"Boutique" opamps from chinese vendors. Anyone tested?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Status
Not open for further replies.
PS: +1 Abbey about line cords. If a different one makes an audible difference that suggests a design flaw in the SKU. I don't remember writing about funny line cords in my old magazine column because the snake oil crowd hadn't gotten that creative yet.

Really? Make the earth path in that cable sufficiently low impedance and the faint background hum that was annoying without music becomes silence.

Or the noise from dimmer packs that keeps breaking through is reduced enough to be inaudible.

Is that audible? You tell me. Is that snakeoil? You tell me.

Just look at various parasitic currents, the equivalent network of ground and earth connections via mains and line cables and tell me again "cannot make a difference".

Having mains cables that have a big lump in the middle with a DC trap and a safe earth loop breaker can massively quiet down systems.

These days SMPS type power supplies dump massive amounts of noise in the near audio audio band supersonic range into the mains, common mode, which in turn make into the ground of equipment supplied and ultimately are demodulated and cause "birdies".

There are active circuits that can cancel this noise. Add them into a mains cable, plug in and go Depeche Mode...



And yes, most of this is easily audible and measurable. And spare me the "it's all bad design".

It often can be massively improved just by using a low impedance earth connection mains cable. I like steel armoured multicore control cables for robotics and industrial use, we get a shield, many cores of which we can connect many paralleled for super low impedance on the earth path. Depending on the country "premium" mains plugs can reduce contact resistance, generic IEC's are often quite high contact resistance.

I am flabbergasted that people just handwave to make it all go away, because they don't "LIKE" the idea that cables are components in a complex system. This is cargo cult science of the worst kind.

The whole snakeoil brigade can only go on killing their snakes and getting oil from them to sell because the very people who could inject sense and put all of it onto a sound (pun intended) foundation.

Thor
 
Indeed, access to cheaper fab allows the use of multipe layer PCB's. When used wisely, it has a potential for improved performance. When used without proper understanding, it can result in degraded performance. Unwise copper pours have a tendency to mix "grounds" that shouldn't be.

Yes - takes a bit of technique and care in ECAD to handle 'separate' 0V Copper Pours etc.
And also knowing where to avoid the extra capacitance introduced eg at high impedance nodes. Also on board rf filtering needs some attention to avoid capacitance bypassing the filter.
 
I covered some of these exact same topics in my old "Audio Mythology" magazine column back in the 1980s.

I read some of these in back issues.

Much of what you wrote was biased and incomplete. Some of it was completely wrong. Nothing I read actually was on the money based on my understanding of electronics. Excessive oversimplification caused incorrect conclusions.

Like in one column you claimed that the premise of differences in cables was that they altered the signal passing through them in a material way. This specifically was speaker cables, but the principle goes wider.

First I think we can agree that cables with inappropriate electrical characteristics WILL alter signals in audible ways. You even admitted that in the column.

However that is NOT in general where differences accrue (one hopes at least).

The issues are a bit more subtle and do not manifest if you test the cable in isolation. But, considering any cable by itself in isolation is pointless anyway, because nobody uses cables that way.

Cables invariably make connections in systems. So any meaningful analysis must consider the rather complex system. Your columns never really did that. As such they completely failed to expose any myth and actually merely created obfuscation of real issues.

Even something like this:

1692805402337.png
barely scratches the surface for a very, very simple case.

Thor
 
Now some people missed that "96dB dynamic range" is actually a bit misleading.

