Output transistor temperature changes with load characteristics.. - 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 10th August 2011, 06:31 PM   #11
diyAudio Member
 
Join Date: Jan 2011
Pafi thank you very much for the explanation. I had already accepted the loss but in simulation I saw nearly 100A current spikes through the transistors. this looks big stress on transistor and risky. I'm struggling because of this.
by the way I didnt see 100A current spikes (nearly 10A spike) through transistors on the real circuit but I thought because of the capacitance of the probe I missed the sharp spikes!? is that possible?

thanks again for the explanation I will try the ways you adviced
  Reply With Quote
Old 11th August 2011, 04:57 AM   #12
Denniz is offline Denniz  Russian Federation
diyAudio Member
 
Join Date: May 2011
Currently I'm using IR IGBT along with 1N5400 with a full success and going to try out copacs as a next step.

So...

Quote:
Originally Posted by Pafi View Post
1N5400 must be a joke, but not a good one. IGBT and dead time also.
... Should be explained, suppose.
  Reply With Quote
Old 11th August 2011, 07:54 AM   #13
Denniz is offline Denniz  Russian Federation
diyAudio Member
 
Join Date: May 2011
Quote:
Originally Posted by volkantr View Post
hi there,

My problem is about the output transistors. When I connect single 8 ohm resistor as load, the circuit can operate @100W and output transistors' maximum temperature is about 65C degree and steady. When I connect a filter suitable for 8 ohm load and then operate again @100W, the output transistors' temperature continuously increase more than 100C degree in a short time.
I've had the same story with my very first trial and after all ran out of that using:

1. 2110 instead of 20124 (more control can be done)
2. According "deadtime" scheme (took one used in Crest 3000)
3. Putting "power" diods as recovery ones

One mention: All above is imperical consequence.

Ah, and PWM I used was just 20 kHz.
  Reply With Quote
Old 11th August 2011, 09:02 AM   #14
Pafi is offline Pafi  Hungary
diyAudio Member
 
Join Date: May 2004
Location: Budapest
Quote:
Originally Posted by Denniz View Post
Currently I'm using IR IGBT along with 1N5400 with a full success and going to try out copacs as a next step.

So...

... Should be explained, suppose.
I don't think such an evident thing should be explained, but if you wish:

1N5400 is not fast at all, it's standard recovery diode, probably much slower then the body diode of IRF6645. Vf (@8A) is also higher then IRF6645. So it is absolutely not OK.

IRGB4045 is also worse than IRF6645 in almost every aspect: much higher conduction loss, slower diode, much slower turn off... the only benefit is higher VCE, but it is not needed.

You can use almost anything, with some results, but here the question was not just a not exploding device, but a better solution than the original (which is already quite good)!

Quote:
Ah, and PWM I used was just 20 kHz.
20 kHz says: there are problems with your design. If you had been reached low switching loss (and proper EMC), you could have used much higher freq. with better result. A modern amplifier shouldn't run under 100 kHz even if it's only for sub.
  Reply With Quote
Old 11th August 2011, 09:24 AM   #15
diyAudio Member
 
Join Date: Jan 2011
Quote:
Originally Posted by Pafi View Post
And finally: sometimes you have to accept that a loss cannot be eliminated. Then increase cooling!
I had already accepted the loss but in simulation I saw nearly 100A current spikes through the transistors. this looks big stress on transistor and risky. I'm struggling because of this.
by the way I didnt see 100A current spikes (nearly 10A spike) through transistors on the real circuit but I thought because of the capacitance of the probe I missed the sharp spikes!? is that possible?
  Reply With Quote
Old 11th August 2011, 09:26 AM   #16
Pafi is offline Pafi  Hungary
diyAudio Member
 
Join Date: May 2004
Location: Budapest
Quote:
Originally Posted by volkantr View Post
Pafi thank you very much for the explanation. I had already accepted the loss but in simulation I saw nearly 100A current spikes through the transistors. this looks big stress on transistor and risky. I'm struggling because of this.
by the way I didnt see 100A current spikes (nearly 10A spike) through transistors on the real circuit but I thought because of the capacitance of the probe I missed the sharp spikes!? is that possible?

thanks again for the explanation I will try the ways you adviced
You're welcome!

The lower current spikes can be caused by stray inductance and band-limited current probe also. What is your current probe? (I have only a homemade transformer, unable to measure over 50 MHz.)

Stray inductance in source connection can slow down current slope very effectively. For example Vg=10V, Vgs=5V, Ls=5 nH means 1000 A/ns. trr=31ns @ 100 A/ns, but slightly decreases with current slope rate, so maybe it's only 18 ns. trr=ta+tb, where ta is the time when current rising (fully opened p-n junction), and this time is maybe less then 10 ns, so current spike may be below 10 A, as you measured. But I guess it's a little more in reality.
  Reply With Quote
Old 11th August 2011, 09:44 AM   #17
Denniz is offline Denniz  Russian Federation
diyAudio Member
 
Join Date: May 2011
Quote:
Originally Posted by Pafi View Post
I don't think such an evident thing should be explained, but if you wish:

1N5400 is not fast at all, it's standard recovery diode, probably much slower then the body diode of IRF6645. Vf (@8A) is also higher then IRF6645. So it is absolutely not OK.

