• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Review and Measurements of IRS2092 IRAUD350 Class D Amplifier

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
17,361
Likes
14,739
Location
Seattle Area
#1
This is a review and detailed measurement of a very common power amplifier platform based on International Rectifier/Infineon's IRS2092 class D amplifier IC. The unit was kindly sent by a member to me for testing. Here is what it looks like:

IRAUD350 Class D Amplifier IRS2092 Review.png.jpg

The products seem to be based a reference design by International Rectifier although I can't find an exact match there. You can get everything from single amplifier board fully populated for USD $40 to the fully finished version I have including volume control for USD $298 from China. Just google for IRS2092 and IRAUD350 and you will get countless hits. Some come with volume control and others not.

Power specifications are all over the map. I think I saw one advertised to produce some 700 watts! Manufacturer specs are vague saying since the IC supports +- 100 volt, the can be used to create very high power amplifiers.

Due to use of linear power supply, the loaned amplifier is quite long and very heavy in front where the toroidal transformer is housed:

IRAUD350 Class D Amplifier IRS2092 Teardown.jpg


The unit is quite appealing to the audiophile. We have some fancy parts such as ELNA premium capacitors. Anyone want to guess as to whether they are real or not? Can't imagine they are at this price but the markings look good on them. At a glance, the construction seems nice and tidy. Closer inspection indicates otherwise.

For starters, when I opened the unit prior to start up I found one of the larger power supply capacitors was dangling by a thread! It had good bit of solder on its pins but likely due to oxidation on the capacitor pin, the solder was not holding. Since they did not glue down the caps as they show, the vibration in mail caused it to break free from its solder joint. I soldered the pin and did the same with a few other ones.

On safety front, there is of course no regulatory certification. I was pleased to see a safety ground wire to the chassis. They used a black color instead of proper green. Its gauge is too small likely to cause your home breaker to pop meaning if there is a short form hot to chassis, this wire may burn up leaving the case fully energized. Hopefully the smell of it will keep you away from touching it, maybe not. Then again I checked and they forgot to sand off the anodization from the chassis so this wire is not doing anything! The top lid likewise is insulated from the chassis due to the same finish acting as a barrier between the two. THE UNIT IS UNSAFE TO OPERATE AS BUILT!!!

IRAUD350 Class D Amplifier IRS2092 Teardown AC Mains Ground.jpg


There are other issues. See how a part of the insulation on the black wire is burned off. That doesn't do any damage here but shows sign of sloppy soldering. You can see it much better here:
IRAUD350 Class D Amplifier IRS2092 Teardown shoddy soldering speaker terminal.jpg


I should note that the red wire is connected to the ground terminal on the PCB instead of black. Fortunately they seem to have also wired the inputs reversed so phase is maintained.

Very suspicious are these transistors which seem to have been hand soldered and in really poor way:
IRAUD350 Class D Amplifier IRS2092 Repaired Transistor.jpg


I think they are acting as regulators and are the hottest components as you see later. Likely they were damaged and they replaced and sold them as new.

Note that in some other variations of these designs the transistor is through-hole and mounted on the same heatsink as the output mosfet transistors.

It is one thing to buy a DAC without regulator certification but when it comes to power amplifications, you really have to be careful. Very high currents and voltages exist in these products and I highly suggest buying from reputable companies.

Anyway, let's get into measurements and see if better news awaits us there.

Measurements
Based on feedback from my last amplifier review, I decided to warm up the unit for an hour before taking measurements. Here is how it progressed during that time with respect to distortion:
IRAUD350 Class D Amplifier IRS2092 Warm Up measurement.png


One channel (in blue) keeps getting better with time but the other first shoots up in distortion and then back down. I tried to put the time to good use by opening the lid and taking some infrared shots but the moment I did, I got the spikes in distortion. I understand the lid encapsulating some of the noise but can't figure out why one channel got worse with the lid open but then then other got worse! Anyway, while the difference is not huge, allowing the unit to warm up improved its distortion metric by about 2 to 3 dB. Given the safety concerns I have, I am not sure you want to leave the unit on all the time but there is the data.

