Another homebrew AC-controller - Page 11 - DIY Electric Car Forums
Go Back  

DIY Electric Car Forums > EV Conversions and Builds > Controllers

Register Blogs FAQ Members List Social Groups Calendar Search Today's Posts Mark Forums Read

Reply
 
Thread Tools Display Modes
  #101  
Old 08-11-2013, 08:59 PM
PStechPaul's Avatar
PStechPaul PStechPaul is offline
Senior Member
 
Join Date: May 2012
Location: Cockeysville, MD 21030
Posts: 2,926
PStechPaul will become famous soon enough
Lightbulb Re: Another homebrew AC-controller

An idea I had was to measure the DC link bus and DC current to determine the actual power consumption, and then assume that torque is proportional to the current drawn by the motor (which can be obtained at various combinations of voltage and frequency, within a certain range of slip that avoids breakdown torque). You should be able to find a "sweet spot" where you get the torque you need for the least power in. This maximizes efficiency and minimizes reactive current which results in low power factor. The output power can be estimated from the shaft RPM and the torque (which is indirectly measured by the current).

The problem with using current to measure torque is that the reactive current does not contribute to power and thus should be subtracted from the total current. That should be something that can be derived from the phase angle of the current with respect to voltage applied, which of course is a vector operation, and perhaps just another way to express the space vector control algorithm.

So you might be able to tweak the applied voltage and frequency (or slip), and see if the power factor improves. Eventually (within a few cycles) the optimum combination should be obtained for the given set of conditions, but in an EV they can change quickly. However, this tweaking is mostly advantageous under constant load conditions where efficiency is more critical. You can jump into a more responsive algorithm when maximum performance is required even at the cost of lower efficiency.
Reply With Quote
Sponsored Links
Advertisement
 
  #102  
Old 08-12-2013, 04:03 AM
Siwastaja's Avatar
Siwastaja Siwastaja is offline
Senior Member
 
Join Date: Jul 2012
Location: Finland
Posts: 571
Siwastaja is on a distinguished road
Default Re: Another homebrew AC-controller

Quote:
Originally Posted by cts_casemod View Post
In fact, keeping the slip constant and increasing voltage does increase torque, but in a much more efficient way. However if the voltage is too high, like when using torque boost its quite nice for acceleration, but the motor will drain excessive power at part load with a very low poor power factor.
It shouldn't be. Have you measured this? Keeping optimum slip and increasing voltage should generate torque with all the current. Power factor gets only poor when you have a combination of having voltage but not producing torque -- which only happens with low slip.

Automotive traction is easy because you don't need separate concepts for physical load and torque command; you always generate your physical load instantly by instructing torque, as there always is a large mass to accelerate.
Reply With Quote
  #103  
Old 08-13-2013, 07:27 AM
jhuebner's Avatar
jhuebner jhuebner is offline
Senior Member
 
Join Date: Apr 2010
Location: Germany
Posts: 643
jhuebner is on a distinguished road
Default Re: Another homebrew AC-controller

Quote:
Originally Posted by cts_casemod View Post
Hum.. Very familiar... Did you had to use a motor with the same color as mine too? Haha

Nice setup. I am working on something similar. Slip control. But I am having an issue with the voltage.
What do you use for encoder feedback?
This should give you the idea:

The shaft coupler also has the pulse encoder function.

Quote:
Originally Posted by cts_casemod View Post
So far I have modified a controller to work with slip control, but it uses the same V/Hz profile. What I want to do is change this.. In fact, keeping the slip constant and increasing voltage does increase torque, but in a much more efficient way. However if the voltage is too high, like when using torque boost its quite nice for acceleration, but the motor will drain excessive power at part load with a very low poor power factor.

I am curious to know what you did to tackle this issue
I use a regular V/Hz profile scaled by the throttle input.

Quote:
Originally Posted by Siwastaja View Post
It shouldn't be. Have you measured this? Keeping optimum slip and increasing voltage should generate torque with all the current. Power factor gets only poor when you have a combination of having voltage but not producing torque -- which only happens with low slip.

