Speed Controller – error in Circuit

[Michael GilliganParticipant @michaelgilligan61133]

… and for convenience; a direct link to Andrew's album.

No excuses, now

MichaelG.

[martin perman 1Participant @martinperman1]

Andrew,

I for one like to know how stuff works and on this point I have just spent the afternoon stripping apart my daughters old microwave into component parts. I may not always understand the logic of how something works, I'm refering to electronics here, but as a mechanical engineer I apply my knowledge of hydraulics and pneumatics to electronics and muddle through and if I can fathom it I ask.

 

Martin P

 

Edited By martin perman on 22/05/2015 17:48:36

[duncan webster 1Participant @duncanwebster1]

I've found all this discussion very interesting, and as a result I've added a transistor driver stage to the FET I'm trying to control from an Arduino. The end result is to control one of those ex radar blower fans to drive a street organ. 24V at about 1.5A

I see the argument for not having a flywheel diode, but I don't want to blow the FET. Is it worth trying an RC snubber as used on AC loads? Start with a low R and increase it until the spikes get to say half the Vdss.

[Les Jones 1Participant @lesjones1]

Hi Duncan,
The only argument that seems to make sense to me for NOT having a flywheel diode is to save a few pence. I suspect the motor you are using is series wound so it's characteristics will be very different to a permanent magnet motor. You do not say if you application is just switching the motor on and off or for PWM speed control.

Les.

[duncan webster 1Participant @duncanwebster1]

It's variable speed using the Arduino pwm function, which I think runs at ~500hz. The motor runs in the same direction if you reverse the supply, so it isn't permanent magnet, don't know whether series or parallel field. Makers plate says

Airscrew Weyroc

type G.16,C4.113

24V DC

It was made in 1977, I bought it new about 9 months ago. If I block the outlet it seems to speed up a bit, but it's difficult to tell. I've tried contacting the makers, they never got back to me.

What I'm trying to do is measure the pressure in the windchest and adjust the motor speed to keep tye pressure reasonably constant. I emphasize 'reasonably', as I've seen one just driven from a lamp dimmer with no feedback and it seemed OK. I've got the pressure sensor working and the Arduino pwm responding , I'll have to invert it to suit the transistor drive, as that itself acts as an invert.

[TrevorGParticipant @trevorg]

Hi Duncan,

Les is right you may save a few pence on the diode but you are getting no other benefit from not fitting the diode. The faster the PWM the greater the energy saving by fitting a diode ( and the cooler your FET will run).

Its worth checking if you can separate the armature from the field winding. If it was serial connected it would have poor speed stability so I would expect parallel wound. If that is the case constantly power the field winding and connect the armature to the controller for best results.

On the subject of the diode I have produced another trace (shown at two different timebase speeds so you can see both the big picture and the detail)

Sorry the current trace is a bit noisy but it is a homemade probe (unfinished project) however it shows the effect of the diode quite nicely.

The motor in this test is running at part speed on a 7V supply at 1A
Key:
Purple=drive into FET Gate which shows clearly the point at which the FET is turned off
Yellow=FET Drain voltage
Blue=current into motor(=FET+Diode current)

The important thing to notice is that current continues to flow INTO the motor after the FET has been switched off i.e. the motr is getting MORE power it is NOT being braked.

For clarity I have included the relevant part of the circuit showing the mointoring points.

Trevor

[Les Jones 1Participant @lesjones1]

Hi Duncan and Trevor,
I have found that I have one of these motors which I bought for use a s blower on a small furnace. It is definitely series wound. (I confirmed this this by lifting one brush off the commutator while measuring the resistance between the motor terminals.)

Les.

[TrevorGParticipant @trevorg]

Surprising Les, given that Duncans motor did not appear to slow down under load. However, your test seems conclusive.

Just goes to show you can't rely on a subjective assesment like the motor sound. Need to do a measurement.

Trevor

[jason udallParticipant @jasonudall57142]

Re blower fan.
Gents, I am sure you know that “fan” loads go up by iirc forth power of rpm…thus fans “self” regulate to some extent
For your organ..I would look at a sort of “gasometer” say a storage device vents when too much pressure ( think bell jar rising out of bath of water…pressure relates to height of rise and high enough exposes vents in side)
As to half bridge drives…
Plenty of chips with already optimised silicon..l297 say…
Free wheel diodes built in.thermal shutdown.
Hhay..can even buy as module on pcb with heatsink and screw terminal connection. ..
Btw l297 has enough half bridge to do two motors fwd and reverse

[Les Jones 1Participant @lesjones1]

Blocking the output of a centrifugal fan puts LESS load on the motor as all it is doing is rotating a fixed quantity of air. You can notice this with a vacuum cleaner. If you block the inlet. (Or outlet if accessible.) the motor speeds up.

Les.

[jason udallParticipant @jasonudall57142]

Les. Yep. Called dead heading I think..see the same in “centrifugal” pumps

[duncan webster 1Participant @duncanwebster1]

Thanks everyone, diode and transistor drive for the FET it is! I've learned a lot from the interchange. I had thought that the current flowing when FET off was because the motor was acting as a generator, but obviously not as that would entail current going the opposite direction

I've just checked the motor again, this time on full welly. If you block the output it definitely speeds up.

