New Posts New Posts RSS Feed - Harshness, Clicks and Pops
  FAQ FAQ  Forum Search   Events   Register Register  Login Login

Welcome to the Graham Slee Audio Products Owners Forum

 

Open to all owners plus those contemplating the purchase of a Graham Slee HiFi System Components audio product and wishing to use this forum's loaner program: join here (Rules on posting can be found here)

This website along with trade marks Graham Slee and HiFi System Components are owned by Cadman Enterprises Ltd


Harshness, Clicks and Pops

 Post Reply Post Reply Page  123 9>
Author
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Retired

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 16298
Post Options Post Options   Thanks (2) Thanks(2)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Topic: Harshness, Clicks and Pops
    Posted: 31 Jan 2021 at 11:26am
Through Big-Tech, Moore's law has hammered home that subtle artifacts make a difference - sometimes explosively.

I watched a video explaining "n -pn" leakage, that in nanometer technology, it caused Latency, so off isn't off entirely.

They knew about this before they embarked on transistor stuffing many years ago, but I'm surprised by how such things are ignored in audiophile circles. They seem to be more interested in the magical stuff of zero scientific connection. It could be due to a lack of scientific education.

Leakage current exists in common or garden silicon, too - things like power transistors, discrete small-signal transistors, and the transistors that make up a silicon chip such as an op-amp.

Could the Early voltage be related? Maybe, but the point I'm making here is old technological understanding gave way to a new technical understanding, with minimal in the form of transitional fossil evidence.

To put it another way, the language changed. It might mean much the same as it used to do, but we can't catch them up to ask.

James Early discovered that on-ness increased with on-ness, but Moore's suffers from off-ness being less than off-ness.

Early has little effect on Moore's because switching is "different" to analogue. All I can discuss is what I do, not what they do. The point is that transistors are not the clean devices many take for granted while indulging in their other-world esoteric pastimes.

One interesting word popped-out of the video, which is Latency. In the word Latency, we can see another word, Late. We can add the letters n and t to obtain Latent from Late. We can observe Latent if we are near a brick or stone building in the summertime. Early in the day, with temperatures rising, the building feels cool. Near dusk, with temperatures falling, the building feels warm. That is Latent Heat.

Therefore, Latency can describe a stickiness, where things don't change the instant we'd like them to change.

Big-tech discovered that Latency is the spanner in Moore's law as far as silicon is concerned. It brings Moore's law to an end. Looking back, they showed the valve (tube) as having significant Latency, followed by the bipolar transistor, the FET, and then nano-transistors. Even though the valve has the speed needed for high power-high frequency work to this day, it suffers the worst leakage. This leakage is such that it doesn't switch off sufficiently, so power is wasted.

As junctions between materials become narrower, leakage becomes a problem, and now they're at 3D - building upward - so that 2D can be shrunk further.

Can any of this benefit audio? Some say the same characteristics cannot exist at audio frequencies - that they live at GHz and THz frequencies. As if by a miracle, they do not exist at MHz or kHz. Instead of a complete building on a summer's day, shall we use a single brick? Latent heat might not be easily felt by hand, but we can establish more accurate measurement, and using it, we can find there is Latency of temperature. Therefore, some of what exists up there can exist down here.

During my research for this part of this topic, I could not find any correlation between the words Late, Latent, and Latency, which I thought rather odd.
That none should be able to buy or sell without a smartphone and the knowledge in how to use apps
Back to Top
Godra View Drop Down
Senior Member
Senior Member
Avatar

Joined: 29 Apr 2016
Location: Mont-Royal
Status: Offline
Points: 179
Post Options Post Options   Thanks (0) Thanks(0)   Quote Godra Quote  Post ReplyReply Direct Link To This Post Posted: 01 Feb 2021 at 12:30am
Pretty interesting read Graham. Thanks for sharing!  Like you said, it's funny how all things digital are always advertised as perfect. I think we are misled in this case. When we think about digital, we instantly have in mind that there is only two value that sums up everything : '0' (false-no voltage) and '1'(true- voltage). This suffers no ambiguity. This lack of ambiguity is to me the hallmark of all thing digital. Though,  in reality, from what I have read on digital to analog converter , it seems that some '1' can be seen as '0' and conversely...
Clearaudio Performance DC with Clarify tonearm + Virtuoso V2, Tannoy Legacy Eaton, Accession MM and BAT VK300Xse
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Retired

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 16298
Post Options Post Options   Thanks (1) Thanks(1)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 01 Feb 2021 at 5:53am
It could be said that a nano-transistor has no capacitance; instead, it's one of those cases where it's so small it can be neglected.

