TongXin (a.k.a. Nano) motor project

As some will know, I fitted a Crystalyte 405 to the 20" front wheel of my recumbent a few weeks ago. This works OK, but delivers way more power than is either needed or legal (unless restricted, which is what I've tried to do).

I spotted that Rennaisance Bikes in Vancouver had some TongXin motors on sale, so decided to risk buying one to fit to a spare folding bike. I ordered the motor and 36V controller last Friday evening, it was despatched from Vancouver on Monday and arrived here (near Salisbury) yesterday (via economy rate TNT!). Best bit was the price, just $200 for the hub and controller.

First impressions are that it looks an amazing little thing, very, very light and if anything it looks to be better finished than the Crystalyte. The hub is a sensorless (3 wire) one and the 36V 15A controller is also tiny, little larger than a cigarette packet and is completely sealed. The hub freewheel seems very low drag too, which is a major difference to the Crystalyte.

It initially looks to be a good option for a budget ebike, provided that all that's needed is a bit of power assistance. Being so small, I doubt that it's going to be very powerful or be much help on steep hills, but only a test will tell!

Hopefully I will get this laced up into a wheel this weekend so that I can try it and report back on performance.

Jeremy

Two questions:

What is speed of the motor? 190rpm?

What size wheel are you going to put it in?

The motor is a 260 rpm one, according to the label, model number 120SWX24, although the curves I have seem to show that max hub rpm is about 215 at no load, 36V. TongXin only publish one set of motor curves, so I have no idea how these relate to the motor I have. I suspect the curves are really for the "190rpm" motor.

The motor is going into a 20" rim and I'll probably run it at 36V. This should give reasonable hill climbing performance, I hope, and not provide any real assistance on the level at speeds over about 15mph. If this is how it seems to work, then I may put it on the recumbent as a replacement for the Crystalyte 405, as all I really need on that is hill climbing assistance

I wish we had some really solid data on these motors - trying to make sense of the garbled Chinese information is really challenging at times!

Jeremy

Jeremy,

Good luck with your Tongxin. I've got one too, which is currently being made up into a wheel, for my Christmas project bike. They do look pretty!

I'll post details when I've got mine working, and look forward to hearing how you get on with yours.

Regards

Frank

...TongXin motors on sale, so decided to risk buying one to fit to a spare folding bike. Jeremy

Jeremy,

If you fit a Tongxin to a folding bike, you could send it to AtoB magazine to get a 5-star rating and try to break their range record!

Frank

I have to own up (shamefully, perhaps......) to not being an A to B subscriber. What's the nature of the range competition and why is it about folders?

The TongXin is indeed a pretty little motor. I dropped it into my LBS this morning for them to lace it into a nice Velocity Razor 406 rim, with 2mm stainless spokes. I should get it back next week, just in time for it to be my Christmas project too!

Jeremy

if you want the 405 motor to have half the power(torq and speed) you can put a 100kohm resistor onto the throttle sensor...

the motor responds to throttle resistance of between 350Kohm 30Kohm.

350k the motor will make a tiny sound and not turn,

270k, the motor will do about 3kph on flat but not start from standstill

100k is about 50% power

50k is about 75% power

30k is the motor on full power

if you look at the throttle connection coming from the controller, on the back there are 3 metal strips. the one on the left is voltage sensor(!), in the middle is neutral, and on the right is 5V.

you can put some kind of potentiometer, it has to be very safe or you will be crashing at full speed, basically it's probably a good idea to have 100% resistance at standby, jumping straight onto 270k as a minimum setting, and 30k max. I'm going to make myself something that I can use with gloves in the winter because it's freezing!...

to be honest I read something that said that the tongxin might just be just as powerful as the crystalyte but goes about 20% further. apparently these two guys did a race and the crystalite was struggling in the last 5 miles although it might be another model.

Edited December 14, 200718 yr by giguana

The AtoB reference is joke about their slightly over-the-top praise of the Brompton-Nano, which puts a Tongxin motor, in a rack bag on a Brompton folder, with power from a 7Ah battery. The combination gives assistance up to speeds of c13 mph. They put a fit cyclist on it and discovered he got a range of c50 miles - presumably because he spent most of the time pedalling above the cut-off speed.

