What to make of these measurements: Tortuga LDR3.V25

[BDWoody]

Hi All,

Stumbled on to this thread and while a bit dated thought I'd share some thoughts as the designer/manufacture of the Tortuga LDR preamps.

I've read the neurochrome.com report with great interest as you can imagine. An impressive piece of work and I mean that sincerely. A bit amazed that the author took all that time and effort but nevertheless top notch work.

I could quibble around the edges about the report but I really don't dispute the overall arc of the results. I mean it's measurement after all. Either you measured it right or not. One clarification worth noting is that the report was done on our older V2 model board which was replaced with our V25 design which has significant changes over the V2. The V2 had some unfortunately noisy op amps in the design which are gone in the V25.

But my main point regarding the report isn't that it's factually incorrect or technically flawed. It's that it's lacking in perspective. Most discussions here and elsewhere starts with an implicit premise that ANY level of measured distortion is discernible to the human ear and that therefore a few points less distortion means something sounds better than a few points more distortion. So we chase specs and measurements and are impressed when the specs are better and the noise measurements are lower.

But does lower distortion really matter? By that I mean can you actually honestly really tell the difference when you're listening to music? Human ears/brains are not distortion measuring devices. Below a certain threshold level we can't hear the distortion. Yes, it's there! You can measure it. But so what? I've dug into this topic and while you can ALWAYS find a few exceptions the vast majority of scientific studies on the ability of human beings to discern distortion conclude that anything ~1% or less can not be detected by the human ear. Can you find someone somewhere who claims otherwise? Of course you can. But the consensus is that it doesn't matter below a certain threshold and that threshold is WAY above most measurements.

So yes, you can argue that LDRs are inferior attenuation devices because of their higher distortion characteristics compared to the other technologies. Yet, when you build an LDR based attenuator and actually listen to it your human experience of that moment tells you you like what you're hearing. And then along comes a smart engineer with sophisticated measuring equipment and tells you you shouldn't like it and that you're being fooled.

If measurement was the definitive metric of how good something sounded we'd test every audio component, rank them by test results, publish the results and then buy the equipment with the lowest distortion we can afford. And we'd put every bottle of wine through a similar process and only buy the one with the best specs. And we'd put every car on a dynomometer and buy the one with the best performance. But we don't do this. Not because measurements aren't useful (especially to engineers in the design process) but because beyond a certain point, the measurements don't matter and are not definitive proxies for the human experience.

Nelson Pass said it best....“The ear is not a microphone, the brain is not a tape recorder, and measurements are limited in describing subjective quality. I like to have low distortion and so on, but these things take a back seat to what I experience when I listen. There are plenty of products which have great specs – I will not be offended if you buy those.”

Peace,
Morten

Uh huh

 

[tomchr]

Morten: Thanks for taking the time to write.

But does lower distortion really matter? By that I mean can you actually honestly really tell the difference when you're listening to music? Human ears/brains are not distortion measuring devices. Below a certain threshold level we can't hear the distortion. Yes, it's there! You can measure it. But so what?

True. Had the LDR volume control measured, say, -100 dB THD, consistent bandwidth vs volume setting, etc., my conclusion would have been: "This may not be state of the art (i.e. better THD can be had), but it still provides good performance without colouration". But >10% THD at line level??! C'mon!

I've dug into this topic and while you can ALWAYS find a few exceptions the vast majority of scientific studies on the ability of human beings to discern distortion conclude that anything ~1% or less can not be detected by the human ear. Can you find someone somewhere who claims otherwise? Of course you can. But the consensus is that it doesn't matter below a certain threshold and that threshold is WAY above most measurements.

Yes. Scientific studies tell you people's preferences and behaviours on average. They do not tell you how you will behave (or how to behave). You may not share the preferences of the participants in the study. Individual variability is a thing.
I do agree that research shows distortion below a certain threshold is unlikely to be audible. Depending on which study you read, that limit is around -80 dB for THD. Your volume control exceeds -20 dB, i.e. provides distortion that's more than 1000x higher than the audible threshold, so your argument that the distortion of the LDR volume control is inaudible, thus, irrelevant falls apart pretty quickly.

So yes, you can argue that LDRs are inferior attenuation devices because of their higher distortion characteristics compared to the other technologies. Yet, when you build an LDR based attenuator and actually listen to it your human experience of that moment tells you you like what you're hearing.

So do that! Set up an experiment in a controlled environment. Let a significant number of participants listen and indicate their preferences. Run the stats. If a statistically significant number of participants prefer the LDR volume control over, say, a stepped attenuator, great! You now have a selling point. If, on the other hand, a statistically significant number of participants prefer the attenuator, then you should admit to yourself that the LDR doesn't sound as good as you thought. And if there is no clear preference between the two, you can possibly conclude that the LDR doesn't add anything audible (or that your sample size was inadequate; or that the effect size is small; or any number of other things).

