and even some VESC and ODrive versions are open source to get inspiration from.
But if you want to sell it for essentially the cost of producing it, then you’ll quickly either go bankrupt, or spend large amounts of your own cash so that you can provide people boards under effective cost…
If you can sell this idea to some venture capitalists who will then fund our cheap boards, I say lets go for it :slight_smile: but based on my own money it seems like a very unattractive proposition.

I feel that there are really many designs of this type out there on our forum already, from the Lepton to the Krakatoa, from SimpleFOC Shield to MKR FOC boards… there’s loads of stuff you can order made at JLCPCB if you’re happy to place the order yourself, assume the risks involved and then get into the software layers you need for it.

4 Likes
8

Don’t get me wrong, I understand it’s not easy, but again back to the nurdles vs well built building blocks vs complete but overpriced toy car that you can’t have…

I advocate here only for a step forward beyond the boards you speak of. And sorry to beat a dead horse, basically I’m too lazy to get on with doing it right now, but I will. I have gotten some grant money to pay for the time and boards. I just wanted to discuss the subject to make sure I’m not barking up the wrong tree.

Also I want it to be a community project, the next flagship board, not yet another nurdle on the pile. The MKR board is kind of a nurdle… it takes a lot more than that to make a motor driver, as we know here. There seems to be a tendency in the technical community to flop back and forth between “this is so easy we’ll just leave it for the end user to mop up” and " that’s incredibly expensive because it’s a ton of work". It appears that my task is to find a golden mean.

A well designed board would not come with “risks” when you order it from jlcpcb. The design would be well refined and there would be code that is known to work well, other people who have used it for similar things and so on.

9

I really think at the end of the day it is just remarkably difficult to make something that has wide appeal. Especially depending on user application, the voltage ranges, current requirements, form factor, connectors, communication protocol is all a ton of work.

I will say that people are working on making things like what you are asking for, but because of exactly the reason that I mentioned above, it seems not exactly what you want! I don’t think it’s possible to design something suitable for everyone.

I have been working now nearly 2 years on trying to achieve the goals for even just my own personal project, and have made now 5 entirely unique, fully custom motor control PCBA (all brought up and work correctly at a fundamental level) which all run simpleFOC, designed for my own purpose, and they still leave me wanting just a few extra things here and there! Not to even mention the difficulty of architecting the control layer on top of sfoc…

10
  1. Complete set of features for a particular use case;
  2. Price/cost effective;
  3. User friendly;

You can pick ONLY one.

Also, having “one board fit all” is completely unrealistic. Even my socks are different when I go hiking or jogging.

Cheers,
Valentine

1 Like
11

I am pretty close to exactly what I want out of all three of these on my lemon-pepper boards, actually! But then it really is only suitable for exactly what I designed it for, so it loses appeal to wider audience for general motor control purposes.

1 Like
12

Yeah but there is the board on sale right now on the simplefoc website. Surely it’s possible to do at least slightly better?

13

You see, here it starts already, how do you define „better“? For some people the offered board may be perfect, for others to big or not powerful enough or using the wrong CPU… There is simply not „the“ board.

1 Like
14

A bit of a cheeky comment, but this feels like grabbing a piece of soap in the shower. Every time I feel like I got a good hold of it and try to “just squeeze a bit harder to get it right” it pops right out of my fingers and out of the bathtub, then plops out of reach on the nice dry floormat by the door. Every time! Damn it! :japanese_ogre:

Valentine

15

I have no difficulty defining better, though. Higher performance to cost ratio for the purpose at hand, which is 90% of what people here do on the forum, in this class of current. Does what it needs to do at lower cost, with less hassle. No?

I like the idea of “pythonic”. The python guys have imo a nice philosophy that has worked out well. “batteries included”, " at least one clear winner choice", stuff like that.

