I have advised a few people that the irs2092 will work with irfb4227's without a buffer.
I found this to be true for my first batch of boards.
On my most recent batch the dreaded 2092 "click" started on driving the amp hard.
This turned out to be the mosfets gate capacitance slowing down the mosfet switch on period.
So for even one pair of 4227's you need a gate buffer. I use the tc4420.
This seems to drive two pairs ok as well.
I found this to be true for my first batch of boards.
On my most recent batch the dreaded 2092 "click" started on driving the amp hard.
This turned out to be the mosfets gate capacitance slowing down the mosfet switch on period.
So for even one pair of 4227's you need a gate buffer. I use the tc4420.
This seems to drive two pairs ok as well.
...I am using the 20957 with totem drivers (FZT853 and 953) for IRFB4115 and it is working great.
Of course you do not need such high voltage BJTs, but should be fast high current switching types. FZT853 and 953 are simply in my assortment, because I do not want to store the same thing in multiple voltage levels...
I made the pcb tracks as short and as thick as possible.
The problem was simply the irfb4227's were to capacitive at their gate for the 2092. The gate and so the mosfet switched on too slowly and the irs2092 saw this and started resetting. With the buffer the problem went away completely. The over current settings were set high so it couldnt have been an over current causing the reset.
The problem was simply the irfb4227's were to capacitive at their gate for the 2092. The gate and so the mosfet switched on too slowly and the irs2092 saw this and started resetting. With the buffer the problem went away completely. The over current settings were set high so it couldnt have been an over current causing the reset.
Yes, but the usual ringing tail after the switching event (if not suppressed) can trigger the overcurrent protection too, or cause parasitic toggling of the comparator due to cross-talk, which severely degrades efficiency and reliability. I associate this behavior with thru-hole parts and/or not using a proper ground plane (not low enough power-to-signal cross-talk, usually appears with amplifier near clipping into low impedance). Only amplifiers free from parasitic switching can be considered reliable.
Very Old thread, but here goes.
Been fightin some with IRS2092 and IRFB4227 and have experienced same as stated above.
OCP tripping at a much much lower point than set for.
Knew it would probably be difficult to drive directly, but noticing that iNUKE 6000 uses IRS20957 driving IRFB4227 directly, I thought that ti should be possible.
But finding the dificulties, I have been digging.
Sure my reference Vss for the lower point could do with a much better layout, I'm still thinking it should be ok.
One thing making this even more difficult to pull off is the very low Rds,on og 19mOhm making for a very low tripping voltage, and therefore making very little room for error.
Checking out the iNuke 6000, the OCP is set to more than 150A .... guess they found the same problem, and just cranked up the trippoint to compensate.
But have also been reading in the doc for IRAUDAMP9 and around rhe use of buffers is says:
"This buffering action is very necessary to speed-up the switching times of each mosfets in order not to exceed the OCP voltage monitor time. The IC commences drain-to-source voltage monitoring as soon as the HO /
LO go to high state but after the leading edge blanking time."
In AN-1138 it is stated that the blanking time is 450nS, ..... would seem long time enough for switching more or less anything ... I very sure my on switching time with 4227 was a lot faster ... so are we back to the layout and parasitics?
And why does IR say you need buffers to overcome this??
Anyway driving the large Qg load of a 4227 or a pair of these would seem to require more current than the IRS2092 is rated at.
But from Diodes/Zetex design note 80 we have:
The average Gate current during the switching event can be calculated thus:
IG = Q/t,
where:
IG is the average gate current
Q is the total gate charge (QGS + QGD)
t is the switching transient time (ton or toff)
For IRFB4227 Qg = 70 nC, with Iout,IRS2092 = 1 A
=>
t = 70nC/1A = 70nS ..... much less than the blanking time
But using a buffer to gain more output current seems like a good way to go anyway, and I'm sure there is more to this than this ....
Another question is whether there is any output current protection on IRS2092??
1A at 12 V amounts to a load of 12 ohm ..... with an internal gate resistance of say 2 ohm then the minimum Rg sould be 10 ohm.
I'm running my current design with half that!!!
Las think on my mind:
Thet is the upper limit for using IRFB4227??
Couldn't you get at least 1000W in 4 ohm (+-90Vdc supply) safely?
Why would yo have to go to 2 pairs? ..... and double the challenge of driving them ....
