Analog audio tape recorders - any experts here? Let's talk bias.

[cvanc]

Got a question that I've long wondered about.  What does the output signal do as you adjust bias from zero all the way up to "too much"?

Let's assume a clean mid-frequency sine wave is what we are recording.  (400Hz, 1KHz, something like this.)

Does the sine wave exist at the output across the entire bias adjustment range?  Or does the sine wave literally not exist at one or both ends of the adjustment range?

As you sweep the bias level what does the output look like on a scope/measure like on a distortion analyzer?  Does too-low bias look/measure differently than too-high bias?

Hope this all makes sense, thanks for any insight.

[bob91343]

Early cheap tape machines would occasionally use dc bias.  But the vast majority of units superimpose a high frequency, maybe 40 kHz, on the recording signal going to the record head.

The purpose is to average out the basic magnetic nonlinearities.  You adjust it to get the best performance or, failing to make a big deal of testing, you adjust it to a level that the manufacturer deems is best.

It's not critical, and once adjusted will likely not need readjusting.  If you use a different tape formulation you may need to make a small readjustment.  If the service manual for the machine doesn't help, you probably should try to minimize distortion, or maximize output level.  But if you do that, you may compromise frequency response.

The simplest way is with a machine that plays back as it records so you don't have to keep rewinding and retesting.  Cheaper machines have only one head, doubling as record and playback.

[PKTKS]

I am by no means expert ...

but I dealt with tape recorders quite a lot in 70/80/90s

Your question greatly depends on the fab. The bias in general workaround some
specific issue with the HEAD itself  like optimizing GAP reluctance
and providing proper magnetic HEAD level (field polarization)
(or proper linear level of the head core)

In late 90 the VHS systems pushed the BIAS to full FM modulated carrier
so that VHS heads would have proper magnetic levels

As said above some heads require a carrier to achieve that optimum
levels - so if you change the bias you should care about having maximum
output on the HEAD itself - the bias is there to do that.

As far as i know - as things radically changed over 80/90/00
and as industry creeps were postponing digital recording ..
to the maximum

till the age of the PCs just ditch them all...

Paul

[Tom45]

The response curve of the combination of recording head and tape magnetic layer isn't linear. The purpose of the bias is to raise the level of the magnetic field to get into a more linear region of the combined response curve.

So what do you see with different bias levels?

With good design, the bias should be set to the most linear region. So more or less bias would move into a less linear region with a resulting increase in distortion. At the extreme there may be an onset of clipping.

That is the theory. I haven't ever experimented with this, nor have I seen reports by any one else having experimented with different bias levels.

[bob91343]

The problem lies in the measurement.  It's hard to see small changes in response and distortion in a tape recorder due to other anomalies.  Tape was never a great medium, although companies like Ampex refined it very nicely.  (I still have my Ampex 400A and it works like new).

Rather than reinvent the wheel, my take on it is to set the bias as the manufacturer suggests.  For some, it's low distortion (hard to measure), for others it's best frequency response (hard to measure), and for some it's maximum output (or sometimes extra bias to reduce output by 1 dB or so).  Finally, the manufacturer has gone through all this, presumably, and gives its recommendations.

By this time in history, there are few recording heads still in use that haven't been worn down to the point of poor performance.  Still, it may pay to make some careful tests.

Having said all this, and at the risk of being trampled on, I submit that more modern recording techniques are superior and one probably shouldn't be using tape if the goal is high quality recording.  In my experience, the worst problem is flutter, with dropouts and broken tape also annoying.

Yet again, my Ampex sounds wonderful.  Except to verify its excellent performance, I haven't used it in many years.  It's bulky and only records in mono.  Uses several 12SJ7 tubes; I was lucky enough to obtain a few spares, although it still has all original tubes working fine.

[mikerj]

I've noticed a resurgence of interest in magnetic cassette tapes, but I don't understand why.  Is it younger people who never grew up with it's many downsides and regard it as a novelty, or the older generation looking back with rose tinted spectacles?

[bsdphk]

A good source of information is the patents etc. around Bang&Olufsons "HX professional" which they later sold to Dolby.

The BeoCord 9000 autocalibrates bias (and other recording parameters) in a surprisingly simple way, so I think as a first order model, the bias is there to flatten the frequency response curve.

[GLouie]

I am also not an expert, but offer my experiences.

Your questions are too broad to give specific answers. Tape machine manufacturers vary wildly in the amount of bias adjustment they offer. The tape in question, whether a consumer or studio machine, the signal level - will all interact.

