# Non-Isolated power supply for three phase inverter

Status
Not open for further replies.

#### tgrandahl

##### New Member
Hello All,

I am currently working on designing and building a basic Three phase AC motor drive using V/F and PWM. I will be controlling the drive with a FPGA and I am ready to begin testing with hardware. The motor I am currently designing for is 2.2Kw 3HP spindle motor that is rated 24,000RPM @ 400Hz 220V.

I am using and Fairchild Semi SME6G15US60 IGBT module for switching the output three phase. With a International Rectifier IR2136 3-phase bridge driver for driving the IGBT's.

I am running into many questions when considering power supply design. I need to rectify the incoming line power and feel that isolation from mains is not a viable part of the design for the following reasons.

-Isolation Transformers are too large and expensive when using this much power.
-Boost / Buck regulator would be nice but too inefficient.

-The DC bus, three phase output, and other non-isolated voltages will be enclosed once the project is finished. Not accessible for touching...

-The logic side of the drive can be opto-isolated from the power electronics. So all logical inputs / outputs, controls, and processing can be run off a separate isolated power supply.

I feel confident in the safety of the design however I understand the potential danger of the situation and would like some feedback if possible.

Thanks,
Tyler Grandahl

#### tgrandahl

##### New Member
perhaps a rough drawing would help,

#### Attachments

• vfd psw drawing.JPG
64.7 KB · Views: 2,178

#### schmitt trigger

##### Well-Known Member
Yes indeed, you can run your project on a non-isolated supply. All VFDs run that way.

Be sure to include:
-fuse the 3 AC input lines
-fuse the DC output line
-EMI filtering at the 3 AC input lines
-Transient voltage suppression
-provide lots of clearance between the exposed conductors and chassis or mounting screws

#### smanches

##### New Member
Just curious, was this motor out of an airplane originally since it wants 400hz power?

#### tcmtech

##### Banned
400 hz is actually a common drive frequency for smaller sized motors that need to run at much higher RPM's than the standard 60 cycle VFD's run at.

However a 2.2 KW factory made VFD is cheaper to buy than build. I get them used on line a few times a year for under $100 with shipping. You will never make one with all the features and abilities a standard off the shelf factory VFD has! I promise! So standard 'how do I make an inverter' response applies here as well. Just buy one. Its far better and will cost less than what you can make even a basic one for. #### tgrandahl ##### New Member 400 hz is actually a common drive frequency for smaller sized motors that need to run at much higher RPM's than the standard 60 cycle VFD's run at. However a 2.2 KW factory made VFD is cheaper to buy than build. I get them used on line a few times a year for under$100 with shipping.

You will never make one with all the features and abilities a standard off the shelf factory VFD has! I promise!

So standard 'how do I make an inverter' response applies here as well.
Just buy one. Its far better and will cost less than what you can make even a basic one for.

Yes this is an excellent point and true if your only need is the end use of the device. It also explains why you don't see a lot of DIY VFD projects out on the Internet.

I am finishing up my third year as a Electrical Engineering Technology student at RIT. I am building my own drive for the educational ass kicking it provides. This is currently an area of great interest to me, I want a project that can evolve with me. So I am starting with a drive that uses simple concepts and technology and can be updated as I get into DSP and play with newer tech.

I do actually need a drive for the spindle on my mill so its not a total waste of money. Although i'm barley over \$100 so far.

Thanks for the feedback its much appreciated and always helpful with trying something new. Its good to have found this forum, I wish I had sooner. There look to be a lot of fun projects and knowledgeable people around here.

~Tyler

#### smanches

##### New Member
I'm actually in the middle of building one as well for a pellet stove combustion blower. Again, for the learning experience that has an actual use in the end.

Using a PIC for the main control for a PFC and an h-bridge. Everything is fully isolated except for the mains supply through the PFC/h-bridge to the load. Still have a couple feedback loops to finish, and the high side bootstrap. But I have to finish putting my workroom back in order before anything gets worked on.

I love the power electronics though. Turning any power source into any power source makes me feel, well, powerful.

#### tgrandahl

##### New Member
smanches,

Very cool, I feel like motor controllers are an untouched gem in terms of DIY projects. Using an H-Bridge I am going to assume this is for single phase motor?

I am really interested in what type of feedback your implementing and how your doing it with a PIC.

What type of H-Bridge were you using? I took the easy route for now and just got a "driver" IC for my IGBT's. I just need to provide external bootstrap diodes from vcc and caps.

