Objective comparison of analogue and digital techniques

[Radio Wrangler]

There are a few differences between analogue and digital techniques which seem to be frequently overlooked.

Let's consider the good old LP record. I've got quite a bit of money invested in these things, and some of the stuff was never available in any other format. But if we wanted analogue to be the future and we wanted to make it even better (I'll use the same phrasing for the digital stuff, just to be fair) how would we do it?

The speed of the stylus in the groove, and the amount of deflection allowed, the resilience of vinyl and the masses involved and the slew rate of the cutter, plus the sizes of particles of muck all conspire to limit the upper frequency response and the signal to noise ratio. The lower frequency response comes down to the cartridge compliance and the effective cartridge/headshell/arm mass.

If we wanted to improve this system without shifting compromises between parameters, it is going to cost size. If we want more signal to noise ratio then we scale-up the whole record so the dust particles are smaller in proportion. If we want higher frequency extension, well, CD4 records proved you could get up to 45kHz and beyond. Unfortunately they also proved the HF content didn't last very long. Lasting HF would need a faster stylus-groove speed and thus a bigger record. Quadrupling the area of a record can give us 6dB more signal to noise, and will shift the HF content versus lifespan curve upwards by an octave.

Maybe something stiffer and less wearing than vinyl could give more HF extension that lasted without the size increase? But we've not made any progress on the size of dust in homes. Rumble is also going to limit how much LF you want.

How about magnetic tape? Size matters again. Tape speed versus head gap size sets the HF response. Tape speed versus head length shapes the LF response. Want more bandwidth? Then it costs proportionally in tape length for a given piece. Signal to noise ratio improves with width. There are possibilities with FM recording and helical scan, but they just move tradeoffs around. Want to store twice as much information? You'll need twice the volume.

Analogue IS analogue. What is stored is a literal scaled model of the original.

Digital techniques have two tricks up their sleeve:

1) Codes

This is the power of numbers. So fundamental, no-one mentions it, few people notice it. Let's say I want to represent a signal sample and it happens to be 0.123 Volts. Let's say the maximum we need to handle is 1v and we opted for 1mv resolution. That's a thousand possible different voltages and that will take 10 bits to represent it.

Looking at the doubling I mentioned in considering LPs, say I doubled the size of a CD and doubled the number of bits stored on it, but kept the sampling rate the same, I could have 20 bits for each sample and my resolution would improve by a lot more than 2. It would improve by 1000. It would improve disproportionately.

This is what allows us to write the number 1,000,000 and not have to write '1' a million times. Digital methods can exploit weighting, while one molecule of vinyl, or one magnetic domain has just the same importance as another.

Si if I want to improve a CD by 6dB in dynamic range, I need to increase it from 16bit samples to 17 bit. so the size goes up by 6.25% which is better than 100%.

There is a price for this trick. Some bits are worth more than others, so an error in a heavyweight bit will make a much, much bigger bang than the pop from the same error on a less important bit.

If I want to do the same as I did to the LP and give it an extra octave, the number of samples doubles and the area doubles. combine this with the 6.25% for that extra bit, and the CD size goes up by 112.5% compared to 300% for the LP (a quadrupling of area)

It gets a bit better because the CD pits can now be made smaller and much more data put on DVD or Bluray discs.

2) More coding

Errors will always occur, and numerical data is much more affected than proportional analogue data because of weighting. However, there are mathematical techniques which allow errors up to some proportion to be detected and fixed if some extra data is 'wasted' in redundancy. This error correction is surprisingly good. A moderate increase in data size allows a lot of errors to be corrected. The likelihood of a very large number of errors happening is very remote, so errors can usually all be fixed unless something big has gone wrong... big physical damage that would have ruined an LP anyway.

This error fixing fixes the risks associated with bit weighting, but it also makes digital recordings able to be copied for an indefinite number of generations without progressive degradation. This means that CDs, hard drives or whatever can give up perfect data provided they are read and copied before they have degraded too much for error correction to work.

The CD standard was set a long time ago and they packed as much as they could onto a chosen size of disc. 44.1 kilosamples/second and 16bits can definitely be increased nowadays. But were they enough? are they still enough? Everyone has their own opinion.

The LP/microgroove standard was set an even longer time ago and has served us well.

I don't see anyone bringing out a new LP standard in the forseeable future.

CDs are morphing into computer files and downloads. The use of multiple standards is possible and happening. What we seem to be being swamped with are lossy compressed formats like MP3, but there is some non-lossy high resolution, high sample rate stuff.

I think the LP/45/78/Cassette/Open-reel are what they are, and it's a matter of personal taste whether these are good enough for each listener. Most of the best gear for using these formats has already been made.

It's also a personal matter whether you think the CD standard is enough or not. But CD is not all there is to digital, it's just one example, but I have a lot of money in a CD collection, and lots of it not available on CD.

Digital techniques have the potential for further progress, and with the shift to downloads, they get the opportunity as well. If only the ungodly don't make MP3 triumph over all.

I think most of the best CD players have also already been made, if you think of a machine with a slot or tray, and outputs ready to go to an amplifier. So whether you like analogue only, or whether you thing currently available digital stuff is good enough, then some of the best gear for both camps is out there waiting to be restored.

Me? I'll play it both ways... or all three.... or who expected the Spanish inquisition...

