Ungie, thank you for pointing out that app note. I did not see that the totem pole driver circuit is suggested by IR for that driver chip. Apparently the current sensing is using a reference voltage and measuring voltage drop across a resistor to program a trip level. The over current trip that resets the operation of the '2092 when driving too much capacitance at gates is what I thought would be affected, but is clearly not important for protection of the amp, just the chip itself... using the external driver looks like a good way to stretch the power capability of '2092 designs pretty far.
I came across a similar problem.
In my case it was bad with huge thumps and siren type noises on power down.
I contacted IR and they said it was due to me using too much smoothing capacitors. As I didnt want to reduce smoothing I added a small VCC monitor PIC microcontroller that holds the 2092 in reset if the VCC goes below 15 volts.
...no doubt that's an option. But only in case you want to play with a uC.
Much simpler one would do it with two transistors and a zener.
In the same way as IR is monitoring the main rail, one can also monitor the supply of the IRS.
Take inspiration here. Page 23, figure 27
http://www.irf.com/technical-info/refdesigns/iraudamp7d.pdf
...no doubt that's an option. But only in case you want to play with a uC.
Much simpler one would do it with two transistors and a zener.
In the same way as IR is monitoring the main rail, one can also monitor the supply of the IRS.
Take inspiration here. Page 23, figure 27
http://www.irf.com/technical-info/refdesigns/iraudamp7d.pdf
I found I needed an opto isolator as the voltage I was monitoring was around 40 volts lower than the reset pin voltage.
Also I found shorting out the reset pin blew up the 2092.
It is better to discharge the reset capacitor through 220R.
!!
...have to check this in detail, because I am right now playing with a private postfilterfeedback design using the IRS2092 - and wanted to short the cap at the SD pin... (like I am used to do with the IRS20957)
Will keep you updated.
So far all my trials to kill the IRS2092 by shorting the reset cap, did not cause any defects.
Even with brute force shortening the IC simply resets and automatically restarts as described in the data sheet.
...stressed it with 110 events, brute force shorting by a low Rdson MosFet BUZ111S.
...also stressed it over 20 times by shorting it with a screw driver (causing uncontrolled transients), but it always works like intended.
========================================================
@ kaip & spare-parts
Hot transistors:
You can try to increase the dead time by adjusting R13 and/or R19.
Resetting at high volumes:
Might be phantom tripping caused by HF resonances of your switching stage.
Can be cured only by optimized layout and snubbering and partially by component choice.
Edit: ...another improvement might be decoupling of the rails with small caps right at the IC.
Lifting the overcurrent setting of the IRS might also help, but does not
cure the root cause, only mildening one of the symptoms on the expense of reliability.
Stroll through the forum, there are multiple passages in multiple threads on this topic.
Even with brute force shortening the IC simply resets and automatically restarts as described in the data sheet.
...stressed it with 110 events, brute force shorting by a low Rdson MosFet BUZ111S.
...also stressed it over 20 times by shorting it with a screw driver (causing uncontrolled transients), but it always works like intended.
========================================================
@ kaip & spare-parts
Hot transistors:
You can try to increase the dead time by adjusting R13 and/or R19.
Resetting at high volumes:
Might be phantom tripping caused by HF resonances of your switching stage.
Can be cured only by optimized layout and snubbering and partially by component choice.
Edit: ...another improvement might be decoupling of the rails with small caps right at the IC.
Lifting the overcurrent setting of the IRS might also help, but does not
cure the root cause, only mildening one of the symptoms on the expense of reliability.
Stroll through the forum, there are multiple passages in multiple threads on this topic.
Last edited:
Although I like digital electronic and PICs I had to keep it simple.
As said I am restricted to my small PCB size and through hole components.
Heat was solved with other MosFETS.
Now am curious about the IRFI4019/20 so will be using this in a new version.
Optocouplers are nice, but too big here.
The solution with the transistors from the IR application note is perfect.
It is an OverVoltage and UnderVoltage "protection" and has the added advantage
that you can have a simple external mute pin that can be shortened to GND.
Here is my new version: (have not build it yet)
As said I am restricted to my small PCB size and through hole components.
Heat was solved with other MosFETS.
Now am curious about the IRFI4019/20 so will be using this in a new version.
Optocouplers are nice, but too big here.
The solution with the transistors from the IR application note is perfect.
It is an OverVoltage and UnderVoltage "protection" and has the added advantage
that you can have a simple external mute pin that can be shortened to GND.
Here is my new version: (have not build it yet)
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
I definitely had problems if I shorted the CSD pin.
Maybe it was a static discharge that killed it when I shorted it out with a screwdriver ? The 2092 is definitely touchy about attaching scope probes while turned on. Even IR use a 100R in series with the reset capacitor. I use 220r.
I found the irfb4019's to work very well with the 2092.
I thrashed the amplifier and couldn't get it to reset.
However I did find I had to increase the OC setting to help.
The 2092 can be a minefield It is so touchy about layout and decoupling and scope probes.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.