controlling 250V DC/AC with solid state

[Simon]

I need to have a control board switch DC or AC loads, due to the currents of up to 20A relays don't seem to be feasible for DC so I'm thinking solid state. SCR's don't seem to be practical at DC but I could be wrong but could a mosfet do the job ? for DC it would work fine as as the DS channel acts like a resistor and will pass current both ways will it work in AC too ?

[SeanB]

Look up some IGBT AC switches, basically a bridge rectifier, a big ass IGBT capable of handling the voltage and current, A VDR and snubber across the IGBT and bridge and finally an opto IGBT driver to provide the little wimpy voltage to turn it on and off. Handles DC, AC and anything in between.

[nctnico]

With an IGBT you'll be looking at 20 to 30W of losses. MOSFETs in series is also an option but parallel diodes may be necessary besides proper drive circuitry. I'd use a relay. Simple and reliable.

[Simon]

With an IGBT you'll be looking at 20 to 30W of losses. MOSFETs in series is also an option but parallel diodes may be necessary besides proper drive circuitry. I'd use a relay. Simple and reliable.

as i said in my op i have looked at relay's, find me one that will do 140V 20A DC (NOT AC) that does not cost the earth.

Another option may of course be to use a bridge rectifier in the case of AC systems

[nctnico]

An IGBT solution won't be cheap either. It wouldn't surprise me if a €20 relay turns out to be cheaper. Anyway, if you can use a bridge rectifier you could get away with a MOSFET. Be sure to use proper filtering and overvoltage protection 

[Simon]

find me a relay then, I have found relays that will handle 110V AC but only 28V DC

[Lurch]

Just to clarify, what is the current and voltages that you will be using?

A relay or contactor will need a different rating for AC and DC loads, but these ratings will also be dependent on the relevant voltages and load types so it isn't just a case of e.g. 250V/20A AC/DC. Last panel I assembled with 110VDC 16A switching was huge. You will need to de-rate some part of it for DC.

[Simon]

I'm powering a 1.5KW heating element and a 1 KW brushed motor both at 110-140V DC, yes as i said earlier any switching device has much lower capability at DC than AC, so to get anything that will cope on DC means humungus AC ratings so it's worth looking at solid state, I was thinking an SCR at first but was hoping to make it work for both AC and DC. Rectifying AC is not a huge obstacle and the motor only needs to run for a few seconds so we can cut out the heater when it is on to reduce the load greatly and save over designing for a few seconds of unnecessary simultaneous operation

[Lurch]

That's a decent relay you'll need there then! The ratings, and associated DC derating, will be related to inrush and arcing at switch on so you may not necessarily need to turn anything off, or rather, turning stuff off may not actually make much difference WRT switch/relay/contactor sizing.

[SeanB]

If you are using a bridge to supply the motor with DC just use a SSR in the AC feed to the bridge then, or even the AC rated relay, as it will only be switching AC. Otherwise if you have to use DC look at SCR commutation using another SCR and a capacitor to pulse the anode negative enough to switch it off.

[Simon]

If you are using a bridge to supply the motor with DC just use a SSR in the AC feed to the bridge then, or even the AC rated relay, as it will only be switching AC. Otherwise if you have to use DC look at SCR commutation using another SCR and a capacitor to pulse the anode negative enough to switch it off.

You are sort of looking at this from the wrong end, I do not have a problem with A/C control of the motor or heater they are in fact both A/C devices the problem is controlling them with DC because of the issues with mechanical contacts. But while we are solving with DC problem I am trying to ensure that the same circuitry can also switch A/C or we could go down the route of having a separate power control board so that just this part can be swapped over for A/C or DC operation or for A/C operation relays can be used and then be replaced with the control board for DC.

An SCR sound like the easiest solution I'm just a bit apprehensive about how to turn it off with DC but then at the end of the day I will not be designing the nitty-gritty of this circuit as it is being subcontracted by my employer but in order to speed things up and ensure we don't get lost in unlikely solutions I'm looking for some options to talk to our subcontractor about.

The problem we have is that the whole unit needs approving for rail and we could be getting requests for A/C or DC versions so I am trying to avoid a retest for the second batch should they be A/C units.

[Lurch]

You are sort of looking at this from the wrong end, I do not have a problem with A/C control of the motor or heater they are in fact both A/C devices the problem is controlling them with DC because of the issues with mechanical contacts. But while we are solving with DC problem I am trying to ensure that the same circuitry can also switch A/C or we could go down the route of having a separate power control board so that just this part can be swapped over for A/C or DC operation or for A/C operation relays can be used and then be replaced with the control board for DC.

I would expect that if it is rated for 20A@120VDC then it will most certainly not be a problem to also be used for 20A@400V AC

[SeanB]

OK, so your incoming supply will go to a bridge rectifier, and then the 2 loads will need power switching. Then you will need a heatsink at least for the bridge ( 15A units will need it for sure, and if you are doing rail 40A would not be overkill) and then 2  IGBT or power mosfet, or even bipolar transistors will do, select on voltage drop at the current of 10A or so to get the lowest cost device that runs the coolest. Those get a heatsink with the bridge, and if you are willing to make the bridge out of individual TO220 or TO247 diodes then all fit on one heatsink on edge.

[nctnico]

I'd let the subcontractor deal with the problem. Maybe he/she already has a clever solution but goes with the (more expensive) one you are providing. There is nothing more irritating than a customer with an idea on what the solution should look like.

[Simon]

I'd let the subcontractor deal with the problem. Maybe he/she already has a clever solution but goes with the (more expensive) one you are providing. There is nothing more irritating than a customer with an idea on what the solution should look like.

So far anything that they have produced i have had to fix and taking up a discussion about something that needs doing generally involves them making it sound like it needs to be more complicated than really need be. I know a customer with an idea can be a pain, but a supplier with no idea or too big an idea is worse when we are paying.

[Simon]

You are sort of looking at this from the wrong end, I do not have a problem with A/C control of the motor or heater they are in fact both A/C devices the problem is controlling them with DC because of the issues with mechanical contacts. But while we are solving with DC problem I am trying to ensure that the same circuitry can also switch A/C or we could go down the route of having a separate power control board so that just this part can be swapped over for A/C or DC operation or for A/C operation relays can be used and then be replaced with the control board for DC.

I would expect that if it is rated for 20A@120VDC then it will most certainly not be a problem to also be used for 20A@400V AC

So go and find me a relay rated for 200V DC that will switch a 20A load..........

[nctnico]

I'd let the subcontractor deal with the problem. Maybe he/she already has a clever solution but goes with the (more expensive) one you are providing. There is nothing more irritating than a customer with an idea on what the solution should look like.

So far anything that they have produced i have had to fix and taking up a discussion about something that needs doing generally involves them making it sound like it needs to be more complicated than really need be. I know a customer with an idea can be a pain, but a supplier with no idea or too big an idea is worse when we are paying.

Why not hire someone else? There are dozens of small companies to choose from. Just put an add on a forum.

