Rectifier and IGBT to Make 5500W Power Supply

[cobyG]

Hello Everyone,
I am currently working on a school project which requires me to design and make a 5500W,240V single-Phase linearly variable power supply capable of powering a low inductive water heating elements.
So far my design consists of the very simple schematic seen below and will be driven by a PWM generated by a MC.  The schematic is purposefully vague and uncompleted, cause hopefully you guys help me figure out how complete it .  The carrier frequency of the PWM is set to 30Hz, and below you can see the simulations performed on Matlab to determine the linearity of this frequency at different duty ratios(since the circuit does not have Zero crossing detection).
I have acquired some large, heat sink mount IGBTs, 3120 opto gate drive and 35A full bridge rectifiers, and have access to a large selection of passives, so sourcing additional parts should not be difficult.

So Before I start putting things together and letting the smoke out, I figured it would best to ask.  I have asked my professors and TAs, but their answers just did not seem right.
Questions:
1.  If using 120VAC as input (for testing reasons), can I tie the negative DC side of the bridge to earth ground?  will this cause ground currents and overload the bridge?
2.  Somewhat continuing the last question, how do I drive the gate if the Emitter(Directly connected to the negative of the bridge) is not earth ground?  Do I just common it to a floating DC source?
3.  Is it better to change the design so that the the load is on the AC side of the bride, and just short the Collector and Emitter?
4.  Is an RC snubber across the IGBT good enough for this configuration?

Thank you in advanced!

[viperidae]

If you tie your negative to ground, you'll get a short from active/hot through the bridge, back through the ground wire, then from the ground bus bar to the neutral bus bar in your fuse/meter box.
Neutral and earth are tied together in your meter box aren't they?

You should never connect any mains current to the ground wire.

[cobyG]

Yes, Neutral is diffidently connected to earth ground at the main disconnect.  Does that mean that I should float the power supply driving the gate?

[David Hess]

Wouldn't this be an ideal application for a zero crossing initiated TRIAC?  The controller should modulate it for a whole number of cycles to produce the correct output level which presents an interesting challenge.

[IanB]

Indeed. The details given in the OP are very vague, but if we presume this is a standard resistive water heater that will run from a regular AC mains supply, then it has a slow response time and significant thermal inertia. As a result, controlling power in whole (half) cycles would seem to be very direct and efficient.

[David Hess]

I mentioned it because many soldering irons and industrial temperature controllers work in exactly this way.  A TRIAC is the power element and it is modulated using PWM over whole numbers of power cycles with triggering only at zero crossings.

I do not see any advantages to using an IGBT and many disadvantages.

[Jay_Diddy_B]

Hi,
Assuming that you are in the USA, the center of the 240V will be the neutral, which is connected to ground at the distribution transformer. You cannot ground any of the circuit.

I suggest that you build a scale model of your circuit using a transformer rated 36vac and 2A and use a 33 Ohm load. Use 3x 100 Ohm in parallel.

The transformer will give you isolation from the line voltage and power dissipation will be around 40W. This will be much safer.

Alternative is to do simulation in SPICE.

Here is an LTspice model of your circuit:



And the results:



This is for the PWM generator starting at the zero crossing of the line cycle.
If the PWM generator should be a multiple or sub-harmonic of the line frequency to get predictable results.


This is the input current for a 37% duty-cycle:




This type of load is going to cause very annoying flickering on lamps connected to the same circuit.

I have attached the LTspice model

Regards,

Jay_Diddy_B

[Phoenix]

I am currently working on a school project

What level of school is this?

[floobydust]

Heaters this large are inductive and pulsing DC on them makes a large back EMF and there's no need for DC here. I thought you get etching of the heating element too.

I use TRIAC's (SSR) with zero-cross and simply turn them on for an integer # of complete line cycles, so there is no DC offset seen by the distribution transformer.
About 1-3 seconds to repeat, as the the thermal mass here means you don't need msec PWM.

The sims looks good. I don't see symmetry in the 37% (supply current) waveform though.