IRGB4045 is also worse than IRF6645 in almost every aspect: much higher conduction loss, slower diode, much slower turn off... the only benefit is higher VCE, but it is not needed.

You can use almost anything, with some results, but here the question was not just a not exploding device, but a better solution than the original (which is already quite good)!



20 kHz says: there are problems with your design. If you had been reached low switching loss (and proper EMC), you could have used much higher freq. with better result. A modern amplifier shouldn't run under 100 kHz even if it's only for sub.
Thank You very much for an explanation, Pafi.

You're absolutely right in every aspect!

The only thing is "body diod" - doesn't it is just a parasitic PN structure inside a transistor?

So, why should one rely upon this "diod" as recovery one?
  Reply With Quote
Old 11th August 2011, 10:08 AM   #18
diyAudio Member
 
Join Date: Jan 2011
Quote:
Originally Posted by Pafi View Post
You're welcome!

The lower current spikes can be caused by stray inductance and band-limited current probe also. What is your current probe? (I have only a homemade transformer, unable to measure over 50 MHz.)

Stray inductance in source connection can slow down current slope very effectively. For example Vg=10V, Vgs=5V, Ls=5 nH means 1000 A/ns. trr=31ns @ 100 A/ns, but slightly decreases with current slope rate, so maybe it's only 18 ns. trr=ta+tb, where ta is the time when current rising (fully opened p-n junction), and this time is maybe less then 10 ns, so current spike may be below 10 A, as you measured. But I guess it's a little more in reality.
I use tektronix p2220 voltage probe to see the current pass through 0.1ohm resistor (as you can guess resistor is between the supply and transistors). I'm not using current probe.
As you say there may be measurement errors to see the real current values.
because of this I cant understand the techniques, I applied to reduce the spikes, is vital or not...

by the way as you said in previous message irf 6645 has a fast body diode so applying shunt diode technique is useless I think because I couldnt find a diode that has better power and timing specs and lower Vf?!
  Reply With Quote
Old 11th August 2011, 11:01 AM   #19
Denniz is offline Denniz  Russian Federation
diyAudio Member
 
Join Date: May 2011
Quote:
Originally Posted by Pafi View Post
1N5400 is not fast at all, it's standard recovery diode, probably much slower then the body diode of IRF6645. Vf (@8A) is also higher then IRF6645. So it is absolutely not OK.
To conclude:
1. Leaving a design without recovery diod, one increases, let's say, power dissipation across transistor - that is an answer to Volcantr's very initial question, suppose.
2. 1N5400 just works and it shows on oscillogram and power transistors temperature substantial decrease - why should one ask for more?
3. Overall, when one TESTS a device for a minute or so and when the same device RUNS for an hour - the power reserve, that IGBTs along with robust recovery diod give, conquers all. Imperical fact, either.

But, should be added, that currently I'm on the very curious way to try a TRUE "Transistor plus Diod" structure from IR - they call it "copac".
  Reply With Quote
Old 11th August 2011, 02:01 PM   #20
diyAudio Member
 
Join Date: Jan 2011
Quote:
Originally Posted by Pafi View Post
You're welcome!

The lower current spikes can be caused by stray inductance and band-limited current probe also. What is your current probe? (I have only a homemade transformer, unable to measure over 50 MHz.)

Stray inductance in source connection can slow down current slope very effectively. For example Vg=10V, Vgs=5V, Ls=5 nH means 1000 A/ns. trr=31ns @ 100 A/ns, but slightly decreases with current slope rate, so maybe it's only 18 ns. trr=ta+tb, where ta is the time when current rising (fully opened p-n junction), and this time is maybe less then 10 ns, so current spike may be below 10 A, as you measured. But I guess it's a little more in reality.
Btw I tried shunt diode by using SS26 and MRA4007T3G diodes seperately (these diodes may be not suitable again but I have only these parts now) and it didnt work.
I played with the gate resistors and increase it near 33ohms but again it didnt work.. one last way, playing with dead times, may be it will work but now the mosfets increadibly hot! If I execute the circuit after 1 munites, the fets are smoking (without cooler) is that normal? only placing a cooler can be solution? it seems a bit risky to me. With single resistor as load with the same conditions it works with no problem (no cooler again)

Also I wonder, there is a lot of application "full bridge drive with inductive(reactive) load". Is everybody confronts with the same issues? may be I'm struggling at the false floor.?

I feel like loosers but for the last two weeks.. I'm struggling with this and I couldnt go forward a teeny step
  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
NFB and load characteristics niko084 Solid State 51 8th March 2011 05:51 AM
Transistor and Temperature lumanauw Solid State 10 12th September 2004 09:44 PM
MC load, normal input characteristics peranders Analogue Source 34 1st October 2002 10:50 PM


New To Site? Need Help?

All times are GMT. The time now is 01:02 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