PA AMP 4 x IRS 2092 Full bridged 2 x 1000W RMS 4 OHM - 200 USD - real power or Fake ?


2010-12-11 5:13 pm
Dear Friends

I visit yesterday Frankfurt Fair Germany Prolight and Sound 2016

It looks IRS 2092 Class D Mosfet Driver coming popular for PA Amplifier Manufacturer. With IRS 2092 many companies offer big power PA amplifier with low cost.

Different PA Amplifier Manufacturer offer 1000 W RMS 4 OHM with fullbridge IRS 2092 Class D Mosfet Driver Design.

Ist 1000W RMS possibile ?

can IRS 2092 working stable in PA Amplifier now ?

Attachment picture 1000W RMS x 2 with DSP - Sales price 200 USD FOB China

Im not sure this amp working stable and scare to buy this product

Any ideas ?


  • 200 USD DSP Internal Magnetic Audio.jpg
    200 USD DSP Internal Magnetic Audio.jpg
    193 KB · Views: 1,009
  • 200 USD DSP Magnetic Audio.jpg
    200 USD DSP Magnetic Audio.jpg
    128.4 KB · Views: 986
Last edited:
By design IRS2092 only supports pre-filter feedback, which limits "high quality audio" use of these amplifiers to sub-bass and mid-bass duties. Used full range they could be regarded as "mid quality audio" because HF response is load dependent and shows higher THD and higher output impedance than usual class AB schemes. Difficult to get HF right for anybody requiring good trebble for pleasure.

Another factor is the origin: China. They just use the lowest cost passive parts and simplest design that does not fail during their tests, aggressive circuit simplification. When circuits designed that way fail, even just due to aging, there is usually a big chain reaction where many parts fail, tracks fuse, SMD parts vaporize and sometimes holes are burned in PCB. A hell to repair. No real sense of long-lasting fair-failing industrial engineering, this is Occidental or Japanese thinking, more parts and consideration required.

By design there is nothing in parts like IR2110, IRS2092 or IRS20957 preventing them to achieve full ratings. Only good design and manufacturing is required. My personal mark with 2x IR2110 and 8x TO-220 IRFB4227 is 6000W/2r (implemented in a product selling for a few years now). With 2x IRS20957 and 4x TO-220 IRFB4615 I recently achieved 1600W/5.6r in first PCB revision without any significant problem related to power output.

Also, there is nothing in parts like these preventing an amplifier to achieve high audio integrity within the expected frequency range (IRS2092: good on LF/MF only). It all depends on class D modulator used, optimization of switching, layout and power supply. Companies that have not yet developed and/or optimized their own modulator schemes resort to parts like IRS2092 which has built-in modulator (nearly all unknowns already solved, maximum simplicity) or copy circuits without proper understanding, which can have unpredictable results.

By the looks this amplifier seems like a collection of datasheet circuits. IRS2092 datasheet circuit. LLC resonant SMPS (non-PFC) direct from datasheet circuit. Some big company audio DSP implementation direct from application note, and with most software already provided by manufacturer. Safe design approach with little research, little risk and no innovation or exceptional features at all.

In comparison, during that time others have been developing things like class D modulators with HF responses flat within +/-0.1dB at any load, intelligent limiters with soft clipping that allow extra loudness without unpleasant distortion (like classic amplifiers), voice coil temperature monitoring and limiting without extra wires to get theoretical maximum dB out of speakers, or single stage PFC power supplies that do more power than the LLC using less parts and space (featuring induction cooking IGBT haha). All this requires a lot of research and effort to develop new or custom schemes, so cannot be offered at the same price.

It must be understood that when a product like the amplifier shown in the pictures is bought, nearly no research or innovation is being funded, more likely some slavery is being funded. It can be useful to do lo-fi parties from time to time, but it's not state-of-the-art, and it's not likely to contribute to future technological innovation.


Disabled Account
2005-02-14 3:41 am
It is not a dream. One pair mosfet IXFH94N30P3 can deliver plenty output.

