Things can be easy, however...
The Sound of Audio Amplifiers: Can you hear a difference between Amps? | Audioholics
The Sound of Audio Amplifiers: Can you hear a difference between Amps? | Audioholics
If distortion becomes excessive, it can become fatiguing too. Not only harmonic distortion, but especially intermodulation distortion, where the bass (high amplitude part of the spectrum) and the non-linearity of the tube cause the midrange and the treble to become amplitude modulated. While some find that a bit of tube distortion sounds good, too much of it will definitely make your ears protest after a while of listening.
Problem is, while your preamp and the IRS may measure well individually, it is the low input impedance of the IRS which causes the preamp to produce much more distortion at a given output level than it would with a normal load impedance. Did you measure the individual components, or the combo of the tube pre and the IRS?
Well... I have not supersonic ears, by definition, noone can not ear ultra sonics.
and I can not measure that with my UMIK 1, and not sure my speakers can produce ultra sounds.
I compared REW graphs between TPA 3116 and IRS 2092, and I didn't see something strange... But I have to say I am not as good as you on audio.
=> I don't know how to explain what is "audio" fatigue
what I suspect, and I may be wrong, there is a paradox to reproduce songs with very high fidelity without airflow, or something we can define globally as atmosphere and maybe it can cause fatigue : It is like optical illusion for audio or like watching for a ongtime cubism paints....
This is why having few distorsion on the signal path may be benefits to help the brain to remember the sound comes from speakers...
This is only my opinion and maybe a wrong one (and surely).
Did you check for mislabeled polarity on speaker outs? I just tested my TDA7498's and they are incorrectly labeled from factory. If I listened, they sound phasey and wierd - very fatiguing. A false sense of spatiousness. I swapped the +/- on right channel and BAM - the stereo phantom image locked and phasiness went away, and fatigue went away. I double checked to make sure driver cones move out on kick drum and measured freq response in stereo to make sure bass was not cancelling. Check speaker polarity.
I assume I've done this checks. I am not the only one with listening fatigue with IRS2092.
Maybe I have to prototype again my IRS 2092 and make some tests
Strange , A very happy IRS2092 listener here , 5 x L15D LJM in my HT setup with Tannoy mains, centre and surrounds, it's my favorite amp for more than two years now,listen to it for a couple of hours a day , just watching ( Digital) TV and Netflix and the odd CD's and LP's and even the old cassettes .
Never ever had the fatigue feeling . stopped switching over to my other amps once ore twice a week , because it just sounds right for me and my family .
Must admit that in stereo mode , listening to vinyl I might have some second thoughts ,but switching all the cables does not convince me enough to go to all the trouble anymore .
Cheers ,
Rens
If I want to get a sense of the circuit schematic for this amp, is it close to an IRAUID 7 or whatever? Can someone please post schematic they think this uses or is it following IR's factory recommended schematic from data sheet?
Ok - thanks. This is a pretty amazing chip and all the peripherals for $20. The distortion figure is very low for a 50v supply.
Have you read this?
Trevor Marshall - Class D Audio Amplifier Design - TDA7498 Output filters
Although it is not about the IRS2092, it is a stark example of how small details (in this case component selection) can make or break an implementation of a circuit. Although the IRAUDAMP7 as built in the labs of IR seems to perform well, there is no guarantee that the performance of the clone boards will be identical. Did anyone thoroughly verify the performance of a cheap board?
For example, the Sure electronics board operates at 200 kHz, instead of the 400 kHz of the original IR implementation. This may be an indication that the output inductor is not as good as the one originally specified by IR and can't handle the high switching frequency.
Trevor Marshall - Class D Audio Amplifier Design - TDA7498 Output filters
Although it is not about the IRS2092, it is a stark example of how small details (in this case component selection) can make or break an implementation of a circuit. Although the IRAUDAMP7 as built in the labs of IR seems to perform well, there is no guarantee that the performance of the clone boards will be identical. Did anyone thoroughly verify the performance of a cheap board?
For example, the Sure electronics board operates at 200 kHz, instead of the 400 kHz of the original IR implementation. This may be an indication that the output inductor is not as good as the one originally specified by IR and can't handle the high switching frequency.