Olive & Toole performed such experiments at Harman Kardon. You can find the results in various AES papers and also in Toole's book on sound reproduction. They consistently found that equipment that measured better was associated with a more positive listening experience for the participants.

If the subjectivists would apply some scientific rigour in their work and provide valid data that showed a statistically significant preference for their product vs a product that measures well, I would take them a lot more seriously. But all I get is, "many people like it" (when the salesman is hovering over them or after they've blown their money), "measurements aren't everything", and "Nelson Pass says..." Sorry. I can't take that seriously.

And then along comes a smart engineer with sophisticated measuring equipment and tells you you shouldn't like it and that you're being fooled.

I have never said that you "shouldn't like it" or "that you're being fooled". That's a flat-out lie. Throwing me under the bus like that does not lend credibility to your arguments. Thanks for recognizing my smarts, though.

Nelson Pass said it best....

Appeal to authority doesn't work for me either.

Tom

 

[Stretchneck]

Also though I would add to this (slightly overheated) thread.

I fine these results quite encouraging:

https://www.thevinylsource.co.uk/ldr-techical-test

Obviously measurements are not everything when it comes to sound quality (for example the Topping Pre90 I had, whilst sounding good, is not as nice to listen to as the Audio-gd Master 9 I have) it does help to know that LDR’s can measure very well.

 

[solderdude]

Any potmeter or DACT are factors better than this. Why bother with something that distorts the signal ?

 

[Stretchneck]

Apparently not - I have read and spoken with a number of people on different forums now. In particular there are a lot of folks on the artofsound forum that have done back to back listening comparisons with LDR and other high end passive (goldpoint, khozmo with z-foils etc) and whilst these where all very good the general thrust was that LDR and slagleformers that tended to be favoured in listening tests.

Now I like to buy kit that measures and reportedly sounds good. The above link does demonstrate that this particular LDR measures well, so I don’t see an issue in this case.

 

[solderdude]

the general thrust was that LDR and slagleformers that tended to be favoured in listening tests.

All sighted tests I'm sure.
-80dB distortion (LCD) versus -150dB distortion (DECT, relay and potmeter, digital attenuation) is quite a difference.

Have one send to Amir, bet you the seller won't just like the coffee one never was sent. Of course both know it will measure poorly and thus not get a thumbs up.

It measures relatively well for an LDR volume control. It measures like crap as a volume control.

Of course if one likes it one likes it.

 

[Stretchneck]

We’ll see when I get it in for demo

 

[solderdude]

I'm pretty sure you'll like and even prefer it.

 

[tomchr]

When you get one for demo, ship it to Amir for testing. If it measures well, great! If not, then we have two independent measurements verifying it. Either way the world benefits. Knowledge is power. More power, more better.

Tom

 

[tomchr]

And then along comes a smart engineer with sophisticated measuring equipment and tells you you shouldn't like it and that you're being fooled.

I've never actually said any of those things. All I did was to make the measurements available. You can infer whatever you want from that, but you are responsible for your own inferences.

Nelson Pass said it best....“The ear is not a microphone, the brain is not a tape recorder, and measurements are limited in describing subjective quality. I like to have low distortion and so on, but these things take a back seat to what I experience when I listen. There are plenty of products which have great specs – I will not be offended if you buy those.”

Appeal to authority...

Pass also says, "We should no more let numbers define audio quality than we would let chemical analysis be the arbiter of fine wines." My question for Nelson (and you too, I suppose) is then, "Would you drink fine wine from a dirty glass?" If I'm enjoying a good wine, I would prefer it served in a clean glass. I'm looking to taste the wine. Not wine + dish soap + bleach + whatever else was in that glass before my wine. Same with my enjoyment of music. I'm looking to hear the music. Not the music + volume control.

Tom

 

[mansr]

Pass also says, "We should no more let numbers define audio quality than we would let chemical analysis be the arbiter of fine wines." My question for Nelson (and you too, I suppose) is then, "Would you drink fine wine from a dirty glass?" If I'm enjoying a good wine, I would prefer it served in a clean glass. I'm looking to taste the wine. Not wine + dish soap + bleach + whatever else was in that glass before my wine. Same with my enjoyment of music. I'm looking to hear the music. Not the music + volume control.

Also, modern wine making involves quite a bit of chemistry.

 

[Deleted member 2944]

I'm not sure how apt the dirty glass analogy is, but the basic premise is correct.
However, a subjectivist would retort that if the wine tastes better in a dirty glass, then he wants it in a dirty glass.

