Problems with IRS2092??????

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I couldn't get it (irs2092) to stop resetting on high power when it really shouldn't be. (L-25D board) I am going to have to disable the overcurrent protection and replace it with a fuse. The irs2092 application note says to disable the low side ocp, connect the ocset pin directly to the vcc pin. That doesn't solve it for the high side protection. But disabling the high side protection is more obvious. looking at the amp7s drawing makes it pretty obvious. If we remove R18 from the amp7s drawing, the CSH pin should be solidly held to the VS pin. This should tell the amp that there is no current flow and disable high side ocp. I think what tends to happen on these designs is that depending on the configuration, the mosfet may not turn on as quickly as it is supposed to. Some mosfets will take more time to turn on. If it takes too much, then it will trip the ocp because ocp is measuring voltage across the fet. LJMs l25d board uses larger fets that probably have a larger gate charge requirement and take more time to turn on. There is also probably some extra inductance in the circuit because of trace length imposed by using separate fet cases causing the DS voltage to appear higher than it is. I have had to scrap with this problem for a while, and it has made me rather dissatisfied with the l25D board. It must be working right for some people. Something else that might raise the DS voltage would be an output filter inductance that is too small, causing current rush spikes .. or maybe any capacitance on the fet side of the output inductor could do the same thing. Anyways ... totally disappointed with the irs2092 ocp. It's such a promising idea.
 
I had exactly the same problem.

The CSH pin needs a voltage divider on it to reduce the voltage into that pin.

I also later had a problem driving high gate capacitance mosfets and I used TC4420 gate drivers to get around that.

Decoupling is also important, as close to 2092 as possible with decoupling across b+/b-/gnd close to mosfets.

The ocset pin also needs as close to vref as possible to reduce the resetting.
 
In the case of the voltage divider on the high side, we should be able to divide the voltage down as much as we need to, but the problem on the bottom is that vref sets a maximum of 5 volts. If anything is a little slow, the 5 volts is still too low (especially if the transistors are getting hot). With ocset set to vref, this unit is still tripping when it shouldn't be, even though it is tolerating more current and really almost acting satisfactorily. I'm pretty sure the larger fets just don't quite turn on fast enough. I can't change the board, so using fet drivers is out of the question. I might be able to set up to voltage divide off of vcc rather than vref. That might actually solve the problem. Unfortunately, this is not a tactic that could be used generically because people have different power supply voltages. Maybe a good solution would be to use a 12V regulator to create a 12 volt reference voltage for ocset. I have some 15v regulators, but I don't know how much voltage they can hold. Maybe all this needs is some good old zeners to set a voltage reference. Don't need a lot of current.
 
I am using a 1l over 4k7 divider on the ocset pin and not getting tripping with a 750 watt amplifier.

If you have set ocset to 5 volts then there is a problem somewhere else.
The 2092 looks at the DS voltage of the lower mosfet and if it is above ocset then it resets.
It could be your not charging the mosfet gate fast enough or it is simply a slow mosfet.

Which mosfets are you using and are you using a gate driver chip or transistors ?
 
These guys are IRFB4020 They have about twice the current rating of the IRFI4020 that is quoted in the amp7s diagram. I thought that would imply twice the gate charge (Qg) but looking at the specs, it shows the IRFB4020 and the IRFI4020 with 18 and 19 nano-coulombs respectively and both with a gate resistance of approximately 3 ohms. So I don't see how the switching could be significantly different from the reference design.

Of course I am running low impedance load (4-6 ohm) which accounts for some of the issue. The reference design does not allow 4 ohm load, but its using the weaker transistors. The ds voltages are similar between both transistors, and the original board had the same current limit resistors as the reference design, telling me that while the transistors can carry the load, the ocp is designed to trip before they do. So I raised both resistances which didn't help much. Then cut them out, and that helped as the board I changed now trips less easily than the one I didn't, but they still trip when they shouldn't. So getting desperate, I took out the trip timer capacitor, now when it trips, I just hear a tiny little crackle, but this is not the right solution. ...

To get the amps to reliably run stable, I have to use about 6 wraps of wire in each of their signal ground lines because their switching rf energy seems to mutually interfere, causing a sporadic, noisy distortion. But running without the tiny coils doesn't seem to affect the trip behavior. After thinking about it, I realized that putting those coils in there can't impact the top end cross-out because the inductance just won't be high enough to compare to the input/output impedance. Interestingly though, if I raise that inductance quite a bit higher, the boards will error out a lot harder and I have to power them down and back up to fix the error.

The top end crossout is important because supposedly the output filter has resonance between 30 and 60 khz, and resonance can cause tripping. The board is in fact using the supposedly optional top end cross out resistor/capacitor pair that is intended to stop those resonance conditions from happening. Knocking the high frequencies down by 6 db doesn't seem to have an impact on tripping, so I'm pretty confident that I am not causing resonance in the output filter.. especially since that resonance is dampened heavily by a 4 ohm load. ... Gotta be something else.

It just might be some kind of a dynamic interaction between the amp and the speakers crossover reactances causing sporadic high currents. A resistive dummy load would help me consider that possibility. In any case, an amp has to be able to deal with crossover reactances or it's not up to par. Only thing to do I can think of at this point is to try pushing OCSET up higher than 5 volts. The amp is not overheating. I added some extra heat sink, and it is running warm but not hot so long as I keep it right side up, so whatever is happening is not causing a lot of transistor current at a partial on state.

