Inconsistent and unexplainable(?) results when using same PEQ filters in two devices

[damaged_goods]

I have a setup where I have my Yamaha AVR running as a preamp into a HT bypass input on my integrated amp, which then acts as power amp for the front channels when watching movies. For two channel music listening I have a WiiM Ultra streamer connected to one of the line inputs on the integrated amp. So the same L/R speakers and power amp section are used in both cases, but with different pre-amps.

The AVR, which is now some 8-9 years old now, has a somewhat limited PEQ functionality with 6 bands for L/R with only a few predefined frequencies to choose from. The lowest possible frequency to set is 62.5Hz.

I have a huge room mode peak at around 45Hz that I can correct with PEQ on the subwoofer (or on the front speakers when using WiiM's PEQ). For L/R I have manually tuned the EQ and bumped the mid bass, which is attenuated due to limitations of the room (geometry/asymmetry size, etc).

The weird thing is that after applying PEQ on the AVR, the 45Hz peak is diminished, even with no subwoofer/active bass management and no corrections done below 62.5Hz. I first thought that perhaps the bumped up mid bass would effect low bass too, due to interaction with the room, but when applying the exact same PEQ filters in my WiiM Ultra, I get a completely different result, where the 45Hz peak is still there, and the effect on the mid bass also looks very different. Increasing the mid bass on the WiiM also results in quite audible distorsion, even after compensating by lowering pre-gain to avoid any clipping, which is not the case with the AVR.

Does anyone have any idea of what Yamaha might do in their filtering that has this kind of effect? I find it very peculiar...

Here's the in-room frequency response with no EQ applied for both the AVR and the WiiM Ultra. As you can see the graphs are matching. No big differences.

Here's the graph for the EQ'd response on the AVR, where the 45Hz peak is mysteriously gone:

And here we see the same graph as above, but also the response from my WiiM Ultra with the exact same PEQ filters and a completely different result. (Again, no subwoofer enabled in any of the measurements. Just L+R stereo.)

And here's the PEQ filters applied (on both L and R channel):

 

[Sokel]

Can you post the comparison of WiiM vs no EQ?
They seem somehow similar apart from the skirt of 45Hz.

Freq shift at WiiM 's EQ maybe? Or no EQ applied at all?

Edit: I would measure the outcome electrically if I was you.
Edit2: is that positive gains you apply there???

 

[Keith_W]

Are you applying PEQ's by uploading biquads, or are you manually typing in the PEQ values of f0, Q, and gain? Because, biquads are not transportable. They are generated for specific sample rates and different devices implement biquads differently. Whereas PEQ's should be the same. To diagnose the problem, show us the PEQ's on the AVR, and the PEQ's on the Wiim.

 

[damaged_goods]

Can you post the comparison of WiiM vs no EQ?
They seem somehow similar apart from the skirt of 45Hz.

Freq shift at WiiM 's EQ maybe? Or no EQ applied at all?

Edit: I would measure the outcome electrically if I was you.
Edit2: is that positive gains you apply there???

The first image is without EQ for both WiiM and the AVR. If that’s the comparison you’re asking for? Or do you mean with and without EQ in WiiM on the same graph?

I don’t really have equipment for proper electrical measurements, only a multimeter.

The gains in EQ are positive indeed. I have adjusted down pre-gain in the WiiM Ultra accordingly. For the AVR I have not compensated for the positive gain, but even with automatic room correction there’s positive gain up to 6dB applied, so I must assume that there’s enough headroom or that pre gain is automatically decreased.

Also, I can get very good results in WiiM too, by attenuating the 45Hz peak (by a lot, like 10dB) and applying just a bit positive gain to the mid bass correction.

I think the results with the WiiM EQ makes perfect sense, it’s the AVR results I can’t understand… why would applying positive gain to the mid bass completely eliminate the 45Hz peak?? In the WiiM I have to add a filter with -10 dB to eliminate that same peak, independent of how I EQ the rest of the frequency spectrum.

 

[damaged_goods]

Are you applying PEQ's by uploading biquads, or are you manually typing in the PEQ values of f0, Q, and gain? Because, biquads are not transportable. They are generated for specific sample rates and different devices implement biquads differently. Whereas PEQ's should be the same. To diagnose the problem, show us the PEQ's on the AVR, and the PEQ's on the Wiim.

I’m manually typing in the PEQ values, although in the case of the AVR, frequency and Q values are only selectable from a fixed set of values to choose from.

Here’s what it looks like in WiiM:

And here’s the AVR interface. There’s no way of showing a summary of all filter values, so the screenshot shows only filter #4. But the displayed graph visualize the result of all four applied filters, and it looks similar to that in WiiM, as you see.

 

[Sokel]

Something strange is at place.
Any chance of something automatic is at play on your AVR? Because the response looks much nicer there

There's also a great difference in levels. That 12Hz bump should be aligned at the comparison between AVR and WiiM.

 

[damaged_goods]

Something strange is at place.
Any chance of something automatic is at play on your AVR? Because the response looks much nicer there

There's also a great difference in levels. That 12Hz bump should be aligned at the comparison between AVR and WiiM.

This is my thinking too. But all DSP filters except PEQ are disabled. Tone controls are bypassed, YPAO volume and Enhancer is off. And if I set PEQ to “through”, which means bypass, I’m getting the result of the first picture in my original post, which aligns perfectly with the result from my WiiM Ultra when no EQ is applied there. So it’s definitely some magic in the AVR PEQ that cause the strange (but good) result.

I can achieve a similarly good result with WiiM, but then the parameter values are completely different. I need to attenuate the 45Hz peak with a very narrow -10dB filter, and I don’t need to add that much positive gain to the mid bass either.

Here’s the PEQ filters I use with WiiM, I’ve added a little more bass overall though, so it’s a less flat response than with the AVR, but that’s on purpose.

 

[damaged_goods]

@Sokel Mystery solved, sort of.

When I configured my custom PEQ filters in the AVR I copied one of the auto room correction profiles I had from before and changed the values there. But it seems those custom changes are applied on top of the existing filters. If I create a new custom profile from scratch and set the same values then the frequency response is the same as with WiiM.

It’s quite weird that the auto calibration can correct front channel frequencies that do not exist in the manual calibration, but this is a mid-tier AVR model that is getting close to ten years old, and I guess room correction has come a long way since then. It’s at least nice that I can partly adjust the auto correction manually to achieve a better result.

 

[Sokel]

@Sokel Mystery solved, sort of.

When I configured my custom PEQ filters in the AVR I copied one of the auto room correction profiles I had from before and changed the values there. But it seems those custom changes are applied on top of the existing filters. If I create a new custom profile from scratch and set the same values then the frequency response is the same as with WiiM.

It’s quite weird that the auto calibration can correct front channel frequencies that do not exist in the manual calibration, but this is a mid-tier AVR model that is getting close to ten years old, and I guess room correction has come a long way since then. It’s at least nice that I can partly adjust the auto correction manually to achieve a better result.

Glad it did, it was kind of obvious.
Take good care measuring the outcome with REW, use long sweeps (watch levels though! )

Combining filters has its own effect, so compare distortion too after correcting.

 

[markstein]

There are different implementations of filter "Q", see e.g. the link below for further details. So, the same PEQ filter settings may not lead to the same result on different devices.

Parametric Equalizer "Q" Definitions and Bandwidth

For audio PEQ filters, the difference between various definitions of the 'Q' parameter and their relationship to the half-gain bandwidth is discussed.

www.andyc.diy-audio-engineering.org