[Kremmen]

Well, for example the GA 75 type DC contactor from my onetime employer ABB can do that and more. The series has much larger ones as well.
If you google 'ga dc contactor site:au' the first link will be to the relevant product page. As a general note you will find that ABB can provide a solutio  for almost any electrical or power electronics application. No idea about the prices though - other than that the Chinese will be cheaper

[Lurch]

So go and find me a relay rated for 200V DC that will switch a 20A load..........

DC contactor brochure;

http://www05.abb.com/global/scot/scot209.nsf/veritydisplay/f25bae3eb7bde2e0c12578cd0029dca9/$file/1SFC101004B0201.pdf

Specifically;

http://www.ng.abb.com/product/seitp329/12daed844051a599482570f30045c9e5.aspx?tabKey=2&gid=ABB1SFL557025R7011&cid=9AAC100109

[AlfBaz]

It's mostly about the breaking capacity get a 3 phase contactor run one leg through one pole and series the other through two poles

[Lurch]

It's mostly about the breaking capacity get a 3 phase contactor run one leg through one pole and series the other through two poles

Bad idea.

[Kremmen]

Series connection is actually a technique to use but you won't turn a regular AC contqctor into a DC one quite that easily. A proper DC contactor is qualitatively different in that there are special arrangements to extinguish the arc when contact breaks.  You won't find those in a regular contactor.
Here is a link to the ABB contactors catalog where you will find the DC ones from page 207 onwards: http://search-ext.abb.com/library/Download.aspx?DocumentID=1SXU000023C0203_01&LanguageCode=en&DocumentPartId=&Action=Launch

[Simon]

So go and find me a relay rated for 200V DC that will switch a 20A load..........

DC contactor brochure;

http://www05.abb.com/global/scot/scot209.nsf/veritydisplay/f25bae3eb7bde2e0c12578cd0029dca9/$file/1SFC101004B0201.pdf

Specifically;

http://www.ng.abb.com/product/seitp329/12daed844051a599482570f30045c9e5.aspx?tabKey=2&gid=ABB1SFL557025R7011&cid=9AAC100109

and how much do they cost ? probably hundred of pounds ? so cheaper to use solid state devices

[Lurch]

So go and find me a relay rated for 200V DC that will switch a 20A load..........

and how much do they cost ? probably hundred of pounds ? so cheaper to use solid state devices

See, now you're just being awkward!

[Simon]

No I have a product to get designed, I can't justify spending money on solutions that have cheaper alternatives, at the moment for all I can we can use a mosfet for the 110-150V DC and then use relays for ac if it's the cheapest solution and the customer will have to pay for testing two solutions although I'm trying to avoid it.

[Lurch]

the customer will have to pay for testing two solutions although I'm trying to avoid it.

Why use 2 separate products? Maybe this is me not understanding exactly what you are testing or why but can't you just make a single board with a relay and MOSFET's and then just test that rather than making 2 separate boards?

[Simon]

the customer will have to pay for testing two solutions although I'm trying to avoid it.

Why use 2 separate products? Maybe this is me not understanding exactly what you are testing or why but can't you just make a single board with a relay and MOSFET's and then just test that rather than making 2 separate boards?

We are making units for trains so they have to undergo stringent tests. The power available in trains seems to vary from 110V AC/DC to 240V AC/DC or at least that is the range we want to cover to try and make one control board that will cover all possibilities. If we can't use the same circuit for both AC and DC we may have to make separate power switching boards that get swapped for AC or DC versions with a common main control board. As it is we can't have that sort of current on a PCB so need off board switching anyway.

[Codemonkey]

You've not said how much money you consider to be too much. There are plenty of relays and contactors that will switch a 140V 20A DC load without any problem, just as there are many solid state relays that will do the job. I can't imagine trying to re-invent the wheel here being any cheaper than just buying a mass produced off the shelf component.

I used to design emergency lighting systems that ran at ~110VDC or AC, with loads of anything up to about 200A and we never had any problem finding a relay/contactor to do the job.

[Simon]

Well the current AC design for 240V has standard relays, I guess they cost at most a couple of pounds. I''m after the next cheapest option.

[AlfBaz]

Series connection is actually a technique to use but you won't turn a regular AC contqctor into a DC one quite that easily. A proper DC contactor is qualitatively different in that there are special arrangements to extinguish the arc when contact breaks.  You won't find those in a regular contactor.

I must be missing something. Whilst some contactors don't have major arc drawing shields most of them have some sort of metallic flutes to achieve some degree of it. Secondly you are breaking the load in 6 places, 2 on the (pick one) negative leg and 4 on the positive. Are you suggesting that at 120V 20A DC the energy will be sufficient to keep arc's going on all six break points or are you envisioning the arc drawing out of the contact, ionizing the external air and flashing across the outside of the contactor?

I'm not trying to be obstinate I am genuinely interested. I've seen this technique used more often than I care to mention and am puzzled at these concerns

[Simon]

I had considered a capacitor between contacts so that the "pulse" from breaking contacts goes through/charges the capacitor as an alternative path to the breaking contacts but what do you do with the charged capacitor, as soon as you close the contacts again you will have a massive short on the capacitor and probably do more damage unless extra contacts/relays can then cut the dead load off and discharge the capacitor ready for the make.

[Lurch]

I must be missing something. Whilst some contactors don't have major arc drawing shields most of them have some sort of metallic flutes to achieve some degree of it. Secondly you are breaking the load in 6 places, 2 on the (pick one) negative leg and 4 on the positive. Are you suggesting that at 120V 20A DC the energy will be sufficient to keep arc's going on all six break points or are you envisioning the arc drawing out of the contact, ionizing the external air and flashing across the outside of the contactor?

I'd say a little of both. There's also the inrush currents to consider, which under fault conditions could be pretty major and you'd end up with a massive burning thing where your contactor was. There's also the utilisation category, and if you just slap a standard AC-1/AC-3 contactor in and run 120VDC@20A through it and anything ever goes wrong it will be your fault regardless. You could technically get away with doing it as long as nothing ever drifted outside of the perfect conditions you tested it in, but I wouldn't want to put my name to it.

[AlfBaz]

I'd say a little of both. There's also the inrush currents to consider, which under fault conditions could be pretty major and you'd end up with a massive burning thing where your contactor was.

Inrush currents? Not quite sure how they pertain to the problem at hand which is using a contactor designed for AC to break DC and under fault conditions the problem is not merely constrained to dc and should be handled by protection up stream

There's also the utilisation category, and if you just slap a standard AC-1/AC-3 contactor in and run 120VDC@20A through it and anything ever goes wrong it will be your fault regardless. You could technically get away with doing it as long as nothing ever drifted outside of the perfect conditions you tested it in, but I wouldn't want to put my name to it.

I was hoping for a better explanation. You've simply expanded on "Bad idea" without adding the reason why. If you choose the appropriately size contactor for the application I still cannot see where the problem lies
As for the "utilisation category" it's used to de-rate  the device depending on the load and again if something goes wrong you need to analyse the hazards wether you are use AC or DC and are going to have to implement other protection features such as monitoring for voltage at the contactors output when no power is applied to the coil and possibly energising  a shunt trip on an upstream CB that's feeding the rectifier.

