bigmoose
1 MW
Here is the spreadsheet again to play with. I have added a few more motors. It is in a zip wrapper, just unzip it to get the xls spreadsheet.
View attachment MotorCurrentsV3.ZIP
View attachment MotorCurrentsV3.ZIP
99% vw 100% PWM, does it make a difference?
- Thread starter zombiess
- Start date
zombiess
10 MW
Arlo1, I have started construction this week. But yes, I am still chasing some work with the China controllers right now as well because I still find them use full to learn with. I am concentrating mostly on power stage design. I am running a marathon with all of this so I will be slower because there is an absolutely huge amount of information to digest and process. I spent almost 3 months working with high hopes who was very generous to tutor me before I posted up the driver design. I received the boards about 1.5 weeks ago and am populating lebowskis controller board right now. I hope to be able to communicate with it tonight.
I am also working on a simple ir2110 driver layout to be fed by a china controller. Total opposite of what I did with high hopes, but this is a low cost experiment that is still helping me learn and have fun... maybe even make a little money too. Some people may see it as a waste of time but I learn in my own way and one of those ways is trying to fully understand this simple controller and its limits before going off into deeper waters. No matter what though, I am having fun and learning which is what is important to me. I doubt I would have gone down the road of matching FETs and gaining all that knowledge without questioning the norm and just accepting the general consensus on this forum that irfb4115 FETs suck and can't handle high current like the irfb4110 FETs can. Turns out I learned that even in the same batch there can be a huge variance in fets and that variance is greater in irfb4115 parts vs irfb4110.
Another issue is cost and complexity. China controller is cheap and simple. If I build a working 300a 135v power stage I could sell to the masses for $400 vs someone buying the cheapest sevcon, what do you think most people would buy? I am excited to get lebowskis controller running, but its setup is still way too complex for the average user right now. I've gone way off topic now, but please do some more tests on the china controller you dislike because as you see you just learned more info.
I am also working on a simple ir2110 driver layout to be fed by a china controller. Total opposite of what I did with high hopes, but this is a low cost experiment that is still helping me learn and have fun... maybe even make a little money too. Some people may see it as a waste of time but I learn in my own way and one of those ways is trying to fully understand this simple controller and its limits before going off into deeper waters. No matter what though, I am having fun and learning which is what is important to me. I doubt I would have gone down the road of matching FETs and gaining all that knowledge without questioning the norm and just accepting the general consensus on this forum that irfb4115 FETs suck and can't handle high current like the irfb4110 FETs can. Turns out I learned that even in the same batch there can be a huge variance in fets and that variance is greater in irfb4115 parts vs irfb4110.
Another issue is cost and complexity. China controller is cheap and simple. If I build a working 300a 135v power stage I could sell to the masses for $400 vs someone buying the cheapest sevcon, what do you think most people would buy? I am excited to get lebowskis controller running, but its setup is still way too complex for the average user right now. I've gone way off topic now, but please do some more tests on the china controller you dislike because as you see you just learned more info.
speedmd
10 MW
Remember your friends here on ES with some possible big discounts.
$400 for all the r&d and all the work zombies puts into that is a HUGE discount already.speedmd said:
Remember your friends here on ES with some possible big discounts.
That's why I don't bother I charge out $60 a hour at my shop that would make a 24 fet cost you over $700 with the work that goes into it.
HighHopes
10 kW
- Joined
- Mar 28, 2013
- Messages
- 930
you don't want to know my rates ;P
Lowboski - current measurement in motor drives is also an interesting subject because it sits on a very important subject of control theory and of course connects the output to the input in a critical way. as for filtering, i don't think it is all that important to filter heavily with 2nd or 3rd order. seems odd to say that, but that's my experience. what matters is to watch the group delay because that affects a lot the controller bandwidth. even an unbalanced 3-phase measurement can easily be compensated (within reeason) inside the controller but adding all the measurements together and then subtracting this from each phase. for perfect balanced 3-phase you would subjtract zero.. no effect. it is just a trick to compensate for uneven measurement. also, like i've said, if you synchonize your ADC to sample only when the mosfets are guaranteed NOT to be switching on/off then there is not too much noise so you do not need to filter too much. if somehow a big tranient got through, the integrator in the regulator will limit its affect and also the machine itself is also a gigantic 1st order filter too so that big transient just doesn't have a big effect. anyway, you could write a hole book chapter or two on the subject of current measurement in motor drives, its a deep subject because of the balance that must be found between torque control including during acceleration, stability, achieve specification.