Very generally, assuming a normal level (-10 to 0 dB) you should get a lower level, more distorted audio signal recorded with little bias, then as you increase bias, you get a higher higher output level and less distortion. As you keep increasing bias, the signal level will go down again and you will get more distortion. Many manufacturers recommended "overbiasing by X dB" meaning to start with low bias, increase to the audio peak output, then continue to increase bias until the audio level drops by X dB (typically 1-3 dB). For many years, this was close to the sweet spot of sensitivity, distortion, and frequency response. It also assumed some pretty high distortion (3%) at modest levels over 0dB. My feeling is that the distortion would be a bit different from underbiasing vs. overbiasing. There is a lot of hysteresis when trying to magnetize tape with audio; it's different with signal level, type of tape, and whether a studio machine or a cassette deck.

As someone mentioned, it is a bit harder to measure all of the important audio parameters while adjusting bias, but IMO, you are much better off doing so. Bias changes all sorts of things to the audio signal at different levels and over the audio signal range. A little tweek of the bias might improve the THD, but change the frequency response at 10kHz (requiring some HF EQ) and the modulation noise at 30Hz. My feeling is that a bias setting is the best balance of all factors to give optimal performance.

It took maybe 50 years to get analog tape recorders to the point they finally reached, before digital took over. The Audio Engineering Society e-library has a wealth of documents about analog tape, by the people who developed the industry. All docs are free to AES members, probably too expensive to randomly look at by non-members.

[floobydust]

There's an art to setting bias, and the physics behind it are complicated. I've worked on a couple big 2" tape 24-track machines in recording studios.

Ampex 1967 Biasing in Magnetic Tape Recording is a good discussion.

Optimum bias amplitude depends on tape formulation and speed, record head gap.
It affects distortion, noise and freq. response. It's difficult because you can trade-off noise for distortion or high frequency response - all three parameters interact.
Recording engineers can optimize bias on certain tracks, for high freq. response on the drum kit, or noise for vocals etc. So you will find all kinds of personal rules and preferences for setting bias. Pros will use an audio analyzer and check THD verses frequency to determine optimum bias, checking noise as well.

With a DC bias, such as a permanent magnet like Sony uses on their cheap cassette products, you end up operating on 1/2 the B-H curve and the result is high distortion with low sensitivity. It's better than no bias at all, but you can still see a distorted sine-wave.

For AC bias, basically you set bias amplitude to give maximum output (highest tape sensitivity) at 1kHz.
This is not the best for low noise though and popular method is to apply slightly too much bias (overbias) to the tape.
With the overbias method, you set bias for max. output level with a 1kHz tone, then using a 10kHz tone dial bias up until output drops a few dB, depending on the head gap and tape formula. Ampex 456 was 3-4dB. Then you have to check and adjust freq. response, another 48 adjustments 

Too much bias will give you tape noise, a poor S/N ratio.
Mag tape noise is hiss plus low frequency "bias rocks" as they are called in the recording industry. It's a bumping, crackling noise sounding like rocks being hit together.

The "bias rocks method" is to record a low freq. sine wave 20-50Hz, and on playback use a HPF and listen only to the bias rocks and adjust bias for minimum noise.

[rhb]

One minor comment.   The purpose of the AC bias is to reduce the noise.  So if you want to adjust the bias level, short the input and measure noise after recording.

[schmitt trigger]

In the late 70s, I worked as an assistant engineer at a pair of recording studios. My job was equipment maintenance, like repairing cables, cleaning the mixing console slide pots and maintaining the tape machines. The bias alignment was exactly as Floobydust mentioned.

The only extra steps were that the machine’s transport and heads had to be thoroughly cleaned and degaussed. Then a master tape with reference tones would be run to ensure that all the playback channels had the same gain. Then adjust the recording bias.

At the time, for PRO use you could use either BASF or 3M “Scotch” tapes, and you could run them at either 15 or 30 IPS, depending on how much the customer was willing to pay. For each setup the bias level had to be adjusted, and on a 24 track machine the process would take a while.
The studios owned 8, 16, and 24 track machines, again depending on customer’s willingness to pay. Thus I spent a good deal of time setting up the machines.

Did it matter? Subtle but certainly  audible to the principal engineer.
I know because the only time I was scolded by him was when I adjusted the bias for something different than the plan, and he noticed.

[cvanc]

Hey everybody, thanks so much for the meaty conversation.  Very interesting stuff.

Floobydust, that old Ampex paper is exactly the kind of theory I was looking for - thanks!

I had not heard of this "bias rocks" noise before.  Do any of you know if there's a recording of it online somewhere?  I'm curious to hear what it sounds like.