[SeanB]

Just remember the fuses and circuit breakers you use have to be DC rated as well. For breakers that means big and massive arc chanbers and for fuses HRC sand filled units. AFAIK the only ones rated for DC are Diazed fuses, and a few ABB motor control fuses. In any case they are not the cheapest units.

[peter.mitchell]

I'd say a little of both. There's also the inrush currents to consider, which under fault conditions could be pretty major and you'd end up with a massive burning thing where your contactor was.

Inrush currents? Not quite sure how they pertain to the problem at hand which is using a contactor designed for AC to break DC and under fault conditions the problem is not merely constrained to dc and should be handled by protection up stream

There's also the utilisation category, and if you just slap a standard AC-1/AC-3 contactor in and run 120VDC@20A through it and anything ever goes wrong it will be your fault regardless. You could technically get away with doing it as long as nothing ever drifted outside of the perfect conditions you tested it in, but I wouldn't want to put my name to it.

I was hoping for a better explanation. You've simply expanded on "Bad idea" without adding the reason why. If you choose the appropriately size contactor for the application I still cannot see where the problem lies
As for the "utilisation category" it's used to de-rate  the device depending on the load and again if something goes wrong you need to analyse the hazards wether you are use AC or DC and are going to have to implement other protection features such as monitoring for voltage at the contactors output when no power is applied to the coil and possibly energising  a shunt trip on an upstream CB that's feeding the rectifier.

Yeah, i'm not seeing it either, I mean, Inrush currents? At the same RMS voltage, AC would provide significantly more inrush current, if it closed near the peak of the cycle. As for current derating, it isn't an issue because the RMS heat would be the same... The only issue i can see is quenching the arc, which was the purpose of using 3 phase contactors with the contactors in series.
THAT SAID, if you are using a 3 phase contactor, and in series, i would say use one rated for 3x the current, as generally, current is expected to be shared across the phases/contacts equally, rather than the total current going through each contact individually.

The loads you have specified, a heating element, and a DC brushed motor should have no issue being switched in this method, however, as you will need approval from the rail regulatory body, you probably won't pass with this solution.

Over rated, DC capable SSR is the solidstate way to go, but they are very expensive, a massive DC contactor would be cheaper, by far.

[Lurch]

Yeah, i'm not seeing it either, I mean, Inrush currents? At the same RMS voltage, AC would provide significantly more inrush current, if it closed near the peak of the cycle. As for current derating, it isn't an issue because the RMS heat would be the same...

I thought I saw some mention of an inverter somewhere, maybe I have misunderstood something along the way so I'll retract that if it's all wrong!

however, as you will need approval from the rail regulatory body, you probably won't pass with this solution.

I think that's what I was trying to get at.

[Kremmen]

Series connection is actually a technique to use but you won't turn a regular AC contqctor into a DC one quite that easily. A proper DC contactor is qualitatively different in that there are special arrangements to extinguish the arc when contact breaks.  You won't find those in a regular contactor.

I must be missing something. Whilst some contactors don't have major arc drawing shields most of them have some sort of metallic flutes to achieve some degree of it. Secondly you are breaking the load in 6 places, 2 on the (pick one) negative leg and 4 on the positive. Are you suggesting that at 120V 20A DC the energy will be sufficient to keep arc's going on all six break points or are you envisioning the arc drawing out of the contact, ionizing the external air and flashing across the outside of the contactor?

I'm not trying to be obstinate I am genuinely interested. I've seen this technique used more often than I care to mention and am puzzled at these concerns

Of course any relay or contactor will make and break any voltage/current until it is destroyed or decayred beyond repair. Unless you grossly overload the unit, breakage won't happen right away, and maybe not for a long time. A relay / contactor is an electromechanical device so it doesn't have a similar abrupt breakdown behavior as say a semiconductor switch. So you may get away with abusing a device for some considerable time. That doesn't mean it is a sound design principle however. When i was in that business it was a strict no-no to even contemplate applying devices beyond their approved ratings. Depending on the design culture YMMV and so on.

A proper DC contactor has in addition to the arc snubber chambers, a set of permanent magnets to guide the arc for maximum snubbage. These are significantly lacking in an AC contactor where they would not function as intended due to the oscillating polarity of AC currents and voltages.

And yes, they will be expensive - even exorbitant in price compared to the cheapo €1.25 Chinese AC contactors that barely manage the decent minimum in AC, for a time. But then we are not playing in the same ballpark either. Whether a DC SSR is any cheaper i have no idea - this is a question that should be evaluated early in the product life cycle to avoid expensive/painful dead ends halfway through the design schedule.

[peter.mitchell]

Well, it looks like you'll probably have to roll your own SSR if you want to use an SSR; if it has to do 25A, 150~VDC, 240VAC I can't find any.
The back to back FET ones only go to 300VDC, which isn't enough for 240VAC, and the higher DCV ones are IGBT, which don't work with AC.
This is a digikey listing of relays/contactors, it is weird though; that OMRON one looks like a bit of an outlier...

[Zero999]

Are you intending to use the same component for both 120VDC/AC and 230VAC?

I found a 200VDC 40A SSR in RS, otherwise if you want the same component for all systems, you'll have to make your own.
http://uk.rs-online.com/web/p/solid-state-relays/7360879/

[peter.mitchell]

Are you intending to use the same component for both 120VDC/AC and 230VAC?

I found a 200VDC 40A SSR in RS, otherwise if you want the same component for all systems, you'll have to make your own.
http://uk.rs-online.com/web/p/solid-state-relays/7360879/

I believe he was intending to use the same device/setup with both supplies, to save manufacturing steps and part selection confusion thus the need to work at 110-140VDC and 240VAC,

[Simon]

Uh i seem to have missed some replies/emails. Yes as you see a solid state device that may do two costs a lot: £63, we need two, I have found a mosfet that will do 250V with about 0.0175 ohm of full on resistance that raises to 3 times that when heated and it costs under £5. We have spoken to the subcontractor and they are going to redesign the whole thing, we will have one main control board and then separate switching gear so that we don't have large currents in the board and we can change the switchgear as we get new project options thrown at us. If they want AC again a simple relay solution will do it. They seem to think that they can do a universal AC/DC solution....... I'll let them think about it before finding out they came to the same conclusion as me.

[oldway]

I'm powering a 1.5KW heating element and a 1 KW brushed motor both at 110-140V DC

I don't think it's possible to switch on a 1KW brushed motor at 110-140V DC without a current limiting resistor.
Inrush current would be very high and would probably damage the motor and the SSR.

[Simon]

it is something at the back of my mind, it is a vacum cleaner motor, do they put starting resistors in vacum cleaners ? or is the DC a problem ?

[SeanB]

Dc is the problem. Take your 1kW motor and measure the DC resistance and calculate the current it draws until it gets up to speed. The current will be in the multi kW range for the first second or so if the supply allows it. With an AC motor the coil inductance limits the peak current to the current flowing for the first half cycle to magnetise the core.

[Simon]

hm, would an inductor in series be any use ? as the current tries to rush in it will opose it and effectively give some "soft start". Alternatively can gradually increasing PWM be used to soft-start ?

[SeanB]

PWM would work as soft start. Inductor would be bloody big and even more expensive.

