Tube Phono Measurement Anomaly - Seeking Input

[L0rdGwyn]

Hi!

I am experiencing what I believe is a measurement issue which has me a bit perplexed, so figured I'd come to the measurements experts

I recently completed a tube / FET hybrid phono stage, schematic below. It is being fed from the balanced outputs of my turntable, balanced input to a Lundahl LL9226XL 1:10 MC SUT transformer, through a triode-strapped D3a gain stage, then an all-in-one RIAA EQ, then a triode-strapped EF86 gain stage, followed by a FET output buffer.



The issue I am having is taking measurements to determine adherence to the RIAA curve such that I can make necessary adjustments. The heart of my measurements setup is a MOTU M4 audio interface. I have approached the measurements in two different ways:

1) Using a inverse RIAA filter with built-in voltage dividers to provide either MM or MC level voltages via an on-board switch
2) Running an unfiltered signal into the phono pre across a voltage divider, then applying the inverse RIAA curve via soundcard calibration in REW

In both instances, I am seeing a significant LF rolloff below 500Hz with the inverse RIAA applied. This would result in a very tinny sound which subjectively is not there, the bass is very full, I suspect it is a measurement anomaly. Below is the DSP applied inverse RIAA result in REW



Just to be sure, I did the following to rule out in-circuit high-pass points without a change in the LF rolloff:
-Bypassed the input transformers and fed MM level signal to the input of the D3a
-Added parallel capacitance to the 0.1uF coupling cap
-Added parallel capacitance to the 1.5uF output cap

I am using the line-level balanced output of the MOTU M4. Loopback at line level is flat as can be. So it seems there is some other high-pass being applied inadvertently. At this point, I need a second set of eyes, apologies if the answer is obvious, this is my first phono design. I appreciate the help.

 

[DonH56]

Sure there's not a 100 Hz rolloff, like a HPF cutoff for a sub, in there someplace? Not in your schematic, but someplace in your measuring setup...

 

[LTig]

What's the input impedance of the Motu?

 

[L0rdGwyn]

Sure there's not a 100 Hz rolloff, like a HPF cutoff for a sub, in there someplace? Not in your schematic, but someplace in your measuring setup...

No, not that I can see, I rechecked everything in REW, no high pass filter applied. I also saw similar albeit less pronounced rolloff when taking measurements via a real-world inverse RIAA filter in Arta.

What's the input impedance of the Motu?

Looks to be 1Meg for single-ended, per the spec sheet.



Unfortunately, I do not have another balanced output source on hand to compare to.

 

[L0rdGwyn]

Here is another data point I acquired today, which I think proves to some degree that the measurement is flawed.

The smooth green curve is the pseudo-RIAA phono stage output with the software applied inverse RIAA. The signal input was taken from the MOTU M4 balanced line out and voltage dropped across a resistive divider. As you can see, it starts to roll off rather than continue to extend upward as a true RIAA would, hence my LF rolloff when the inverse RIAA is applied.

But the two noisy curves are signal sweeps taken from a CBS STR100 test LP, the lower frequency of which is from 20Hz-200Hz (3dB higher than reference, per the test record's description), the higher sweep from 40Hz-20kHz. Obviously these will take into account cartridge alignment, tonearm resonance, etc. which still needs some work, but the point is the LF response continues to extend upward as it should. I may need to consider attempting the measurement on a different ADC, not sure what else to try at this point. Feeding the phono input from a Topping D30 DAC rather than the MOTU M4 DAC yields the same result.

 

[LTig]

Shall I supply you with a flac file which contains an anti RIAA sweep? Just tell me the sweep time, number of channels, sampling frequency and bits. Then you can skip the anti RIAA stage.

 

[L0rdGwyn]

@LTig thank you for your offer, however I am able to bypass the inverse RIAA filter by applying the curve in REW via soundcard calibration, so no need for the sweep! I do appreciate it though.

Brainstorming some potential causes, I am wondering if the DAC being loaded down by the phono stage at low frequencies. The phono is set up for my MC cart with a 470ohm reflected impedance across the input transformer. The listed output impedance of the MOTU M4 is 100ohm. If so, perhaps I need to increase the phono input impedance for the sake of the measurement, or else place a buffer circuit in between the DAC and phono input.

 

[LTig]

@LTig thank you for your offer, however I am able to bypass the inverse RIAA filter by applying the curve in REW via soundcard calibration, so no need for the sweep! I do appreciate it though.

That's a good idea. Mind to share the calibration file?

Brainstorming some potential causes, I am wondering if the DAC being loaded down by the phono stage at low frequencies. The phono is set up for my MC cart with a 470ohm reflected impedance across the input transformer. The listed output impedance of the MOTU M4 is 100ohm. If so, perhaps I need to increase the phono input impedance for the sake of the measurement, or else place a buffer circuit in between the DAC and phono input.

Better lift one leg of the 470 Ohm resistor. Or skip the input transformator at all.

 

[L0rdGwyn]

@DonH56 @LTig @SIY I think I was correct, the phono with the 470ohm input impedance was loading down the DAC below 500Hz. I instead used a single-ended signal across a voltage divider and fed it directly to the D3a input across the 47K load resistor, bypassing the input transformer. I then applied the inverse RIAA calibration in REW, and got the below.



A lot more garbage above 10kHz with the single-ended input (and alligator clips), but the point is this is a believable frequency response. Down ~1.5dB at 20Hz, will need to investigate if this is a real LF rolloff or if the measurement can be further improved, regardless we are in the ballpark! Thanks for helping me get the gears turning. Seems measuring a balanced input MC phono with the ever-popular soundcard measurement rig isn't so trivial after all

@LTig happy to provide the calibration file, you simply use a tab-delimited RIAA .txt file and add it to the "Soundcard Calibration" in REW, attached. Can be applied universally via the preferences or to individual measurements.