Based on this test, they rated the Brompton-Nano as one of the two (now three) top electric bikes on the market

Frank

Giguana,

I've already done something similar with the 405 controller, but as it's a voltage signal from the throttle (the controller isn't current driven) I used a potential divider to change the throttle range. I also biased the zero up slightly to make the controller response a little more linear.

My throttle sensor, like most I think, isn't a potentiometer at all, it's a Hall effect sensor with a voltage output, ranging from about 1V for zero to about 4V for full throttle.

This works OK, but the motor still draws a lot of current and provides a fairly non-linear response, with more power than I need. It's also a heavy motor, much, much heavier than the TongXin.

Frank,

Thanks for putting me in the picture! If my installation provides good power below 15 mph and little or no drag above this speed, then it will be ideal for the recumbent. I can pedal the 'bent fast on the level or up gentle hills, but really need a bit of help to keep some speed up on steep hills. The 'bent is less stable at low speeds, so changing down to the granny gear doesn't work too well, in fact the wobbles when trying to ride at very low speeds are a bit of a worry in traffic.

Jeremy

Reports on the Nano in smaller wheels with 24 volts have been of good climb ability, and I think that might be all you need for your 'bent. There's not only the saving in battery weight, but also the longevity of the roller drive. It's usage with 36 volt power which is most likely to shown up this innate problem.

.

Thanks for that tip, Flecc. I was wondering if it would be OK at 24V, but the very small size of the motor made me wonder. I will try it on a couple of my test 12V SLAs to see how it goes, then look at getting some decent NiMH packs for it, perhaps.

Interestingly the LBS (who are Powabyke dealers) thought that the hub was a dynamo when I took it in! They were surprised that it was so small and light for a hub motor.

Jeremy

Definitely worth trying at that. Reading between the lines in feedback, you could be pleasantly surprised at the torque around 10 mph in a 20" wheel.

That Powabyke motor is so far behind the times now, not surprising the Nano would give them a surprise.

.

the hall effect sensor is a kind of potentiometer, you can replace it independently with .5Mohm variable resistor of some kind, which means you will much more control than a hall effect. with a hall effect sensor there is about 2 mm of difference in between slow pace and jogging pace, so if you just want to admire a roadside you have to really concentrate on it, but with a potentiometer you have gradual and complete control of all the power levels. A massive improvement!

As I said before, the Hall effect sensor is a voltage output device, it has a low apparent source resistance (near zero in fact, albeit with a max source current limit) and therefore behaves slightly differently to a resistive potentiometer.

A resistive potentiometer will simulate a voltage source, but will be non-linear if required to source any current, due to the difference in current between the two parts of the track (obviously the part "above" the wiper is sourcing the standing current plus the current to the controller, the "lower" part is only sinking the standing current).

Provided that the total potentiometer resistance is low in comparison to the input resistance of the controller, say around the 5k ohms to 10k ohm source resistance that is recommended for these controllers, then this non-linear effect is negligible. If the potentiometer resistance increases though, the non-linearity gets progressively worse with increasing source resistance.

Here is a worked example to demonstrate this, with a 100k ohm controller input resistance (mine seems to be about 80k ohms as far as I can tell) and your suggestion of a 500k ohm potentiometer.

The resistive potentiometer will deliver the following percentage voltages for the stated percentage rotation:

0% rotation = 0% voltage

25% rotation = 13% voltage

50% rotation = 22.2% voltage

75% rotation = 38% voltage

100% rotation = 100% voltage

Can you see how the loading effect of the controller input resistance makes a high resistance potentiometer highly non-linear?

The Hall device will happily source the current needed to provide a linear output, due to it's low output resistance, although the mechanical design of the Chinese ones I have does introduce some non-linearity due to the mechanical arrangement. I've partially compensated for this by summing a small zero throttle voltage to the output.