Thanks for the heads up Timpert. I will link this in my 7492 thread (which also discusses 7498) - http://www.diyaudio.com/forums/class-d/275505-tda7492-amp-12.html#post4373885. I checked my cheap Chinese ($11) 7498 and it uses metalized film caps for the output filter. It even has a big 5watt 2.2ohm resistor on the output snubber. This amp runs very cool to the touch - unlike my cheap 7492's which are noticeably warm and even hot on one of the inductors. So far though I am very happy with the performance of the black 7498. It had tremendous reserves compared to the 3116 as it is a 100w design. I will need to get a real sound card so that I can test distortion on dummy loads with my amps. Now I am curious how they are doing. Although when I test them with speakers and microphones, the distortion at HF's is exceedingly low. So I am lucky not to hit the 4% HD problem due to x7r resistors as Trevor Marahall pointed out.
Last edited:
Just a correction - X7R is a dielectric material for ceramic capacitors, it is not a resistor. Because the material is piezoelectric, the capacitor mechanically deforms when voltage changes, and thus the capacitance changes. See for example here:
http://product.tdk.com/capacitor/mlcc/en/faq/pdf/31_singing_capacitors_piezoelectric_effect.pdf
and here:
http://www.avx.com/docs/catalogs/cx7r.pdf
http://product.tdk.com/capacitor/mlcc/en/faq/pdf/31_singing_capacitors_piezoelectric_effect.pdf
and here:
http://www.avx.com/docs/catalogs/cx7r.pdf
Yes, that is typo on my part, I meant capacitor - the caps in my 7498 board are the little rectangular metal film types. I see why many of the amps I have use metal films on the output LC filters now. There are a few designs by TI where ceramics are used on the output filters as recommended by the factory. Take the TPA3116D2 EVM for example, the output filter cap is 0.68uF 50V X7R SMT part. See Fig. 2, page 5 for example for C21-C24.
http://www.ti.com/lit/ug/slou336b/slou336b.pdf
But on YJ blueblack, YBDZ, or the SMAKN amps that I have, these have been replaced by big metal film caps. Not sure I can hear or measure any difference though for HD in this case.
http://www.ti.com/lit/ug/slou336b/slou336b.pdf
But on YJ blueblack, YBDZ, or the SMAKN amps that I have, these have been replaced by big metal film caps. Not sure I can hear or measure any difference though for HD in this case.
TI doesn't recommend ceramics in the outputfilter, TI's EVM outputfilter uses ceramics. TI recommends using filmcapacitors if designer want to bring down THD.
Come to think of it, only the EVM and my DUG v0 amp uses ceramic SMT's for the output filters, oh, and my first deadbug flipchip TPA3118 - because I followed the EVM schematic. However, that amp had probably one of the first bootstrap snubbers in place aside from the EVM.
I just noticed something on which I am not certain where to post, so I'll just throw it in here. If you look at the datasheets of the TPA3116, TDA7498 and IRS2092, you'll notice them all having a similar oscillator topology. The TPA also has its feedback resistors connected to the input summing nodes, just like the IRS. Only the TDA is equipped with an input buffer, giving it an input impedance of around 60k Ohms, regardless of the gain setting. Not only makes this the easiest chip to drive out of the three (the hardest being the IRS in the IRAUDAMP), it also makes sure that the amplifier doesn't inject a (small) AC current at the switching frequency into its own input. This would, at least on paper, make the TDA7498 the least critical one of the three when it comes to drive requirements.
Supportforum TI.com, optimising tpa3116 for THD, you have read it befor:
EVM is a good starting point. Select inductors that does not saturate at peak current. Any loss of inductance of more than 3% up to the peak load current can increase THD. Use at least X7R or better capacitors (MKP/MKT).
You may also use AM0, AM1, and AM2 to select higher PWM freq which can provide slightly better THD+N at the expense of higher idle current.
reg,
Paul Chen
Applications Engineer
You may also use AM0, AM1, and AM2 to select higher PWM freq which can provide slightly better THD+N at the expense of higher idle current.
reg,
Paul Chen
Applications Engineer
I have exactly the same experience! I had earlier a Sure 2092 board with linear Power Supply. It had more sharp highs, more punchy bass overall more precise sound than tpa3116, but after 30 minutes of listening I noticed something "strange" (clinical?) in its character and has become fatigue to listen.
Tpa3116 has more tubish (smooter) sonic character and extremely good tonal balance. Very easy to listen to several hours. It has more laid back sound and a bit smaller soundstage compared to irs2092 (and Hypex UCD 400) but sounds very natural.
egra
I will have to do some long term listening on mine to see this fatigue you all are talking about. If it sounds like it has more highs, those are probably source of fatigue. It may be in the sharpness of the transients which produces HF ringing?