So, What Is the Characteristic in Tubes that People Want to Emulate, Anyway?

[Lee Flier]

 

If I could have only a tube pre without a transformer, or a non-tube pre with a transformer, I'd probably take the latter all things being equal.

 

 

Yeah, me too... I'd say "tube mic, transformer pre" is the most consistently winning combination. Not that I'd want to use a tube mic for everything either, but for vocals and acoustic guitar anyway, it's pretty hard to beat that combo, assuming the mic and pre are decent quality to begin with of course.

[Lee Flier]

 

I think that part of the "warmth" people perceive is the more "ear-pleasing" harmonics that we were discussing earlier, which sounds less "harsh" to the ear. If this weren't the case, we could simply EQ a sound to sound "warmer" by having more bottom end, rolling off the top end, and that would be the end of that. But when people perceive "warmth", it seems to me that it's much more than simply an EQ adjustment.

 

 

Yeah, exactly... you shouldn't have to roll off the top end for it to sound pleasing to the ear. But that's what most people do now in order to avoid harshness in the top end.

[UstadKhanAli]

 

Yeah, exactly... you shouldn't have to roll off the top end for it to sound pleasing to the ear. But that's what most people do now in order to avoid harshness in the top end.

 

 

Yeah, good point about that too. If you need to roll off the top end to stop something from being harsh, there just may be something wrong with the sound in the first place! And agreed about the previous post as well with the "tube mic, transformer pre" being the most consistently winning combo. I end up with that most of the time, although I obviously also use every combination I have.

[Anderton]

 

Yeah, exactly... you shouldn't have to roll off the top end for it to sound pleasing to the ear. But that's what most people do now in order to avoid harshness in the top end.

 

 

In the real world, high frequencies are pretty delicate. If you look at the spectral response of most acoustic instruments, there's not a lot going on in the high end. Also, they're most likely to be absorbed in the listening environment. Given that a lot of recorded sounds are "hyped," there's an unnatural amount of highs. With a lot of mastering jobs, just shaving off the very highest frequencies gives a much more pleasing, and realistic, overall sound.

[MikeRivers]

 

I think that part of the "warmth" people perceive is the more "ear-pleasing" harmonics that we were discussing earlier, which sounds less "harsh" to the ear. If this weren't the case, we could simply EQ a sound to sound "warmer" by having more bottom end, rolling off the top end, and that would be the end of that. But when people perceive "warmth", it seems to me that it's much more than simply an EQ adjustment.

 

 

Oh, I'm certain that it is, but then what are harmonics but higher frequencies coming out that didn't go in? What you get with tubes, at least with triodes (like the 12A*7 popular as the first stage of in instrument amplifier) is 2nd harmonics occurring first. Transistors generally don't generate 2nd harmonics when they overload, but go right to the odd order harmonic series that makes up a square wave.

 

I think you can get "warmth" with EQ, but not as quickly as with a guitar amplifier coupled with a good guitar and a good player. In other words, it sounds like a good (hopefully appropriate) electric guitar sound.

 

But even or odd harmonic distortion surely isn't what people talk about when they talk about a "warm sounding" mic, be it tube, solid state, or dynamic. (we'll dismiss peizoelectric mics - they're never "warm" unless they're roasting on a crackling fire).

 

When people talk about a "warm" microphone, it's always because of frequency response irregularities, and those can be created equally well with the right EQ though it's not always trivial since there are other differences that influence the sound of a mic which are equally as important as on-axis frequency response. For example, a mic that picks up harsh reflections and perhaps emphasizes them due to irregular off axis response won't be perceived as "warm" while one used in a good room with smooth off-axis response is likely to be "warmer."

 

But tubes don't make a mic warm other than in body temperature.

[Lee Flier]

 

In the real world, high frequencies are pretty delicate. If you look at the spectral response of most acoustic instruments, there's not a lot going on in the high end. Also, they're most likely to be absorbed in the listening environment. Given that a lot of recorded sounds are "hyped," there's an unnatural amount of highs. With a lot of mastering jobs, just shaving off the very highest frequencies gives a much more pleasing, and realistic, overall sound.

 

 

... but prior to the digital era, we didn't have this discussion. And contrary to popular belief, that's not because tape rolls off the high end or can't pick it up in the first place. It's because the high end sounds good, hyped or not.

