tape emulation

[mlaargh]

Your analog tape machine or what ever you use is prolly full of 8 cent transistors. I bet the resistors biasing said transistors only cost about 1/3 of a cent. Cost is only sometimes an argument.

[soundguy]

cost is not the argument, its design.

You can use 8 cent transistors to build a circuit that will be insanely awesome.

when you buy an 8 cent dip opamp, you are buying an 8 cent circuit that you now have to design around. If you dont like the sound of the dip package, you are stuck designing around it. Thats a really great scenario to be stuck with.

The real point is there is no fucking reason to have an active buffer stage on a converter, let alone one that uses some off the shelf dip crap.

dave

[helstab]

Your analog tape machine or what ever you use is prolly full of 8 cent transistors. I bet the resistors biasing said transistors only cost about 1/3 of a cent. Cost is only sometimes an argument.

great forst post.

[mlaargh]

Maybe I'm not as well versed as I could be in digital recording equipment, but why would one use a 2$ op-amp for a job that an 8 cent one can do? It's the same as replacing all your 8 cent transistors with over spec'ed, more expensive ones... It seems that the root of your comment is regarding the nessesity of an active buffer... I haven't seen a schematic of them design in question and am too lazy to UTFSF, but in my experience, people are in the equipment manufacturing market to make money - thus would avoid unnessesary stages (more components = more money) in thier designs. Is this "active buffer" also functioning as a D/A converter or a LPF? As stated above I dunno... I do know that when run in spec, you can get very good audio reproduction out of a very cheap op-amp. Is it the LM741 we speak of? I don't know the THD ratings for the audio spectrum off the top of my head, but the CMRR seems as high as any similar configuration you could build with an expensive transistor array.

[soundguy]

even an expensive audio devices dip package is way less money than a transformer. There are lundahl transformers with amorphous core material that sound probably more transparent than the slewing you'll get from ANY opamp, discrete or not.

I think you are not looking at the comparison properly. Buying a more expensive transistor to replace a cheap transistor is a non issue to a large degree. Replacing cheap transistors with expensive ones is folley. You can use an inexpensive device as part of a well designed amplifier to sound awesome. When you buy a prepackaged opamp, you are stuck with what you get, which may not be what you want out of an active stage. If you use discrete components you can design that active stage to be whatever you want it to be as opposed to whatever chip was available from the parts distributor. Not everyone likes the sound of a chip with 100 transistors in it. In the case of a converter however, there is no reason at all to be using an active buffer in the first place, a transformer will do the job just fine and dandy. Just more evidence that "pro audio" still isnt taking digital recording seriously, not only cant you get a converter with a passive buffer, but so far as I know, with all the boxes that have active buffers, they all use off the shelf opamps. Please correct me if Im wrong about that. It is kinda frustrating to put a bunch of care into a recording only to have all your work go through a cheap buffer circuit from a chip datasheet... All the specs and THD ratings and slew figures and all that shit dont add up to much for me, its what the things sound like. AD opamps look like they'd be the most awesome thing on paper but none of those specs clue you in to what the sound stage of the amplifier is like which is really the only thing I think anyone should really be concerned with. Neve amps spec for shit ona distortion analyzer, doesnt seem to stop people from fieinding like mad for them.

dave

[Family Hoof]

Uh oh... don't really like where this is going with the IC arguements, but I had to step in and say I very much agree with Dave concerning the record/repro electronics having a huge effect. Why do you think different tape machines sound so different when you're using the same media at the same settings? (and some much more desireable than others) I think the mic pre analogy is a good one here, because just think about the similarities between a tape head and dynamic microphone. They are both passive transducers which output a weak signal needing tons of carefully tailored amplification in order to reproduce it correctly. Not to mention, all that phase shift and distortion from the pre and de-emphasis EQ circutis, in addition to the type of EQ you choose (IEC or NAB) are also going to be huge cotributing factors to the "analog sound". I would take greater care in these stages if building the aforementioned device.

Now getting back to the original topic - the part about magnetic transfer affecting the signal. Let's think for a minute about what exactly is happening to the audio during record and reproduction on magnetic tape. It is an alternating current inducing magnetic analog of itself - whose domains are being aligned by a high frequency bias signal - this magnetism is then inducing an alternating current in the repro head and the audio may be heard once again. Okay, now what would you say are the key sonic coloring elements in this process? I would say the composition of the magnetic materials and the effect of the HF bias are most dominant. And how can we exploit them? In addition to overall design and choosing components wisely, we can experiment with saturation and under/overbias. If you ask me, two tape heads face to face is an awful lot like a transformer. Were I to design such a circuit, I'd start experimenting there... but I'm not interested in doing a box like this so we'll leave it at that and let a more qualified designer take over.

[Family Hoof]

thus would avoid unnessesary stages (more components = more money) in thier designs. Is this "active buffer" also functioning as a D/A converter or a LPF?

In the case of a converter however, there is no reason at all to be using an active buffer in the first place, a transformer will do the job just fine and dandy.

This is a VERY interesting topic and deserves its own thread(s). I must credit soundguy with first bringing the issue to my intention. I hardly know anything practical about the design of PCM A/D converters, but I do know that when recording audio there are three things required from the analog input circuitry in order to let the digital part of the circuit do its job. These are 1) block DC, 2) anti-aliasing filter (steep low pass filter to get rid of all content above half the nyquist limit), 3) provide a degree of isolation from the outside world, aka buffering. In my opinion/experience the filter design is going to be by far the most crucial element and cause the largest (destructive) audible effects if done poorly. It's got to be as close to a brick wall fitler as we can possible get, and this is asking for a ton of trouble in the analog domain. As for DC blocking, it is basically inherent in the design of the filter/buffer amp so not much thought given to it... although, I do believe a DC blocking transformer might have the potential to sound better than a servo or simple electrolytic cap. Now we come to the aforementioned issue of buffering. Assuming you get your filter right this is going to be the next big thing to worry about. It is hard to come up with a definitive arguement for one approach or the other, but you cannot deny that it's totally rediculous no one has even offered the transformer option as of yet... or discrete amps as far as I know. Lundahl makes parts that will work for this. Someone who knows what he/she is doing needs to go and try it out.

