Sine filter for PWM inverter output 3

[Skogsgurra]

Does anyone have access to design rules for sine filters for PWM outputs? Or - almost equally good - typical values for L and C versus U, P and f-max? Topology and voltage drop at rated load and frequency also appreciated.

 

[DougMSOE]

One of the best filters for this is called a parametric transformer. I built a small one in college about 50 va.
Not very efficient but very effective. Ramp, square, PWM, etc. as the input, but always a sine wave output. Most of this literature will be found in the 'old' section of the library,not much on the web. Good luck.

 

[Skogsgurra]

Thanks Doug, very interesting - in many respects.

My web search brought me to some "New Age Designs" claiming "overunity", which seems to be a new word for Perpetuum Mobile. It would be extremely nice if it works, but there is one catch - the output from an ordinary frequency inverter needs to be variable and the circuits shown are working with a fixed frequency.

There is one thing that pussles me; You say that you built one such circuit at school and that it was not very efficient. All hits on the web claim that the circuit is "overunity" which is another way of saying that efficiency is more than 100 percent. And that is not bad at all!!! ;-)

Still waitng for some down-to-earth thumb rules for sine filters...

 

[cbarn24050]

Hi, the standard LC filter works well. Just a choke and a cap to each side of the dc link (one set for each phase). Electrolitics work but you have to make sure you stay within the caps rated ripple current.

 

[ScottyUK]

cbarn24050,

What happens to electrolytics connected to AC without a DC bias?

Fizzzzzzzzzz..............Ka-booom!

I'm amazed you didn't learn this trick in college. We sure did!

Additionally, electrolytics have such lousy tolerance control, high ESR, and ESL, that you could not seriously consider them for a power filter.

Use polypropylene film types for this sort of application. They have better impulse rating, are normally self-healing, have low ESR and ESL, and are designed for AC service.

 

[Skogsgurra]

Very good gentlemen, but I humbly asked for topology *and* design rules. If anyone have such available.

BTW, connecting the electrolythics back to the DC busbar does bias them so they dont explode. Ripple current is still an issue, though.

 

[jbartos]

Suggestion to the original posting: Please, would you elaborate in the "sine filter" specification in standard filter area terminology. Please notice that "sine filter" term is not defined in IEEE Std 100-2000 "Dictionary." This does not necessarily mean that the "sine filter" term is not known in the industry. E.g. visit

http://www.transcoil.com/documents/KMG-Brochure.pdf

for MotorGuard Sine Wave Filter
or

http://transcoil.com/documents/I_O_M_Manual.pdf

for MotorGuard Sine Wave Filter and its specifications what they typically include

 

[cbarn24050]

Hi Scotty, you should read my post more carefuly.

 

[Skogsgurra]

Sorry jb,

Why should I rephrase my question "in standard filter terminology"? (I guess that you mean number of poles, filter type, passband ripple, slope and the like, but I do not think that this would lead to the kind of answer I am looking for).

I asked for design rules for sine filters for PWM inverter outputs. There are many such filters available - and I have used several brands - but since they seem to perform quite differently with respect to losses, resulting waveform, dynamic performance, space and cost, I am interested in getting information from as many quarters as possible so that I can find out what design parameters have been important for different producers.

So I repeat: I humbly asked for topology *and* design rules. If anyone have such available.

 

[Highspeed]

As far as I know there aren't any special rules for Sine filters for Motors
Basically the filter topology will be derived from
1-how far you pwm f is from the frequency to achieve
2-If your Pwm'ed signal is Sinus or not
3-If you are running at constant speed or variable speed
4-What distorsion is to be achieved
When the filter topology is designed then to make it work is another story.
I did find that working ahead the filter (ie having a high freqency pwm and already sinus shaped signal) was helping a lot but thats kind of obvious)
I am not sure I helped but if you are more precise on your application I could try to be more efficient.
Good Luck
HS

 

[jOmega]

Hi Skogs ..

I think the problem in getting the response you asked for is that the parametrical and topological detail are within the domain of "proprietary" information ... and manufacturers of such filters are not inclined to give away the fruits of their labor and the keys to their stores.

I know you can appreciate their concern.

