Simple BLDC controller

[Jeremy Harris]

are the fet drivers good enough for 6 of these fets?

IXFN340n06 SOT227
http://ixdev.ixys.com/DataSheet/98751.pdf

Should be, but those FETs don't have a very good spec when compared to the IRFP4368 TO247 package ones I'm using.

The Ixys FET has a max current of 100A (package limited), a max voltage of 60V and an Rdson of 3mohm maximum.

The IRFP4368 has a max current of 195A (package limited), a max voltage of 75V and an Rdson of 1.85mohms maximum.

The only characteristic that's better for the Ixys FET is the junction to heatsink thermal resistance, at 0.23 deg per watt versus 0.53 deg per watt for the IRFP4368. This thermal advantage is partially lost from the higher Rdson though.

The parts I was waiting for have now arrived, so with luck I may get the board up and running later today.

Jeremy

[nieles]

yes i know they are not the best fets, but i like the fact i can mount them very easy. (isolated base plate)

the 100A limit is probalbly because of the screw connection at the terminals when the fets are in 24 7 use.

[Ricky_nz]

The parts I was waiting for have now arrived, so with luck I may get the board up and running later today.

Jeremy

Good timing.
I've got enough parts to start bringing up my power board this weekend too with the same MOSFETs although I would really like to have the capacitors that are still in transit I think I can make do.

I look forward to seeing how well your simple to build low cost controller works .

I use to do the toner transfer thing myself but with bad eyesight the hole drilling was hard for me. I just used cheap HSS 0.8mm and 0.5mm drills . The HSS drills were cheap enough I brought packs of 10 and you could get about 200-300 holes before the fibreglass killed them.
Still I had better luck with HSS rather than with reduced shank carbide bits. The HSS just flex a bit rather than breaking.
The carbide ones cut really nice but just too fragile, especially with a less than perfect drill press.

Ricky

[Jeremy Harris]

Thanks Ricky. I think yours will be more versatile, but I want to try and get some hands-on experience with the power stuff before trying something more adventurous with the control end.

I managed to get the controller finished this evening and have powered it up to check that all looks well. Quiescent current is around 20mA and all the voltages look OK. I'm in desperate need of a nice glass of wine right now, so will put off the final testing with a motor until tomorrow.

Meanwhile, so you don't all think I'm slacking, here are a couple of photos of the finished unit.

Controller ready for test - 1.JPG (90.15 KiB) Viewed 533 times

Controller ready for test - 2.JPG (83.83 KiB) Viewed 533 times

There's only one shunt fitted at the moment, to limit current to 20A for testing. I'll add more shunts as I get a feel for how the thing works.

Jeremy

[liveforphysics]

Loving it!
Great work Jeremy!

[Arlo1]

Yes very awesome!

[Alan B]

Nice progress, Jeremy!

[deVries]

I'm in desperate need of a nice glass of wine right now, so will put off the final testing with a motor until tomorrow.

Meanwhile, so you don't all think I'm slacking, here are a couple of photos of the finished unit.

Wish I could ship you a nice case of wine, so here's to this kind of "slacking"...

Have you already wound your Colossus too?

[Ratking]

Nice work Jeremy

There's nothing this community needs more than a easy and powerful controller. Make it cheap as well and you got a winner.
It's so good to see that this forum just make things that isn't available or expensive.

Keep up the good work, I will follow this one closely

Ratking

[Jeremy Harris]

Cheers folks. I should get back on it tomorrow afternoon (UK time) and see if I can get it up and running.

Jeremy

[AussieJester]

Fantastic work Jermey, this one looks simple enough even a nub like myself with board and components in hand could assemble
if there are some kits made Jermey i would definitely be interested in a couple of them.

Look forward to the testing process now...

KiM

[Jeremy Harris]

I started testing it this morning with mixed results. I was getting some really odd behaviour from it and thought at first that I'd got the Halls and phase connections messed up somehow. After half an hour or so of head scratching I found the problem, a stupid error on my part. I missed the small print in a note on the datasheet for the controller chip that quietly mentions that the hi side drive outputs are active low, whilst the lo side drives are active high. I now need to bodge up some inverters into the hi side drive lines to get things working in the right sense. This should be easy enough to do, as I socketed everything, so I can just make a daughter board that will have the inverters and the controller chip on it and that will plug into the existing socket. It's still annoying, though.

Jeremy

[Alan B]

The pesky details in the fine print again.

Good progress, Jeremy!

[oldswamm]

Funny how the real world shows up those 'little' details.
Do they make a version of the driver with an inverted high sides?

I once made a prototype board that didn't have any jumper wires, cut traces, or extra components soldered to the traces (I think it was an lm317 regulator board).

Since it came up, I thought I'd explain the lazerjet technique in more detail, so anyone who tried it would be more likely to succeed.
A lazerjet works by transferring plastic powder to paper electrostatically.
It then melts the powder into the paper.
To make a transfer, you need a paper that the melted plastic doesn't penatrate well, but still bonds to.
Then you melt it to the pc board with a cloths iron.
When you wash off the paper, you have the carbon/plastic compound left on the board.