Automotive traction is easy because you don't need separate concepts for physical load and torque command; you always generate your physical load instantly by instructing torque, as there always is a large mass to accelerate.
Yeah, thats what I found. What I have repeatedly read about is saturation. So at some point your extra current won't be producing a useful magnetic field but unuseful heat. I have no idea when that happens though.

Quote:
Originally Posted by PStechPaul View Post
An idea I had was to measure the DC link bus and DC current to determine the actual power consumption, and then assume that torque is proportional to the current drawn by the motor (which can be obtained at various combinations of voltage and frequency, within a certain range of slip that avoids breakdown torque). You should be able to find a "sweet spot" where you get the torque you need for the least power in. This maximizes efficiency and minimizes reactive current which results in low power factor. The output power can be estimated from the shaft RPM and the torque (which is indirectly measured by the current).

The problem with using current to measure torque is that the reactive current does not contribute to power and thus should be subtracted from the total current. That should be something that can be derived from the phase angle of the current with respect to voltage applied, which of course is a vector operation, and perhaps just another way to express the space vector control algorithm.

So you might be able to tweak the applied voltage and frequency (or slip), and see if the power factor improves. Eventually (within a few cycles) the optimum combination should be obtained for the given set of conditions, but in an EV they can change quickly. However, this tweaking is mostly advantageous under constant load conditions where efficiency is more critical. You can jump into a more responsive algorithm when maximum performance is required even at the cost of lower efficiency.
As the software is open source, you may try it out
Reply With Quote
 
  #104  
Old 10-02-2013, 01:48 AM
jhuebner's Avatar
jhuebner jhuebner is offline
Senior Member
 
Join Date: Apr 2010
Location: Germany
Posts: 643
jhuebner is on a distinguished road
Default Re: Another homebrew AC-controller

I found that I might have oversized my caps quite a bit.
http://www.ecicaps.com/pdf/whitepape...inal_Rev_4.pdf

By this paper I calculated the actual capacity need for a voltage ripple of 1%. I then end up at around 300F.
Caps like that cost around 80$ at digikey, bringing down the price significantly.

The reason why capacities are so large when elcaps are used is the ripple current capability. For example a typical inverter elcap @1200F has a ripple current capability of 6-7Arms. But the ripple current in my case is more like 40Arms. So I'd need around 7200F of elcaps.
Film caps have a much larger ripple current capability and thus can be sized by the voltage ripple requirement.
Reply With Quote
  #105  
Old 10-02-2013, 02:43 AM
cts_casemod cts_casemod is offline
Senior Member
 
Join Date: Aug 2012
Location: Croydon, London
Posts: 665
cts_casemod is an unknown quantity at this point
Default Re: Another homebrew AC-controller

Quote:
Originally Posted by jhuebner View Post
I found that I might have oversized my caps quite a bit.
http://www.ecicaps.com/pdf/whitepape...inal_Rev_4.pdf

By this paper I calculated the actual capacity need for a voltage ripple of 1%. I then end up at around 300F.
Caps like that cost around 80$ at digikey, bringing down the price significantly.

The reason why capacities are so large when elcaps are used is the ripple current capability. For example a typical inverter elcap @1200F has a ripple current capability of 6-7Arms. But the ripple current in my case is more like 40Arms. So I'd need around 7200F of elcaps.
Film caps have a much larger ripple current capability and thus can be sized by the voltage ripple requirement.
I used 6x47uF 900V on my new inverter. Each is rated for a ripple current of 100Amps. Before I was using 4700uF*6 which was just a bit over the required capacity to operate 60HP @ 5KHz @ 320V.

