How to Bias Analog Tape Recorders

The following is a description of how to set correct bias current, for any combination of speed, tape type, record head gap and bias frequency. You do not need to know anything about any of these things to set optimum bias, nor do you need any special equipment.

My father, David E. Blackmer, taught me this trick in the ’70s using a subsonic, around 5Hz and listening to the noise pump. When I was doing 3M field service for Aengus Engineering in the 70s I would set bias listening through headphones to the output of the distortion analyzer bridge of my Ferrograph Recorder Test Set. In the mid 80’s, I saw John Stephens set up Roy Thomas Baker’s 40 track at The Cars studio, SynchroSound, in a similar way, listening to the pumping. I have refined the test some so that, with equipment normally on hand in a studio, one may also listen for distortion as well and not accidentally blow out the monitors.

High frequency bias does three things, it reduces odd order crossover distortion by keeping the highly non-linear middle of the applied vs retained magnetism graph busy, it sets how deep into the oxide layer the recorded signal penetrates and it gets you in away from the rocky surface of the tape. (You can actually see “rocks” on the tape surface as little dimples with dots in the middle.) One wants to involve as many of the oxide particles as possible in the linear retention range. Going a little past this bias point (the peak) causes erasure of the high frequencies, but gets you in past the surface. (This erasure effect is what you are actually watching when you overbias by so many db at 10kHz.) The signal gets thin and spitty sounding in the underbiased condition because there are fewer of the oxide particles involved in retaining the signal. Also the roughness of the tape surface as it bounces the tape on and off of the record head is more strongly coded into the asparity.



This step gets done after setting tape path, repro gain and HF EQ, before setting the record gain and record EQ.

Set a good clean sine oscillator (not a function generator) to about 30Hz at OdbVU, feed this to channel 1 of the tape machine and set the machine to input mode. From the output of tape channel 1 feed a channel of parametric EQ. Patch from the EQ out to the in of a second channel of EQ, Patch out from this 2nd EQ into a limiter and out of the limiter into a monitor channel. Now, with the monitor gain set low and watching the limiter’s input meter, engage the first EQ and dial in a very narrow notch, reducing the 30Hz to a minimum. Bypass the 1st EQ, engage the 2nd one and repeat the process. Now engage both; the 30Hz should pretty much go away. Now you will be listening to all of the artifacts, distortion plus noise, of the recording process plus any residual distortion that your oscillator produces. In effect you have created a distortion analyzer, with your ears, brain and opinions as the detector. The odd order distortion products 90, 150, 210 Hz are very easy to hear.

It is important that the oscillator be a good clean sine wave to start. If necessary, you can reduce an oscillator’s distortion by first feeding it through another parametric EQ set to sharply bandpass the fundamental and reject the mids and highs. You can even notch out the worst harmonic, probably the 3rd at 90Hz. You will need to reduce the oscillator output level the same amount as the EQ boosts the fundamental.

Once you set your EQs do not change the oscillator frequency.

At this point disengage the EQs and set the limiter so that it’s attack and release are fast and it triggers at some safe level well below 0dbVU. Set the compression ratio as high as it goes. Re-engage the EQs and put the tape machine in record, while monitoring the reproduced output of channel 1. Watching the gain reduction meter, make sure that the EQed signal is not triggering the limiter, which is only there to protect you during rewind or accidental change of oscillator frequency. If the EQed signal triggers the limiter it will give you wrong ideas about where the minimums are. I generally put some broadband boost on the mids and highs of the listen EQs to make the distortion and spitting more audible. Set the monitor volume so you can hear the hiss and distortion clearly. When you rewind the tape the limiter should trigger and the monitors not get loud.

Adjust the bias pot of channel one, watching the channel 1 repro meter for the center of the peak, which may be quite broad. In the under current direction the signal will get spitty, noisy and thin sounding. In the overcurrent direction the high frequency componants will drop. This is how you know which way is which. There have been Japanese machines made where counter-clockwise was more bias.current.

While increasing the bias current slightly past the peak, listen to the quality of the noise and distortion and set the bias where these are least objectionable. You may notice in some situations that the minimums for noise and distortion may occur at slightly different bias settings. Try for best overall compromise. The reproduced 30Hz signal should only reduce a little bit, on the order of 1/10 to 1/4db. You can then go to 10kHz and find out how much overbiased you are there. You can then write this 10k overbias number on the tape box and take sheet.

Now step the pair of patch cables through each of the tape machine’s channels and adjust in turn.

Using this method on a channel by channel basis scientifically proves to your direct experience that the bias is doing exactly what it is intended to do, reducing asparity noise and distortion. The Xdb over at 10k method can’t do that & involves taking a lot on faith if you are unsure of the tape type. As an example Scotch 250 at 15ips liked about 1-1.5db over at 10K, while Ampex 456 liked 2-3db, or more, over.