Capacitance is another "show-stopper" for semiconductors - it gets bigger as the device gets bigger. It begins to get troublesome in discrete transistor circuits, becoming far worse in power amplifiers.

The larger the area of each layer of silicon that come together to form collector, base, and emitter, or source, gate, and drain, the larger the capacitance.

The junction between the gain elements - collector to base, or drain to gate - is worsened by voltage gain - whatever its capacitance, it is multiplied by its gain. Power transistors come off worse due to the physical size of the junctions.

Valves (tubes) suffer significantly from capacitance considering their high impedance - big Rs combine with Cs to roll-off the highs. It led to the development of the tetrode (one extra grid) and the pentode (two extra grids) to split the inter-electrode capacitances and reduce the total capacitive effect.

The choice for a low-distortion preamp stage was always the pentode. It could be used for accurate in-loop EQ because it didn't run out of high-frequency gain as a triode does.

A triode inside an EQ network would suffer poor high-frequency accuracy, and failing to control transients, would output considerable distortion and exaggerate click and pops. The triode was often relegated to the cheaper (mono) record player, where its 5% THD didn't matter much. Using a dual vacuum tube and passive EQ in between, sufficient gain could be had from a simple tube. Incredibly, over recent years, this circuit has found its way into the audiophile market.

The valve takes its place at the head of the latency hierarchy, followed by power transistors, small-signal transistors, and then chip-level fabricated transistors in such as operational amplifiers (opamps).

The 1960s silicon revolution rapidly supplanted the valve in small-signal applications. When combined inside innovative configurations, the discrete transistor, available in both NPN and PNP, could do more and do it without as much noise. Impedance had come down an order of magnitude. Large value resistors generate more noise than small value resistors, and so noise is reduced in that ratio.

However, the capacitance was still a problem, and without the ability to add "grids," other techniques had to be found. Oscillation had been a problem for valve voltage followers (cathode followers or common-anode amplifiers) - these being used as output buffer stages. The transistor suffered so much from it that it was almost reinvented as a passive amplifier. The difficulty caused by capacitance resulted in negative resistance.

Semiconductors are not only used for amplification or switching; they are also used for oscillators to generate the frequencies needed for radio communications. Oscillators exist in both transmitters and receivers. In a receiver, they are used to demodulate carrier frequencies.

Transistor output buffer stages (emitter followers) bear a more than a striking resemblance to the Clapp version of a Colpitts oscillator. A buffer stage in a preamp often follows a volume control, and the stage might be oscillating at some considerably high frequency, well beyond our hearing range. A give-away is "early wear-out" of the volume control, which results in a crackling sound. This is simply the accidental oscillator changing and stopping and starting in frequency as the pot is turned. Not all crackling volume controls are due to this, mostly if they've done good service for a few years - that's just debris accumulation.

The discrete transistor designer might find more headaches than bargained for, and for what reason? Do discrete stages work better than a suitable alternative fabricated onto a silicon slice? The operational amplifier design has had to pass the tests of not oscillating or working within a specification where oscillation is guaranteed not to occur.

The opamp might not contain romanticism but has reduced latency and better frequency control within limits. Not all opamps are better than discrete in all applications, but they beat discrete hands down in some applications. Much depends on the designer's experience. There's a place for everything, and all applications are different.

We should now be aware that leakage, the Early effect, and now capacitance, have a significant bearing on how any particular circuit performs.

