My class d IRS2092 project

[nigelwright7557]

Had my first crack at class d today using the IRS2092.
I used the circuit in the datasheet.

Found the IRS2092 wont start with a speaker connected so looking at making the over current protect work a little higher.

[luka]

My also don't want to start... it looks like overcurrent protection, but cant be too sure. And I also used circuit from the datasheet, even made pcb as they said it perfect. All I don't have the same is irfp4227's, pretty big fets I guess

[nigelwright7557]

Originally Posted by luka

My also don't want to start... it looks like overcurrent protection, but cant be too sure. And I also used circuit from the datasheet, even made pcb as they said it perfect. All I don't have the same is irfp4227's, pretty big fets I guess

When i get time i will have a play around with the over current setting, see if i can raise it a bit.

There does seem to be a few conditions that have to be met before the 2092 will run. All the power supplies have to be right, the input has to be zero, the output has to be zero as well.

I am using the 4227 too, seemed sensible as it is ver yquick and i can add heatsinks for higher power.

I am slowly getting into the 2092 and understanding how it works and how to set it up.

[luka]

My best guess would be that dead time is not big enough(and freq. is really low), this is how it looked to me, even if DT at max, but never looked if it really was those XXns...
At first I thought that fets are way too big for the driver, so I added zxgd3001e6 driver, for each fet. And gates really do open fast, clean signal, but it always cycles on/off if I remember, no music I think, coz its that fast...not 100% sure but I think there was sound comming out of it and really good sounding too, but freq was never was what was set to, only about 100k or so.

Does yours also work at lower freq then set?

[nigelwright7557]

Originally Posted by luka

My best guess would be that dead time is not big enough(and freq. is really low), this is how it looked to me, even if DT at max, but never looked if it really was those XXns...
At first I thought that fets are way too big for the driver, so I added zxgd3001e6 driver, for each fet. And gates really do open fast, clean signal, but it always cycles on/off if I remember, no music I think, coz its that fast...not 100% sure but I think there was sound comming out of it and really good sounding too, but freq was never was what was set to, only about 100k or so.

Does yours also work at lower freq then set?

I am using maximum deadtime until I get the 2092 running OK.
I can look at rise and fall times later.

All the voltages are coming up Ok so it must be the overcurrent protection cutting in. It makes a burst of noise then resets itself.

I am using the 4227 MOSFETs and these arent really recommended as the gate charge is too much by about 80%.

[Workhorse]

Originally Posted by nigelwright7557

All the voltages are coming up Ok so it must be the overcurrent protection cutting in. It makes a burst of noise then resets itself.

Your drum core might see itself into saturation which causes steep increase in the ripple current also, which in-turn causes the chip to hiccup under drastic conditions. Another cause might be the current sensing is tripping too early.

[luka]

Originally Posted by nigelwright7557

I am using maximum deadtime until I get the 2092 running OK.
I can look at rise and fall times later.

All the voltages are coming up Ok so it must be the overcurrent protection cutting in. It makes a burst of noise then resets itself.

I am using the 4227 MOSFETs and these arent really recommended as the gate charge is too much by about 80%.

Look at both gates at the same time, to see if there is enough dead time, no shoot-throughs.
Like workhorse said, core could be a problem too, but same core wasn't a problem on other amps, so I would say in my case, that's ok
I wonder what would happen if I would dissable overcurrent

[odnaizutra]

maybe you can start doing another pcb using 0.1uF very close to the MOSFETS from rails to ground, a 0.47uf also very close to the MOSFETS from rail to rail and a snubber 0.1 uF+2r2 from rail to rail, them put 2092 as close as possible to switches.Change the MOSFETS and use other with less than 40 nC of qg.Maybe 4019 ... I am not sure , but irf640n can be an option to begin tests.10 to 15 r as gate resistor, switching frequency of 350 to 400khz not more or less at beggining.Try to connect -b of 2092 direct at the source of low side fet and OUT of high current path from power supply. This track is very critical to ocp.Be sure that all voltages are present in order to not disturb uvp of each block.Remember that vb capacitor has to be charged before 2092 initialization.
Adjust the high side current shutdown to very high value and afterwards to the desired value , do this at low side too.Observe that the way to adjust them are different!
Maybe you will find that the calculated values of ocp are different from the real life...

[odnaizutra]

As said before , but i had some problems when not using t 106-2 at inductor.

[giovyy]

@nigelwright7557
where you purchased IRS2092 in DIP package?
Thanks