arlo1 - designing for cost is new subject for me. i'm not very good at it as you can tell i'm working on a 144Vdc, 30kW motor drive for small EV and it is a real challenge. i keep asking myself, how can i do this for less cost while not sacrificing reliability (i'm willing to sacrifice performance & some design specs where a compromise is acceptable). on the other hand, heck i'm just building one .. so what do i care if it costs $100 more? i am not even worried about designing for maneuverability since it will be hand assembled. so it is just a cost vs. reliability ratio and what i find is that there are some price breaks. seems to me that there is hardly any cost difference between 1kW and 10kW motor drive. after 10kW there is a big jump where you might as well design for 25kW as the price is about the same. the next jump is at 80kW. i don't know for higher. it is related to the power switch, how many you can reasonably parallel before you must move to the next technology packaging (i.e. let someone else parallel).
anyway, i've railroaded zombies post enough - sorry!
Lowboski - current measurement in motor drives is also an interesting subject because it sits on a very important subject of control theory and of course connects the output to the input in a critical way. as for filtering, i don't think it is all that important to filter heavily with 2nd or 3rd order. seems odd to say that, but that's my experience. what matters is to watch the group delay because that affects a lot the controller bandwidth. even an unbalanced 3-phase measurement can easily be compensated (within reeason) inside the controller but adding all the measurements together and then subtracting this from each phase. for perfect balanced 3-phase you would subjtract zero.. no effect. it is just a trick to compensate for uneven measurement. also, like i've said, if you synchonize your ADC to sample only when the mosfets are guaranteed NOT to be switching on/off then there is not too much noise so you do not need to filter too much. if somehow a big tranient got through, the integrator in the regulator will limit its affect and also the machine itself is also a gigantic 1st order filter too so that big transient just doesn't have a big effect. anyway, you could write a hole book chapter or two on the subject of current measurement in motor drives, its a deep subject because of the balance that must be found between torque control including during acceleration, stability, achieve specification.
arlo1 - designing for cost is new subject for me. i'm not very good at it as you can tell
i'm working on a 144Vdc, 30kW motor drive for small EV and it is a real challenge. i keep asking myself, how can i do this for less cost while not sacrificing reliability (i'm willing to sacrifice performance & some design specs where a compromise is acceptable). on the other hand, heck i'm just building one .. so what do i care if it costs $100 more? i am not even worried about designing for maneuverability since it will be hand assembled. so it is just a cost vs. reliability ratio and what i find is that there are some price breaks. seems to me that there is hardly any cost difference between 1kW and 10kW motor drive. after 10kW there is a big jump where you might as well design for 25kW as the price is about the same. the next jump is at 80kW. i don't know for higher. it is related to the power switch, how many you can reasonably parallel before you must move to the next technology packaging (i.e. let someone else parallel). anyway, i've railroaded zombies post enough - sorry!
spinningmagnets
100 TW
If it was reliable, was plug-and-play with a limp-home mode...and was easy to program with a simple user-interface...you could charge a lot more than $400...