More to follow.
That none should be able to buy or sell without a smartphone and the knowledge in how to use apps
Back to Top
Speculator66 View Drop Down
Regular
Regular


Joined: 28 Nov 2020
Location: UK
Status: Offline
Points: 44
Post Options Post Options   Thanks (1) Thanks(1)   Quote Speculator66 Quote  Post ReplyReply Direct Link To This Post Posted: 01 Feb 2021 at 2:55pm
When I was buying my first system - many years ago now - I bought a Wharfedale Linton amplifier.
My mother heard it and said she would like one as well so one was secured for her.

Not long after she got it, she complained to me about hissing down one channel.
It turned out that the amp worked fine for 15 mins or so before the hiss started.
It was duly returned to the Rank Wharfedale Service department, not once, but twice.
They said that they couldn't find anything wrong with it.

So I, as a young lad beginning to learn about electronics stepped in and looked at it.
She was absolutely right and my thoughts turned to capacitors and transistors.

On taking the cover off I was greeted by a beautifully simply design - i'm not talking electronically but layout. There were four plug-in boards, two for each channel.
My simplistic mind said, swap the boards and see which one changes the channel it's hissing on.

I identified the bad board within only a few minutes and then started to replace the transistors one at a time. Second time lucky - there were only six anyway.

So, can transistors cause noise problems such as hum and hissing ?No doubt about it.

That amplifier is still working away in my workshop - currently through some old KEF 104ab Mk II's and it is now ca. 50 years old. Never had a problem since. 
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Retired

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 16298
Post Options Post Options   Thanks (1) Thanks(1)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 02 Feb 2021 at 7:13am
Transistors used to be different from op-amps because, at first, there were just transistors made from ingots of doped germanium, and so, op-amps were discrete assemblies of a number of them.

Today, transistors do not have superiority over op-amps - they are equivalents. When you see somebody boasting the superiority of transistors over op-amps, they are being incredibly big-headed about being able to assemble a number of elements better than a chip manufacturer. Then again, we all make mistakes!

To realise your mistake, you would need considerable experience in developing circuitry with both. An op-amp is not a ready-made amplifier; it is a greater part of one than a single transistor.

If you need a complex solution, the op-amp out-performs a board full of transistors. If you need a simple solution, one or a few transistors might be more suitable.

It was an awkward process making transistors from an ingot of germanium (or silicon). Still, by adopting a similar approach to how printed circuit boards are made, it became possible to print and etch transistors by the one, or by the many, on a silicon substrate.

The ingot method was problematic in attaching the wires. The planar process solved that problem. Bill Shockley's ingot method produced tiny single transistor slugs that could be clamp-wired either end, but the difficulty was sticking the base wire into the middle of it. You can't solder or weld a wire to a rock!

Eight of Shockley's team moved down the road and developed the print and etch method, which allowed the deposition of metal, and thus the ability to accurately and reliably spot weld the wires. The company they formed was Fairchild, which could make a single transistor on a chip, or many transistors on a chip, and a combination of N-doped and P-doped transistors, on a chip.
That none should be able to buy or sell without a smartphone and the knowledge in how to use apps
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Retired

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 16298
Post Options Post Options   Thanks (2) Thanks(2)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 02 Feb 2021 at 10:45am
As a youth, leaving alone my disco work on ceramic cartridge stages, my first home-made magnetic cartridge phono stage used a 741 op-amp, and the EQ was in its feedback. 50% of its improvement over my amplifier's built-in stage was psychological - because I'd built it!

The 741 was the most easily implemented op-amp of its time. The "designed for audio" NE5534 did not make its debut until the late '70s and didn't hit the hobby shops until much later.

I'd spotted a passive interstage preamp in a Tandy book about valves but could see any two gain stages could be used and set about making it with 741 op-amps. The result was revealing! Over the years, I had difficulty understanding how 741's were able to sound so good.

Pouring over valve datasheets many years later, the reason for the 741's success jumped out of the page. The RCA design was made for a cheap but low noise, twin triode valve (today's equivalent is the 12AX7). Remember that I previously remarked on triodes' capacitance limiting their high-frequency performance? This valve also had very poor distortion. By comparison, the 741 was light-years ahead.