Things to watch out for: To much heat/time/force, and things will smear. As someone pointed out, you have to let it cool, so it's well bonded and to avoid the thermal shock. If you can find the controls for the printer, set darkness, or whatever they call it, to max.
Been a long time, hope I haven't forgotten anything, as I intend to use the technique for my own prototypes.

Luck with the changes, Jeremy.
Bob

[Jeremy Harris]

Funny how the real world shows up those 'little' details.
Do they make a version of the driver with an inverted high sides?

That was the first thing I checked, O sort of hoped that On Semi would make a pin compatible version of the NCP5181 with an inverter in the high side, but no such luck. There are such devices around from other manufacturers, I believe, so for version 2 I'll try those instead.

Since it came up, I thought I'd explain the lazerjet technique in more detail, so anyone who tried it would be more likely to succeed.
A lazerjet works by transferring plastic powder to paper electrostatically.
It then melts the powder into the paper.
To make a transfer, you need a paper that the melted plastic doesn't penatrate well, but still bonds to.
Then you melt it to the pc board with a cloths iron.
When you wash off the paper, you have the carbon/plastic compound left on the board.

Things to watch out for: To much heat/time/force, and things will smear. As someone pointed out, you have to let it cool, so it's well bonded and to avoid the thermal shock. If you can find the controls for the printer, set darkness, or whatever they call it, to max.
Been a long time, hope I haven't forgotten anything, as I intend to use the technique for my own prototypes.

Pretty much sums it up. The only other thing to watch for is using the right sort of paper. You can get it to work with ordinary paper, but the results are a lot better using coated paper that's intended for glossy photo printing in inkjet printers. The snag is that the glossy paper can jam when used in a laser printer, unless the print settings are changed to accept coated paper. On my old Epson Acculaser C900 I have to set the paper in the advanced settings page on the driver to "Coated (back)" and then it works fine. I use the cheapest economy glossy photo paper that Staples sell and it works pretty well. I've heard of people cutting pages from glossy magazines and using that with good results, too.

Luck with the changes, Jeremy.
Bob

Many thanks, Bob. The new daughter board layout is done and printed up, I just need to etch the board, drill it and put it together.

Jeremy

[amberwolf]

The best paper for the laser-toner transfer, from what I've tried myself (without actually etching any PCBs from it, just trying the transfer part and then rinsing the paper off) is clay-coated papers. Usually those are the glossy photo magazines, or the almost-card-stock super-shiny-glossy advertisements you get in junk mail. Certain types of "throw away" photo books also use this paper, such as sample-photo books from various touristy places, print-shop sample books, etc.

The only catch is that I have only ever reliably gotten these papers to not jam in Hewlett-Packard LaserJet series printers that are the fairly big ones designed for more continuous duty and longer service life, like my LJ2xxx and LJ4xxx series units, and the old LJII and LJIII, etc. Even then, once the pickup rollers no longer look like elephant skin (all wrinkly) then they don't pick it up and pass it thru very well, and start jamming after a few pages if you don't wipe the rollers off with a very very slightly damp rag.

[ev_nred]

so have u done any more testing on it?
thanks,
ev_nred

[Jeremy Harris]

I'm afraid I haven't been able to test it yet. I made a new daughter board to hold the controller chip and some inverters for the hi side drives yesterday. Having etched it and drilled it I came to fit the components. Here's a photo that shows my error (it's not that easy to spot):

Duff daughter board.JPG (182.27 KiB) Viewed 654 times

Somehow I managed to screw up the printer settings (I've no idea how) so that the print came out with a scale of 1 to 1.1. The result is a perfect board that's about 10% too small for the components to fit....................

The new board is etching as I type this.

Jeremy

[Thud]

AHH the joys of prototyping.
I make stuff in a mirror image to often to be funny. (customers will say top or bottom "shown" & be wrong often).
Jeremy, just a word of encouragment & thanks for letting us tag along on this.
I am very keen to see results & am hoping for all the best.
I have to admite that I am confused as to why the Hi outputs would be active low & vise-versa. Is there a technical reason for such function?

I assume the mother board could be refined to acomodate the required components or perhapes a different controller chip. Its all a couple steps over my techincal knowledge. Anyway, good luck & thanks for updates.
(I am quite anxious to attempt to build one of these )
T

[Jeremy Harris]

When we first switched to CAD/CAM the guys in our drawing office didn't notice that the (US) software they were using defaulted to inches. They drew up a torpedo tail nut that should have been around 30mm high, maybe 40mm diameter and sent the punched tapes away to the CAM part of the operation, a workshop about 200 miles away. Two weeks later a pallet turned up with a 30inch high lump of perfectly machined stainless steel...................

anyway, I've made up a new board, I'll populate it after lunch and see if I can resume testing.