As I've mentioned on this forum before, you can't out-argue a subjectivist from their subjective evaluation. There's no room for discussion there. It's an incontrovertible conclusion.
Objective testing of LDR attenuators will certainly show increased distortion of various levels, but it won't change the mind of an LDR owner who likes the sound his unit imparts.

Dave.

 

[tomchr]

Heh. Yeah. Maybe some do prefer a little dish soap in their wine. Who am I to judge.

And you're absolutely right. You're not likely to change anybody's mind. That's also not my point. Life's too short.

Tom

 

[Stretchneck]

When you get one for demo, ship it to Amir for testing. If it measures well, great! If not, then we have two independent measurements verifying it. Either way the world benefits. Knowledge is power. More power, more better.

Tom

Cheers - I'll use my own ears first. Question...ALPS RK271 within your preamp design... How does the ALPS affect performance? Would using a Muse 72323 based controller or a higher quality stepped attenuator make any difference?

I also wonder how your design would fair against a Mezmerise B1 buffer and Muse 72323/stepped attenuator build (I'm following what people say here in the fact that a buffered passive tends to sound better).

Other than LDR, these are the only options I'm considering, I only need unity gain.

 

[tomchr]

Cheers - I'll use my own ears first. Question...ALPS RK271 within your preamp design... How does the ALPS affect performance?

It doesn't. At least not that I can measure.

Would using a Muse 72323 based controller or a higher quality stepped attenuator make any difference?

Yep. Both would increase cost. The MUSES72323 would be higher noise. If the impedance of the attenuator is the same as the Alps pot, they'll have similar noise but the attenuator will track better channel-to-channel. If you tend to operate below 9 o'clock on the volume control, you'll probably find it more difficult to get to the volume you want with an attenuator as the step size towards the minimum position tend to be rather large.

The only time I've ever found the channel imbalance of a pot to be troubling was with a tube headphone amp with quite high gain. This resulted in the volume being set well below 9 o'clock, which is where the tracking is the worst.

You can see my Preamp Guide for more information.

I also wonder how your design would fair against a Mezmerise B1 buffer and Muse 72323/stepped attenuator build (I'm following what people say here in the fact that a buffered passive tends to sound better).

Mine would be ~60 dB lower distortion. Maybe even lower, depending on whether you use the BJT version or the JFET version of the Mezmerise.

Tom

 

[Stretchneck]

Mine would be ~60 dB lower distortion. Maybe even lower, depending on whether you use the BJT version or the JFET version of the Mezmerise.

Tom

Thanks Tom.

Sorry, you'll need to clarify the ~60dB distortion figure to me. 60dB is a pretty large difference. The original Pass B1 (upon which the Mezmerise is based) is quoted as .003% distortion at worse apparently. So where does the 60dB difference come from?

The MUSES72323 is -124dBV, so is therefore noisier than your buffer, and the channel separation is also somewhat poorer... so I get that.

I recognise that your buffer measures extremely well, but does it change the sound character at all? What's you opinion with it in and out of the chain?

I ask because I tried a Topping Pre90, which also measures extremely well, but whilst very decent sounding, does still alter the sound. My master 9 sounds better (although it has a higher noise floor that then Pre90 and add a bit of fuzz), but the best sound I have found is actually no preamp at all (but with lossy windows volume control).

Thanks for the advice on volume pot vs attenuator... I guess I am just thinking that if the volume pot is in the signal path than it will affect the sound. Have your tried any other attenuators? People have said that even a z-foil stepped attenuator does take away a little (any resistive element will I guess), hence my consideration of LDR or MUSE. If you tell me that your solution based on the ALPS pot is completely transparent to source, you can tell no difference if it's in the chain, then I'll give it a go. I need to order some other bits from you anyway

 

[tomchr]

Sorry, you'll need to clarify the ~60dB distortion figure to me. 60dB is a pretty large difference. The original Pass B1 (upon which the Mezmerise is based) is quoted as .003% distortion at worse apparently. So where does the 60dB difference come from?

You didn't provide any measurements of the B1 in your previous post so I looked at the schematics and from that estimated that the preamp will probably hit 0.01% (-80 dB) THD. That's just my guess based on my understanding of how devices work and my experience with them. I provided my estimate accordingly. The MUSES chip is supposedly limited by the external opamp, so if you're using something like the LM4562 or OPA1612 you should be able to get low distortion. However, whether JRC's claim holds with a modern opamp remains to be seen. Either way, I made the assumption that the volume control (be it Alps RK271, MUSES72323, or a stepped attenuator) will be significantly lower distortion than the B1, which for any of those volume controls would be a valid assumption, so a B1 fitted with any of those volume controls will be dominated by the distortion of the B1 and not that of the volume control.