[Zero999]

For a soft start, you could connect a resistor in series for a second, then bypass it with a suable low resistance switch.

Isolation can be achieved with a photovoltaic opti-coupler.
http://www.vishay.com/docs/83469/vom1271t.pdf
http://uk.mouser.com/ProductDetail/Vishay/VOM1271T/?qs=%2fha2pyFaduiCbdPYawMPDbvzpZNW3JFS40Vo2Z60Tow%3d

[Simon]

Well I suppose given that the power mosfet I found or something similar that our supplier may want to use is less than a fiver so no harm in having 2, one with a resistor in series and one that gives a direct connection once it has started up.

[oldway]

The power on must be done in 2 steps:
- first, power on with a current limiting power resistor in serie with the motor.
- second, after current drops below a reference value, a second Mosfet will short de resistor.
As the power resistor cann't stay in circuit fot a while without overheating, you must have an overload trip after several seconds if current did not drop below the reference value.
You should then wait a time sufficient to provide enough cooling of the resistor, then try to power on again.
After three times, you should stop and show a warning.

[IanB]

it is something at the back of my mind, it is a vacum cleaner motor, do they put starting resistors in vacum cleaners ? or is the DC a problem ?

While we are talking about trains, they certainly do (did) put series resistors on DC traction motors on trains when starting up. There are various steps of series resistor when going from standstill to full power. I don't think you could apply full power to the traction motors when the train is standing still. There would be fireworks.

(I say "did" above because modern train designs no longer use DC traction motors. They rather use AC motors with electronic drives.)

[Simon]

Well I can't see a problem with a mosfet with resistor in series and another mosfet that has the source in common with the other one but the drain connected to the top of the resistor so that first the resistor mosfet turns on and then it changes for the full power resistor.

[oldway]

You could use 2 SSR of the serie 1-DC of CRYDOM for doing this, 01 of 7A (D2D07) for starting with resistor and 01 of 12A (d2d12) for working.

[Simon]

I don't know how high a value a resistor we could get away with to ensure that on direct powering it does not draw too much power, does the current to speed up from a low to high speed depend much on the speed you start at ? We could use say a resistor that will only let 5A through with full voltage across (start up) it that would drop to 2.5A once the speed stabilizes before the direct powering cuts in so only 125ish watts for any amount of time. 12A is cutting it a bit fine for constant running current. We could see up to 140V at full charge.

[IanB]

I don't know how high a value a resistor we could get away with to ensure that on direct powering it does not draw too much power, does the current to speed up from a low to high speed depend much on the speed you start at ? We could use say a resistor that will only let 5A through with full voltage across (start up) it that would drop to 2.5A once the speed stabilizes before the direct powering cuts in so only 125ish watts for any amount of time. 12A is cutting it a bit fine for constant running current. We could see up to 140V at full charge.

The start up current drawn by a typical DC motor depends on the load it sees when stationary.

With the railway traction motor I mentioned above, the load at start up is enormous--there is this giant heavy train to start moving. You may as well try to move a brick wall. The motor on start up will be stalled and unable to turn. You have to ramp the power up slowly to keep the motor from burning up.

With a free running blower motor the situation is very different. The power consumed by a blower varies approximately with the cube of the speed, so when the motor is stationary the opposing load is nearly zero. All the motor has to do is overcome inertia to speed the thing up. It will speed up very quickly. Now with a DC motor as soon as it starts turning it will generate back EMF that opposes the supply voltage and reduces the current draw. So if the motor speeds up quickly the start up current will reduce very quickly.

In short, you may not need any current limiting on start up. The best way to determine if this is the case is to put the thing on the bench and measure it under expected operating conditions. Uncertain predictions and guesswork will never serve you well in industrial engineering. Hard data and reliable calculations are what you need.

[oldway]

My Nilfisk 1000W 220V vacuum cleaner motor has a resistance of 12.8 Ohms
A 110V motor would have a resistance four times lower (for the same losses) , this would be 3.2 Ohms.
Inrush current would be 140V/3.2 = 43.75A.
This is far too high for the motor.
Nominal current is 9.1A at 110V (1KW)
15A inrush current seems to be a good choice, as motor start with no load.
With 140V dc, total resistance must be 9.34R.
Then current limiting resistor should be: 9.34 - 3.2 = 6.14 Ohms.
A 5 Ohms power resistor would do the job.
Inrush current: 140V/(5 + 3.2)Ohms = 17A
Seems to be OK.

[IanB]

Inrush current would be 140V/3.2 = 43.75A.
This is far too high for the motor.

I think this depends on how long the inrush current lasts? (What we really care about is the integral of power dissipation over time, not the peak current per se.) If the inrush current reduces to a lower value in a fraction of a second then it may not be an issue. We also may consider the resistance of the power supply wiring and the supply impedance. Voltage drops in the supply wiring may even prevent a current as high as 40 A from occurring.

[Zero999]

it is something at the back of my mind, it is a vacum cleaner motor, do they put starting resistors in vacum cleaners ? or is the DC a problem ?

No they don't put resistors in vacuum cleaners and the surge shouldn't be that much greater with DC than AC.

I forgot it's a vacuum cleaner motor. It will be fine, as long as the MOSFET is suitably rated to handle the surge and the circuit breaker doesn't cut out too quickly.

[oldway]

No they don't put resistors in vacuum cleaners and the surge shouldn't be that much greater with DC than AC

For sure it is.
In DC, surge is only limited by resistance who is very low to reduce losses.
In AC, current is limited by impedance, that's resistance AND inductance.
Inductance is high because we have the armature (more or less 1mH) in serie with two field coils.
As the air gap is large, this inductance normally do not saturate.
For this reason, surge in AC is readily low in comparison with surge in DC.
Vaccum cleaners are not intended to work with dc power grid.
They even often use a triac power control who don't work with dc voltage.

With surge overload, problem is not overheating, but bad comutation: you can have a "flash" between motor brushes, what can damage the motor.
Overload of more than four times the nominal current is not acceptable in a DC/AC universal motor.

[Zero999]

No they don't put resistors in vacuum cleaners and the surge shouldn't be that much greater with DC than AC

For sure it is.
In DC, surge is only limited by resistance who is very low to reduce losses.
In AC, current is limited by impedance, that's resistance AND inductance.
Inductance is high because we have the armature (more or less 1mH) in serie with two field coils.
As the air gap is large, this inductance normally do not saturate.
For this reason, surge in AC is readily low in comparison with surge in DC.
Vaccum cleaners are not intended to work with dc power grid.

You're right there will be a larger surge on DC than AC but not as much as you may think. There is magnetic coupling between the field and armature which forms a transformer with a shorted secondary when stationary.

Either way, measure the DC resistance and calculate the maximum surge from that - there's no way it can exceed that.

They even often use a triac power control who don't work with dc voltage.

The intention is to use the motor on its own so that isn't an issue.

With surge overload, problem is not overheating, but bad comutation: you can have a "flash" between motor brushes, what can damage the motor.
Overload of more than four times the nominal current is not acceptable in a DC/AC universal motor.

What do you mean? There will be arcing whether it's AC or DC.