Jason, the idea of some kind of hubble bubble pipe bursting into life during quiet bits of music will take some time to fade! If I can't get the electronic one to work something like a weighted flap valve is the back up plan, the water idea certainly means very little pressure rise, just the bouyancy of the bit immersed changing. I reckon it would have to be passing all the time even at peak demand from the pipes to avoid bubbling, which would entail controlling the water level.

[MuzzerParticipant @muzzer]

Turning off a tap against a lot of water flowing produces "water hammer", which is caused by a rapid increase in the pressure across the tap as it tries to suddenly bring the column of water behind it to rest. You can't suddenly stop a large mass that's moving at a decent rate. The current flowing in an inductor (such as a motor winding) behaves in a similar way, so if you try to turn it off suddenly, there will be a rapid rise in voltage (pressure). Trouble is, if you try to turn a power semiconductor switch off slowly to reduce or control the voltage transient, you will make it very hot, so the skill is to switch it as quickly as you reasonably can. You have to manage this kind of switching event very carefully and it forms the very basis of just about any switch mode power electronics circuit.

The PWM circuit in the article is rather like a water pipe (current in motor) with a fast tap (the FET) – but with no relief for the water hammer (the snubber is too small to do anything). It's subjecting the switch to the resulting voltage transient. "Avalanche breakdown" is where the switch breaks down and conducts, rather like a pressure relief valve. However, as the switch is passing both current and voltage simultaneously, it sees a high transient power dissipation which isn't good for it unless in moderation. Modern MOSFETs are inherently able (and optimised) to withstand some limited repetitive avalanche energy which is usually specified in the datasheet, whereas bipolar switching transistors which were the only commercially viable option 20 years ago don't like it. When they start to break down through overvoltage, there is a runaway effect called secondary breakdown that results in device failure.

I've been working full time in power electronics since the early 80s and I learnt very early on how the basic topologies work and why you need a flywheel diode in a buck converter. The last thing I want to see is enthusiasts trying out electronics for the first time and ending up with circuit failures. I think we owe it to them to give good advice and circuit examples that are simple but robust and well designed.

On a lighter note and in the spirit of enjoyment and enthusiasm, here's an interesting example of an hydraulic system that delivers water to a reservoir some height above the source. It's pretty much an hydraulic analogue of a "boost converter" which can boost a low voltage (eg 12V) up to a higher voltage (50V or more). It's the basis of many circuits such as power factor correctors (PFCs) used on the front of many VFDs for boosting mains voltages up to 400Vdc or more. Many of us on this forum are fairly intuitive and practical, so hopefully will find this kind of electrical / mechanical analogue interesting. You can't extend the mechanical model to explain all concepts in electrical circuits but it's a start. Once you begin looking into electronics, you find there's some pretty interesting stuff going on….

**LINK**

Murray

PS – here, the FET is valve 4 and the diode is valve 5.

Edited By Muzzer on 25/05/2015 17:47:16

[jason udallParticipant @jasonudall57142]

Earlier I mentioned the L297…

I of course ment the L298..

[Anonymous]

Posted by Les Jones 1 on 25/05/2015 12:39:16:

The only argument that seems to make sense to me for NOT having a flywheel diode is to save a few pence.

When every penny counts, as on controllers for small lathes, strictly non-essential components tend to get missed off. In higher volumes, >100,000, even fractions of a penny can be important.

Andrew

[VersabossParticipant @versaboss]

Posted by Neil Wyatt on 20/05/2015 14:06:02:

Unfortunately the circuit drawing on page 32 of MEW 229 shows TR2 drawn as an NPN transistor.

A BC327 is in fact a PNP device and should be shown with the emitter and collector the other way round.

So finally the issue 229 landed in my postbox also, and after reading a bit diagonally through that article I'm still a bit lost.

I'm sure someone with sufficient electronics knowledge would have no problem, but I'm not among these…

So please Neil, do you mean that the emitter of BC327 should point upwards to the BC337 (and the arrow changed naturally), or stays as drawn and just the arrow drawn correctly?

A corrected drawing in the next issue is good, but a clear description even better…

Regards. HansR.

[John HaineParticipant @johnhaine32865]

Schottky diode overkill surely? They have two benefits. One, the forward voltage is lower than standard junction diode, but only by quarter volt or so. But this will make very little difference in this circuit as Trevor's oscillograms show.

Second, when they go reverse biased they turn off very quickly compared to junction diodes as the stored free charge is small. This is very useful in a high frequency rectifier, where combined with the low forward drop it gives good efficiency. But in this circuit the diode is reverse biased most of the time and turns on only to clamp the inductive kick.

I reckon an ordinary 1n4148 would do the job perfectly fine, the main thing is to protect the MOSFET against avalanche.

[TrevorGParticipant @trevorg]

Will a conventional silicon diode work ? Yes. But a schottky diode will do a better job for the reasons that John mentions.