Hope you can get back to your project again soon, I know how frustrating it is not having anywhere to work. (Try moving every 3-6 Months!) are you following this project in a thread here?

I am back in my home state for the weekend but I hope to get back to work and get some things spinning soon.

#### dknguyen

##### Well-Known Member
YOu should look at RC hobby DIY electronic speed controllers (ESCs) then. THey do it there all the time...brushless style.

Last edited:

#### tgrandahl

##### New Member
Well I was able to get my motor spinning last night (for a little while), this was an exciting big step.

I had done some prior testing using a bench power supply to power the dc bus and some power resistors for loading the outputs. All looked well up to about 30V as high as my bench supply will go.

I moved up to 120V mains supply using a variac to slowly ramp up the DC bus voltage. At this point the 15V supply for the gate driver and 6V supply for the FPGA were all powered by isolated bench supplies tied to the V- of the non-isolated dc bus.

My Variac is old and beaten, the contact arcs slightly at some points as it slides upwards. I tried to prevent this as much as possible however after awhile everything just burned out, I noticed a spike on the bench supplies at this time and it was as the variac was arcing. The motor was spinning although with great slip the bus voltage was not yet over 100V.

while the motor was spinning I was able to hear the switching frequency (13.3KHz) as well as what sounded like a consistent arcing. I was not able to determine the exact source of the sound however I believe it could have been the IGBT pack. The arcing was occurring at a much lower frequency and was independent of the Variacs arcing.

I burned out the driver, IGBT pack, and FPGA dev board as far as I can tell. :-( I am assuming this is from spikes in the bus voltage from the arcing across the Variacs coil. However can anyone think of what the second arcing noise could be? I have yet to implement any transient voltage suppression as Schmitt trigger had suggested, could this be due to lack of that?

I certainly plan on using more isolation next time between the FPGA and gate driver however what else can I implement to make sure this does not happen again?

#### smanches

##### New Member
Sounds like something shot through the base of the IGBTs, through the drivers and into the controller?

Don't have much experience with IGBTs, but when using MOSFETS, it's a good idea to put a 15V zener on the gates to help prevent this. Any transients coming back through gate get dumped by the zener and save the driver.

Also, using optocouplers between the controller and drivers is good. Adds a bit of on/off delay, but it's only delay, not switching speed.

Do you have an filter choke & cap on the DC input? Basically make a surge suppressor before the IGBTs.

Can't think of any other reasons it would fail in that way. It happened to me once too. Lost everything in that flameout.

#### smanches

##### New Member
Do IGBTs suffer from self-latching? I know mosfets do, which is why you either want a pull-down resistor on the gate, or better, make sure you use a push-pull driver to sink the gate capacitance while it's off and keep it at 0V.

Could this be what happened to you. The h-bridge self latched, burned out and shorted the gate, then shot through the gate?

#### tgrandahl

##### New Member
Smanches this is my first encounter with IGBT's as well, its basically just a MOSFET controlled BJT, so they are latching and are being controlled using a push pull driver.

The 15V zener could be a good idea is this what you use for driving your MOSFETS?

I am not sure exactly what happened to the gates but it does not seem good, there is between 10ohm and 30ohm resistance between the base and emitter of most of the gates. I am assuming something nasty shot through here?

Could the inductive kick from shutting the windings off be shooting through or is this from the arcing of the Variac coming through the DC bus?

There is currently only a large filter cap on the DC bus after the bridge rectifier. I'm trying to think of how exactly to go about sizing a choke for this setup.

#### tcmtech

##### Banned
Most of the smaller VFD's I have dont have a filter inductor of any kind. Some bypass capacitors over every switching device and thats about it.

Your IF2136 Driver IC shares a common line with all of the control circuits. It likely dumped a load of noise back through to the control system that way. Better bypass and surge protection is needed.
Being line driven there is also the possibility that one of your power sources for the control system is set up with bypass circuits that take the low voltage and couple it to earth ground.
Your drive frequency may be wrong and creating a harmonics mess on the system too. I have played around with designing small VFD's before as well.

The standard issue three phase brushless motor driver IC's are easy to make into high power VFD drive units. I have just taken the low level outputs and ran them to high/low driver IC's that then drove the high power and High voltage IGBT devices.
Full potentiometer simultaneous control of frequency and PWM and one switch forward reverse with no high or low frequency limits unless I put them there.

