how do simple mains inverters work? modifiable?

Thread Starter

daveyjones97

Joined Oct 21, 2011
46
thanks kubeek, you the man!
i dont really need to run the motor at high speed but do want as much torque as it can give. im dont mind forgetting my universal motor if some sort of 2kw 5000rpm not too big motor would make the controller side a lot easier.
if a change of motor is going to be a lot easier and better would it be better to start a thread like "cheap 72vdc 30A 2000w motor controller suggestions please" ?
i dont need this to be any harder than it needs to be and the only parameters are upto 72vdc source, 30A +/- 5 and 1500 - 2000w. oh between 2500 and 7500 rpm motor speed would be nice but not essential, i can engineer my way through gearing soloutions.
thanks
 

kubeek

Joined Sep 20, 2005
5,794
Finding a motor that runs on your battery voltage would make things a little bit easier because you then won´t need to find the right inductor, but this amount of power is still not a trivial job.

So I still have to ask, do you have a multimiter and at least access to borrow an oscilloscope once in a while? Did you ever solder? Without this it´s gonna be a long way.
 

Thread Starter

daveyjones97

Joined Oct 21, 2011
46
sorry, family chaos distracted me. i have a 30w and100w soldering iron and a uni-t ut105 "portable automotive multipurpose meter". i have soldered simple wire connetctions with regular solder but generally circuit boards seem to use some sort of higher temp solder and trying to remove parts from a razor e300 esc (long story, on here) was unsuccessful.
i would think about buying a better iron and scope as electric vehicles are the future imo and id like to have some understanding for the future.
i have rewired an old car before but i realise thats no helpful with this sort of thing.
is it best to start a new thread for a controller with those specs and then get a suitable motor, or is there hope for the universal motor method? thanks as always
 

Thread Starter

daveyjones97

Joined Oct 21, 2011
46
just for clarity, yes i would buy a scope if it was required. i have no access to anything helpful offline. of course then i would need advice on a scope....:rolleyes:
 

thatoneguy

Joined Feb 19, 2009
6,359
You can get the DSO Nano from Seeedstudio for $89, about 60kHz effective bandwidth, but that's enough for what you are doing.

It's not scientific grade, but it is very portable, about the size of a cell phone, and has allowed me to fix stuff using only it and a voltmeter to get to the problem quick. A bench scope isn't something I always have with me, some seller on eBay had them for $50 (ver. 1), so now I have a few of them. Just like flashlights and multimeters, 1 for car, 1 for home, 1 to loan.
 

Thread Starter

daveyjones97

Joined Oct 21, 2011
46
thanks, i never moan about buying tools i find you need the most specialist stuff when you least expect it. im totally prepared to spend $100 or less new or second hand for a scope as its clear i'l need one. i can always stick it on ebay if/when this project fails. anyone got any reccomendations? il need to research portable ones
http://www.ebay.co.uk/itm/VELLEMAN-..._Measurement_Equipment_ET&hash=item27c1cef118 excessive, or a wise investment?
 
Last edited:

tom66

Joined May 9, 2009
2,595
So this is my idea, controller keeps M1 on until current on sense resistor reaches set current, then M1 switches off very fast and dumps the current through D1 into the load. After some dead time the cycle repeats.

I think it will be doable with high enough frequency to keep L1 small, but M1 and D1 will need to withstand the max current and peak voltage. Values in schematic are wild guesses.

So guys, is this doable at 30A and ~250V output, in terms of physical size of L1 for example?
That's one way to do it, but...

If OP wants 1000W at 240V equivalent AC, they need a 340V bus. This is dangerously high voltage. 340VDC would be required at at least 3 amps.

L1 needs to be much larger.

Vout = Vin / (1-D)

for zero current, this gives us a duty cycle of about 79%.

Inductance is then calculated:

Lmin = ((Vout - Vin + Vdiode) * (1 - D)) / (i_load * frequency)

A good frequency is around 50kHz. Increasing this to a higher value is okay, but designing a very high frequency switcher is not easy, I suspect, for the op. Increased frequencies cause greater switching loss, but in high voltage MOSFETs the major loss is on-resistance.

We can neglect Vdiode as it is about 1-2V. So at 3A load, Lmin = 1.4mH. This is a minimum inductance. Tolerances, saturation losses, core losses, winding resistance, minimum input voltage etc. all need to be taken into account.

What will the current through this inductor be? Well, you can work it out through complex equations which I can't remember or you can work it out by assuming efficiency is 90% (typical for a good boost converter) and this gives us 14A rms. At 79% duty cycle, this means a saturation current of at least 39A. Such an inductor will not be cheap. It will likely weigh several kg too.

Next, select output capacitor value. Too low capacitance will create high ripple voltage and ripple current. A 10µF 400V capacitor will go pop very quickly, due to the several amps of ripple current passing through it. Remember, the 3A current to the load... it has to go somewhere and it essentially goes straight into the capacitor (essentially a dead short at 50kHz) during the OFF time which then discharges into the load during the ON time.