[Zero999]

SCR's don't seem to be practical at DC but I could be wrong but could a mosfet do the job ? for DC it would work fine as as the DS channel acts like a resistor and will pass current both ways will it work in AC too ?

SCRs can be used with DC. The only trouble is getting them to turn off is tricky.

The most suitable method here would probably be to use two SCRs and a capacitor. You'd need optically couple the SCRs somehow for isolation.

http://www.sentex.ca/~mec1995/tutorial/triacs/triacs.html
http://www.onsemi.com/pub_link/Collateral/HBD855-D.PDF
http://www.most.gov.mm/techuni/media/EP02024_6.pdf

MOSFETs can be connected back-to-back for AC.

[IanB]

In DC, surge is only limited by resistance who is very low to reduce losses.

No, surge is limited by inductance. The motor is an inductance, that as you said, does not saturate. As with all inductors the current is zero when voltage is first applied and then rises as the magnetic field builds up. However, since the motor is unloaded it will start turning immediately and will speed up rapidly. In doing so the spinning motor will generate back EMF that will counter the build up of current. So the start up surge current will be far less than the current through an equivalent resistance.

[oldway]

You're right there will be a larger surge on DC than AC but not as much as you may think. There is magnetic coupling between the field and armature which forms a transformer with a shorted secondary when stationary.

Either way, measure the DC resistance and calculate the maximum surge from that - there's no way it can exceed that.

I don't agree. There is no shorted secondary when stationary. This is only the case with AC asynchroneous motors, not with DC/AC universal motors...

What do you mean? There will be arcing whether it's AC or DC.

You should read some technical papers about armature reaction to understand why there are comutation problems with high overloads.
That's what happens when you start a DC/AC universal motor in DC without current limiting resistor.
In AC, it does not happen because current is limited to a far lower value by inductance.
http://web.uettaxila.edu.pk/CMS/SP2012/etEMbs/notes%5Carmature%20reaction%20&%20commutation.pdf

[oldway]

SCRs can be used with DC. The only trouble is getting them to turn off is tricky.

That is a very big trouble
You also have to use fast turn-off thyristors...
Not good at all 

[oldway]

In DC, surge is only limited by resistance who is very low to reduce losses.

No, surge is limited by inductance. The motor is an inductance, that as you said, does not saturate. As with all inductors the current is zero when voltage is first applied and then rises as the magnetic field builds up. However, since the motor is unloaded it will start turning immediately and will speed up rapidly. In doing so the spinning motor will generate back EMF that will counter the build up of current. So the start up surge current will be far less than the current through an equivalent resistance.

I don't agree: time constant is very short compared with mecanical inertia of the motor so current reach the static value (u/r)well before motor can generate a back EMF.

[Simon]

Well i will have to measure the motor but I found a mosfet that will do err I think around 70A with a peak of 300A (not sure how long it can last for) I'll have to look at the datasheet when i get back into work tomorrow. I can measure a similar 240V motor to get an idea but I'll have to wait for the actual motor to arrive to be sure.

[Simon]

How about using the mosfet as the limiting resistor ? if the gate voltage ramps up over a period (say 1s) it will act as the resistor and as the full on drive, so one mosfet not 2 and no resistor.

[SeanB]

That can work, but be careful, as the instantaneous power dissipation of the die can very easily be exceeded. Though it might be rated for 300A peak pulse ( likely a limit from the bonding wires) the instantaneous power can limit the amount of power it can handle as a slowly switching device, it can have hot spotting and detonate.

[Simon]

I'll check the datasheet tomorrow see what power it can take, if the motor resistance is not too low and won't make the start current higher than 70A then we won't get huge spikes through the mosfet but we can use the mosfet to create some resistance to help ease the start. We're running off the batteries on the train kept topped up by alternators, I don't know how long the cabling is and how thick but no doubt it will also act as a starting resistor.

[calexanian]

Using a Bridge Rectifier is a simple solution if the forward voltage drop is not an issue. You can use any BJT or FET but insure the driving source is isolated so you are not shorting across anything.

[Simon]

Yes the problem is we are carrying up to 20A so no mean feat for a bridge and quite a lot of heat to dump in a case, also the AC option will be more likely when we are talking 240V this means we will need a mosfet that can handle even more voltage as the peak will be 340V on a good day so we need to look at mosfets for 5-600V which will start to get costly plus the cost of the bridge and heat sink, we could just revert to relays.

[Zero999]

You should read some technical papers about armature reaction to understand why there are comutation problems with high overloads.
That's what happens when you start a DC/AC universal motor in DC without current limiting resistor.
In AC, it does not happen because current is limited to a far lower value by inductance.
http://web.uettaxila.edu.pk/CMS/SP2012/etEMbs/notes%5Carmature%20reaction%20&%20commutation.pdf

All right, I don't have time to read the entire paper but assuming you're right about the increased arcing when overloaded, in this case the motor is driving a fan so the overload lasts for under a second, is it really a big problem? It think this is a case of test it and see what happens.

[SeanB]

It is on a train, not really a place for suck it and see. The reason a train car is 30 tons in weight is because the other cars are 30 tons in weight. Trains are not a place for gentle, you have to expect horrible spikes, surges, lightning and such, and that is for a light bulb.

[Simon]

The motor is a hand drier motor so no real mechanical load, as it happens I am using a vacum cleaner motor (off the shelf part and is the same type of motor) that is made to take some stick as it has to suck through a filter where as I have no such filter. the mechanical load of a fan increases I think to the square of the speed so the actual load during start-up is near non existent.

[Zero999]

I don't agree. There is no shorted secondary when stationary. This is only the case with AC asynchroneous motors, not with DC/AC universal motors...

I can tell you that's not true. The armature winding is magnetically coupled to the field. Connect 12VAC to the field coils and you'll measure a small AC voltage on the armature when it's stationary. If the armature is short circuited, it will vibrate a bit and will spin in either direction as it operates as an induction motor.

In the case of a vacuum cleaner, the armature is connected in series with the field so the stall current will be different, depending on the phasing.

You should read some technical papers about armature reaction to understand why there are commutation problems with high overloads.
That's what happens when you start a DC/AC universal motor in DC without current limiting resistor.
In AC, it does not happen because current is limited to a far lower value by inductance.
http://web.uettaxila.edu.pk/CMS/SP2012/etEMbs/notes%5Carmature%20reaction%20&%20commutation.pdf

That article concerns shunt wound machines, not series wound.

The motor is a hand drier motor so no real mechanical load, as it happens I am using a vacum cleaner motor (off the shelf part and is the same type of motor) that is made to take some stick as it has to suck through a filter where as I have no such filter. the mechanical load of a fan increases I think to the square of the speed so the actual load during start-up is near non existent.

I thought there's no point in discussing this so I thought I should test it.

I tried a standard 230VAC vacuum cleaner motor + bridge rectifier connected to the mains and it works perfectly. There's hardly any difference in arcing, whether it's run off AC or DC. The motor starts as usual (I didn't notice any massive arcs) and operates in exactly the same manner as it would do if run directly from the mains.

This is with an unfiltered DC supply. I don't have access to a smooth 230VDC supply otherwise I would test it. Unfiltered DC will have a lower peak voltage but 100% duty so the two effects could cancel one another out. Try powering it from a variac and add a large filter capacitor.

