Class-D Amp with IRS2092 - Page 2 - diyAudio
Go Back   Home > Forums > Amplifiers > Class D

Class D Switching Power Amplifiers and Power D/A conversion

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 27th January 2008, 08:55 PM   #11
Eva is offline Eva  Spain
diyAudio Member
 
Eva's Avatar
 
Join Date: Oct 2003
Location: Near the sea
Send a message via MSN to Eva
Ouch!!

Aren't you doing any math/physics to figure out optimum component values? I become quite puzzled when I see people replacing 10nF by 1uF and asking if it won't be "too slow". There are optimum values that give best parasitistics cancellation and noise rejection with little sensitivity loss, but you have to calculate them or find out the ones producing best waveforms with oscilloscope.

Discontinuous operation is harmless, it actually results in less EMI.

I would just use 200V diodes. Qrr may be 3 o 4 times lower. The RHRP is the hyper-fast version, I think, while the RURP is the ultra-fast.

Have you considered the implications of using two separate filter inductors for SMPS regulation? You should have used a single coupled one.
__________________
I use to feel like the small child in The Emperor's New Clothes tale
  Reply With Quote
Old 27th January 2008, 09:02 PM   #12
mag is offline mag  Switzerland
diyAudio Member
 
Join Date: Jan 2008
Location: Chiasso
Ok it was just a quick try... The only capacitors that I have now in 0603 package are 1nF, 10nF and 1uF so let's try with one of these. For sure I will do some calculations/simulations to find out the best value

I have used two separate inductances mainly to simplify the PCB layout.

What are the benefits of using one common core for both positive and negative rail?
  Reply With Quote
Old 27th January 2008, 09:12 PM   #13
Eva is offline Eva  Spain
diyAudio Member
 
Eva's Avatar
 
Join Date: Oct 2003
Location: Near the sea
Send a message via MSN to Eva
When current draw is not equal in both rails, one inductor may be working in discontinuous mode and the other in continuous mode. In these circumstances, the voltage in the discontinuous mode rail will tend to be pumped to the maximum unregulated value. If the discontinuous side is the only being sensed, then the voltage in the continuous mode side will drop instead. Your current design should only be used in bridge mode. The coupled inductor avoids these problems almost completely.
__________________
I use to feel like the small child in The Emperor's New Clothes tale
  Reply With Quote
Old 3rd February 2008, 11:43 AM   #14
mag is offline mag  Switzerland
diyAudio Member
 
Join Date: Jan 2008
Location: Chiasso
Hi Eva,
I am still trying to make this current limit protection work, and it is not an easy task.

I have made some simulations and seems that even a 5nH series inductance in series with the shunt resistor creates very huge voltage spikes between the mosfet sources and gnd. I think that a TO247 package has at least 10nH of parasitic inductance so the situation is even worse.

I have noticed that after some minutes at idle the shunt resistor becomes hot. How is it possible? it is a 2mohms and the parasitic inductance can not heat up... The idle current consumption on the primary side is 1.27A.

The shunt part number is Riedon FPR 2-T218, it is rated 2W without heatsink and 30W with heatsink.

thank you
  Reply With Quote
Old 3rd February 2008, 12:04 PM   #15
Eva is offline Eva  Spain
diyAudio Member
 
Eva's Avatar
 
Join Date: Oct 2003
Location: Near the sea
Send a message via MSN to Eva
If a resistor becomes hot, some current must be flowing and the spikes are a furter proof. Find out why.
__________________
I use to feel like the small child in The Emperor's New Clothes tale
  Reply With Quote
Old 3rd February 2008, 12:24 PM   #16
mag is offline mag  Switzerland
diyAudio Member
 
Join Date: Jan 2008
Location: Chiasso
Hi Eva,
this is the waveform I see accross the resistor terminal at idle

The scale is 1V/div - 1us/div

there is an oscillation of around 4-5MHz. Can it be caused by the shunt inductance + some parasitic capacitance? In any case I think that the main problem are the spikes as high as 2.5V.

I have tried to put a capacitor in parallel to the shunt but the situation does not change too much.