David

[sp10mk11]

Fair comment David and thanks for putting a rational case for each system.
Gary

[Radio Wrangler]

I forgot one sentence, If I doubled A CD in size it would change from 16 bits to 32 and so would increase its resolution by a factor of 65536 which is an awful lot. That 1000 to 1 ratio was just for that 10-bit example. Sorry.

Both systems have different things going for them.

Unfortunately mankind has a fixation with ranking and packing orders. When asked by someone whether the Land Rover or the Jeep is better off-road, I'm inclined to direct them to the John Deere catalogue

One thing about audio amazes me. There aren't any cases of "I can't hear any significant difference" or slight differences. They are all reported as being life changing and forcing people to have to listen to all of their record collection again.

I designed and built some equipment for myself. It is very definitely to a professional standard in looks as well as sound, but it pulls the rug away from under some people. Without any badges and without a recognisble house style, they don't know what they're supposed to say about the sound

David

[kalee20]

That's a good comparison of what is inherently possible with the two techniques.

You can also compare in other ways, such as, Digital is so versatile, but analogue is more fun. (And EMC emissions are much easier with analogue, too).

[MotorBikeLes]

As far as I am concerned, the biggest difference between Analogue and digital is that analogue is generally easier to repair if you know the theory of its various parts. digital is much more difficult (for me). Almost 25 years ago, I needed to get a job to keep a roof over my head and food in my belly. I was offered a poorly paid job repairing CD players, as none of the firm's techs knew how or wanted to know. I explained that I did not understand them, but would have a go. It took a while to understand much of what David wrote in a few sentences, but I eventually got there. I even became quite good at it, but have forgotten most of it today. It helped that I understood optical polarisation and stuff like that, and I even taught myself to remove and replace 80 leg chips WITHOUT DAMAGE using a combination of a Weller solder gun and a little 10w antex iron, plus a miniscule "walking stick" I made by bending a fine "beading" needle around a thin nichrome rod heated by electrical means.
I think a simple superhet radio is the easiest to understand, but some of the old stuff many of you love is so difficult to work on.
Decent quality PCBs make life much easier.
Les.

[Restoration73]

It is possible to replay vinyl (and others) using a laser, thereby eliminating any physical
contact with the groove. I believe turntables were produced for this purpose.

[Radio Wrangler]

I seriously considered trying that back in about 1980. At work we had a National Physical Laboratory waveguide-beyond-cutoff precision attenuator whose carriage position was measured by a laser interferometer.

Could I rig up something similar to play records?

Maybe. Then I calculated the amplitude of wiggle associated with a small signal at maximum frequency, and then compared it to the wavelength of light I'd be using, and scared myself. A colleague also commented on the size of molecules in the record. Oooo - Errrr......

Later, when some people actually built these the signal to noise ratio was found worse than the classic stylus method. All records are dirty. The interferometer reads the dirt particles, the stylus bulldozes the dirt out of the way. They didn't catch on even though they would have been reassuringly expensive.

More recently, people have used high resolution scanners to photograph records and used software to turn the photos into sound. Do-able, but not competitive with styli.

For the meantime I have a B&O 4000 turntable and a Revox A77 which are absolute classics to handle the analogue world. My 22 year old Sony ES CD machine died a couple of years ago needing unavailable drawer mech parts, so it got replaced with a Sony 5300ES and that does the digital domain for me.

Oh, the new Sony is awful. It sounds superb, but when you put a disc in it spends a minute making inscrutable decisions before it will play anything, It puts you right off using it.

I've seen the prices the premium hifi brands ask for their audiophile streamers, and I think I'll handle the future by building a nice DAC and using a computer. That ought to cover almost any format.

David

[kalee20]

Originally Posted by Radio Wrangler

Maybe. Then I calculated the amplitude of wiggle associated with a small signal at maximum frequency, and then compared it to the wavelength of light I'd be using, and scared myself. A colleague also commented on the size of molecules in the record. Oooo - Errrr......

I seem to recall reading, that the diaphragm of a typical moving-coil microphone, when picking up a sound at the limit of its sensitivity, moves about a tenth of the diameter of a hydrogen atom.

[jamesperrett]

Originally Posted by Radio Wrangler

One thing about audio amazes me. There aren't any cases of "I can't hear any significant difference" or slight differences. They are all reported as being life changing and forcing people to have to listen to all of their record collection again.

I can think of one fairly recent example where various microphone amplifiers were blind tested while operating well within their performance envelope and the results were completely inconclusive. The article is at

http://www.soundonsound.com/sos/oct1...es/preamps.htm

People seem to like to be told that there are huge difference between gear - I'll bet there's something psychologically satisfying in thinking that you made the right purchase based on sound quality.

[G6Tanuki]

Originally Posted by kalee20

I seem to recall reading, that the diaphragm of a typical moving-coil microphone, when picking up a sound at the limit of its sensitivity, moves about a tenth of the diameter of a hydrogen atom.

From memory of a neurology lecture some 35 years ago, the generic mammalian eardrum is likewise able to 'hear' down to the thermally-excited background hiss at normal temperatures.

Fortunately, we also come equipped with a few hundred millennia of well-evolved signal-processing hardware that serves to extract the signal from the noise! I'm reminded of the time i listened to a seriously-ancient radio-amateur working in a CW contest; he turned-off the HRO-5's BFO and instead used the incoming signal to beat against the background noise.

Sometimes you can use the 'threshold effect' of simple non-linear diode detectors working at low signal levels to your advantage.