With 65k values, we can only use 32k values per polarity, so a "16 Bit Audio System" is actually a signed 15 Bit Audio system. So it actually cannot do 96dB, only 90dB.
Do you think that a digital signal constituted of a suite of positive MSB is an invalid signal? I know it is not bound to happen often, but mathematically - which is the esseence here - it seems valid to me.
if we define SNR like LP, CD only manages 76dB.
How do you arrive at this figure? Magically losing 20 (or 14) dB... Actually it's quite difficult to qualify dynamic range with only one figure, particularly on systems that use intense pre/de-emphasis like tape or vinyl.
Logic (which seems mostly absent in audio) would suggest that one should design speakers such that can be easily placed in normal living environments, preferably visually blending in and having a directivity, frequency response etc. conductive to offering high quality audio without the requirement for acoustic treatment etc.
The only "speakers" capable of it are called headphones.
I don't think anybody has invented the "Krummlauf" loudspeaker yet.
Siegfried Linkwitz did a fair bit of work in this area - something for which I'm even willing to forgive him for his part in foisting this absurd and unnecessary even order crossover onto the world, where a 3rd order butterworth would have been a MUCH better choice for MOST applications
As much as I respect S. Linwitz work on x-overs, I'm not a fan of his home music contraptions, particularly so-called "dipoles".
The benefits of 4th-order LR x-overs have been demonstrated enough in concert systems, and Mr. Lipshitz and Vanderkooy have perfected the concept.
 
What else is there than resistance, inductance at contact quality? Maybe shielding...

Add in capacitance parameters. Screening / Shielding is definitely 'a thing' both ways.
Cable configuration may be the dominant factor especially if conducted emissions have a significant H field component. L&N twisted ? Or not ? Twist Pitch ? (Okay - this is essentially an inductance effect but not 'simple' wire inductance)
This is primarily about susceptibility to emi so depends hugely on the DUT.
 
Indeed, access to cheaper fab allows the use of multipe layer PCB's. When used wisely, it has a potential for improved performance. When used without proper understanding, it can result in degraded performance. Unwise copper pours have a tendency to mix "grounds" that shouldn't be.

Even for audio I found mostly that continuous copper planes for ground offer improved objective performance.

Where very high impedance nodes are present, the stray capacitance may cause issues, but this can usually be compensated easily.

"Grounds" don't mix.

Currents "mix" if the layout is poor. If the impedance the currents pass through is zero, there is no problem if currents mix. In reality the impedance is never zero of course.

But even in this case a contiguous copper planes offer the lowest impedance and thus the lowest error voltages.

As a rule, signal currents should not really flow in ground planes, loops should be closed locally. In effect a local star ground that also happens to be connected into the ground plane.

The only flow of signal currents should be between stages, where it is minimal. Simple logical layout keeps those loops contained.

Split's in planes are often worse than not having them. Current follows always the lowest resistance path, by the time you have layout where a split is feasible and nothing crosses the split, you can take the split out anyway, as current will not flow there.

Power supplies should deliver DC, not AC so at the worst the current flow will cause a DC offset. Again, the lowest resistance is afforded by a plane, so again errors are minimised.

Now something you should not do is to lignal lines that carry other signals (I2C, SPI etc) under high impedance nodes.

With a ground plane and SMD you can place these on the "other" side of the PCB and ground plane isolated from the high impedance nodes they pass under... Again, contiguous ground plane wins.

I'm with Mr Ott on that one, splits are a losing proposition. They seem sophisticated and special, but if really needed they actually indicate poor layout.

Sometimes other considerations force "poor layout" but you will know and THEN and ONLY then the split planes become necessary but you will definitely know when that is the case.

Thor
 
Add in capacitance parameters. Screening / Shielding is definitely 'a thing' both ways.
Cable configuration may be the dominant factor especially if conducted emissions have a significant H field component. L&N twisted ? Or not ? Twist Pitch ? (Okay - this is essentially an inductance effect but not 'simple' wire inductance)
I didn't neglect shielding, as you may see, but we are talking of mains cables, not signal cables. Most mains cables have similar construction, the main significant difference being the conductor cross section.
 
I didn't neglect shielding, as you may see,

Indeed you did not neglect shielding. You wrote "Maybe shielding..." and my meaning was to confirm that it (or its absence as with standard mains cables)is a factor.

but we are talking of mains cables, not signal cables. Most mains cables have similar construction, the main significant difference being the conductor cross section.

Yes. Understood. But I was extending into how an "optimised" mains cable might be constructed.
We've probably (?) all seen DIY efforts using eg bundles of CAT5/6 cables (in addition to DIY speaker cables constructed from same). And expensive 'special' mains cables claiming "Special Powers" :oops:
 
Do you think that a digital signal constituted of a suite of positive MSB is an invalid signal? I know it is not bound to happen often, but mathematically - which is the esseence here - it seems valid to me.