Jeremy

Edited December 14, 200718 yr by Jeremy

Hey Jeremy

I too find that the throttle on my crystalyte controller to be not very linear. How would I go about making it more linear?

I'll post the final details once I've got mine tweaked to be a bit better than it is now. The key to this apparent non-linearity is really the way the controller works, I think. Looking at the brushless controller circuit used in the Crystalyte it seems that the throttle input is really giving a motor speed control, rather than a motor power control (or more accurately a motor torque control). I'm reasonably sure this is why they seem non-linear, as the way to produce best throttle response might be to arrange for the controller to provide increasing motor torque with change in throttle position.

This would overcome the problem I had of needing to move the throttle a long way to get enough torque to move off, only to find that the motor then gave more power than was needed.

For interest, here are the voltages measured from the Hall throttle (with a 5V supply), relative to percentage throttle position:

0% = 0.89V

25% = 1.45V

50% = 2.42V

75% = 3.55V

100% = 4.24V

Ideally, the voltages from the throttle should be 1V, 1.75V, 2.5V 3.25v and 4v, as the controller is expecting to see a 1V to 4V input. What I've done initially is attenuate the range, so that the controller only sees a maximum voltage of 2.5V and sum a voltage at the lower end to bring the minimum up to about 1V. This isn't perfect by any means, which is the reason I'm going to do some more experimentation before recommending a good mod.

Jeremy

that is interesting about the non linear behaviour of the voltage controllers, you are right I have my figures are little bit wrong, I have to press a very hard on the wires on the electronics kit I have because otherwise the figures are skewed...so I have done some measurements with my 406 with the standard 20A xlyte controller...this is using an assortment of ceramic resistors.

kohm = estimated rotations every second / kph drainbrain reading

690=0.8 / 6

570=0.9 / 13

503=1.0 / 14

470=1.2 / 14

320=2.0 / 18

270=3.0 / 22

220=4.0 /26

147=5.0 /27

100=6.0 /30

80=8.0 /38

47=12.0 /44

33=12.0 /44

I think resistor values above 600k would be too weak to be useful, the plan is to make the potentiometer fall to zero after about 600 by cutting the resistance thread...I am thinking of using a slide potentiometer mounted with a spring, with some safety cut-off breaks.

you can make a graph of the values, I'm about to do so, but that's a pretty fair estimate of a linear taper. that said I want a nano!

Edited December 15, 200718 yr by giguana

Apart from any concerns about linearity I wouldn't fancy using a pot because of concerns about reliability. Only the highest quality industrial spec. rotary pots would be suitable in a wet dusty environment, and such pots usually require a high operating force. A slide potentiometer would be virtually impossible to seal against dirt and moisture.

By contrast a hall effect sensor is inherently waterproof and if the external connections are sealed will work perfectly even when immersed, which given Fleccs testing methods may be important

Edited December 15, 200718 yr by Ian

Actually a linear response might not be what you need. It may be better to have more throttle movement for a smaller speed change near the lower end, if starting off and controlled slow motion is important. But then I know another application (hovercraft) where you only use the top third of the range. My point is that the ideal mapping between the throttle position and the controller input voltage is unlikely to be exactly linear. All you want to start with is to make sure that it does cover the full range and is reasonably well behaved.

More important might be what the controller is setting? Is it rpm, torque or power? A quick check with a scope on my Torq shows that the throttle control varies the width of the pulses sent to the motor. The repetition rate is fixed at about 16 kHz, and the current drawn within the pulse is set by the motor and the rpm. I don't know if the controller does then adjust the pulse width according to rpm or current but I haven't seen it doing it.

Jeremy, is it possible that your observation of it running away after getting in motion is a battery effect? Once the motor is turning the current drawn drops; if the battery voltage then goes up significantly it might give the effect of more motor power instead of less.

HTH

Nick

Hopefully I will get this laced up into a wheel this weekend

Hi Jeremy,

Interesting project. I take it you bought the motor only, not built into a wheel. Are you doing this yourself, or is there a local shop that does it?

Nick

You might be right about not wanting a linear throttle response, Nick.