[MikeRivers]

... but prior to the digital era, we didn't have this discussion. And contrary to popular belief, that's not because tape rolls off the high end or can't pick it up in the first place. It's because the high end sounds good , hyped or not.

 

Ah, but tape sounds so warm! And one reason is because the magnetic saturation produces similar harmonics to overdriven tubes, coming on slowly like tubes. And not only is there tape to saturate, but the tape decks that we admire for warmth (really, not the Sony rescued from the thrift shop, even if it has tubes) usually have transformers, in, out, or both.

 

Most people were happy to move away from tube electronics in tape decks however, for lower noise, lower distortion when you don't want it, and better stability. With a tape deck, to get warmth, you don't drive the electronics into non-linearity, you drive the tape into non-linearity.

[UstadKhanAli]

What if we're not talking about saturation on tape? I think it produces highs in a more ear-pleasing manner. There's a reason why tape is so popular to emulate with DAWs - tape has an ear-pleasing quality. And I agree with Lee: tape didn't roll off the highs. It reproduced it quite well.

 

For those who don't know me who think I'm an analog zealot, I'm not. I use a DAW primarily and own a digital keyboard and a virtual modeling keyboard in addition to my analog synths. But I also have to call it like it is.

[Anderton]

 

... but prior to the digital era, we didn't have this discussion. And contrary to popular belief, that's not because tape rolls off the high end

 

 

Virtually all tape recorders had a pre-emphasis filter at the input, and a de-emphasis filter at the output. That doesn't mean the frequency response was any less, but back in those days, recorders used analog filters which many people think add a certain inherent character and warmth. I think they were also necessary to filter out any residual signal from the bias oscillator. The brickwall filters in PCM don't have the same character, and I've often wondered whether if it was possible to create a true analog filter to provide the same brickwall filtering, whether people would like the highs of PCM better.

 

There were a lot of frequency manipulations possible with tape, and most good engineers tweaked the pre/de-emphasis and bias to get a high end they liked. You can't do that with digital.

[UstadKhanAli]

 

Virtually all tape recorders had a pre-emphasis filter at the input, and a de-emphasis filter at the output. That doesn't mean the frequency response was any less, but back in those days, recorders used analog filters which many people think add a certain inherent character and warmth. I think they were also necessary to filter out any residual signal from the bias oscillator. The brickwall filters in PCM don't have the same character, and I've often wondered whether if it was possible to create a true analog filter to provide the same brickwall filtering, whether people would like the highs of PCM better.

 

 

I would love to hear if this is possible or if someone is exploring this.

[Lee Flier]

Me too.

 

I'm not so sure that all the character that I like comes from tape, again. You can definitely get a lot of that silky high end mixing a digital recording on a decent analog console. But I definitely think there might be some unflattering things going on with brickwall limiters (in converters) that could be made better.

[Anderton]

I would love to hear if this is possible or if someone is exploring this.

 

In a way, that's what DSD is about - the clock frequency is so high you can basically just twist two lengths of wire together, hang them from the hot output to ground, and have enough filtering Well maybe that's not quite true, but I'm one of the people who thinks the primary reason why DSD sounds better than PCM is due to the extremely relaxed filtering requirements - no brickwall needed. And the main quality to which I'd ascribe "sounding better" with DSD are the highs, which I feel DO have that silky, smooth quality.

[MikeRivers]

 

What if we're not talking about saturation on tape?

 

 

Then why else talk about tape? Flutter? A lot of bass players like the low frequency "head bump" because the bass plays back louder than recorded due to fringing at the edge of the head.

 

 

I think it produces highs in a more ear-pleasing manner. There's a reason why tape is so popular to emulate with DAWs - tape has an ear-pleasing quality. And I agree with Lee: tape didn't roll off the highs. It reproduced it quite well.

 

 

Right about the high frequency response. A decent tape deck will be reasonably flat to about 28 kHz at 15 ips, but that may or may not be better than ruler flat to 22 kHz for a 44.1 kHz digital system. But the distortion with A/D and D/A conversion is quite different from the distortion through what amounts to a ferrite core transformer (the tape and heads). So, like tubes, it's not a lack of distortion, but rather the presence of a kind of distortion that we find pleasant.

[Anderton]

You can definitely get a lot of that silky high end mixing a digital recording on a decent analog console.