The frequency is set to 110kHz (oscillator running at 220KHz)
I will try some more test an I let you know
Attached Images
File Type: jpg 03022008013.jpg (64.1 KB, 6269 views)
  Reply With Quote
Old 3rd February 2008, 01:06 PM   #17
Eva is offline Eva  Spain
diyAudio Member
 
Eva's Avatar
 
Join Date: Oct 2003
Location: Near the sea
Send a message via MSN to Eva
Leakage inductance of current measurement shunts introduces a zero at F=R/(2*pi*L). This effect is cancelled by adding a pole at the same frequency. An RC lowpass does the trick, F=1/(2*pi*R*C).

Ringing is another story. It comes from LRC resonators made out of circuit parasitistics. A TO-247 may easily exhibit 20nH total parasitistic L, while 12nH is the typical figure for a TO-220 with short leads. Then it comes PCB trace inductance. Transformers are evil because they exhibit complex parasitistics due to capacitance between turns of different windings, non contiguous turns of the same wnding and distributed leakage inductances.

Spikes may be a sign of cross-conduction or too fast hard switching. They appear bigger than real due to the zero in the unfiltered shunt frequency response. They also cause energy to be sotred in parasitistic RLC tanks and to be progressively released during a long period of time resulting in ringing.

Leading edge blanking allows to get rid of spikes. Consider keeping current limit inactive during the first, say, 500ns of the cycle.
__________________
I use to feel like the small child in The Emperor's New Clothes tale
  Reply With Quote
Old 3rd February 2008, 01:24 PM   #18
mag is offline mag  Switzerland
diyAudio Member
 
Join Date: Jan 2008
Location: Chiasso
Thank you,
so with the values involved a 1uF capacitor in place of C109 will create a pole that cancels out the zero created by the shunt.

I think also I have some cross conduction problems. If I look at the mosfet drain waveforms I see that they can potentially overlap. I can not see it clearly on my scope because of the very very reduced bandwidth (it is 40 years old an 2MHz BW....)

In case crossconduction occurs increasing the dead time may help.
I don't see any strange current consumption peaks due to cross conduction but by now I am using a 3A current limited power supply. I have not yet connected a car battery instead of the power supply to avoid to blow everything up just because of stupid mistakes...

I will try with the 1uF cap and let you know
  Reply With Quote
Old 3rd February 2008, 01:35 PM   #19
mag is offline mag  Switzerland
diyAudio Member
 
Join Date: Jan 2008
Location: Chiasso
I have tried with 1uF and the spikes after the low pass filter are now around 20mV. It seems to be ok but what about the speed of the current sense circuit?

The crossconduction problem seems to be quite confirmed. The mosfet starts to heat up after about 10 minutes (no heatsink).

I will try to increase the dead time
  Reply With Quote
Old 3rd February 2008, 08:09 PM   #20
mag is offline mag  Switzerland
diyAudio Member
 
Join Date: Jan 2008
Location: Chiasso
I have tried increasing the dead time but the results is always the same.

Now I have around 400ns of dead time but the results are basically the same as before.

The shunt and the mosfets still heat up at idle (with no heatsink).

Attached you will find the gate waveforms, it seems that the gates are driven very fast and with a good sqarewave.

BTW: I am using the UCC37322 as gate driver IC. This IC can sink or source up to 9A peak. The mosfets have 300nC QG (one pair of 150nC each).
Attached Images
File Type: jpg gate_2us_5v.jpg (78.4 KB, 6078 views)
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

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
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
IRS2092 customizable DarkOne Class D 55 30th July 2014 10:15 PM
IRS2092 Amp Design: Please help! Devendra Class D 11 17th March 2011 09:10 AM
Another amp with IRS2092 mag Class D 23 18th June 2010 07:25 PM
IRS2092 Amplifier zox2003 Class D 22 31st August 2009 09:49 PM
Any suggestion for IRS2092 ja2dhc Class D 4 2nd August 2008 10:53 AM


New To Site? Need Help?

All times are GMT. The time now is 07:20 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2