One thing I noted in your original post is a lack of definition as to the fundamental frequency range over which the VFD would operate..... and the switching frequency. With regard to the latter... does the switching frequency remain constant.... or is it variable; and if so, over what range.. One other consideration not set forth in your original post... is the results you would expect to obtain .... yes, you want the voltage applied to the motor to be sinusoidal .... but must it be 100% pure... or is some distortion acceptable? and if so, how much.

I also have a curiosity as to what precipitates the need for such a filter.... that is, what's driving the quest ?

Unless you are seeing absolute precision, I think that a Pi type type filter would accomplish the task with acceptable results.

Kind regards,

j

&#937

 

[Skogsgurra]

jO, your constructive comments are much appreciated.

The reason for my question is that many manufacturers of PWM inverters claim that they can deliver "pure" sine waves by adding what they call "sine filters" to their inverters. Different manufacturers have different price tags and differing sizes. Some need filters that are almost as big as the inverters and certainly heavier, while others have less weight. Common for the manufacturers I have talked to is that they never disclose such data as waveform distorsion. Some say that there is no voltage drop in their filters and some admit that there is. ABB has, for example, introduced their "Step Up Filter" which - in fact is a filter and an auto-transformer that restores the voltage level.

I have not specified frequency range and carrier/switching frequency simply because I am talkin about ordinary motor drives and I expect those guys (and girls) that have anything to contribute will know those parameters. Also, I have not included residual distorsion because I have no idea what could be expected under all operating conditions. All I can do is refer to manufacturer's specs - and hope that they are good enough.

Finally. EMC considerations, motor losses, motor sound, bearing problems and insulation issues (in small motors that simply can't be delivered with special insulation) are some of the driving forces behind the quest.

So, I still humbly ask etc...

 

[rodar]

Try this paper.
It has a topology and calcultations

http://decision.csl.uiuc.edu/~ksinha/teleops/files/papers/Wave/jouenj97.pdf

Rodar

 

[Skogsgurra]

Thanks rodar. A very good paper with calculations that help understanding this technique. The filter type treated is, however, a du/dt filter and not a sine filter. The du/dt filter takes care of most of the problems mentioned earlier, but you will still have far from sinewave in cable and motor. So, even if I am very grateful for this link, I wonder if you have an equally valuable link to a similar site where sine filters are dealt with.

 

[rodar]

Try dis 1

http://gdansk.bradley.edu/iecon02/limited/cdrom_backup/html/pdf/SF-002563.pdf

It concerns filtering to the fundamental
Rodar

 

[Skogsgurra]

Thanks, rodar!

Very useful link.

Perseverance pays!

 

[ScottyUK]

Rodar,

What an informative paper. Thankyou - an ex-colleague will receive that link on Monday, to his pleasant surprise.


cbarn,

Apologies, only read half your post re. electrolytics to DC bus.

 

[jOmega]

Rodar:

"Dis 1" is also a Du/Dt filter ... Not a Sinus filter ...

But a good paper, none-the-less.

jO

 

[Skogsgurra]

jO,

"Dat 1" is sine enough for me - excuse me fine enough...

 

[jbartos]

Comment on skogsgurra (Electrical) Jan 10, 2004 marked ///\\Sorry jb,
Why should I rephrase my question "in standard filter terminology"? (I guess that you mean number of poles, filter type, passband ripple, slope and the like, but I do not think that this would lead to the kind of answer I am looking for).
///Normally, the standard terminology is used to facilitate an effective communication in specific engineering areas, avoids ambiguities and removal of posting in this Forum (as mine for example).\\I asked for design rules
///There are various design rules, e.g. design procedure, specification, description, guide, handbook, etc. None of them explicitely have been posted yet. Are you abandoning these?\\ for sine filters for PWM inverter outputs. There are many such filters available - and I have used several brands - but since they seem to perform quite differently with respect to losses, resulting waveform, dynamic performance, space and cost, I am interested in getting information from as many quarters as possible so that I can find out what design parameters have been important for different producers.
///Often, design parameters and specific values, included graphs tend to be proprietary.\\So I repeat: I humbly asked for topology *and* design rules. If anyone have such available.
///More detail information what is actually requested would help.\\\