With the increase to 440V either I would have to get some 500V ones, which is both expensive and risky on high temperatures, or I would have required about 10 of these for operation up to 100HP, in series (Rated for 800V)

But it is difficult to find info on this subject. Most literature refers to rectifier input which goes all the way down to zero. Also, it is important to refer to the impedance of a lithium battery. On my case my pack has 6mOhm + Cable losses. A good electrolytic has at least 20-30Mohms, so not only paralleling them is required to handle the ripple current but also, because an electric current travels trough the path with less impedance, which in this case would certainly be the battery and not the capacitor.
__________________
The Electric Polo @ http://www.diyelectriccar.com/forums...tor-78701.html
300A AC POWERTRAIN + PFC CHARGER
LiFePO4 440VDC 8KW
AC INDUCTION
Reply With Quote
  #106  
Old 10-02-2013, 03:30 PM
valerun valerun is offline
Senior Member
 
Join Date: Nov 2010
Posts: 1,534
Blog Entries: 1
valerun is an unknown quantity at this point
Default Re: Another homebrew AC-controller

film caps are the way to go. That, plus a laminated copper bus. We use PowerRing caps (datasheet attached). Rated Ripple current 500A continuous for a 1,000uF 600V device.
Attached Files
File Type: pdf power_ring_1000uF600V-700D348_jan-20-web.pdf (529.6 KB, 77 views)
__________________
Electric Fiat Spider, BMW 3-series (x2): http://tinyurl.com/7x7oaa2; BMW X5: http://tinyurl.com/7yu23op
DIY 10kW Charger: http://tinyurl.com/73oz2oe and http://www.emotorwerks.com/tech/electronics
Android EV Dashboard: http://tinyurl.com/kc9r8gc
Reply With Quote
  #107  
Old 10-02-2013, 05:55 PM
PStechPaul's Avatar
PStechPaul PStechPaul is offline
Senior Member
 
Join Date: May 2012
Location: Cockeysville, MD 21030
Posts: 2,926
PStechPaul will become famous soon enough
Smile Re: Another homebrew AC-controller

That's a very impressive capacitor. How much does something like that cost? I could not find any pricing on their website:
http://www.sbelectronics.com/

I found similar capacitors at Digikey, such as this 380 uF 400V, for $88:
http://www.digikey.com/product-detai...957-ND/2704611

Here are the specs:
http://www.kemet.com/kemet/web/homep...F3303_C4DE.pdf

Rather impressive also with 100A RMS at 10kHz, 5700A peak, 810 uOhms ESR and 40 nH ESL. This compares to the Power-Ring device 1000uF 600V with 125 uOhms and 5 nH.

How do you determine the size of capacitor needed? One way might be to determine the energy storage requirement based on the switching frequency and allowed ripple. Assume 400 VDC and 100 amperes with an allowable ripple of 5% or 20 volts, and 10 kHz. The energy is 400*100*100uSec or 4 W-sec. Since E=0.5*C*V^2, C=50 uF. But since we need 380V minimum, the energy difference between the two voltages needs to be used. Thus, we can use dV = I * dt/C, which results in 500 uF.

It may be useful also to consider the size of an inductor that would perform similarly. E=0.5*L*I^2, so 4 Joules at 100A would be 800 uH, and for 5% ripple it would be 8 mH. A 100uH 10A inductor is about $3 and to get 800 uH at 100A would take 800 pieces so the capacitor seems to be the clear winner.

There may be better ways to do this, and my calculations might not be correct, but they seem reasonable, at least for the capacitor.

I just found the reference posted previously for sizing the capacitors, and the formula is:

C = V / (32*L*dV*f^2)

Using their example with L=100uH, the capacitor for 5% ripple would be 400/(32*100*20*10^2) = 62.5 uF. But obviously the inductance affects the result significantly, and with 10 uH the capacitor would be 625 uF. The motor would be the load inductance, and that might be estimated from the voltage, current, and PF. So a 240 VAC 10 kW motor might have a 0.8 PF which would have 41.6A resistive and 8A reactive at 60Hz so X(L) = (240*0.2)/(41.6+8) = 0.96 ohms and L = 0.96/(2*PI*60) = 2.5 mH. These are very rough calculations and I'm not sure they apply directly to a three phase induction motor inverter. The example seems to be for a DC motor.

