diyAudio

diyAudio (http://www.diyaudio.com/forums/)
-   Class D (http://www.diyaudio.com/forums/class-d/)
-   -   Yet another IRS2092 problem (http://www.diyaudio.com/forums/class-d/189690-yet-another-irs2092-problem.html)

LukeFrog 26th May 2011 04:20 AM

Yet another IRS2092 problem
 
1 Attachment(s)
Hi Everyone

This is my first time posting, I have read many threads on this forum and have found them very helpful :) I was wondering if anyone could help with a problem I'm facing.

I'm currently building a class D amp based loosely around the IRAUDAMP7. We are using the 2092 chip, and have just powered it up for the first time. For some reason, the LO output is working fine, outputting a nice square waveform, but the HO output isn't doing anything at all. It just seems to be sitting at Vs.

Does anyone have any idea of what might cause this or if it's an easy-to-fix problem?

luka 26th May 2011 06:00 AM

why are R16, R20 0R resistors?
the main reason I can see is, that you have no Vb capacitor, to hold the voltage, and coz you don't have it, IRS can't open the fet above + supply voltage, which means that fet is 100% closed

you need capacitor between pin 13 and 15, without it, it will never work

Also I think, and you will see that by yourself, input filter, like in IRAUDAMP7, p30
also 100n input cap seems low in values for full range
also capacitors for suppy filtering, all you have, vcc, +5, -5...

And don't even get me started about output filter, I feel this are not the real values
And there are several things still missing, which mush be on board, note that this is not AB amp, if you are designing it like that

LukeFrog 26th May 2011 06:09 AM

1 Attachment(s)
Thanks Luka

My apologies, the schematic uploaded was an earlier version. We do have a cap across Vb and Vs (see attached).

R6 and R20 have been meddled with as we thought the high-side current sensing could be the problem. We have tried the recommended values for R2/R16 and R6/R20 as described in the application note for the chip and it didn't seem to do anything. They were then made open in the hope that it would set the threshold so high that we could eliminate it as the cause. But to no avail. We're still scratching our heads.... :/

luka 26th May 2011 07:03 AM

You really have 1n4001 for D2, D6?? that will not be ok, you need fast diodes there
ok, so, that is the voltage betwen pin 15 and 13, should be Vcc- diode forward drop

Again, none of diodes MUST NOT be 1n400x series, they are slow, 50Hz/60Hz type diodes

What is the freq. of the LO that you said it works
Again, that is real value of the C31

LukeFrog 26th May 2011 07:32 AM

Hi Luka

The diodes used for D2 and D6 are fast recovery diodes. The schematic lacks some minor part number discrepancies due to the fast approaching deadlines of the project. Suffice to say, the diodes are sufficient in this context.

When we put a 100Hz sine wave into the input, we get a 100Hz square wave out of the LO pin that ranges from the negative rail to Vs. However, when this was done, the FETs were not in place, as the circuit would immediately draw large amounts of current and heat up rapidly. So the output may not be correct because the switching frequency feedback from Vs. So we've put it down to the high side not working.

I should also mention that this is for a subwoofer (<200Hz) so high frequencies are not of a concern to us.

We appreciate your help, this is our first D class Amp, so it's a steep learning curve!

luka 26th May 2011 08:36 AM

You don't need any input to get oscilation, which should be in few 100kHz range, not what you put on input
And you need to have fets in the circuit, that is like having no motor in car, coz it wants a lot of gas.
you can limit current with outside resistors, say 10-22R per rail
Idle current of working circuit is very low
and for start use the biggest dead time you can

Now its time you post some measurements, of the voltages you have

nigelwright7557 26th May 2011 08:54 PM

I feel we are wasting our time helping you without a fully up to date circuit diagram.

Is your inductor really 10mH ? it should be around 20uH !
The input capacitor should be 10uF.

I can promise you the 2092 does work if everything is right plus decoupling and pcb layout kept really short.
I have been through the painful exercise of getting one working too.
Its not like class AB which is much more forgiving.
I first fell foul of decoupling the output mosfets and decoupling b+/b-, b+/gnd b-/gnd.
Then my 12 volt supply was a bit lacking and needed extra decoupling.
I then used a power inductor instead of a proper class d inductor, in the end i used a t106-2 core with a couple of meters of 18swg enammeled copper wire.

nigelwright7557 26th May 2011 09:05 PM

Your overheating could be incorrect deadtime.
You could begetting shootthrough when both mosfets are on at the same time.
Set the deadtime to max until you get the 2092 running OK.

LukeFrog 27th May 2011 01:07 AM

Hi guys

We tweaked our HIGH side current sensing resistor values and put the FETs in and it works!

We have changed our dead time selected to be the max (45ns) at the moment.

The output filter now has a capacitor value of 2.2uF and inductor value of 30mH which removes all of the switching frequency noise.

Works good for now, next step is to make the second channel and join them to make a full bridge output.

Thanks for all the help guys :)

luka 27th May 2011 05:55 AM

you mean 105ns?
And I bet you don't use 30mH inductor
at what freq. does it operate now?


All times are GMT. The time now is 05:30 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