This essentially means capacitor sizing will be dictated by the ripple current rating. Motors don't really mind that much about having an unstable DC input voltage, but capacitors do mind about ripple current and control loops must be stable. So, my favourite capacitors for such an application are Rubycon MXC's because they are high reliability, high temperature devices. No reasonable sized individual capacitor will withstand 3A ripple current. But we can put several smaller ones in parallel. If we decide to design for 4.5A (no harm in overengineering!) then 3 x MXC's rated at 560µF 400V (1680µF). This is a huge amount of capacitance and can easily do great harm if messed around with, the energy is simply massive (about 194J, enough to run the electric motor for 194 ms.) Also, such capacitors run at around £6 each.

Input decoupling must not be left out. If you are pulling 18A peaks from the battery, you're going to create huge spikes on the input to the converter which will not do the converter any good - the MOSFETs will probably be killed quicker by these than the output spikes of 340V. It's not so easy to size this capacitance, but estimate 680µF 100V Rubycon ZLH x 8 (ultra low impedance capacitor), or 680µF 100V Panasonic FC x 8 (low impedance but not as good as the ZLH series.) Again such energy is very large, about 28J. These capacitors are cheaper, probably about £2 each, but I haven't checked. Some high frequency polycaps would also be a good idea, maybe 1µF/100V each, but not crucial.

MOSFET sizing - choose the lowest on-resistance 400V device you can get hold of. Placing several in parallel is highly recommended. Heatsinking is crucial since these will contribute the greatest loss. A nice device: http://www.onsemi.com/PowerSolutions/product.do?id=2SK4221 but $10 each - or - http://www.onsemi.com/PowerSolutions/product.do?id=2SK4117LS at $2.21 each, but not as good.

The one thing that is easy is diode sizing. MUR460 is capable, but I'm not entirely confident it will like being run at 3A continuously. Consider getting a 5A TO-220 diode rated for at least 500V reverse. Mounted on a heatsink. In this application, it doesn't really make much sense to use synchronous rectification, as the losses in the diode are about 5W, compared to probably nearly 25-30W (rough guess, assuming 4x2SK4117LS in parallel) in the MOSFETs.

Unfortunately it's not as simple as just putting in any old values. :(

I would suggest acquiring a lower rated voltage motor or experimenting with the motor on 72V with a simple H-bridge or single MOSFET driver.
 
Last edited:

shortbus

Joined Sep 30, 2009
10,045
@daveyjones97, I know you want to DIY this project but have you had a look on Ebay-UK at the prices of motor and controller packages? You should also consider the fact that this type project often takes time and smoked parts before the bugs are worked out. :) Not meaning to rain on your parade ,but just stating some truths.
Hope that you don't take offense at this, just giving some advice from personal experience.

http://www.ebay.co.uk/sch/i.html?_f...3&_nkw=e-bike+motor&_sacat=See-All-Categories
 
Last edited:

thatoneguy

Joined Feb 19, 2009
6,359
thanks, i never moan about buying tools i find you need the most specialist stuff when you least expect it. im totally prepared to spend $100 or less new or second hand for a scope as its clear i'l need one. i can always stick it on ebay if/when this project fails. anyone got any reccomendations? il need to research portable ones
http://www.ebay.co.uk/itm/VELLEMAN-..._Measurement_Equipment_ET&hash=item27c1cef118 excessive, or a wise investment?

That looks decent, pretty high sampling rate, and the bandwidth is adequate for most of what you'll be working with.
 

tom66

Joined May 9, 2009
2,595
Not dual channel, wouldn't touch it.

For that price I can buy a very nice analog oscilloscope which is quite old but more than powerful enough for the task at hand. I could get a 50 MHz dual channel Tektronix for £100.
 
All those portable ones that look like mp3 players or phones are NOT real oscilloscopes, but rather toys that lets you get the feel of what a oscilloscope does.

These toys are NOT intended as real test instruments but rather intended as just TOYS.
 

thatoneguy

Joined Feb 19, 2009
6,359
All those portable ones that look like mp3 players or phones are NOT real oscilloscopes, but rather toys that lets you get the feel of what a oscilloscope does.

These toys are NOT intended as real test instruments but rather intended as just TOYS.
I agree, but they work great with kids to get an idea of what is actually happening, they catch on faster.

For an audio band single channel scope, they are a decent deal at $70, and easy to carry around.

In my non-mobile workshops, I do have much nicer equipment, but when somebody wants to see if they are working with square wave or triangle wave, there's nothing wrong with a toy scope. It far outpaces the homebuilt ones that use an LCD and a 32 bit PIC or AVR for capture/display.
 

Thread Starter

daveyjones97

Joined Oct 21, 2011
46
ok thanks guys. im wary of crossing the line from determination into stupidity and to my eyes the motor controller idea is just about the wrong side. i have been collecting schematics but tbh if i can buy a controller and motor for cheap enough thats what im going to do. when i break the controller i buy il do more oscope research and get one then. theres probably a thread somewhere on scopes, so will look for that.
thanks for the advice as always.
 
Top