I should however remind you the function of this diode is not just to clamp the inductive spike but also to redirect the energy into a productive use i.e. powering the motor. The schottky diode will recover more energy than the silicon diode.

For many of the uses of this type of controller the difference between the two types of diode is unimportant and either could be used. However, the schottky rectifier is better and these days is often cheaper the a conventional silicon rectifier of comparable rating.

For example for a motor running at 10A I would probably use a IN5822 schottky diode( cost 11p) Vf 0.5V @10a but you could use a silicon diode IN5401 (cost 14p) Vf 1.0V@10A.

Incidentally a !N4148 is a small signal diode rather than a rectifier and will not survive this usage (Vf=1.1V @0.2A).

Trevor

[John McNamaraParticipant @johnmcnamara74883]

Just My few Amps worth.

Electronics is not my thing, Although I have built many projects over the years from published art.

Reading through this post has been enlightening it is clear there is a rich source of expertise within the membership.

How about debating and publishing a circuit right here that embodies the group consensus. It may take a few iterations before the dust settles but the result should be rather good, Professionally designed in fact.

I have enjoyed the debate so far.

Regards
John

[Tony RayParticipant @tonyray65007]

Aw its gone awfully quiet, I was enjoying this !

general comments then some questions

General,

Yes do publish electronic from time to time, I find it interesting that this article has elicited so much debate whereas the 'wood butchery' does not.

If a comnon design could be a greed i'd like to see a group buy on PCB's

Questions

Firstly I can build circuits to a good standard but only have a basic knowlege of components & design – I can't be good at or have time for everything.

I have killed two 'bought a PWM speed controller on fleabay for £3.50, Its 555 driven FET with a schottky diod e and what look like a power diode for emf protection and but I'm reversing the polartity to the motor to allow reversing ( It goes through a zero position rotary switch) will that protection diode only work half the time i.e when the polarity is swtiched one way ?

If I buit the Inchanga design am I likley to blow that up too ?

My motor is 24v 1A brushed paravalux running @ 5000 is RPM but greared down and I'm running it at 24V as a powerfedd on my SX3 mill

Any help gratefully received

Tony

[John RuddParticipant @johnrudd16576]

Tony,

Without knowing the failure mode of the two speedos you have killed its difficult to determine whether the covered design will fail the same way….having re read my posting and further looks at the schematic, Inchanga's design seems robust enough, but I'm no expert….

As for the Schottky diode, as long as it is is connected from the Drain to the supply rail and all you are doing is switching the motor round then the diode should not fail,IMHO.

You state the motor used is 24v dc rated at 1 amp, but is that it's unloaded current, full load running or stalled current….these make a difference….

 

Just looking at the text of the article, it would appear that the device chosen is quite conservatively rated, so you shouldn't have any issue

Good luck with the build.

Edited By John Rudd on 15/07/2015 14:12:40

Edited By John Rudd on 15/07/2015 14:14:47

[TrevorGParticipant @trevorg]

Tony,

It would be interesting to see the circuit digram on the fleabay design.

All depends on how confident you feel on electronics. But if you feel able to trace out the circuit and put it online it would be interesting.

If it is the one I have just seen describesd as 6V/12V/24V 10A it should be possible to repair/uprate. What form did the failure take? Does the motor run only at full speed only or does it fail to run at all? I always doubt the specs on these Chinese controllers the are often hopelessly optimistic. As for tthe claim of over current protection this would need a microcontroller rather than a 555.

As John mentioned The 1A could mean idle current, full load current or stall current. For example I have a motor with 2A idle, 20A full load and 100A stall current. If you have the Parvalux model number it can be checked on the manufacturers website.

When you start a motor , until it starts to rotate it will pull the stall current, (if your power supply is up to it!) this will then drop off very rapidly as speed increases. If you then reverse the motor the instaneous current will be TWICE the stall current falling to the stall current as the motor stops before beginning to rotate in the reverse direction. This is a great way to fry your controller so it is a good idea to stop the motor before reversing it. Of course having a power supply too feeble to supply these huge currents can also offer some protection.

Trevor

[Neil WyattModerator @neilwyatt]

Inchanga has sent me some calculations on how his design copes with the back emf which will be going into the next 'scribe a line'.

Neil

[MuzzerParticipant @muzzer]

Any chance of seeing them before you commit to print again?

Murray

[TrevorGParticipant @trevorg]

Posted by Neil Wyatt on 15/07/2015 20:41:08:

Inchanga has sent me some calculations on how his design copes with the back emf which will be going into the next 'scribe a line'.

Neil

I to would like to see this. The issue has never been about back emf (just another name for the voltage generated by the motor due to its rotation) but rather about the inductive spike that occurs during switching. As both the back emf are and the inductive spike are well understood it is difficult to see what new insight there might be.

The inductive spike is well understood and intensively modelled by switch mode power supply designers. The buck regulator type of power supply, which is dependant on the capture of inductive spikes, is everywhere you look these days.

Trevor

[Author]

Posts