No. In the analog realm we map (for arguments sake) the range of +2.8V to -2.8V into 65k values. Digital silence is 0x0000, but negative maximum is 0x8000 and positive maximum is 0x7FFF. 16-bit samples are represented as 2's-complement signed integers, ranging from -32768 to 32767.

How do you arrive at this figure? Magically losing 20 (or 14) dB...

No magic. For LP the "0dB Reference" is 5cm/S, however the permitted maximum signal is 25cm/S, though it possible to cut higher and to tracker higher signal levels with quality equipment. The 25cm/S limit was selected to allow for cost constrained low grade consumer gear to reliably track.

So for LP the maximum signal level is +14dB with reference to the arbitrary "0dB" reference, or we may say LP has 14dB Headroom.

16 Bit Digital (actually all digital) uses 0dBFS as reference. To make it comparable to LP we need to set the reference level to -14dBFS.

Secondly, the 96dB dynamic range comes from a milkmaid calculation of "6dB per bit", however, above we noted that a 16-Bit digital audio system is actually a signed 15 Bit system.

Thus the correct formula for the theoretical SNR/DNR needs to subtract the sign bit and thus arrives at (6.02dB *(16-1)b)+1.76dB giving 92.06dB if we want to be pedantic, which after we correct the reference level to -14dBFS gives 78dB SNR with 14dB Headroom. I omitted the 2dB for simplicity.

Actually it's quite difficult to qualify dynamic range with only one figure, particularly on systems that use intense pre/de-emphasis like tape or vinyl.

True dynamic range for LP, Tape etc. (but also modern digital systems) is frequency dependent.

But the figures we are quoting are usually the result of using 1 kHz signal at nominal signal level (0dB) and an integrated measurement of the noise floor (weighted or not).

The only "speakers" capable of it are called headphones.
I don't think anybody has invented the "Krummlauf" loudspeaker yet.

Well, again. We can make highly directional speakers, which can be placed near boundaries and have a radiation pattern that stabilises the virtual sound source locations from our two channel signal across a wide lateral listening window and allows an acceptable vertical listening window.

This still requires a relatively fixed listening area to correctly perceive the stereophonic sound field, but it has been done and does not need headphones.

As much as I respect S. Linwitz work on x-overs, I'm not a fan of his home music contraptions, particularly so-called "dipoles".

You should perhaps read the reasoning.

But I agree, I prefer cardioid systems, more precisely hypercardioid systems to dipoles and they need to have the correct amount of toe-in so that when the listener moves off-center listening positions, the level from speaker now further away is increased, retaining a stable phantom sound source location even if the listener moves.

This can get a listening window nearly as wide as the distance between the speakers.

The benefits of 4th-order LR x-overs have been demonstrated enough in concert systems, and Mr. Lipshitz and Vanderkooy have perfected the concept.

They are vastly inferior in this application to crossovers that use elliptical filters and have a 10th order higher slope in the actual crossover region and they have been repeatedly tested in blind listening tests as inferior in subjective terms to the classic 3rd order butterworth.

For situations where lobing is a non-issue use 3rd BW, for situations where it is, use much higher order to gain a much better result, such a 10th order elliptical filter (EL10).

LR4 is either too much or not enough, the widespread use notwithstanding. I have noted that we now often see state variable subtractive LR8 - these sound even worse in terms of the summed result, but offer steeper slopes, yet they have much greater complexity and still less steep slopes than a EL10.

BTW, this thread has taken a most extreme veer from chinese boutique op-amp's.

Thor
 
I read some of these in back issues.
good.. it doesn't show..
Much of what you wrote was biased and incomplete.
back at ya...
Some of it was completely wrong. Nothing I read actually was on the money based on my understanding of electronics. Excessive oversimplification caused incorrect conclusions.
?
Like in one column you claimed that the premise of differences in cables was that they altered the signal passing through them in a material way. This specifically was speaker cables, but the principle goes wider.