We can talk about the reality of costs, profits, and big manufacturing all day, but that's just not very enlightening at this point, IMHO. I'd much rather talk design ideology and pure musical mojo (with regards to perceived sonic performance), but maybe that's just me. Then again (to make another Neve reference), the summary of Rupert's AES speech this year was "fuck specs, use your ears". Coincidentally, this coincides with the reintroduction of transformer i/o to his highly regarded designs.

[edited for clarity]

[mlaargh]

aN ITERESTING DISCUSSION, INDEED. OH SHIT CAPs lock was on. As it is sleepy time for me, I'll be back tomorrow evening. I think some interseting points were made, and hopefully have some interesting points of my own to submit [insert mom zing here].



Tomorow, chaps.


What's the deal with avitars here, anyway? I saw no interface for submitting/linking said image in my profile editor.

[alissa]

digital recording is abotu 25 years or so away from being ready for any kind of serious use.

uh oh.

maybe you should try cryogenically freezing your recording until then.

[wenzel.hellgren]

can someone please explain WTF an active buffer is?

[Family Hoof]

can someone please explain WTF an active buffer is?

Okay, let me give it a shot. I apologize in advance for making an ass of myself and confusing everyone.

First understand the difference between passive and active. We're basically talking about energy here. The audio signal has some energy of its own and if we want to add more energy the basic laws of physics say it will have to come from somewhere else, such as a DC power supply. An active circuit is contributing gain and therefore involves some sort of amplification - regeneration of the signal. A passive circuit contributes no gain, only loss, and outputs a "modified" version of the original signal.

Now a buffer is self-explanatory. Just ask yourself, what is a buffer in geography, in a computer, or anything else in life? It's some sort of go-between to ease the transition from one thing to another. Well, no big surprise a buffer in electronics is the same thing. Our audio signal has to go through many different circuit stages and processes in getting from point A to point B, and we want each stage to perform optimally, keeping our signal healthy. A buffer will help prevent one stage from being adversely affected by another stage downstream, or vice versa. As stated earlier, the sensitive PCM sampling process in A/D conversion needs not disturbance from somewher else in the signal chain if we want it to do a good job, so some sort of input buffer is needed.

An amplifier (an active device) makes an excellent buffer for two reasons. 1) Because we know that electric current always takes the path of least resistance (impedance), and backwards through an amplifier is rarely this path. 2) Because an amplifier can amplify - increasing signal strength in order to drive it through whatever it encounters ahead.

A transformer, albiet passive, also makes an excellent buffer, the primariy reason being that there is no physical connection between its input and output. Ever wondered why they're called "isolation transformers"? Galvanic isolation - look it up. Also, common mode rejection (cancelation of content that is common to both the + and - copies of the signal in a balanced connection, i.e. ground noise and EMI/RF interference) is often an important feature of a buffer,nd a well designed transformer will be better at this than most any amplifier on its own.

[joel hamilton]

Even if I wanted to debate the "awesomeness" of the support electronics in a Studer A827 (which i most certainly would not, if I was a tech i would own an A80, or an ampex 2" machine instead of my 827), the sound of the machine is recorgnizable, and very much a part of what i do. i prefer ITS sound to the sound of the PTHD converters I use.

If I had the space, I would have a bunch of racks of scully UTC based line amps from 3 of their 8trk machines, just to use as a tonal shaper before hitting my console.

xformers= good. If I only used protools, i would be building a box full of good xformers to interface with my console which has an xformer on every function anyway...

Funny that people try to make things without transformers and then look for "tape emulation."

Just use stuff with xformers in it. i wind up with SOMETHING patched on almost every channel of a mix because of this. If I am mixing out of HD on the console (my usual deal these days) having something on almost every channel, and a very insane high headroom console that is discrete, xformer based: good sounding to me.

Hooray.

Brad: I cant wait to see what you come up with!

[wenzel.hellgren]

The real point is there is no fucking reason to have an active buffer stage on a converter, let alone one that uses some off the shelf dip crap.

dave

If the anti-alias filter is in the active buffer stage, that is a pretty damn good reason for them. Got any transformers with anti-alias filters in them?

[soundguy]

do you actually want info or do you just want to argue? seems like the later.

many people have modified converters by removing the active buffers in cheap converters and replacing them with transformers. Its very easy to do. Unfortunately, if you dont want to rip apart your gear and modify it, there arent manufacturers that offer this type of product prepackaged.

It is *possible* to build a box the way I described, this isnt some kind of good sound fantasy Ive dreamed up because I have nothing better to do. Its a solution that comes to mind because everything on the market is made with the same cheap crutch because its cheap to do, not because there isnt an alternative. A converter should sound good, shouldnt it?

dave

[Family Hoof]

If the anti-alias filter is in the active buffer stage, that is a pretty damn good reason for them. Got any transformers with anti-alias filters in them?

No, the filter is a separate stage. Dave is simply talking about off the shelf IC op amps. You could use op amps as part of your filter design, but AFAIK there's no such thing as an op amp or transformer with the filter built in. That would be kinda retarded and totally goas against the argument that if you want to design something well yourself, you shouldn't start with limitations imposed by a monolithic integrated circuit designed by someone else. That is the heart of this debate we're having here.