If there are low frequency “rocks” in the noise that won’t go away then the record or play head may be magnitized, the bias or erase signal may be distorted (bad transistor or resonators out of tune), the record amp bias trap may be out of tune (which can load and distort the bias waveform), or there may be actual DC getting to the head (leaky cap, conductive contamination on printed circuit board). Anything that produces even order distortion (one side of the wave affected differently than the other) on the bias or erase waveform effectively adds DC.

Here is an interesting effect relating to tape types vs optimum bias current vs audio signal level. Many tapes, like the old Scotch 250, as you raise and lower the level of the audio signal, have their optimum bias points stacked straight up and down. Other tapes, such as Ampex, now Quantegy, 456, have different optimum bias currents for different levels. Try changing the level of the 30Hz tone while aligning 456 and see how the noise and distortion minimums vs bias current move. It is my belief that this tape type produces more of the sonic qualities that we generally identify as “tape”. If you tune the bias for a 456 type tape at a low signal level, and then raise the level up near its headroom you will find the optimum bias point to have wandered so that it occurs at a different current. Whether you set your bias at a high or a low VU with this type of tape depends on where in level your signal spends most of its time and what you want it to end up sounding like. A subtle effect.

The analog tape recording process actually samples at two rates: twice the bias frequency (as every half cycle drops to the level where the signal is retained near the trailing edge of the record gap) and at the random rate of the asparity noise. Asparity noise is produced by the statistical distribution of oxide particles in the record head gap vs time. This noise is different from the simple fixed thermal noise of an amplifier, which stays at one level, regardless of the level of the signal. If you put a tone through a tape recorder and watch the output on a spectrum analyzer, you will see that as the signal level is raised the noise rises up around it in a mountain, with peaks at the odd harmonics. The truth of the matter is that tape recording is shaped noise. It is a non-linear transfer process that produces noise-like sidebands for every frequency in a complex signal. This is a major componant of the tape sound.

Another distinctly aspect of tape is the approach to saturation. The tape headroom sound and the mechanism that produces it are very different from the way an amplifier overloads. As level is increased, the tape, again by a noise-like process of statistical distribution, gradually runs out of magnetic domains that can hold each higher energy level. Also, at high flux levels the present signal is self-erasing the just previous one. These effects together produce a gentle “S” shaped onset of distortion, the famous tape compression.

There are also time delay dispersion effects in the magnetic recording process that do some rather odd things to the time relation between low and high frequencies. As an experiment, fly an impulsive signal, like a kick drum with a good sharp high frequency edge, a strong deep bass thud and a clean, well defined start time, from your Pro-Tools (or whatever) editor into an analog tape recorder. Now fly the reproduced tape signal back into Pro-Tools and try to line it up with the original. Something funny happens; when the two high frequency energy peaks line up there is energy appearing well before the main spike in the tape processed signal.

Think about that.

It’s just this sort of behaviour that makes some swear by 15 ips tape for drum basics, or like to fly digital submixes to tape to liven them up.

Also listen to the background voices in Queen’s “Bohemian Rapsody”. There is a sparkling breathy cloud around them. I believe that this is Roy Thomas Baker’s Stephens 40 track at work. The Stephens has so much headroom in the electronics that it is virtually impossible to make it clip. At Synchrosound I witnessed guitar solos where the needle went to the right peg of the meter and just stayed there, that sounded perfectly amazing. All of the distortion was due to that very narrow track of oxide, pretty much a pencil line. You can weld with the electronics.

As an aside, John French, of JRF Magnetics, who worked on building the original heads, told me that the 40 track head was almost impossible to make, and that they had to make several for each one that worked. Track height and azimuth alignment is so critical that generally one mixes off of the record (sync) head to make things come into focus. The repro head is mainly used for alignment of the bias.

The Stephens transport is simply brilliant, no pressure roller, no capstan, no brakes. The reel motors, tachs and analog computation do everything amazingly smoothly. John Stephens is an electro-mechanical genius.

And now for something completely different…

Many years ago my father David told me of another, purely electronic, way to produce analog asparity by feeding band-shaped noise into the control port of a VCA, so that it’s gain is wandering around at a noise rate. (Of course noise is generally considered to be a Bad Thing in a gain control circuit and is specifically warned against in the literature of  That Corp., the present makers of the DBX-Blackmer VCA and RMS modules, & a great company.) Shaped noise sidebands appear as the signal “lights up” these gain changes. The noise envelope decays with the signal, as with tape.

Tape-like time delay dispersions can be realized with all pass filters, though there is the problem that the function is not a simple monotonic delay, but wanders back and forth in phase vs frequency space. Appropriate adjustable pre and de-emphasis filters are placed before and after the VCA to make it act yet more like a tape machine.

Additional instantaneous control of the gain of the VCA by the waveform itself through a full wave zero time constant detector, with some transfer function processing circuitry, provides a variable saturation law.

This yields a device that can sound more like tape than tape does, with adjustments for all parameters, that produces that breathy sparkle, gentle approach to saturation and more. Another Blackmer box…



Leave a Reply

Your email address will not be published. Required fields are marked *