The motor is a giant hunk of iron and copper. I'd be more worried about the semiconductors which need to be rated to with stand the surge.

EDIT:
The bridge rectifier I used was GBPC808 which is rated for 800V and although the current rating probably wasn't high enough, it survived the test but I only ran it for a minute or so. A MOSFET may not have such a high surge capacity. If it's a problem, try the SCR method but again you're faced with the problem of getting it to turn off.
http://pdf.datasheetarchive.com/datasheetsmain/Datasheets-20/DSA-390489.pdf

[IanB]

the mechanical load of a fan increases I think to the square of the speed

Probably closer to the cube of the speed in fact. The developed head varies with the square of the speed and the flow varies directly with the speed. The power consumed is proportional to the head times the flow.

So yes, the load on start up consists only of the inertia of the rotating parts and the friction from the bearings and gearbox.

[Simon]

well it's a 20Krpm motor so it will laugh at the inertia of a light sheet metal impeller

[Simon]

the motor does achieve full speed very quickly, there is quite a jolt of the motor on start up from the recoil, if it's not held firmly down it will shoot across the benh - yes voice of experience speaking.

[IanB]

well it's a 20Krpm motor so it will laugh at the inertia of a light sheet metal impeller

The motor itself has inertia, which is why it shoots across the bench...

[Simon]

well yes that too, what I meant is that apart from the armature weight it has very little weight to sin up the only real loading comes from it moving air so it really takes off on power up, it's definitely at full speed within a second

[oldway]

I can tell you that's not true. The armature winding is magnetically coupled to the field. Connect 12VAC to the field coils and you'll measure a small AC voltage on the armature when it's stationary. If the armature is short circuited, it will vibrate a bit and will spin in either direction as it operates as an induction motor.

In the case of a vacuum cleaner, the armature is connected in series with the field so the stall current will be different, depending on the phasing.

That you measure a small voltage on the armature when you feed the fields with ac proves that the armature is NOT a  short circuited secondary.

That article concerns shunt wound machines, not series wound.

Armature reaction happens as well with both. DC traction motors (series) have armature reaction compensation.

About starting universal series motor with DC without current limiting resistor:
My Nilfisk 1000W 220V vacuum cleaner motor has a resistance of 12.8 Ohms, inductance is 46mH.
Assuming a cos phi= 0.8, nominal current should be 5.68A. (measured current = 5.3A)
Impedance (50Hz) = 19.3 Ohms.
Time constant (L/R) = 3.59 ms
Inrush current with 220Vac: 11.4A (two times nominal current)
Inrush current with 220Vdc without current limiting resistor: 17.2A  (three times nominal current)
Inrush current with 280Vdc without current limiting resistor: 21.87A (3.85 x nominal current)

[Zero999]

That you measure a small voltage on the armature when you feed the fields with ac proves that the armature is NOT a  short circuited secondary.

It proves the armature acts as a secondary winding. If it's put in series with the field coils, the current induced in the armature will either be added or subtracted to the surge current, depending on the phasing. To find out which, you'd need to test the phasing of the motor, which I didn't do.

About starting universal series motor with DC without current limiting resistor:
My Nilfisk 1000W 220V vacuum cleaner motor has a resistance of 12.8 Ohms, inductance is 46mH.
Assuming a cos phi= 0.8, nominal current should be 5.68A. (measured current = 5.3A)
Impedance (50Hz) = 19.3 Ohms.
Time constant (L/R) = 3.59 ms
Inrush current with 220Vac: 11.4A (two times nominal current)
Inrush current with 220Vdc without current limiting resistor: 17.2A  (three times nominal current)
Inrush current with 280Vdc without current limiting resistor: 21.87A (3.85 x nominal current)

That occurs for under a second so is not going to overheat the motor, which has a huge thermal time constant. During the test I conducted both the and 5A top fuse in the plug survived, so as long as the semiconductors are rated to stand the surge, it won't be a problem.

[Simon]

I found a IRFP4768Pbf it will take up to 93A continuos at 25C, 66A at 100C and will withstand 370A pulses the width of which are "limited by the junction temperature" which is limited to 175C

[oldway]

@Hero999:
Please read what i wrote:

You should read some technical papers about armature reaction to understand why there are comutation problems with high overloads.
That's what happens when you start a DC/AC universal motor in DC without current limiting resistor.

High overload can produce a flash at the collector (this can damage the motor) and wearing the collector and brushes...No concern at all with overheating.
You seems to prove that you know nothing about dc motors...

it proves the armature acts as a secondary winding. If it's put in series with the field coils, the current induced in the armature will either be added or subtracted to the surge current, depending on the phasing. To find out which, you'd need to test the phasing of the motor, which I didn't do.

The surge current is the same in the armature and in the field coils as they are conected in series.

[oldway]

I found a IRFP4768Pbf it will take up to 93A continuos at 25C, 66A at 100C and will withstand 370A pulses the width of which are "limited by the junction temperature" which is limited to 175C

You can also use a PWM drive of the Mosfet instead of a series resistor.
So you provide a "soft start" of the motor applying reduced voltage at start up.
As the motor is highly inductive, you don't need any additional inductor, only a fast free wheeling diode.
NB: series current limiting resistor is a solution to be applied only when you can't use the PWM solution. (relay or SSR)
NB2: to be "bulletproof" in case of output short, i would add a little inductance (no way to protect a MOSFET against short if there is no inductance, they are too fast) and a cycle by cycle peak current limiting circuit.
NB3: PWM has other advantages:
1) there is no current surge at all, you can choose a lower current MOSFET.
2) it's safer for the motor.
3) you can limit the output voltage at 110V by PWM. The max voltage of 140Vdc seems too high for a 110Vdc motor.

[Simon]

surely with PWM you will still get the huge spike, it won't last as long but it will be there. We;d need a lot of "tank" capacitor to help average the current out and not blow a circuit breaker.

[Zero999]

High overload can produce a flash at the collector (this can damage the motor) and wearing the collector and brushes...No concern at all with overheating.
You seems to prove that you know nothing about dc motors...

No, as i said before, I tested it and it works fine. It doesn't matter what the article says, practical results indicate in reality it doesn't make any difference. The arcing on the commutator is virtually the same irrespective of whether the supply voltage is AC or DC. The surge lasts for under a second which is not long enough to cause any damage to the brushes or commutator. I'd also expect the starting torque will be higher on DC, causing more rapid acceleration, thus the surge will not last as long on DC than it would for AC but I didn't test this.

I'd suspect over the long term, there would be less heating with DC than AC as there won't be eddy or hysteresis losses in the field coils but I didn't run it for long enough to see if there was any difference.

The surge current is the same in the armature and in the field coils as they are conected in series.

I didn't say the current in the armature and field coils was different, of course it will be the same.

What I mean is the motor behaves like a transformer with the field as the primary and armature as the secondary. Connect the primary and secondary coils in series on a transformer and the current will depend on the phasing. Try this with a 240V to 12V mains transformer and you'll see, the current depends on the phasing of the primary to secondary winding. Again, I didn't test this with the motor so don't know what the phasing is.