Reading Lukes posts about the controllers he tried to get to work gave me a headache...
zombiess
10 MW
spinningmagnets said:
And this is one reason why I continue to play around with the Xie Chang controllers. I'm trying to work on my design skills and have what should be a decent driver that will be tested with Lebowski's controller, but I also am interested in a low tech high power add on to the Xie Chang controllers I could possibly sell and make money to fund projects like Lebowski's controller. I'm starting to look into control systems now as well, maybe phase current limiting on the Xie Chang... I don't know, haven't made it that far yet, but I have several ideas floating around in my head. I honestly don't want to code my own controller, I'd rather buy someone elses brain and add my own power stage to it, but it has to be user friendly. User friendliness is key and even the Xie Chang controllers are at the edge of that for many new users. It's one of the reasons why I setup my throttle tamer to be tuned with a screw driver vs a terminal or custom windows interface where I could have a lot more options, it's just not user friendly. Most people understand how to operate a screw driver. I very recently helped a customer of mine get the basic settings on his CA V3 working for throttle control, he didn't understand the instructions at first but I knew enough of the basics to walk him through it. I see this kind of stuff every single day working in IT. Some things just can't be made user friendly though.
I really need to compile all my Xie Chang controller info into one location, it's good reference material. Arlos1's videos were very informative on showing the operational limitations they have.
What I want to do next is scope different inductance motors maybe i will just add some inductors in series with the phase wires. Then see how long the big on pulse is...
As well I will try to scope the current with a fast responding sensor. I do like testing what works from a cheep controller so I can learn how they did it to make mine work well on a budget.
HighHopes in a EV I think space is not as big of a deal so if you need to give up space in trade for a cheaper cost then Cool. I think like you start by making it work then make it work better then make it cheaper and smaller. But the start has to be with in reason in terms of price. I'm not rich yet....
Zombies I am excited to see you get something working. Just do it man its so fun. I am heading to the slow season so I should be able to rip on it soon.
We also got a new guy at my full time work so cutting to 4 day weeks is looking like it can finally happen.
As well I will try to scope the current with a fast responding sensor. I do like testing what works from a cheep controller so I can learn how they did it to make mine work well on a budget.
HighHopes in a EV I think space is not as big of a deal so if you need to give up space in trade for a cheaper cost then Cool. I think like you start by making it work then make it work better then make it cheaper and smaller. But the start has to be with in reason in terms of price. I'm not rich yet....
Zombies I am excited to see you get something working. Just do it man its so fun. I am heading to the slow season so I should be able to rip on it soon.
We also got a new guy at my full time work so cutting to 4 day weeks is looking like it can finally happen.
John in CR
100 TW
HighHopes,
I agree wholeheartedly about the big jumps in price as power increases. That's what I love about the systems I'm using now. I can have complete reliability at well over 16kw peak input using a pair of controllers that cost less than $100 ea. I believe much cheaper is possible. There's also the guy who made his own brushless motor, and found it to run much better using 6 or 7 sensorless RC controllers, each powering a stator tooth or two per phase, than it did with a high priced 3 phase controller with hall sensors.
I don't know squat about electronics beyond a bit more than basic user understanding, but when I see tiny RC controllers priced less than $30 that can push 100A, I'm baffled that no one is taking the approach to simply change the motors so they require multiple tiny cheap sensorless drives instead of building monstrous high priced drives.
Take my motors for example, 10 pole pairs and 24 slots. They're wound as two 12 slot motors. That works great, but to push 250-300A peak into the motors with 6 phases, I still need a pair of relatively large boxes. It seems like all the factory needs to do is wind the motors as 8 single slot per phase motors, and run it with 8 quite small controllers ganged together.
Since 3 phases and under 100A is simple and cheap, but 3 phases and 600A is hard and expensive, why not go with 24 phases at a conservative 75A each to get to 600A? Imagine a controller that can take up to 600A from the battery for just $240, and about the same size as a 6 fet controller plus a nice heat sinks.
With each phase consisting of a single tooth, isn't perfect commutation more easily achieved too?
After a certain amount of head banging someone has to look at getting big power the easy way, so the focus can be turned to where it's needed, ie perfecting throttle and braking response. 5 years into this hobby and the only user inputs suck so much that most forgo free additional range through regen, and we have to buy add on devices to make throttle response acceptable. Hell, we can't even buy a decent throttle. It's baffles me.