First I think we can agree that cables with inappropriate electrical characteristics WILL alter signals in audible ways. You even admitted that in the column.

However that is NOT in general where differences accrue (one hopes at least).

The issues are a bit more subtle and do not manifest if you test the cable in isolation. But, considering any cable by itself in isolation is pointless anyway, because nobody uses cables that way.

Cables invariably make connections in systems. So any meaningful analysis must consider the rather complex system. Your columns never really did that. As such they completely failed to expose any myth and actually merely created obfuscation of real issues.

Even something like this:

View attachment 113572
barely scratches the surface for a very, very simple case.

Thor
Barely scratches your hypothetical surface...

It was decades ago and I (just barely) recall multiple different discussions about cables in my old columns. Invariably in every case of credible audible differences there were measurable physical manifestations. I recall citing a classic AES paper about speaker wire from some mid west college professor. Instrument cables had more mundane explanations like cable capacitance, hum rejection, etc.

carry on, I have better things to do than reargue old settled issues.

JR
 
Add in capacitance parameters. Screening / Shielding is definitely 'a thing' both ways.
Cable configuration may be the dominant factor especially if conducted emissions have a significant H field component. L&N twisted ? Or not ? Twist Pitch ? (Okay - this is essentially an inductance effect but not 'simple' wire inductance)
This is primarily about susceptibility to emi so depends hugely on the DUT.

Indeed. And how about we introduce a conductive plastic filler around the individual wires that is also loaded with ferrite powder and is placed below our screen and this in turn is below our magnetic tinned steel "armour"?

How might such a cable affect conducted emissions or indeed conducting emissions into the DUT.

The failure of all this debunking is that it focused only and strictly on the intended signal (or power) passing through the cable and ignores all else. Those are not the cable problems you seek, move along, go about your business.

Thor
 
Invariably in every case of credible audible differences there were measurable physical manifestations.

Did I ever claim the differences were NOT measurable physical manifestations.

Depends what you measure though. What you see on an AP2 and what you see on a 5GHz spectrum analyser are both objective measurements of physical manifestations.

Instrument cables had more mundane explanations like cable capacitance, hum rejection, etc.

So cables DO MAKE A DIFFERENCE? I though you said that for anything but resistance this was a complete myth?

Thor
 
Yes. Understood. But I was extending into how an "optimised" mains cable might be constructed.

Shields are something to consider. Geometry which impact on lumped and distributed RLC models. As mentioned handling (and potential attenuation) of conducted emissions must be considered.

Also, a cable with a lump on it that contains electronic components is still a cable, right?

How about a cable with a Low DCR differential and common mode lowpass at 10kHz, a DC Blocker (to stop toroidal transformers from vibrating) and a earth loop breaker.

This cable has the potential to make quite big measurable and audible differences to many systems compared to generic cables.

We've probably (?) all seen DIY efforts using eg bundles of CAT5/6 cables

I would not recommend CAT5 for Mains. The rating is insufficient.

On the hand, while not formally qualified for mains use and thus technically illegal, RG213/U or RG214/U has more than adequate ratings and using two gives a mains cable where the earth connection has twice the copper section of the actual conductors equivalent 10AWG for Earth and 13AWG for conductors.

It makes a mains cable that offers superior objective parameters to you average "kettle lead".

(in addition to DIY speaker cables constructed from same).

Cat5 for speaker cables is a very good option and can be shown to offer better resuluts as an equivalent crosssection "speaker cable", especially once the latter had time to oxidise, sulphidise and whatever happens to the poor wire once all the softeners in the PVC had time to work.

And expensive 'special' mains cables claiming "Special Powers" :oops:

And we are mixing non overlapping magisteria. Marketing and pricing is nothing relevant to the science behind the subject and to bring it is just an underhanded way to cast FUD.

Let's stick to what is relevant to us here, as we are talking DIY, 'expensive 'special' mains cables claiming "Special Powers"' are not relevant here, nor are strictly speaking expensive cables with magic boxes containing 20 cent of passive components, though their electrical side may have a bearing to the discussion.