I found a IRFP4768Pbf it will take up to 93A continuos at 25C, 66A at 100C and will withstand 370A pulses the width of which are "limited by the junction temperature" which is limited to 175C

No there's no point in messing around with PWM or a current limit resistor, providing the MOSFET can handle the surge it will be fine. In my test I used a 3A bridge rectifier to power a 750W motor for a minute so I'm pretty sure you'll be able to use such a beefy MOSFET to drive your much larger motor.

I thought you were intending to use the same switch for both 230VAC and 120VAC? If so you will need a 400V device.

The question then is whether you use a bridge rectifier or connect two MOSFETs back-to-back.

[oldway]

surely with PWM you will still get the huge spike, it won't last as long but it will be there. We;d need a lot of "tank" capacitor to help average the current out and not blow a circuit breaker.

Why would you get a "huge spike"?
You can drive the MOSFET with very short pulses, and the average voltage will be very low, a few volts.(inductance of the motor is "averaging the voltage" applied on the resistance of the motor)
Current is then limited by motor resistance. (3 Ohms for example).
With a min average voltage of 6V (for example) and a 3 Ohms motor resistance, you will have a 2A start up current.
Is that an "huge spike"?
You must than have an soft start increasing average voltage up to 110Vdc.

If you have to feed the motor with both dc an ac voltage, i would use a relay in ac feeding and a relay + MOSFET (with PWM drive) in dc feeding.
Relay in dc feeding may only be actueted without current, with MOSFET in off state. (actueted, i means changing of state) (NB: i am not very fluent in english)
Also make a contact logic for protection so ac and dc relays cann't never be both on.

[Zero999]

surely with PWM you will still get the huge spike, it won't last as long but it will be there. We;d need a lot of "tank" capacitor to help average the current out and not blow a circuit breaker.

Why do you need a capacitor? The motor has enough inertia that it doesn't make any difference.

The only capacitors and inductors you need are for RF suppression. The motor I tested already had a capacitor (150nF if i remember correctly) connected in parallel with it for EMI suppression but no inductor. I didn't conduct any EMC tests but as this is on a train, I suspect it may need to meed more rigorous EMC standards than a domestic vacuum cleaner.

[oldway]

@Hero999: To deny measured and calculated values and to substitue them by "I feel that", this is not serious at all.
Before saying anything, you should inform yourself a little better on what you say.

Tests should be made of a scientific manner respecting the actual conditions.
You did absolutely no surge current measurement, you have no 110V battery to test such a 110V motor.
For testing, you must have a dc source of low internal resistance, as low as a battery.

In addition, considerations such as "I did two tests and the motor was not damaged" are completely ridiculous.
A collector aged, clogs, the air humidity can change, etc. ....

You're an amateur handyman and this is a circuit that must be installed on a train, this is not the place for low level crafts.

We must never accept a project where we did not respect the rules of technology.

[Zero999]

@Hero999: To deny measured and calculated values and to substitue them by "I feel that", this is not serious at all.
Before saying anything, you should inform yourself a little better on what you say.

Tests should be made of a scientific manner respecting the actual conditions.
You did absolutely no surge current measurement, you have no 110V battery to test such a 110V motor.
For testing, you must have a dc source of low internal resistance, as low as a battery.

In addition, considerations such as "I did two tests and the motor was not damaged" are completely ridiculous.
A collector aged, clogs, the air humidity can change, etc. ....

You're an amateur handyman and this is a circuit that must be installed on a train, this is not the place for low level crafts.

We must never accept a project where we did not respect the rules of technology.

What you say makes absolutely no sense. The surge current last for under 1 second so can't damage the motor in such a short period of time. There's a reason why it's called a universal motor - it can be run off either AC or DC, go and look it up.

I agree, testing is important but you haven't even tested it, just done some rough calculations which didn't take any of the things you've mentioned such as environmental conditions into account. You're right, more rigorous tests are needed so I suggest Simon should go and conduct some himself.

[Simon]

surely with PWM you will still get the huge spike, it won't last as long but it will be there. We;d need a lot of "tank" capacitor to help average the current out and not blow a circuit breaker.

Why do you need a capacitor? The motor has enough inertia that it doesn't make any difference.

I meant a power supply capacitor so that the PWM does not generate repeating high current spikes, oldway keeps banging on about the average voltage. A fast reacting circuit breaker will not give a rats ass about the average voltage but the spikes that are being drawn off the supply. If a tank capacitor helps feed the modfet it will help take the spiky load off the power supply line, we do have to get this through emc tests.

[Simon]

I agree, testing is important but you haven't even tested it, just done some rough calculations which didn't take any of the things you've mentioned such as environmental conditions into account. You're right, more rigorous tests are needed so I suggest Simon should go and conduct some himself.

pft I won't pass that on to my employer, i know what they think of R&D: nothing, our customer is even more ignorant and i think our electronics subcontractor is yet to learn ac versus dc power control as they usually deal with digital control boards.

[oldway]

What you say makes absolutely no sense. The surge current last for under 1 second so can't damage the motor in such a short period of time. There's a reason why it's called a universal motor - it can be run off either AC or DC, go and look it up.

I agree, testing is important but you haven't even tested it, just done some rough calculations which didn't take any of the things you've mentioned such as environmental conditions into account. You're right, more rigorous tests are needed so I suggest Simon should go and conduct some himself.

A flashover is an arc generated around the commutator and eventually causing a short circuit in the positive and negative brushes and damaging the commutator and the bruh holder. 

This arc can be generated by ionisation of the air by comutation sparking anomalies ocurring during high overloads.

I gave you a link over armature reaction to explain what occurs during such overloads, but you did not read it.
http://web.uettaxila.edu.pk/CMS/SP2012/etEMbs/notes%5Carmature%20reaction%20&%20commutation.pdf

Armature reaction move the field neutral line and this results in very severe comutation sparking anomalies.
How many time do you think it is necessary to generate an arc ?
Have you ever seen a flashover in a dc motor or dc generator ?
Remember that whe have to start a 110V motor with dc voltage as high as 140V.

[Simon]

well the motor in use is actually rated for 120V and that has to be designed to deal with at least +/-10% as is the specified variation on the mains so that takes it to 132V not far off the 137v they are giving as a maximum (110V * 1.25 which is the max voltage of lead acid batteries)

[oldway]

I meant a power supply capacitor so that the PWM does not generate repeating high current spikes, oldway keeps banging on about the average voltage. A fast reacting circuit breaker will not give a rats ass about the average voltage but the spikes that are being drawn off the supply. If a tank capacitor helps feed the modfet it will help take the spiky load off the power supply line, we do have to get this through emc tests.

There are no repeating high current spikes in PWM control of an inductive load.
Equivalent schematics of a running dc/ac universal motor is an inductance in series with a resistance and in series with a counter EMF.
The inductance of the motor is "filtering" the current and, if the PWM frequency is high enough, the ripple will be very low...it will be almost dc current.

Current spikes in the MOSFET have the same value as the dc current in the motor.
Somes very short high current spikes could occur if you don't use a fast diode as free wheeling diode.