As long as the Ploss is not too big,and it will for sure decrese with faster switching, and the voltage is not too high (so maybe it should actually be IRFB4229 to be safe on 90Vdc)
/Baldin
Been fightin some with IRS2092 and IRFB4227 and have experienced same as stated above.
OCP tripping at a much much lower point than set for.
Knew it would probably be difficult to drive directly, but noticing that iNUKE 6000 uses IRS20957 driving IRFB4227 directly, I thought that ti should be possible.
But finding the dificulties, I have been digging.
Sure my reference Vss for the lower point could do with a much better layout, I'm still thinking it should be ok.
One thing making this even more difficult to pull off is the very low Rds,on og 19mOhm making for a very low tripping voltage, and therefore making very little room for error.
Checking out the iNuke 6000, the OCP is set to more than 150A .... guess they found the same problem, and just cranked up the trippoint to compensate.
But have also been reading in the doc for IRAUDAMP9 and around rhe use of buffers is says:
"This buffering action is very necessary to speed-up the switching times of each mosfets in order not to exceed the OCP voltage monitor time. The IC commences drain-to-source voltage monitoring as soon as the HO /
LO go to high state but after the leading edge blanking time."
In AN-1138 it is stated that the blanking time is 450nS, ..... would seem long time enough for switching more or less anything ... I very sure my on switching time with 4227 was a lot faster ... so are we back to the layout and parasitics?
And why does IR say you need buffers to overcome this??
Anyway driving the large Qg load of a 4227 or a pair of these would seem to require more current than the IRS2092 is rated at.
But from Diodes/Zetex design note 80 we have:
The average Gate current during the switching event can be calculated thus:
IG = Q/t,
where:
IG is the average gate current
Q is the total gate charge (QGS + QGD)
t is the switching transient time (ton or toff)
For IRFB4227 Qg = 70 nC, with Iout,IRS2092 = 1 A
=>
t = 70nC/1A = 70nS ..... much less than the blanking time
But using a buffer to gain more output current seems like a good way to go anyway, and I'm sure there is more to this than this ....
Another question is whether there is any output current protection on IRS2092??
1A at 12 V amounts to a load of 12 ohm ..... with an internal gate resistance of say 2 ohm then the minimum Rg sould be 10 ohm.
I'm running my current design with half that!!!
Las think on my mind:
Thet is the upper limit for using IRFB4227??
Couldn't you get at least 1000W in 4 ohm (+-90Vdc supply) safely?
Why would yo have to go to 2 pairs? ..... and double the challenge of driving them ....
As long as the Ploss is not too big,and it will for sure decrese with faster switching, and the voltage is not too high (so maybe it should actually be IRFB4229 to be safe on 90Vdc)
/Baldin
Hey luka ... long time 
Hey Nigel
So with the buffers, are you able to control the OCP trip point with good precision?
... that is does the calculated trip point resemble the measured point?
I'm still not sure I understand the reason for the resetting:
- too long switching time due to too low driving current, longer than the blanking time ... seems a bit unlikely with 450nS
- bad layout around the sensing resulting in spikes or too high voltage caused by parasitic
- other
- a combination
I did connect via a 4R7 gate resistor, but it didn't work well.
Would like to understand better before making yet another PCB.
Also the question of how far you can actually push a single pair of 4227, remains
KR Baldin
Hey Nigel
So with the buffers, are you able to control the OCP trip point with good precision?
... that is does the calculated trip point resemble the measured point?
I'm still not sure I understand the reason for the resetting:
- too long switching time due to too low driving current, longer than the blanking time ... seems a bit unlikely with 450nS
- bad layout around the sensing resulting in spikes or too high voltage caused by parasitic
- other
- a combination
I did connect via a 4R7 gate resistor, but it didn't work well.
Would like to understand better before making yet another PCB.
Also the question of how far you can actually push a single pair of 4227, remains
KR Baldin
The 2092 doesn't care if the mosfet has gate drivers or not, it always monitors the output.
What happens is the mosfet starts to turn on slowly because of gate capacitance but doesn't get there in time and OCP cuts in and resets the 2092.
I used tc4420 gate buffers in one design and transistor gate buffers in another design and both worked well.
What happens is the mosfet starts to turn on slowly because of gate capacitance but doesn't get there in time and OCP cuts in and resets the 2092.
I used tc4420 gate buffers in one design and transistor gate buffers in another design and both worked well.