Nearly everybody making phono stages today uses the same basic circuit, and some use three amplifier stages with two passive filters. I don't think they'd use 741 op-amps, but maybe the NE5534 or NE5532 dual cousin.

Now, some declare that their's is a discrete transistor design, which to some might appear smart, but having been there, I beg to differ.

The great thing about the passive interstage is it does help with the harshness, clicks, and pops. The interstage EQ is also the equivalent of a scratch filter - it attenuates any instability in the gain stages to a degree. That degree might not last, depending on the capacitors used in the filter.

But first, we need to understand the filter's behavior. A single-stage filter is similar in many respects to a tone-stack. In the case of RIAA, it is a low-pass filter with a step. It rolls off the highs given a starting point of 50Hz (this is called a pole), and then stops doing so at 500Hz (this is called a zero), then starts rolling off again from 2122Hz (the second pole).

Now, that would be fine if the driving impedance of the gain stage was very low - low, as the output of a 741. A class-A transistor output has considerable collector resistance, just like the original RCA valve circuit had significant anode resistance.

On positive half-cycles, the resistance adds to the filter, and on negative half cycles, it isn't there. And so, what you have to do is average it out on test, using an audio analyser, tweaking values as you go. To avoid this problem, the output of the first gain stage can be made push-pull, resulting in distortion problems due to gm-doubling - or not - which depends on signal level (loudness). The op-amp has this "pre-sorted."

Another problem is the capacitors and their dielectric absorption - because, in a filter, they charge and discharge. Should they be slow or inductive, they will not attenuate immediately to the correct level, and so clicks and pops might be exaggerated. The leading edge of a note might be exaggerated too.

If we also have a single-sided drive signal - a collector pulling down on the negative excursion and a resistor pulling up on the positive, depending on the first gain stage's absolute phase, some instruments sound different. The trumpet is an excellent example of an instrument favouring one half of the cycle over the other.

By setting the filter impedance ten times the collector resistance, this problem is minimised, but the high-value filter resistors add their noise. That's OK if you can put up with a bit of hiss, and if it's a valved stage, you'll probably expect hiss.

The filter capacitors ought to be fast to remove leading-edge exaggeration, so we can employ polypropylene types here. As long as they're not those audiophile wound ones, which are inductive (even if they claim otherwise), things should improve. However, not all polypropylene capacitors are equal, and some still lead to an exaggerated sound.

By and large, the biggest problem is the channel mismatch. Even using 1% components, the driving impedance difference per half cycle moves the leading edges away from the averaged level, such that the stereo image drifts and sometimes rapidly shifts, confusing the sound stage.

I made phono stages this way for many years, gradually learning why the 741 op-amps passive phono stage sounded good when its specification would suggest otherwise. It was blunting the edge.

One further observation was reading the technical hi-fi blurbs, which made much of pure resistive negative feedback in the gain stages. As the gain stages were multi-transistor, there must therefore be excess-phase. Although non-destructive, the lack of capacitive compensation leads to high-frequency oscillation (well beyond the audible range). Although, with passive filters blocking it from escaping the phono stage, it could have a very "dry" sound, which became ever more evident the more prolonged the circuit was left on.
That none should be able to buy or sell without a smartphone and the knowledge in how to use apps
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Retired

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 16298
Post Options Post Options   Thanks (1) Thanks(1)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 03 Feb 2021 at 9:42am
To supplement my last post I include illustrations of passive RIAA stages.

passive riaa eq block diagram

A block diagram illustrating the basic workings of a dual gain-stage with passive RIAA EQ.




The original RCA design with R's and C's transposed.
That none should be able to buy or sell without a smartphone and the knowledge in how to use apps
Back to Top
 Post Reply Post Reply Page  123 9>
  Share Topic   

Forum Jump Forum Permissions View Drop Down

Forum Software by Web Wiz Forums® version 12.01
Copyright ©2001-2018 Web Wiz Ltd.

This page was generated in 0.141 seconds.