Right now, people are always trying to roll their own motor drivers, within the sub-30 amps class, and it doesn’t go very well. It takes a long time and is a ton of work, they end up spending more than $20 for sure in the end. The b-g431b-esc1 boards are a blessing. I’m suggesting something which surpasses them in essentially every way. For the purposes we tend to see here. That is, it does not need to be small, it doesn’t need a built in st-link. It should have inline sensing. But most of all, it’s gotta be reasonably well refined.

That’s the hard part, actually getting it refined, tested, getting all the bugs worked out, well it’s all hard but that’s the part nobody seems to want to follow through with, which thus remains not done. But it’s a real barrier for a scientist or something that doesn’t know much about motor drivers and has to have a solid building block. And $100 usd is too much for many many contexts. It’s possible to do better.

The raspberry pi pico is a good concept imo. You use the same board for dev and production. It works great. Could still use more work, but pretty good.

It’s never done, but progress is possible :slight_smile:

16

Again, this is good example for the dilemma. I don’t need higher performance (whatever that is) than the B-G43B-ESC1, but I would love to have it more robust, with SPI out of the box and real connectors. I am still feeling a bit too novice for such a project but I am planning to build one next winter. Not because I need it (the ESC1s are doing a marvelous job currently), but because I want to try it. Using something readily available would take away half the fun.

2 Likes
17

Just contribute to this undertaking to mod/test the qvadrans. If you try to “diy” you’ll realize it’s actually a lot harder than it looks, get a poor result, spend a lot of time and money on prototypes… that doesn’t seem fun either, to me.

There is a large gap between “something readily available” and where we are going to be any time soon. I’d encourage you to use your skills and knowledge to contribute to the commons instead of producing get another “motor driver”. Ask yourself why there wasn’t one already ready for you when you came to the scene. There are so many littering the internet. And none of them were suitable? Why? They are not done, that’s why. Even if they work a little bit, there is a lot more than needs to be done to get them compatible and bug-free. It’s a long road.

There is this duality where people think it’s easy but then it’s too much work to actually do and nobody does it.

There is always lots of real things to do, there is no need to make stuff up, even for educational purposes. I think this is a bad habit from our educational system.

For instance, why not simply take and test the qvadrans, let us know how well it’s working, publish the code you use so the various features are brought into the code base. Like the work that other guy had to do to get CAN working and the oscillator and so on. There is tons to do, nothing is plug and play here. Getting board files made for arduino. Lots and lots of things that are worth doing and also have value as foundation building. It’s more effective educationally to do real things, generally. We could instead embrace that which is worthwhile and collaborating with others as fun and interesting, instead of the consumerist thing of repeating the same steps over and over again and producing tons of trash and little else, in the process.

18

I guess, I was not clear enough with my statement. My goal is to (re)learn designing hardware and I love it. I have developed software for 45 years or so now, but stopped hardware and embedded design (my favourite actually) 35 years ago when SMD components killed that hobby and other obligations became more demanding. Analog was always a mystery to me. Now, the hardware stuff is easier accessible than ever before, I have a nice hobby project and a bit more time to spend, so it is a good time to start with hardware again. I don’t care about failures and initial poor results (of course, I care about the money involved at some level…) and of course, I do some cherry picking from proven designs and try not to move too far away from the beaten track. As long as I learn from it and have the perspective to have “my” functional design at the end, I am happy with it. The software side of it is way less interesting for me, have done that too much in my life. Of course, I fully understand that people have different motivations and that your objective differs significantly from mine, which is perfectly ok.

2 Likes
19

We always like to have the ultimate “device” which fulfils all our needs and can be used everywhere.
Of course it is not possible, like the search for the ultimate car. You can buy a 200.000€ racing car, but if you want to use it for construction works and transport 2 tons of stones it is totally useless and a much cheaper car could fulfil your needs.
A while ago there was a project for the “ultimate continuity tester” which sounds simple compared to the motor driver boards, but if you go deeper it is not so easy. The interesting approach was to start with only defining the requirements and the specification like they do in large companies by analysing the markets needs.
You can also regard the RISC-V ISA as an example for the specification driven approach.
There is a video for the continuity tester which is worth watching due to it’s philosophy:
The ultimate continuity tester