Sound quality is different issue. Always better is more expensive. But power output you can play i.e. For low and sub, cheap class D Amp can make it. Not only cheap breand, actually big manufacturer did this also. But of course they sell as not cheap. :D
I know that modulation riddle can be quite interesting, including LC in the loop, like ucd and maximizing loop gain while keeping phase margin. If one starts thinking about that, he ll never need any more mind games for years and still keeping his mind busy. so why does the word need any crossword riddles at all?
But, honestly a prefilter feedback like the inukes can be quite good sounding, especially if the load is known, like in active speakers. At a rock concert, nobody would care about some non perfect group delay or amplitude response at 17khz, believe me. if one is head banging in the before row that will easily cause deviations of 20db for your ear at this freq, why worry at all.
and that is where class d shines, multi-kw modules.
The weak link in the chain is anyway not the amp but the speakers.
If a 200 usd amp has failed, there is nobody trying to repair it, when there are many components burnt.
so why even consider easy repair as a design target? take 1 more spare unit for every 4 amps under load to the gig, ensure that a failed amp does not kill the entire show, that is it.
i know that usd 10.000 amps have some bells and whistles for the high-end touring companies, but the party dj takes what is affordable to rock the house. period. he is not interested in saving maybe 800VA reactive by having pfc within the amp, when the chicken heater or kebap guy nearby, same ac line, can consume 5000W up and down easily, by turning the temperature up and down depending on how many guys are waiting for his snacks. if the main breaker falls, all will blame the electrician company who will dare to come only with a 4 times overspecified ac line anyway to avoid any issues. period.
btw, i am curious what model did you design that does 6000 /2r?
The frequency response of a system influences the subconscious brain (the lymbic system) regardless of whatever the conscious brain (cortex) is trained to notice and tell, or not to tell, or to deny. The mood of people in loud music listening environments depends on frequency response down to less than 0.5dB increments, no matter if they can explain or they just deny it, their biochemistry changes.

The "economy" based in single-use technology equipment is also single-use, and it is already starting to fail, just ready to go to the dumpster. Such an economy scheme will never again be possible. If the wave front has not caught your zone yet it will soon do haha
Anyway, Eva, if the amp-speaker combined response has some small bumps and valleys because of higher output impedance in the 5-figure-Hz area (what many "older" guys are hard pressed to hear at all), the engineer can just make up for this by moving one or 2 eq sliders up 'n down. And don t tell me anyone can hear a few microseconds of group delay at those frequencies. I tell you that you can do a monster rock festival with a lowpass at 13 kHz for all signals involved and nobody will complain as long as the mud/beer/drugs balance is "just right"

" If the wave front has not caught your zone yet it will soon do haha"
I can only guess what that means, since I am not a native speaker, assuming that it means that the future of amps is with longevity and repairability, then I am 100% ok with this.
I am not within that amp design business currently, and for personal use I prefer cheap but powerful amps, since I am not a millionaire unfortunately.
The precision of human hearing is not absolute, it is ratiometric, that's why we can hear changes <0.5dB, with respect to another constant reference at other frequency, and these ratios affect mood. When the ratios get altered, the mood the music transmits also gets altered.

The society that seeks any ambient noise just for shaking and calls this "music event" is also single-use, it is not compatible with the process of formation of qualified professional workers. This has the potential to crash entire economies because less and less qualified professionals are formed on each successive generation. After several generations everyone finds "too-complex-to-consider" any job other than serving in a fast-food chain or some merchandising franchise.
eva, looks loke you are some kind of perfect-audio-philosopher. Ok with me, but never forget: The speaker is always the weak link distortion-wise, freq-response-wise, interference-wise, ....etc...then there is the room,...., so any significant improvement there will dwarf improvements on amping technology nowadays.
Not hard for an amp today to do 20hz -20khz at +- 2db - now show me that speaker system that is equally good at all directions, including the vettical interferences, the lobing, etc....
I had problems getting 2092 to drive IRFB4227.
Unless my 10R gate resistor was too big ?
I used both TC4420 and pnp/npn gate drivers in different amps and they worked fine.

The 2092 is a bit fiddly to get working. It is fussy about decoupling and pcb layout.
I even had to decouple the gate drivers to get it working.

Then there is power down issues with huge thumps and siren type noises through speakers.
I had to design a PIC based reset circuit to monitor VCC, if it fell below 12 volts I applied a reset to the 2092. IR recommend a reset circuit if power supply capacitance is high as the slow discharge on power down confuses the 2092.