Hey Baldin, sure is, time flies...Hey luka ... long time
Hey Nigel
So with the buffers, are you able to control the OCP trip point with good precision?
... that is does the calculated trip point resemble the measured point?
I'm still not sure I understand the reason for the resetting:
- too long switching time due to too low driving current, longer than the blanking time ... seems a bit unlikely with 450nS
- bad layout around the sensing resulting in spikes or too high voltage caused by parasitic
- other
- a combination
I did connect via a 4R7 gate resistor, but it didn't work well.
Would like to understand better before making yet another PCB.
Can't say, I don't think I ever hit OCP, maybe did couple of shorts here and there
But it does sounds you are getting in the same situation ChocoHolic was in. You might want to go over his thread
Babysteps of 1kW Class D Rookie Amp
and probably even more
1kW Gen2
There were all sorts of problems like this, solutions also, good read in general
Maybe you can make simple jig, fet, inductor, diode,... and some kinda pulse generator, so that you could see how thermal and switching performance is and have control at the same time
Last edited:
Hi b-force and luka
Yes both TC4420 and TC4452 are a bit of overkill, and also introduces an unwanted 50nS propagation delay.
Yse had my eys on the Zetex dual transistors and had started to incorporate into my design. Low delay, fast switching and high current handling.
And for implementation with lower capacity FETs like IRFB5620 I can just use a lower current version of the ZXGD ..... probably will benefit FETs like these also.
Yes both TC4420 and TC4452 are a bit of overkill, and also introduces an unwanted 50nS propagation delay.
Yse had my eys on the Zetex dual transistors and had started to incorporate into my design. Low delay, fast switching and high current handling.
And for implementation with lower capacity FETs like IRFB5620 I can just use a lower current version of the ZXGD ..... probably will benefit FETs like these also.
Hello everyone, I have created an iraudamp9 starting from the basic project using only a pair of IRFB4227.
I'm powering everything with +/- 52volt, while VCC is derived in the same way that it happens in IRAUDAMP7S (do you think it is not enough?), In this way, during the accented use, VCC also drops to 10volt (this is bad ?)
I used a totem pole driver (BD139 + BD140).
This circuit has been realized on a single PCB integrating also a switching power supply to be able to use it in the car.
The amplifier works, but at low volume it has a bit of distortion, moreover, as in your case, the OCP intervenes too soon.
How could I set the trip higher?
Is there a document I have to read? which? can you tell me? thank you.
I'm powering everything with +/- 52volt, while VCC is derived in the same way that it happens in IRAUDAMP7S (do you think it is not enough?), In this way, during the accented use, VCC also drops to 10volt (this is bad ?)
I used a totem pole driver (BD139 + BD140).
This circuit has been realized on a single PCB integrating also a switching power supply to be able to use it in the car.
The amplifier works, but at low volume it has a bit of distortion, moreover, as in your case, the OCP intervenes too soon.
How could I set the trip higher?
Is there a document I have to read? which? can you tell me? thank you.
Hi Mario
What load are you using?
Even for 2 ohm I think you could and should use lower rated MOSFETS.
E.g. IPP200N15N3 G, would give you 50A and a load of only 1820 pF, and with Vds on 150V should be ok for your +-52Vdc
And it can be driven without the buffer ..... also BD139/140 is not the best for this .... better to use SMD types to reduce trace lengths and parasitics
Gate voltage on 10V should be ok, but why is it dropping? And from what level?
And distortion at lower levels does not sound right at all
You need to read teh app notes and data sheet for IRS2092 ... there you can find the calculations for OCP trippng point
https://www.infineon.com/dgdl/an-1138.pdf?fileId=5546d462533600a40153559a077610d1
page 10
What load are you using?
Even for 2 ohm I think you could and should use lower rated MOSFETS.
E.g. IPP200N15N3 G, would give you 50A and a load of only 1820 pF, and with Vds on 150V should be ok for your +-52Vdc
And it can be driven without the buffer ..... also BD139/140 is not the best for this .... better to use SMD types to reduce trace lengths and parasitics
Gate voltage on 10V should be ok, but why is it dropping? And from what level?
And distortion at lower levels does not sound right at all
You need to read teh app notes and data sheet for IRS2092 ... there you can find the calculations for OCP trippng point
https://www.infineon.com/dgdl/an-1138.pdf?fileId=5546d462533600a40153559a077610d1
page 10
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.