 

Interestingly, a lot of those were bandwidth-limited as well before the "You can hear things at 50kHz and therefore response that goes out to 100kHz is wonderful" weirdness started to appear. I was doing analog circuit design during its prime, and it was just considered a good idea to roll stuff off at the very high and very low frequencies (outside of the audible range, of course). But analog is a very imprecise technology, and who knows if that filtering had an effect or not in the audible range.

 

One might think that going to 96kHz would allow for more relaxed brickwall filters, but that's not really the case as the extra octave only buys you a 6dB shallower rolloff, which isn't really enough to alter a brickwall filter's character in a fundamental way.

 

It may just be (warning: speculation alert!) that it's the accuracy of a brickwall filter that's the problem. When I designed the Quadrafuzz back in the mid-80s, I designed an analog cabinet emulator that followed the specs for high and low rolloff inherent in a guitar speaker/cab combination. Not knowing any better , and being a believer in "best practices," I designed it for minimum phase shift, minimum ripple, etc. It didn't sound very good. Then on a whim I went back and re-designed the same filter structure but with maximum phase shift, maximum ripple, and basically traded off as many good characteristics for bad characteristics as was theoretically possible. After that, it sounded great, and very "cab-like."

[MikeRivers]

 

Virtually all tape recorders had a pre-emphasis filter at the input, and a de-emphasis filter at the output. That doesn't mean the frequency response was any less, but back in those days, recorders used analog filters which many people think add a certain inherent character and warmth. I think they were also necessary to filter out any residual signal from the bias oscillator.

 

 

The filters of which you speak are what's known as "equalization," and there are standard curves so that tapes recorded on one (properly adjusted) machine will play back correctly on another. The bias trap is a resonant circuit at the bias frequency, which is well out of the audio range and doesn't really affect anything but keeping the bias off the output. It's like some D/A converters will have the sampling frequency at the output because it isn't adequately filtered out.

 

But analog filters work by introducing phase shift. You can make a phaseless digital filter, and there are some plug-ins like that, but most of the time people prefer the sound of a more common phase-changes-with-frequency type filter design. The head gap rolloff - where the wavelength approaches the width of the gap in the head pole piece and no longer looks to the head like a magnet - is phaseless, yet we put a phase-shifting filter set around it. The result of phase shift through a filter, as well as the change in frequency response that you want it to accomplish, is that harmonics are not all delayed by the same amount of time, so the harmonic structure of a complex waveform changes. When it gets too bad, the effect is described as "smearing" or "spitting."

 

 

The brickwall filters in PCM don't have the same character, and I've often wondered whether if it was possible to create a true analog filter to provide the same brickwall filtering, whether people would like the highs of PCM better.

 

 

All of the above, taken in small doese. The original "brick wall" filters are part of what gave early digital recording a bad name. There was significant phase shift down as low as about 3 kHz for a 22 kHz brick wall filter, and the frequency response in the last half octave before the steep cutoff was pretty bumpy. Apogee made a better one for the Sony DASH recorder (that was their first product). Modern oversampling converters have all but eliminated that problem, and in fact even garden variety modern converter chips sound very good and don't have the problems resulting from the phase response of the filters. And they sound even better when you can convince the users to not try to fill the track waveform graphic all the way to the top with squiggles.

 

 

There were a lot of frequency manipulations possible with tape, and most good engineers tweaked the pre/de-emphasis and bias to get a high end they liked. You can't do that with digital.

 

 

Same for distortion. You can set the flux level to give you as much or as little harmonic distortion as you want. But when you get a lot of level on tape so that it's going into saturation (not fully saturated, of course) sometimes the playback level coming from the head is high enough to clip the head preamp stage, so you need to back off on that stage's gain. It took a while for people to figure that out when they started using tape that took a lot more energy to saturate it. And the problem worked the other way, too. Assuming that the record amplifier had enough headroom to drive the tape as hard as you wanted, the heads would overheat, sometimes burn out, sometimes soften the oxide on the tape as it went past.

 

Ah, those were the days where you could really be creative with your tools without just buying more plug-ins.

[Anderton]

 

The filters of which you speak are what's known as "equalization," and there are standard curves so that tapes recorded on one (properly adjusted) machine will play back correctly on another.

 

 

I think you have it backwards. Pre-emphasis and de-emphasis were introduced first to reduce noise, and then the curves were standardized to guarantee consistency from one tape machine to another. However, why there are separate curves for NAB and CCIR/DIN is beyond me.