The Mezmerise B1 hits 0.003% (-90 dB) at 2 V, 1 kHz according the measurements in the link you posted just above (thank you for that!). The distortion is higher at higher output amplitudes. So my estimate of -80 dB THD wasn't too far off. Turns out my ~60 dB difference estimate is more like ~50 dB difference (assuming 2 V, 1 kHz). Still a large difference. Recall that my Universal Buffer reaches -140 dB THD.

Did I misunderstand your question?

I recognise that your buffer measures extremely well, but does it change the sound character at all? What's you opinion with it in and out of the chain?

Not according to the measurements.

I guess I am just thinking that if the volume pot is in the signal path than it will affect the sound.

And that may be why you hear a difference. This is known as expectancy bias. And if you do hear a difference then I do believe you. I do believe that you perceive a difference. But I also know from my measurements that the volume control did not change the stimulus that reached your ears. I can come up with several reasons why someone would experience a difference in perception even when presented with the same stimulus. That's all part of the human experience.

You are right in one aspect, though. The volume control certainly has the potential to influence the sound quality. For example, if you built a stepped attenuator with highly non-linear resistors (high voltage coefficient, high temperature coefficient, etc.) the attenuator would certainly distort the signal measurably. In that case, the output of the volume control will not simply be an attenuated version of the input. It will be an altered and attenuated version as exemplified by the LDR volume controls.

Note that some prefer to alter the sound. To them a little extra second order harmonic distortion (or whatever it is) sounds better. They prefer a little dish soap in their wine. That's their choice. If the manufacturers would provide measurements of their equipment, these folks could look at the FFT and go, "that adds the right amount of H2 to suit my taste" and buy that equipment. Those of us who prefer low distortion could stay clear. I'm certainly on the 'clear' side of the equation, but that's me and my preference. There are ~7B of us on this here space rock and it would rather suck if we were all just clones of the same person. At least I prefer diversity. But that's me and my opinion.

Have your tried any other attenuators? People have said that even a z-foil stepped attenuator does take away a little (any resistive element will I guess), hence my consideration of LDR or MUSE.

Nah... I don't have the resources to buy one of each attenuator out there and test them. Sorry. If the manufacturer ships one to Amir for measurements, I'm happy to help interpret the measurements.

In order for me to recommend anything other than an Alps RK271 (or RK097) pot or a stepped attenuator with metal film resistors, the manufacturer will need to provide data. As a minimum I expect the manufacturer to provide the RMS output noise and the THD+N. I would like to see the IMD as well. Manufacturers who use exotic materials or resistors in their attenuators should justify why using theory and/or measurements – not marketing babble.

If I was to go with an integrated volume control, my choice would be either the MUSES72323 or the CS3318. I'm sure both are good, but the CS3318 is better characterized. The MUSES data sheet leaves a lot of open-ended questions. The advantage of the IC-based volume controls is that they lend themselves well to remote control.

If you tell me that your solution based on the ALPS pot is completely transparent to source, you can tell no difference if it's in the chain, then I'll give it a go.

It is completely transparent as far as I can tell and as far as I can measure. Your personal experience may be different.

I need to order some other bits from you anyway

I have no plans for going anywhere.

Tom

 

[Stretchneck]

Superb answer @tomchr - thanks so much for your time. Seems that I need to order a universal buffer ;-)

 

[Stretchneck]

LDR preamp now in my system - in summary, it’s very decent, but not my cup of tea.

The LDR preamp I have has more reverb and longer decays (is this second harmonic?) than the Qutest direct to amp - this gives the impression of a larger soundstage, but masks detail (are these traits common to LDR?) The longer decay on cymbals can also be very distracting as they can end up as more of a continuous shimmer on certain tracks when really they would benefit with more attack start/stop.

The unit also has a reasonably veiled sound compared to DAC direct.

Bass is still nice and punchy with LDR.

I do think it’s a very nice solution, perhaps for a mid-fi system, or if someone wanted to add some extra reverb/decay (tube like?) then this might be a great solution. But for me it doesn’t sound truthful.

Goes to show that each component sounds different despite measurements

 

[LTig]

LDR preamp now in my system - in summary, it’s very decent, but not my cup of tea.

The LDR preamp I have has more reverb and longer decays (is this second harmonic?) than the Qutest direct to amp - this gives the impression of a larger soundstage, but masks detail (are these traits common to LDR?) The longer decay on cymbals can also be very distracting as they can end up as more of a continuous shimmer on certain tracks when really they would benefit with more attack start/stop.

How much longer is the decay?

The unit also has a reasonably veiled sound compared to DAC direct.

Bass is still nice and punchy with LDR.

I do think it’s a very nice solution, perhaps for a mid-fi system, or if someone wanted to add some extra reverb/decay (tube like?) then this might be a great solution. But for me it doesn’t sound truthful.

Goes to show that each component sounds different despite measurements

Nope. The LDR volume control measures very different compared to a simple volume pot.