[Simon]

you still have to start the damned thing !!! that first PWM cycle will still draw an instantaneous 70A to start !

[oldway]

you still have to start the damned thing !!! that first PWM cycle will still draw an instantaneous 70A to start !

Not at all !
http://eprimes.wordpress.com/2012/04/25/fundamentals-of-non-isolated-buck-converter/
The only difference is that whe don't have a capacitor C.
NB: Equivalent schematics at starting of a dc/ac universal motor is an inductance in series with a resistance.

[IanB]

I meant a power supply capacitor so that the PWM does not generate repeating high current spikes

No, because the motor coils have significant inductance. Inductances resist change in current. This means that the first instant you apply a voltage to an inductor the current will be zero. If you PWM an inductor it will act as a current smoothing device.

[Simon]

ok but what will the scope trace of the current draw look like ?

[Zero999]

A flashover is an arc generated around the commutator and eventually causing a short circuit in the positive and negative brushes and damaging the commutator and the bruh holder. 

This arc can be generated by ionisation of the air by comutation sparking anomalies ocurring during high overloads.

I gave you a link over armature reaction to explain what occurs during such overloads, but you did not read it.
http://web.uettaxila.edu.pk/CMS/SP2012/etEMbs/notes%5Carmature%20reaction%20&%20commutation.pdf

Armature reaction move the field neutral line and this results in very severe comutation sparking anomalies.
How many time do you think it is necessary to generate an arc ?
Have you ever seen a flashover in a dc motor or dc generator ?
Remember that whe have to start a 110V motor with dc voltage as high as 140V.

I have read that paper now several times.

Yes I do understand how over loading a parallel wound DC motor can cause commutation problems and eventually magic smoke and fire, long before the windings overheat. The flux generated by the armature opposes that of the field winding, thus shifting the magnetic neutral angle away from the geometric neutral. As a result, the brushes start to interrupt the current flowing in the armature causing arcs to be generated due the back-EMF. If this gets bad enough, a chain of arcs will occur between all commutator contacts, forming a short circuit across the armature. In a generator this would result in catastrophic failure, whilst in a motor, one would hope the circuit breaker would trip and prevent a fire.

I don't believe this is will happen when a vacuum motor starts on a DC current with no inrush protection. There are two reasons for this:

#1 You're right that the surge will be greater on DC than AC but because this is a series wound machine, any increase in armature flux is mirrored by the field, until either the field or armature saturates. The armature flux will not weaken the field flux because the current in the field will increase along with the current in the armature.

#2 The situation in the paper you've linked to describes a motor or generator which is already spinning when subject to overload but isn't the case here. In order for the magnetic neutral angle to be shifted, the armature has to be spinning first. If it's stationary as in the instant the motor is started, the magnetic neutral angle will be equal to the geometric magnetic neutral angle. Once the motor is spinning, the armature will be generating a back EMF and the current will dramatically decline, which will happen more quickly on DC than AC as the starting torque will be higher.

I've only managed to get the arcing you've described by removing the fan and reconfiguring the motor as a parallel wound machine. In my experiment I removed the fan, connected the field to a 12VAC transformer secondary and the armature to the output of a variac. Interestingly the arcing seemed to be worse when I ran it from AC than DC and running it backwards made it worse still, presumably because the brushes are already pre-aligned to give proper commutation in one direction only. This was awhile ago. I used an old 650W Panasonic vacuum cleaner motor with a 12V 50W toroidal mains transformer powering the field.

Practical tests show running an unmodified motor off 230V unfiltered DC doesn't produce any more arcing than normal. I suspect the inductance of the field coils would help to smooth the mains ripple so the fact it's not filtered may not be as significant as I first imagined. If I can find my hall effect sensor, I'll measure the current ripple on an oscilloscope, failing that I'll just measure the inductance and do a calculation to estimate it.

ok but what will the scope trace of the current draw look like ?

Sawtooth.
http://ee.lamar.edu/EELABS/ELEN2107/Lab6.pdf

pft I won't pass that on to my employer, i know what they think of R&D: nothing, our customer is even more ignorant and i think our electronics subcontractor is yet to learn ac versus dc power control as they usually deal with digital control boards.

Seriously?

There must be plenty of 12V SLA batteries around, you can connect up in series to the vacuum cleaner motor and MOSFET, obviously use a fuse with a 150V DC rating an AC fuse won't be able to break the same DC current, an arc could form. It should take long to find out if anything bad is going to happen when the motor is started. It's very important this kind of thing is tested. Remember all that buggering around with the sensors? You don't want a repeat performance or worse, magic smoke and fire.

[Simon]

I already pointed out that we could spend just £500 on some batteries as we don't have any and we need 10 to carry out tests but that was too much. Our customer will be buying a power supply that runs off the mains which will be all the access we have to power for testing, I know, I know, but apparently times are hard and I don't know what excuse they will make up when times are not hard. Of course we'd need to put the batteries in a box to stop some idiot killing themselves with them. and we have no kind of approval or training on high (lethal) voltages, not even me.

I have run a drill motor off rectified mains before and it worked fine, no more sparking than on AC (it was not a speed controlled one).

[oldway]

I have read that paper now several times.

Very good!
I'm sure you have learned something about the DC motors and I am pleased with this.

But this document does not allow you to know all about it.

Some of your statements are true, others are not.

I obviously can not develop it here because this is not what Simon expected and it does not answer to his question.

I don't believe this is will happen when a vacuum motor starts on a DC current with no inrush protection.

This is not right.

The series motors are better protected against flashover that separately excited motors due to the existence of an inductance in series with the armature which limit the di / dt.

But contrary to what you think, they are not safe from flahover.

Indeed, the independent excitation motors works with saturated field or nearby saturation.
Armature reaction also occur with universal motors.
For this reason, traction series motors have auxiliary poles and compensation windings to offset the armature reaction.
Universal motors are DC motors adapted (laminated frame to avoid losses by eddy current) to operate with alternating current but they are not designed to be started directly from a battery.
PWM soft start is far better, avoidind risks of flashover and high current surges in dc bus.
With a weak battery, starting dc motor without current limiting can also induce a voltage dip and undervoltage fault in others equipments.

@Simon: ...staggering...
I have worked in foreign countries, in project management and development in power electronics.
I never see such a thing....It is better to say "no comment" for not being rude.

[Zero999]

How about using an NTC resistor?

I know they're not designed for starting motors and I certainly wouldn't recommend it for a motor driving a load with a lot of inertia, but this motor reaches full speed within second and the surge will be lower than a rectifier and capacitor, which is what they're designed for.

[Simon]

Interesting, I didn't know they had NTC's specifically for inrush current control.

[Zero999]

Interesting, I didn't know they had NTC's specifically for inrush current control.

Yes, the cold resistance, when no current is flowing, is relatively high when it heats up due to the surge, the resistance dramatically drops. They're normally used for rectifiers in switched mode power supplies, rather than motors but the surge is short lived in this case.

Here's an example:
http://www.digikey.co.uk/product-detail/en/AS35%2020010/570-1116-ND/2614399
http://media.digikey.com/PDF/Data%20Sheets/Ametherm%20PDFs/AS3520010.pdf