I haven't looked at the Crystalyte controller waveform yet, but going on the datasheet for the brushless controller and PWM chips it looks very much as if this controller works the same way.

The high side motor drive FETs are supplied with a PWM supply, so it looks very much as if the throttle is directly controlling the effective voltage applied to the motor. I have a feeling that the response might be a bit better if the throttle response could be made more proportional to motor torque, but am really just guessing at the moment.

I went out and measured my throttle voltages this morning and was a bit surprised to find that the threshold at which the motor starts to turn (with the wheel off the ground) is a bit above the 1V that it's supposed to be. The motor didn't start up until the throttle voltage was at 1.35V, which further explains the apparent non-linearity at low throttle settings. I shall now increase the low end offset voltage a bit and see if that makes a further improvement in response.

I haven't checked the battery voltage under load yet, Nick, but will do now, as a shiny new Cycle Analyst (see here for details: The Cycle Analyst Homepage - Ebike Amp-Hour and Watt Meter) was one of the things in my package from Justin at Renaissance Bikes!

Jeremy

Hi Jeremy,

Was confused for a moment there until I realised Renaissance and ebikes.ca are the same. I got a cycle analyst and its pretty good. I was sitting here thinking that if I took a micro and a current shunt and a display I could find out things like..... But I asked on here first and flecc told me such a thing already existed.

The only issues I had with installation were that eZee used a different size bullet connector to the ones normally available and I had to make a bracket to hold the speedo sensor the right distance from the spoke.

Nick

My Crystalyte controller luckily already has the multi pin connector for the CA, so it's just a matter of plugging it in, without needing to use extra shunts etc.

I've been playing with the tiny sensorless TongXin controller this afternoon and found out a few things about it. I haven't been able to connect it to the motor yet, as that's away at the LBS being built into a new wheel, but I was able to power it up on the bench.

My intention was to try and see if I could run it up to look at the throttle response, but this was initially thwarted when I couldn't get it to work at the maximum 30V that my bench supply will go to. It looks like it won't work at 24V, as the controller low voltage cut off seems to be set to about 32V as far as I can tell (I bodged a supply up by adding a 12V battery in series with the adjustable bench supply). 32V seems a bit low for a 36V battery cut off, I would have thought that it would have been set to around 28 - 29V.

The other thing I discovered is that the controller seems to try to start by randomly powering the phase wires initially, looking for a back emf so that it can sync up. It does this for about two seconds then shuts down.

It looks like I will need to wait until I get the wheel back before I can do any more playing about with it!

Jeremy

Hi Jeremy,

I assume you can't get inside it, or if you can, the low voltage cutout can't be changed.

How does a sensorless controller work? Does it look for the back emf on all phases and then decide which phase to send a current pulse to? Does that mean it could not quite get to 100% duty cycle?

Nick

The TongXin controller is completely potted, which has the advantage of making it waterproof, but does mean I can't get inside it to play about!

It seems to work just as you describe, as far as I can tell. Whether or not this stops it getting to 100% duty cycle I'm not sure. From it's behaviour on the bench I would guess that once it has sensed the motor phase direction it should work pretty much like a Hall effect feedback sensor motor. I can't be sure, but I would assume that it uses the two un-powered phases at any instant to measure the back EMF, so it may well be able to operate up to 100% OK.

It's slightly annoying not being able to work at 24V, but I guess I shall just have to see how it performs on the road. I'm still working on an "intelligent" pedelec controller, which may well get around the excess power problem. The only real nuisance of being stuck with a 36V supply is the added battery weight.

I've spent the past hour or two dissecting a very cheap Chinese, switched mode SLA charger. I've now worked out how to make it variable voltage, rather than fixed at 48V, so am adding a 36V/48V switch to it. I'm reasonably impressed with this charger, it's very light, quite well made and is very easy to tweak. It even has a thermostatically controlled cooling fan. The make is Yiyun, they are usually available on eBay for about £18 including postage from Hong Kong and don't usually attract VAT or duty as they are below the threshold.

Jeremy

Edited December 15, 200718 yr by Jeremy