I agree wholeheartedly about the big jumps in price as power increases. That's what I love about the systems I'm using now. I can have complete reliability at well over 16kw peak input using a pair of controllers that cost less than $100 ea. I believe much cheaper is possible. There's also the guy who made his own brushless motor, and found it to run much better using 6 or 7 sensorless RC controllers, each powering a stator tooth or two per phase, than it did with a high priced 3 phase controller with hall sensors.
I don't know squat about electronics beyond a bit more than basic user understanding, but when I see tiny RC controllers priced less than $30 that can push 100A, I'm baffled that no one is taking the approach to simply change the motors so they require multiple tiny cheap sensorless drives instead of building monstrous high priced drives.
Take my motors for example, 10 pole pairs and 24 slots. They're wound as two 12 slot motors. That works great, but to push 250-300A peak into the motors with 6 phases, I still need a pair of relatively large boxes. It seems like all the factory needs to do is wind the motors as 8 single slot per phase motors, and run it with 8 quite small controllers ganged together.
Since 3 phases and under 100A is simple and cheap, but 3 phases and 600A is hard and expensive, why not go with 24 phases at a conservative 75A each to get to 600A? Imagine a controller that can take up to 600A from the battery for just $240, and about the same size as a 6 fet controller plus a nice heat sinks.
With each phase consisting of a single tooth, isn't perfect commutation more easily achieved too?After a certain amount of head banging someone has to look at getting big power the easy way, so the focus can be turned to where it's needed, ie perfecting throttle and braking response. 5 years into this hobby and the only user inputs suck so much that most forgo free additional range through regen, and we have to buy add on devices to make throttle response acceptable. Hell, we can't even buy a decent throttle. It's baffles me.
If I remember correctly, the guy had a 2nd motor controlled by a 2nd throttle, which he used to start (the motorcycle) from a stop and only then "engage" the main motor. The high priced controller was a Kelly, the brushless controller was one of big ones from hobbykinkg. His motor was 36V, so he could use the brushless controllers.John in CR said:
speedmd
10 MW
I think your on to the correct approach here John. Timing, cooling and costs could be big benefactors. It is seriously worth looking at IMO.
They learned this a few years back, losing the distributors in some piston engines and going with individual timing sensors and putting ignition coils right where they were needed. Timing improved improved greatly over the life cycle, and no exposed high voltage wires and all the issues that surround that.
John in CR
100 TW
HighHopes said:
The point isn't about what you can buy. It's about agreeing with you that low power is cheap and easy, even premium quality. Change the motor to use multiples of what is easy instead of going big, heavy, with expensive components and hard to do. Arlo1 is sweating bullets wanting to get to 500A when six 100A controllers is a matter of simplicity.
John in CR
100 TW
Njay said:
I didn't remember him sharing that much detail, since he wanted to sell a book of how to build a motor, but I could be wrong. If he had startup issues, then it was the motor design, not the concept. I know for a fact that a motor with halls wound for dual controllers starts far smoother than a 3 phase. The idea the multiple sensorless controllers would start better is speculation on my part, but wouldn't you agree that it's likely? When I have a pair of sensored/sensorless controllers on one of my 6 phase motors I'll try out sensorless. The sensorless doesn't work very well with a 3 phase and is really just a hobble home in the event of a hall failure deal, so an improvement with 2 working together would be easily noticed.
I have the book (pretty good, actually), but I think he has a video where one can see the 2nd motor and startup process (amazingdiyprojects.com, I think).
I think he said he said he had more range with the rc controllers than with the Kelly. Maybe the Kelly has lower PWM frequency leading to higher current ripple.
I can't comment on your question, I really don't have the knowledge for that.
(pretty good, actually), but I think he has a video where one can see the 2nd motor and startup process (amazingdiyprojects.com, I think).I think he said he said he had more range with the rc controllers than with the Kelly. Maybe the Kelly has lower PWM frequency leading to higher current ripple.