Thor
 
Last edited:
I would not recommend CAT5 for Mains. The rating is insufficient.

Neither would I. Although tbf the DIY cable I saw (online) did account for voltage and current ratings by means of multiple cables and insulation as required.
I wouldn't fancy making it myself ! IIRC wiring into the mains plug was interesting.
My point was that I was thinking past 'standard' mains/IEC cables. Not advocating any particular method.

And yeah - it's far more commonly applied to speaker cables.

 
Neither would I. Although tbf the DIY cable I saw (online) did account for voltage and current ratings by means of multiple cables and insulation as required.

That is still more adventurous than I would be.

BTW, I'm great fan of Cat5/Cat6 and also use it for installations.

I use CAT6 STP for line cables (one pair signal, three pairs ground, shield to shell on XLR. wire tail on Jack and RCA) and installations as instant 4-pair multicore. Paies are bare solid copper in PE with individual foil shields per pair and overall foil & braid shield.

CAT5 in multiple runs makes easy speaker cables (interleaved connections) and it is easy to get the install crew to flood-wire with quad Cat5 to everywhere where you may want to hang tops, add some CAT6STP in case of active speakers, for DMX etc...

300m boxes are surprisingly reasonably priced.

IIRC wiring into the mains plug was interesting.

I'd expect so.

My point was that I was thinking past 'standard' mains/IEC cables. Not advocating any particular method.

Yes, again, we need to consider what potential issues we have in the current loops where (unfortunately) circulating parasitic mains leakage currents and audio signals (ground is a signal) mix in networks with very much non-zero impedance.

If we have 10 interconnected pieces with 1mA mains leakage each that flow to earth via our interconnects and earth connections? How many mV error voltage will we see between audio grounds?

And given I mentioning millivolt and we have signal levels in Volt, what does this do to our SNR, even in balanced systems?

And yeah - it's far more commonly applied to speaker cables.

It works really well there.

STP (Shielded Twisted Pair) is great for line cables and even AES digital. One Cat6 STP cable can carry 8 channels of AES-EBU digital Audio.

Thor
 
Did I ever claim the differences were NOT measurable physical manifestations.

Depends what you measure though. What you see on an AP2 and what you see on a 5GHz spectrum analyser are both objective measurements of physical manifestations.



So cables DO MAKE A DIFFERENCE? I though you said that for anything but resistance this was a complete myth?

Thor
Don't put words in my mouth it's unsanitary.

Notice that I qualified my comment as "instrument" cables (capacitance obviously affects high impedance guitar cable response). It is "speaker" wires that are dominated by simple resistance.

OK here's an anecdote to expand upon that premise. Back several decades ago I was questioned about exactly this relationship. One of my internet "friends" from a sound reinforcement forum was adamant that he could definitely hear the improvement from using a specific brand of funny speaker wire in his system. I indulged him and dug a little deeper into his exact application. It turned out that the length and gauge of his particular speaker wire exhibited enough series resistance to generate a measurable and audible frequency response error from the interaction between the passive loudspeaker's changing input impedance (due to passive crossover) and the speaker's wire resistance. In his case the speaker/wire interaction was actually causing a fraction of a dB boost across the top octave. A subtle but audible difference. 🤔

The first rule of pitching snake oil is that it helps to actually sound different, you can often convince the sheeple that the difference is an improvement. ;)

JR
 
The first rule of pitching snake oil is that it helps to actually sound different, you can often convince the sheeple that the difference is an improvement.

Not typing on your keyboard, but again, did you just claim that cables can sound different?

My point is simple, they do, often for prosaic reasons but these reasons are NOT limited to just how R/L/C influence the signal passing through the cable.

And yes, where there is a difference there is a preference.

If we cannot offer solid reasoning and instead pour scorn on the poor sheeple how stupid and gullible they are to hear a difference (that actually exists) that we claim cannot insist, we just gave them reason to hate and ignore us and to hand their cash to snakeoil man.

Great job helping the consumers.

Thor
 
Status
Not open for further replies.

Latest posts

Back
Top