[Lee Flier]

 

Right about the high frequency response. A decent tape deck will be reasonably flat to about 28 kHz at 15 ips, but that may or may not be better than ruler flat to 22 kHz for a 44.1 kHz digital system. But the distortion with A/D and D/A conversion is quite different from the distortion through what amounts to a ferrite core transformer (the tape and heads). So, like tubes, it's not a lack of distortion, but rather the presence of a kind of distortion that we find pleasant.

 

 

Yes, but I think we need to make it more clear than we often do in these discussions that everything distorts sound. Moving through air distorts sound. Bouncing off walls distorts sound. Our ears distort sound. So whether we're talking about analog or digital or hearing sound live in a room, we can't possibly be talking about anything other than whether the way it distorts is pleasing or not. In fact, you could listen to live music in an anechoic chamber, which will produce the least possible amount of distortion, and most people probably wouldn't like the way it sounds.

 

So you can look at a graph and see that a digital recording has "less distortion" than a recording on tape, but there's no mention of whether we actually have less tolerance for even low levels of digital distortion because digital distortion sounds like crap. I know there have been some pretty decent studies about the types of distortion our ears can tolerate at pretty high levels vs. those we notice and find irritating even at low levels. But this is something that doesn't get mentioned often enough in these discussions. We think of "less distortion" as always being a more desirable thing but I'd argue that it really depends on the type of distortion and whether it's pleasing to the ear or not.

[Anderton]

 

Yes you can look at a graph and see that a digital recording has "less distortion" than a recording on tape, but there's no mention of whether we actually have less tolerance for even low levels of digital distortion because digital distortion sounds like crap.

 

 

I've often thought that what screws with our brains is that digital is less distorted at higher levels, and more distorted at lower levels. The real world just doesn't work that way. Our brain's wiring stretches back way into the distant past, and given the ear's huge dynamic range, we can hear that something "isn't right" even if the levels are so low it's not obvious.

 

This might also be why people like 24 bit resolution better, not because of dynamic range or headroom per se, but because it more closely parallels what we hear in the real world.

[UstadKhanAli]

 

Then why else talk about tape? Flutter?

 

 

Because tape is far more than saturation (distortion).

 

Lee Flier wrote:

 

 

Yes, but I think we need to make it more clear than we often do in these discussions that everything distorts sound. Moving through air distorts sound. Bouncing off walls distorts sound. Our ears distort sound. So whether we're talking about analog or digital or hearing sound live in a room, we can't possibly be talking about anything other than whether the way it distorts is pleasing or not.

 

 

We should discuss the whole quality of tape, not just the saturation. And just to be clear, are we talking about people purposely saturating tape, or the quality of tape in general?

[MikeRivers]

 

I think you have it backwards. Pre-emphasis and de-emphasis were introduced first to reduce noise, and then the curves were standardized to guarantee consistency from one tape machine to another.However, why there are separate curves for NAB and CCIR/DIN is beyond me.

 

 

Yes, increasing the level at high frequencies where there's less energy was indeed intended to improve the signal-to-noise ratio. The standard thing about standards is that you can always make up a new one that you think is better. There's a Nagramaster EQ and the Ampex/MRL 15 ips EQ curve as well, though both are very rarely used but both of which have advantages over the "standard.". Somebody thought he had a better idea. That's how we get new standards.

 

As far as the difference between NAB and CCIR, CCIR uses the same high frequency curve as NAB, but the low end is flat. It worked better with the European head designs. One of the quirks during the 1/2" 8-track craze was that the Otari MX-5050-8 used NAB and the TASCAM 80-8 used CCIR, so moving a tape from one to the other resulted in a difference in the low end. The Otari didn't have a low frequency EQ adjustment and the TASCAM's EQ adjustment range was very limited so the users wouldn't screw it up too badly so you couldn't make it play an Otari tape correctly.

 

And then there's the polarity issue. European recorder manufacturers were wired so that pin 3 went positive with respect to pin 2 when a north-to-south transition crossed the head in the normal "forward" direction, and Ampex wired theirs the other way, hence the "Pin 2 hot" (European) vs. "Pin 2 Hot" (American) standards. Eventually Europe prevailed.

 

Phono EQ was eventually standardized for the LP, but there were a number of different curves for 78s and early LPs. I used to have a McIntosh preamp with a knob to select among about a dozen EQ curves, plus the manual had several more showing how to adjust the bass and treble controls to tweak the closest "standard" EQ.