I can't comment on your question, I really don't have the knowledge for that.
zombiess
10 MW
Arlo1 said:
Try out an EB3 board, they are doing something different, modding the shunt has no effect on phase current, but I don't know how they are doing it. All phases have a small inductive/resistive type trace coming off them through a transistor that the EB2 boards don't. This is one of the things I would like to measure, EB2 vs EB3 phase current overshoot and control. I'd like to see some solid numbers. I can loan you a 18 FET EB3 controller to you if you want to test this. I have an 18 FET EB2 controller but it needs it's FET's replaced, installed into a case / etc. I'm guessing you are using the older EB2xx board in your videos? Do you have any EB3xx boards you can test with?
Most of the ones I've looked at do not measure phase current, only input current. Software calculates the phase current based on duty cycle and the phase current limit kicks in when this reaches the programmed value.
Most of them also keep the low side FETs on for the commutation cycle and all the PWM switching is done with the high side. With no synchronous rectification, the high side body diodes have to dissipate a crapload of heat under high load conditions. This may be what blows them up. Adding synchronous rectification seems like it wouldn't be that expensive.
When pushing the limits of a cheap controller (er, well, any controller for that matter), you try to figure out what the exact failure mode is, then try to work around it.
Another possible FET killer is the high voltage spikes you can get from ringing and inductance in the wiring. These are often 2x the battery voltage, so if you're running 4110s at 72v, you could run out of head room. Good low ESR caps are your friend.
Most of them also keep the low side FETs on for the commutation cycle and all the PWM switching is done with the high side. With no synchronous rectification, the high side body diodes have to dissipate a crapload of heat under high load conditions. This may be what blows them up. Adding synchronous rectification seems like it wouldn't be that expensive.
When pushing the limits of a cheap controller (er, well, any controller for that matter), you try to figure out what the exact failure mode is, then try to work around it.
Another possible FET killer is the high voltage spikes you can get from ringing and inductance in the wiring. These are often 2x the battery voltage, so if you're running 4110s at 72v, you could run out of head room. Good low ESR caps are your friend.
zombiess
10 MW
Fechter, have you looked at an eb3 board yet? I would like someone to help figure out why modding the shunt doesn't alter the phase current the same way an eb2 board does. I reduced a shunt on an 18 FET controller from 1.25 mOhms to 0.5 mOhms and regen stayed the same and my phase current needed to stay exactly the same to get the same acelleration. I initially reduced it by the same ratio as the battery expecting it to behave like an eb2 controller. Result was motor would hit a limit at low battery current and ride very choppy with piss poor acceleration. Things smoothed out at higher speeds if you could get to them on a long enough road.
I have posted macro shots of the eb3 gate drive before.
I have posted macro shots of the eb3 gate drive before.
zombiess said:
That sounds strange. No, I haven't seen one. There must be some other current measurment mechanism.
zombiess
10 MW
Arlo1, please PM me your address and I will ship one to you.
Strange... I know, sounds like it might measure phase current in some manner doesn't it? For the last 2 years I have been posting about this but no one else has ever commented or tried to figure it out, then again I have posted a lot of tech info on these controllers myself that many people made assumptions about but no one tried to actually confirm so it's not a big surprise. Arlo1 has the proper setup to test which saves me the trouble of hacking together a current data logger and then trying to run it on my bike while riding.
If you care to take a look at the circuitry, here are two high res macro shots of the driver section for the high and low sides.
fechter said:
Strange... I know, sounds like it might measure phase current in some manner doesn't it? For the last 2 years I have been posting about this but no one else has ever commented or tried to figure it out, then again I have posted a lot of tech info on these controllers myself that many people made assumptions about but no one tried to actually confirm so it's not a big surprise. Arlo1 has the proper setup to test which saves me the trouble of hacking together a current data logger and then trying to run it on my bike while riding.
If you care to take a look at the circuitry, here are two high res macro shots of the driver section for the high and low sides.
Attachments
Similar threads