I then ran the dashboard at 5 watt of power and got this:

IRAUD350 Class D Amplifier IRS2092 measurement.png


Yuck. Double yuck! Even at such low level of power we have pretty dirty output with 0.02% distortion. It is within the measurements of reference design though: https://www.infineon.com/dgdl/iraudamp7d.pdf?fileId=5546d462533600a40153569ad5ba2bf9

1546647077641.png


Power versus distortion give us this:

IRAUD350 Class D Amplifier IRS2092 4 ohm power measurement.png


We have dual non-linearities. One sets in around 31 watts and the other, at 320 watts.

Reference design (at lower wattage) doesn't show anything as extreme as what I am measuring so something is not right in these instantiations:

1546647959739.png


Thermal stability was good though with the unit happily producing some 150 watts for 5 minutes:
IRAUD350 Class D Amplifier IRS2092 5 min thermal stress measurement.png


Infrared camera shows the regulator transistors (?) to be the hottest component as I noted earlier:
IRAUD350 Class D Amplifier IRS2092 Thermal IR measurement.jpg


We see that the power supply is running pretty cool which is nice. The capacitors are also way away from the heat sink which is nice as far as longevity.

The chassis is at room temperature which is sad because its mass could have been used to keep the amps much cooler. As it is, it is just a sturdy case to hold everything.

Back to measurements, I was not too happy to see such an early roll off in frequency response:
IRAUD350 Class D Amplifier IRS2092 frequency response measurement.png


This likely indicates that there will be filter interactions with different speaker loads, causing their response to change within the audible band. Indeed even changing to 8 ohm load varies the response per reference design measurements:

1546647772322.png


Note the channel mismatch which was also visible in the dashboard. Fortunately this is not as big of a problem with speakers as it is with headphones but still, if you have a DSP, you may want to compensate for this there and use a volume control elsewhere in the chain.

Dynamic range is OK:
IRAUD350 Class D Amplifier IRS2092 SNR measurement.png


International Rectifier reference design runs at around 400 kHz (+- 25 kHz) and that is what we see:

IRAUD350 Class D Amplifier IRS2092 wideband FFT measurement.png


I left the red channel without the AES-17 filter to show much noise this is producing and it is a lot. You will likely get interference on RF devices nearby.

Conclusions
Best way I can describe this amplifier is that it produces a lot of rather dirty power. If you want capacity at a bargain price, and you are not worried about the safety of the unit, this may fit the bill. But otherwise, I have to steer you all away from them. These are NOT high fidelity solutions when the manufacturer's own spec says THD+N is 0.01%. We have DACs that are 100 times better than this. Stop buying $4 cups of coffee, save your money, and buy something more decent.

-------------
As always, questions, comments, recommendations, etc. are welcome.

I am craving some prime rib. Please consider donating some money so that I can afford to get some using:
Patreon: https://www.patreon.com/audiosciencereview), or
upgrading your membership here though Paypal (https://audiosciencereview.com/foru...eview-and-measurements.2164/page-3#post-59054).
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
17,361
Likes
14,739
Location
Seattle Area
#7

Graph Feppar

Active Member
Joined
Nov 19, 2018
Messages
192
Likes
144
#10
Why does amplifier with early high frequency roll off change its high frequency response depending on what load it is connected to?
 

DonH56

Major Contributor
Technical Expert
Patreon Donor
Joined
Mar 15, 2016
Messages
2,464
Likes
2,312
Location
Monument, CO
#12
Why does amplifier with early high frequency roll off change its high frequency response depending on what load it is connected to?
I can guess a couple of reasons:
  1. The output filter is rolling off the response and thus output impedance is rising. That means the load influences the output more.
  2. Possibly related to that, feedback from the output is being reduced, so the control loop has less ability to control the output and force it to match the input. Again causes output impedance to rise.
Note all amplifiers exhibit rising output impedance with frequency as feedback drops due to limited (loop) bandwidth. Not a new thing, but class D amplifiers need an ultrasonic filter perhaps more so than most to suppress switching spikes, and thus often have greater rise than more conventional A/AB designs. It is not bad these days due (IME/IMO) to advances in circuits and devices -- feedforward and multi-loop feedback circuits help close the loop better, and faster devices allow higher switching rates to improve output bandwidth and provide more loop gain at higher frequencies.