 

There is no single optimum EQ curve because they're all a compromise between distortion and frequency response which varies with tape head, cutter head, or phono pickup design. So you pick the fault you prefer and enjoy it for its irregularity.

[MikeRivers]

 

So you can look at a graph and see that a digital recording has "less distortion" than a recording on tape, but there's no mention of whether we actually have less tolerance for even low levels of digital distortion because digital distortion sounds like crap. I know there have been some pretty decent studies about the types of distortion our ears can tolerate at pretty high levels vs. those we notice and find irritating even at low levels. But this is something that doesn't get mentioned often enough in these discussions.

 

 

And well it should get mentioned more often. We can tolerate, and even appreciate (what this discussion is really about) fairly high levels of distortion in music that's correlated with the musical source and in a musical way. What we notice more easily, and of which we're less tolerant, is when the distortion is uncorrelated with the music (or speech or really any other sound).

 

We don't mind 3% THD in a high quality amplifier if the distortion is primarily low order harmonics. But measure an amplifier with crossover distortion (a glitch where the waveform goes through zero) and while the THD number might be low it sounds distorted to us. The reason is that the distortion tends to be around a single frequency or a few harmonically related frequencies, none of which are related to the input frequency. If you look at a spectrum analysis rather than a voltage measurement with the input frequency filtered out, you can see where the distortion numbers come from.

 

For more reasons than are worth talking about here, digital systems CAN have levels of uncorrelated distortion that, while small when measured as THD, contain frequencies that don't sound good along with what we intend to hear.

 

Unfortunately, most of the people who buy this stuff feel good about reading spec sheets and it's rare that you'll find a spec sheet for a guitar amplifier or a D/A converter that has a spectrum plot.

[MikeRivers]

We should discuss the whole quality of tape, not just the saturation. And just to be clear, are we talking about people purposely saturating tape, or the quality of tape in general?

 

I thought we were talking about tubes here.

[Gabriel E.]

This is a 1978 model, the rare 135 watt version that was only made for 2-3 years and is supposedly favored by pedal steel players. It was defective right out of the factory in that wires were never installed to and from the preamp tube on the Normal channel. After many years of use the plate and grid resistors (which are soldered across the power tube sockets) became crispy fried and the filament wires also needed replaced. The neon photo resistor 'roach' in the tremolo circuit also needed to be replaced. The amp is restored and works great these days but it's been many years since I've taken it to a gig. I never had a travel case for it and it's too fragile to travel without one besides which it's boat anchor heavy and way too loud for all but outdoor gigs. For gigging it was replaced 12 or 13 years ago by a Tech 21 Trademark 60 solid state MOSFET amp which was subsequently replaced three years ago by going direct to PA with a Digitech GSP1101. God bless those who still love their old tube amps but I just play local clubs and bars rather than the concert halls and large venues that all you guys play. I don't have any roadies or I'd be right there with ya! To get through my occasional gigs I have to keep it simple, lightweight and reliable so I'm preferring more and more modern gear.

 

 

That's a Twin with an ultra-linear power supply. They're rare for a reason - a perfect example of CBS-era Fender engineering gone awry. When players were clamoring for amps with nice overdrive, Fender redesigned one of their flagship amps to stay clean at any volume. They also made it louder and heavier. A case of tubes doing what they do worst. A nice ss amp like the ones you mentioned will give you great clean sounds without the hassle of that amp. No wonder you prefer them.

[ggm1960]

A nice ss amp like the ones you mentioned will give you great clean sounds without the hassle of that amp. No wonder you prefer them.

For the record I never wrote that I preferred the sound of SS amps over tube amps although it is true that I generally prefer a clean sound at gigs these days. I wrote that I like playing through transistor radios . Of course if I did prefer SS amps I'd be in good company; so did John Fogerty's 60's era hit machine CCR, early Lynyrd Skynyrd endorsed by SS amp upstart Peavey and even......gasp!......occasionally...........The Beatles!

[Raymar]

My brother just bought an all tube Bogner Alchemist 1x12 20/40 watt. I tried it out with his Strat, not impressed, in fact I was quite disappointed with the overdrive channel, although the clean channel was nice and Fendery twangy. I ran my POD 2.0 in front of it and the thing came to life!