[Seekonk]

If this is for school, then the whole idea of the project is to fail.  Great ideas fall apart fairly quickly when you try to implement them. So why are you using an IGBT, just because you have them? That will waste about 20W, not really a design. I get all my hot water at camp by modulating the elements so the panel is kept at the power point voltage.  Nothing gets hot. I hate wasting energy. Think of all the places you are wasting heat/energy.

[cobyG]

Firstly, Thank you to all!  This information is very helpful.
If my responses come across as being argumentative, this is not my intention, i'm just trying to fully understand...

I mentioned it because many soldering irons and industrial temperature controllers work in exactly this way.  A TRIAC is the power element and it is modulated using PWM over whole numbers of power cycles with triggering only at zero crossings.

I do not see any advantages to using an IGBT and many disadvantages.

Indeed. The details given in the OP are very vague, but if we presume this is a standard resistive water heater that will run from a regular AC mains supply, then it has a slow response time and significant thermal inertia. As a result, controlling power in whole (half) cycles would seem to be very direct and efficient.

Wouldn't this be an ideal application for a zero crossing initiated TRIAC?  The controller should modulate it for a whole number of cycles to produce the correct output level which presents an interesting challenge.

TRIACs were the first chose in designing this circuit. Unfortunately I forgot to mention in the first post that the MC driving this circuit is running "parallel" processes and can't handle dealing with measuring zero crossing.  At any given point, the MC measuring 2 pressure sensors, 2 RTDs and driving 2valves and 2 of these heating elements. It's convenient to set the MC to output a PWM and then go do other tasks in "parallel'.
Also, What package besides TO220 do TRIACs come in?  The beauty of my IGBTs is they they except lug terminals and mount directly on a heat sink.  Are there equivalent TRIACs?   

I suggest that you build a scale model of your circuit using a transformer rated 36vac and 2A and use a 33 Ohm load. Use 3x 100 Ohm in parallel.

I have a set up at lab that allows me to work with conditions similar to what you suggested and that indeed is my intention.  Don't have a death wish...
Thank you very much for the Spice sim too! b
 

What level of school is this?

Undergraduate first semester Senior Design.

If this is for school, then the whole idea of the project is to fail.  Great ideas fall apart fairly quickly when you try to implement them. So why are you using an IGBT, just because you have them? That will waste about 20W, not really a design. I get all my hot water at camp by modulating the elements so the panel is kept at the power point voltage.  Nothing gets hot. I hate wasting energy. Think of all the places you are wasting heat/energy.

Very True!  School projects are some what met to make you fail in a safe environment, that way you don't fail as much failing on the job. 
IGBTs were chosen for the price, current and voltage ratings, and most importantly, do not need to be mounted on a pcb.  240V PCB with a TO-220 package scares me.
Also, can not agree more with the idea of wasting energy, but school is an exception .

So my new question to you guys:
1. Would punting the heater in the AC side of the bridge be better?  Let the grid absorb some of transients?
2. If say this design fails, can you please give some suggestions for a power supply that only needs a PWM for the MC to run.

Thanks Again!

[David Hess]

TRIACs were the first chose in designing this circuit. Unfortunately I forgot to mention in the first post that the MC driving this circuit is running "parallel" processes and can't handle dealing with measuring zero crossing.  At any given point, the MC measuring 2 pressure sensors, 2 RTDs and driving 2valves and 2 of these heating elements. It's convenient to set the MC to output a PWM and then go do other tasks in "parallel'.

The TRIAC is normally driven with an optical isolator which itself does zero crossing detection.  There is no requirement for the source signal to monitor for zero crossings.

It is nice to activate the TRIAC for a whole number of line cycles at a time but not strictly necessary.  Pulse width modulating it at your chosen low frequency will work fine.

Also, What package besides TO220 do TRIACs come in?  The beauty of my IGBTs is they they except lug terminals and mount directly on a heat sink.  Are there equivalent TRIACs?

They come in all kinds of packages but only the largest are individually mounted the way you suggest.