FWIWFM, HTH, IME, IMO, YMMV, etc. - Don
 
Last edited:

andreasmaaan

Major Contributor
Patreon Donor
Joined
Jun 19, 2018
Messages
2,612
Likes
1,701
#13
Why does amplifier with early high frequency roll off change its high frequency response depending on what load it is connected to?
It happens because the filter used is passive (made up of inductors, resistors and/or capacitors). The effect of such a passive filter always depends on the load presented to it, in this case by the speaker.

Another way to look at this is that the filter is a network which includes the speaker. All the values (i.e. the values of the components of the filter as well as the loudspeaker crossover components and transducers) interact to produce a low pass filter at a specific frequency and with specific properties.

When designing an amplifier filter, the designer must make an educated guess as to what type of load the speaker will present to the filter. When the real-world speaker doesn't present exactly this load to the filter (i.e. in practice in every case), it will not function exactly as intended by the designer.

Usually therefore the designer will optimise the filter for an 8 Ohm (ish) load, which is about the average load presented by an average speaker in the frequency range of interest.
 

DonH56

Major Contributor
Technical Expert
Patreon Donor
Joined
Mar 15, 2016
Messages
2,464
Likes
2,312
Location
Monument, CO
#14
Note many power amplifiers (of any class) include an output filter... I have not heard of optimizing the filter for a load in a SS design (excepting McIntosh's autoformer); most SS amps target as close to 0-ohm output (ideal voltage source) as they can achieve. IME, so may be wrong. - Don
 

andreasmaaan

Major Contributor
Patreon Donor
Joined
Jun 19, 2018
Messages
2,612
Likes
1,701
#15
Note many power amplifiers (of any class) include an output filter... I have not heard of optimizing the filter for a load in a SS design (excepting McIntosh's autoformer); most SS amps target as close to 0-ohm output (ideal voltage source) as they can achieve. IME, so may be wrong. - Don
I may also be wrong here. I'd certainly trust your more extensive experience over mine @DonH56 :)

I was under the impression that most of these cheap-ish class D implementations included an LC filter on the output.
 

DonH56

Major Contributor
Technical Expert
Patreon Donor
Joined
Mar 15, 2016
Messages
2,464
Likes
2,312
Location
Monument, CO
#16
I think you are (may be) wrong about optimizing for the load, though I imagine some of that happens intentionally or not during listening tests for the design.

All class D amps I know about include an output LC filter primarily to filter the ultrasonic noise (switching hash) but also for load isolation.
Many A/AB amps include an LC filter for load isolation (enhances stability).

Rising output impedance at HF is a consequence of limited feedback loop bandwidth for any amplifier potentially made worse by the extra filtering most (if not all) class D amplifiers require. Look at output impedance plots and they all go up, but many class D amps go up sooner/faster than their A/AB kin.

One other technique used by class D amps is interleaving; parallelling two or more (I have seen as many as eight) output channels interleaved in time so the effective switching rate is much higher. I am sure there are many other tricks behind the scenes; I am not an audio amp expert.
 
Joined
Apr 14, 2018
Messages
31
Likes
27
Location
Washington, DC Area
#17
When i opened this review, first thing that caught my eye was the pink panther. My brain said it does not look right.
Now i know why Amirm framed the picture that way :facepalm::facepalm::facepalm:

View attachment 19927
@sbsj, I think you have the honor of having identified a new testing metric on ASR: the "Pink-Panther-O-Meter". I think it's safe to say that if the PPM (the widely accepted acronym for the Pink-Panther-O-Meter) is showing 180 degrees off its center point (said feline's nose), then we have a stinker of a test subject. My sincere congratulations! :)
 

Blumlein 88

Major Contributor
Joined
Feb 23, 2016
Messages
5,126
Likes
3,785
#18
Many of the early Icepower amps optimize the output filter for flat response at 6 ohm loads. Other loads either peaked or drooped the near ultrasonic region. Not sure about newer versions which I think operate at higher switching frequencies. Tripath also optimize for 6 or 8 ohms in many implementations.
 

andreasmaaan

Major Contributor
Patreon Donor
Joined
Jun 19, 2018
Messages
2,612
Likes
1,701
#20
All class D amps I know about include an output LC filter primarily to filter the ultrasonic noise (switching hash) but also for load isolation.
Many A/AB amps include an LC filter for load isolation (enhances stability).
Yep, this is what I was talking about.

I'd misread your previous post as "not" rather than "note".
 
Top Bottom