Still once I found out I could just buy a used factory built units for less than I could build one I rather gave up on the idea of building my own again.

Modern IGBT's dont have much of a latching problem any more.

#### tgrandahl

##### New Member
Most of the smaller VFD's I have dont have a filter inductor of any kind. Some bypass capacitors over every switching device and thats about it.

So there are just caps before the switching transistors and that's it? I may try to pull a choke from an old PC power supply. I am also expecting to eventually need an inrush current suppression circuit.

Your IF2136 Driver IC shares a common line with all of the control circuits. It likely dumped a load of noise back through to the control system that way. Better bypass and surge protection is needed.
Being line driven there is also the possibility that one of your power sources for the control system is set up with bypass circuits that take the low voltage and couple it to earth ground.

I was powering the control system / the dev board off a isolated 6V bench supply that was not tied to ground. However the 3.3Vand 1.2V regulators on the board were quite hot and so I assumed they had been destroyed by a VCC voltage spike of some sort.

How would a voltage spike possibly from the Variac affect this isolated supply when only the negative sides were tied?

Your drive frequency may be wrong and creating a harmonics mess on the system too. I have played around with designing small VFD's before as well.

Are you referring to the pwm frequency or the desired speed? Had you ever worked with any designs using SVM to minimize this?

#### tcmtech

##### Banned
How would a voltage spike possibly from the Variac affect this isolated supply when only the negative sides were tied?
Being the common and the output tap with the arcing are supplying the power. Bridge rectifiers will pass high voltage spikes without problems.
Are you referring to the PWM frequency or the desired speed? Had you ever worked with any designs using SVM to minimize this?

Referring to the PWM switching frequency. The factory units have several different switching frequencies to choose from just to fine tune a motors operating characteristics.

You got me on what SVM. I cant think of its meaning right now.

#### smanches

##### New Member
What is the stall load current on the motor? These are only 10A IGBTs, and at 2.2Kw, that is right at 10A. If that's the working power, and not stall power, then they are way undersized. Did you happen to have a current meter running? What was the voltage up to when everything blew? What was the frequency at? Were the IGBTs heatsinked well?

The zeners don't help with the driving at all, just a cheap way to protect the drivers and control circuitry in case something like this happens.

Your regulators were hot from the short circuits that occurred in the control circuitry after things blew up. Mine did that too, but thermal shutdown should have saved them.

Still thinking...

#### tgrandahl

##### New Member
What is the stall load current on the motor? These are only 10A IGBTs, and at 2.2Kw, that is right at 10A. If that's the working power, and not stall power, then they are way undersized. Did you happen to have a current meter running? What was the voltage up to when everything blew? What was the frequency at? Were the IGBTs heatsinked well?

The SME6G15US60 Module I am using is rated for 15Amps this provides a little head room, probably not enough. However the motor I am testing with is a 1/2 hp and I am only using one set of of the coils in it so it should be drawing very minimal current. 384W / 220 / sqrt(3) = 1A

I had an ammeter in line with one of the lines. It did not peak over 0.7A from what I remember.

The IGBT's were heat sunk fairly well and the temperature of the module barley rose over ambient.

Your regulators were hot from the short circuits that occurred in the control circuitry after things blew up. Mine did that too, but thermal shutdown should have saved them.

Were you able to save your controller? I removed the regulators and tried providing 3.3 and 1.2 directly from bench supplies with no avail. I'm afraid simply having the common ends tied was enough to destroy everything.

#### smanches

##### New Member
No, controller fried, but the regulators were still good.

Well, that blows that theory about being underated.

Do you have snubber caps on the IGBTS?

I'm running out of ideas now. Back to noise in the system, which is such a nebulous problem.

The common ground is about the only thing different from mine it seems. That bugs me. I don't know enough yet.

Last edited:

#### tgrandahl

##### New Member
smanches,

So yeah I was not using any snubber caps or transient suppression of any kind, so I learn the hard way. This could explain the lower frequency arcing I was hearing, could it have been arcing inside the IGBT pack?

I have never had a real need for snubbers before, so this is new to me. I found an IEEE article I am going to read through.

I am assuming you are using some on your H-bridge, how did you go about sizing these?

#### Attachments

• Snubber Design.pdf
2.1 MB · Views: 478
Status
Not open for further replies.

Replies
6
Views
947
Replies
2
Views
651
Replies
0
Views
1K
Replies
3
Views
2K
Replies
0
Views
1K