Look instead for a panel mount zero-crossing TRIAC solid state relay.  Internally they have the TRIAC, optically isolated zero-crossing driver, and output snubber to prevent false turn-on and other problems.  They do exactly what you want in the form factor you want.  Industrial temperature controllers often use these for the reasons you mention.

[jbb]

Don't have a death wish...
...
Undergraduate first semester Senior Design.
...
School projects are some what met to make you fail in a safe environment, that way you don't fail as much failing on the job.

OK, I know I often go all 'safety safety safety' on this forum, but please be careful.  Unfortunately engineering schools are very variable in terms of safety, and the nature of academic study (tight deadlines, still learning what's going on etc.) can push you into unsafe situations. So I present the following questions:

• Have you received electrical safety training?
• Do you have access to experienced personnel (lab technician, lecturer, TA etc.) who can review your designs before power-up?
• Do you have an emergency stop button in your lab?
• Do you have a defibrillator near your lab?
• Do you have ear and eye protection in your lab? At these power levels, semiconductors can explode and spray shrapnel around.
• Do you have appropriately rated test equipment?
• How are you going to make damn sure that a leak in the plumbing won't dribble (or spray, or outright flood) water onto your control electronics?
• Say you do have a freak accident, get an electric shock, and fall quietly to the floor.  Will someone notice and get you help in time?
• Say you do have a freak accident, get an electric shock, and a friend calls you an ambulance?  (Your profile says 'USA'.) Do you have appropriate health and accident insurance? Often your institution will have an industrial policy which would cover this sort of thing, but you should check.
• Don't trust your life to a power semiconductor switch! (Unless it's in a defibrillator - those things go through a lot of testing!) (OK, not a question)
• Don't trust your life to a relay/contactor! The software behind it might do something weird at any moment. (Also not a question)

I know you want to pass your course, but please remember that accidents can happen and you want to live a long and healthy life.

Unfortunately I forgot to mention in the first post that the MC driving this circuit is running "parallel" processes and can't handle dealing with measuring zero crossing.  At any given point, the MC measuring 2 pressure sensors, 2 RTDs and driving 2valves and 2 of these heating elements.

For now, because you've got a lot going on, I suggest you keep it as simple as reasonably possible. I agree with David Hess that dispatching whole line cycles with a zero-crossing optocoupler is probably a good way to go.

For future reference, there are 2 ways you could try to improve matters:

• Get fancy in your main micro controller with timer Input Captures and an interrupt service routine which does the phase control on the Capture value (advanced)
• Use a second microcontroller to handle the TRIAC.  All it would do is:
  
  • watch the line for zero crossings
  • control the TRIAC
  • listen for commands from the master controller (typically would use a UART for this) (e.g. 'set power to X%')
  • run a timeout counter, so if the master controller stops talking for a while the TRIAC turns off

IGBTs were chosen for the price, current and voltage ratings, and most importantly, do not need to be mounted on a pcb.  240V PCB with a TO-220 package scares me.
Also, can not agree more with the idea of wasting energy, but school is an exception .

So my new question to you guys:
1. Would punting the heater in the AC side of the bridge be better?  Let the grid absorb some of transients?
2. If say this design fails, can you please give some suggestions for a power supply that only needs a PWM for the MC to run.

Whatever semiconductor you use, you will have to deal with 240V somewhere, because you need to drive the gate of the semiconductor. A proper PCB is going to be way safer than just soldering some stuff together on a block of wood. This gate driver board should have the safety isolation built in (e.g. optocoupler) so that your control electronics operates at safe potential.

I would suggest using TRIACs with a larger package than TO-220.  You can probably find a module package with screw fittings if you look.  If not, TO-247 offers a bigger gap between the legs.
TRIACs will have lower loss than IGBTs because you only need one device. That means a smaller heatsink.
If you do use an IGBT, remember they only conduct one way (but often come with a reverse diode added on), so they won't work right with AC. You'd need to use a diode bridge and a single IGBT or two back-to-back IGBTs.

Once you start asking questions about specific circuits, I suggest you post sketches of the schematics (with component numbers please!).  You'll get much better help.