Good and bad daughter boards.JPG (187.76 KiB) Viewed 645 times

I can add the inverter stage easily enough to the main board for the Mk2, if this works OK. I think the chip is designed this way so that it can drive complementary (P channel and N channel) FETs directly. It's almost certainly aimed at low voltage controllers for automotive and industrial use, rather than high power controllers.

Jeremy

[grindz145]

When we first switched to CAD/CAM the guys in our drawing office didn't notice that the (US) software they were using defaulted to inches. They drew up a torpedo tail nut that should have been around 30mm high, maybe 40mm diameter and sent the punched tapes away to the CAM part of the operation, a workshop about 200 miles away. Two weeks later a pallet turned up with a 30inch high lump of perfectly machined stainless steel...................

Jeremy

Hah! Jeremy in the king of helpful anecdote Thanks for taking on this project. An uber cheap, DIY motor controller is fantastic for the community.

[Jeremy Harris]

I finished the daughter board, fitted it and the controller works, albeit with a quirk I need to look at more closely. It draws less no load current when driving my bench hack rewound Towerpro 5330 (fitted with Halls) than a XieChang and during a crude load test (grabbing the motor whilst it's turning) it seems to be at least as powerful, maybe more so on the 20A current limit I have at the moment. It runs right down to a very slow speed, slower than the XieChang will allow, with good low speed torque.

The one snag I have at the moment is that the motor won't start from a standstill when I open the throttle (by applying a voltage to the speed pin). It needs a substantial amount of throttle to be applied and then a flick to start. However, and this is the odd bit, if I switch it off when it's running and then switch it back on without touching the speed setting, it starts up just fine. I think I need to do a bit more investigating as to how the speed circuit works, as I'm sure it must be possible to get it to self-start from zero throttle.

I'm pleased with the power stages. They are working exceptionally well, very clean signals, no ringing or overshoot on the phase or power lines plenty of voltage on the high side FET gate bootstrap drive, even at very low speeds, and excellent "throttle response". Even the gate drives look clean, although I would expect them to as they are being driven with around an amp at initial turn on.

I just need to do some fine tuning to resolve the throttle issue and then sort out some high power testing, which means making some sort of brake first. I'll try and tidy up the circuit diagram and post it later, once I'm reasonably sure that there are no more mods needed.

I've not tried this before, but here's a video shot with my still camera that I've just uploaded to Flickr. With luck you may get to see the controller running..........

http://www.flickr.com/photos/22506359@N04/5260645728/

Jeremy

[Thud]

That is great news.
I am so looking foward to building one of these.

[Jeremy Harris]

Cheers, Thud. I'm part way to solving the start up problem. I was wrong before when I said it always started up when I turned the power on, but not when I just advanced the throttle from zero. It seems to need the 'output enable' line to be held low at power on, or if the throttle is reduced to zero, before it will reliably start. I have a suspicion that this is due to the bootstrap supply not getting kicked up until the first pulse. Flicking the motor gives it a pulse and allows a hi side FET to fire, as does switching the output enable pin on (taking output enable low takes all the outputs to the drivers low).

There's nothing in the data sheet about this behaviour, AFAICS, (oh, what a surprise........) but it seems that the circuit simply can't work without being given a little kick to get it going.

There are a few possible solutions to this. I was always planning to add a small microcontroller to do the supervisory stuff, like looking after the safety aspects (no start unless throttle at zero etc) and measuring the current for use in a "fuel gauge", so I will probably just use this to drive the right signals to the controller. Another way would be to re-design the controller to use an alternative hi side FET drive, using a simple voltage booster running from something like a 555. The simplest way to get around it might be just to fit a comparator to the throttle, so that the output enable pin was held low at zero throttle, then flipped high as the throttle was opened. This wouldn't provide reliable protection from accidentally switch on with throttle applied, because the thing will occasionally fire up on switch on, if there is a glitch that's enough to charge the bootstrap capacitor.

For those interested in a retro direct gearbox, then I can confirm that the controller instantaneously switches from forward to reverse on the fly. I've done it at full throttle (f'ing scary, the motor tried to punch it's way out of the mount!) but wouldn't recommend doing this. Switching direction at modest speeds is relatively painless, the motor takes maybe a tenth of a second to change direction and get back up to speed. The current does peak pretty high when it does this though, even on 24V with no motor load, just inertia, it's clearly banging the current limiter hard.

Jeremy

[deVries]

I'll try and tidy up the circuit diagram and post it later, once I'm reasonably sure that there are no more mods needed.

I've not tried this before, but here's a video shot with my still camera that I've just uploaded to Flickr. With luck you may get to see the controller running..........

Yes! Congratulations Jeremy. Amazing, with all the mystery & mystique of Chinese controllers & Chinese software you throw a British one together in a matter of days, and whoosh... it works, it's alive!!!

Once any tweaks or mods are done could this be a design that Geoff might be able to have made in China by surface mount?

Or, at least, get some boards made & do a BOM parts list like Goodrum does, and like magic... make our very own mystical... or mythical... Jeremy Controller.