The boards also seem to trip more easily once they have warmed up, which agrees with the positive RDS temperature coefficient. It also hints that the tripping is probably genuinely due to VDS. So unless someone has a genius plan, next thing to try is raising OCSET even higher. 6-7 volts should make a difference I would think.
 
OK, I looked at the application datasheet and saw the 10 ohm resistor, went to r-shack and got some. I didn't want to jerry-rig the 10 ohms because I was thinking this thing might need a ground plane. However, I don't see a ground plane on the board unless it's on an inbetween board layer.

OK, so I soldered it up, put the trip timer capacitor back in place and whala, about 4-6 more decibels before it trips (my mixer board has L.E.D.s that tell me what relative level the signal is at). We rock dude!!! Still surprising me is that it trips at all with the OCP resistors removed, but as we know, the low side OCP reference is clamped at 5 volts, so it must be low side tripping. All-in-all, I think there just might be another decibel left in the power supply because I didn't hear clipping at the trip, but this guy is performing, and that makes me smile. I bet it is doing its rated 350 watts into 4 ohms. With the transistor turning on faster, I bet it will be more efficient and cooler. OK, so this thing should have worked out of the box, but didn't. Last thing on the agenda is to figure out how to tune the high side trip. When that is done, I'll return the information to the seller and demand an extra set of boards for my efforts. .. Maybe I should demand two sets because I have probably spent close to $1000 worth of my time to get this thing under control.

... Next stop ... build a class-d amplifier that we can use to invert 5 to 8 kw of power from a solar energy power supply to 120/240 volts. My smeller says that this class D amplifier strategy is superior to the inverter boards I have seen. Cheap power inverters are one of multiple key components of future societal health. Carefully examining the solar power industry shows that they are wasting a lot of money and power, and that the cost of that systematic should be able to be cut in half. Not saying that we can cut the price/watt in half, but between a 30% cut in price/watt and a 30% cut in number of watts necessary, we should be able to chop the system cost in half.

.. and finally, now you know that that resistor needs to be 10 ohms, not 20, and thank you very much for your help.
:)
 
Oh, this thing may also need a thorough test to see if the output inductor handles the heat. That thing gets hot. Anyone ever thought of just using air core inductors for these things? Also, btw, you play guitar? I love guitar. I was extremely pleased when I manufactured a stereo output for the guitar using 3 rather than 2 pickups. So your ears get to hear two separate viewpoints of the strings motion. It adds information and flavor to the sound for sure.
 
Last edited:
Some of the guys building PWM modulators for class E ham radio AM gear are using air cored magnetics for the output filter (Often 4th order) at design load impedances in the order of a few ohms, so air core is possible (I must introduce that crowd to some of the syncronous class D parts, that IR driver for single supply rail D springs to mind as a cool thing for that use).

Usually output inductors getting hot is a sign that you used the wrong core material, or are running too much flux density, there is a **LOT** of difference between something like a ETD49 with a 2.5 mm gap and a FT47 ferrite ring for example, also be a little careful of the iron powder cores (TXXX-2,6,17 and the like) they are sometimes disturbingly temperature sensitive.

Note that an air core inductor tends to have a **MASSIVE** fringing field that can be somewhat reduced by winding it on a -0 toroidal form (this is non magnetic but forms the solonoid into a closed magnetic path), things like steel or aluminium panels within a few coil diameters will completly trash the Q.

Some of the sendust derived high flux density materials may be worth a look.

Regards, Dan.
 
Usually output inductors getting hot is a sign that you used the wrong core material, or are running too much flux density, there is a **LOT** of difference between something like a ETD49 with a 2.5 mm gap and a FT47 ferrite ring for example, also be a little careful of the iron powder cores (TXXX-2,6,17 and the like) they are sometimes disturbingly temperature sensitive.


Regards, Dan.

I use a T106-2 with excellent results. It gets barely warm.

On my first attempt at a class d amp I just used a power inductor and this got to 120 degrees C !
 
-2 is a fairly low permiabilty iron powder core that is not horrible for this sort of thing, certainally far better then something using a -47 or so core (ferrite).

Have you measured how much switching residual is making it out? I am a little surprised you got enough inductance on a -2 core.

Power inductors of the SMPSU variety are sometimes ungapped high permiability materials and using such will want the datasheets studying carefully.

Regards, Dan.
 
dmills;3289293 Have you measured how much switching residual is making it out? I am a little surprised you got enough inductance on a -2 core. Regards said:
There is a few volts of residual carrier on the output but the speakers inductance will reject that frequency that anyway.
I wind on the turns around the core twice to get sufficient inductance.
 
I spoke a little too soon. The added stability for the OCP protection because of the swap to 10 ohm transistor drive seems to be mostly showing up only for the low frequency range. The higher frequency range is also improved, but not nearly so much, so I have more searching to do to get the OCP for this unit under control. I have to wonder why the bottom end is delivered so robustly, but mids and highs are still weak.
 
Circuit diagram 500W amp?

hi,
has anybody reversengineered the 500W amp? Or part of it?
I wonder what those 4 transistors do. Or are those voltage regulators?
And also, I couldn`t find any temperature sensor, I think the fan runs all the time?
The reason I ask, is, bevor I run chance to destroy the board, slipping a probe, or so.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.