Last edited by PStechPaul; 10-02-2013 at 06:29 PM. Reason: Additional calculations based on link reference
Reply With Quote
  #108  
Old 10-03-2013, 03:34 AM
jhuebner's Avatar
jhuebner jhuebner is offline
Senior Member
 
Join Date: Apr 2010
Location: Germany
Posts: 643
jhuebner is on a distinguished road
Default Re: Another homebrew AC-controller

Quote:
Originally Posted by PStechPaul View Post
C = V / (32*L*dV*f^2)

Using their example with L=100uH, the capacitor for 5% ripple would be 400/(32*100*20*10^2) = 62.5 uF. But obviously the inductance affects the result significantly, and with 10 uH the capacitor would be 625 uF. The motor would be the load inductance, and that might be estimated from the voltage, current, and PF. So a 240 VAC 10 kW motor might have a 0.8 PF which would have 41.6A resistive and 8A reactive at 60Hz so X(L) = (240*0.2)/(41.6+8) = 0.96 ohms and L = 0.96/(2*PI*60) = 2.5 mH. These are very rough calculations and I'm not sure they apply directly to a three phase induction motor inverter. The example seems to be for a DC motor.
Yes, I calculated a similar inductance for my AC motor. I still haven't quite understood why the ripple current is independend from the actual AC load current. Any idea?

Here in Germany I'm running into troubles with the home brew inverter. It has to meet EMC regualations. Which seems ok. What isn't, is the price for getting the certificate: around 10000-20000 bucks. Clearly favors big car makers.
Reply With Quote
  #109  
Old 10-03-2013, 04:29 AM
valerun valerun is offline
Senior Member
 
Join Date: Nov 2010
Posts: 1,534
Blog Entries: 1
valerun is an unknown quantity at this point
Default Re: Another homebrew AC-controller

Quote:
Originally Posted by jhuebner View Post
Yes, I calculated a similar inductance for my AC motor. I still haven't quite understood why the ripple current is independend from the actual AC load current. Any idea?

Here in Germany I'm running into troubles with the home brew inverter. It has to meet EMC regualations. Which seems ok. What isn't, is the price for getting the certificate: around 10000-20000 bucks. Clearly favors big car makers.
do you have to get certified in order to sell a kit? I don't think so, no?

Anyway, PowerRing I referenced can be had for ~$200 in volume now, I think. We sell them in our store, too - $249 in single quantities. The main advantage is a natural mating to a laminated copper bus and resulting extremely low inductance. With something like a Kemet cap Paul mentioned, you still need snubber caps on IGBTs (40nH will kick you up over 100V on an IGBT's collector at 500A being switched in 250ns - pretty typical rates - that's in addition to ~50V you are getting from irreducible internal IGBT inductance so things do add up).
__________________
Electric Fiat Spider, BMW 3-series (x2): http://tinyurl.com/7x7oaa2; BMW X5: http://tinyurl.com/7yu23op
DIY 10kW Charger: http://tinyurl.com/73oz2oe and http://www.emotorwerks.com/tech/electronics
Android EV Dashboard: http://tinyurl.com/kc9r8gc
Reply With Quote
  #110  
Old 10-04-2013, 01:54 AM
jhuebner's Avatar
jhuebner jhuebner is offline
Senior Member
 
Join Date: Apr 2010
Location: Germany
Posts: 643
jhuebner is on a distinguished road
Default Re: Another homebrew AC-controller

Quote:
Originally Posted by valerun View Post
do you have to get certified in order to sell a kit? I don't think so, no?

Anyway, PowerRing I referenced can be had for ~$200 in volume now, I think. We sell them in our store, too - $249 in single quantities. The main advantage is a natural mating to a laminated copper bus and resulting extremely low inductance. With something like a Kemet cap Paul mentioned, you still need snubber caps on IGBTs (40nH will kick you up over 100V on an IGBT's collector at 500A being switched in 250ns - pretty typical rates - that's in addition to ~50V you are getting from irreducible internal IGBT inductance so things do add up).
No worries, the f..n certificate is only needed for Germany.

That ring cap is good price for value. Each of the round ones I use is that much.
For lower power application I'd consider a smaller 320F film cap and a less sophisticated bus bar.
Reply With Quote
Reply

Share or Bookmark this

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

 
Support DIY Electric Car
Sponsors

All times are GMT -6. The time now is 07:12 AM.


Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2017, vBulletin Solutions, Inc.
vBulletin Security provided by vBSecurity v2.2.2 (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Ad Management by RedTyger