DIY Electric Car Forums banner

1 - 14 of 14 Posts

·
Registered
Joined
·
971 Posts
Discussion Starter #1
Following up on the ringing discussion on generatorlabs homebrew thread I decided to make a little experiment.

The question being: how much ringing is too much?

So this is the test setup:

Two film caps, 420µF each connected to the DC bus in a laminated fashion. The loop could be smaller, but still.
The Bus voltage is 33V. A rewound 190V ACIM is connected (not on this picture I know). Switching frequency is 17.6kHz.

So here is the first test: how much ringing can I see on the DC bus?
Scope is set to 5V/div

We see an spike to about +-/12V, so about 30% of the bus voltage.

Next the AC output between GND and L1. Scope is set to 10V/div

And we see a bit of ringing as some other phase switches.

Now I completely disconnected the bus caps, no snubbers, no nothing. This is the result:

100V/div so we see 400V spikes which is like 1200% of the bus voltage!
It means that with 100V bus voltage I could kill those 1200V IGBTs with hardly any current going through them.

On the AC side we see bad ringing but the amplitude is small compared to the DC side.
 

·
Registered
Joined
·
53 Posts
Are these tests recent? Or is this the work you originally did when putting your controller together?

As for the ringing you measured on the DC bus; I imagine you just put your probe across the B+ and B- rails but did you have any settings, such as filtering enabled on the channel? I have not checked for ringing on the DC bus. Any measurements I have been looking at have been on the AC side.

This is a little clearer now. You are looking for excessive spikes that could damage the IGBT. I guess I could visualize it as a semi-full dam trying to hold back waves. With no storm the dam is OK. Add some wind (ringing) and the waves rise. If the waves are too high they will crest the dam and cause damage to the village below (IGBT junction).

These phenomenon are not typical in the type of low voltage projects I work with so this is a new arena for me. I am really curious to see what my rails look like now. :rolleyes:
 

·
Registered
Joined
·
53 Posts
I tried to mimic the tests you did.
Scope set to 5v/div
Input voltage is 24vdc
fmax set to 100
Probe attached across B+ and B- rails on controller
I think the sag I see when the motor starts is due to the limitations of the old batteries and they are not fully charged.

I was going to put this up on my thread but I figured I would put it here first to see if I did the measurement as you did.

Here is a snippet: http://youtu.be/HwQYTpMfdX8

Thanks
 

Attachments

·
Registered
Joined
·
53 Posts
What about AC? Can you zoom in a little so we see only one PWM period?
When you say measure AC, where am I measuring across? I am assuming you are saying to measure across one leg on the motor but when you refer to ground are you meaning the B minus rail on the DC bus?

Thanks
 

·
Registered
Joined
·
53 Posts
To be consitent I am posting this video here as well as my controller build thread. This shows the waveforms I get with frequency set to @ 80.

All measurements were on L1 & L2 with respect to B minus on the DC bus.

http://youtu.be/S7SqYOYF0OI

Do you observe any bad ringing components in the waveforms?

Thanks
 

·
Registered
Joined
·
971 Posts
Discussion Starter #8
I'll repost here:
Ok, the ringing is only on the AC side not on DC. So please correct me if I'm wrong but this is not a DC bus problem.

Try using twisted pair cable on the short connection from gate driver to IGBT on one half bridge and see how it compares.

I have seen modules blow up because inside the cable from the terminal to the actual IGBT chip formed a loop.

I wouldn't judge the amplitude of spikes and ringing as alarming but they are definitely a pointer to a design issue.
 

·
Registered
Joined
·
3,141 Posts
It looks like the IGBTs may not be held strongly enough in conduction which allows voltage transients to appear, or (perhaps more likely) the spikes are being induced into the scope probe and/or leads. A quick check for that is to connect the probe to the same ground point you are using as a reference, and any spikes you see are from RFI or ground noise.

Also it would be good to use an analog input and pot to adjust the motor speed more conveniently.
 

·
Registered
Joined
·
53 Posts
it would be good to use an analog input and pot to adjust the motor speed more conveniently.
There is a pot on there. It is shown connected and working in my other threads, however in torque mode the motor ramps up and down due to the fact that no load is on it. For the purpose of these measurements it is simpler to use volt/hz mode and type in a speed.

As for the noise, I could really believe that a lot of the noise is being picked up by the probe leads. If I understood you correctly you are saying to put the probes ground clip and the probe tip, for example, on the DC minus bus and see if I still see noise with both touching the same plane of reference? Is that right? Thanks for the input!
 

·
Registered
Joined
·
53 Posts
Can you post a picture of how you connected the probes?

Id suspect the method is causing the measurement results.
I have the detachable ground clip attached to only one of the probes.
The ground alligator clip is attached to the DC minus bus.
The two probes are attached to 2 of the U,V or W phases.
Connection point are circled in the attached photo.
Thanks
 

Attachments

·
Registered
Joined
·
514 Posts
As i expected, tour measuring with such large loops that stray inductance will influence your results. Additionally the loop allows unwanted magnetic fields to be represented as a voltage reading other than the electrical point you tried to measure. Reduce the pigtail loop area.


Additionally the probes may not be immune to cm currents.
this can be imoroved by adding a ferrite with 2 or more loops of the coax lead.

Hope this helps.

PS try buying a modern digitizing scope with memory larger than a few thousand points, this is always better than an electron beam scope.

Also using Diff probes may be highly benfitial as your scope can be kept at safe voltage levels.
 

·
Registered
Joined
·
727 Posts
The answer to the question: how much ringing is bad? is very simple.

Every design has to meet certain criteria. Ringing is bad if the criteria can not be met.

Common criteria: EMI (electromagnetic interference), SOA (safe operating area), MTBF (mean time between failures), environmental (mostly temperature range, altitude, humidity).
 
1 - 14 of 14 Posts
Top