Working with Tape Heads

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

adamasd

Well-known member
Joined
Jun 17, 2004
Messages
472
Location
Duluth MN
Hello

I am looking for info on working with these things. Any books or websites that are noteworthy?

I have a Multivox tape delay and it seems like a perfect excuse to learn more about designing for tape heads. It has stereo heads throughout with the Left and Right channels bridged for full track operation. Its stock electronics are pretty much budget orientated, not very good and noisy. I figure I can separate the Left and Right tracks and work on my own design for the Left and the leave the other side stock for the time being so I can still use it as a delay.

I am planning on designing the new electronics from the ground up, just as a learning experience. Probably go with discrete transistors so I can improve that area of my electronics knowledge as well. Any direction you can give me would be greatly appreciated.

Thanks,
adam
 
There's a pretty good discussion in chapter 28 of the third edition of the Handbook for Sound Engineers, ed. Ballou, ISBN 0 240 80454 6.

There is also some material IIRC about tape head preamps in the National Semiconductor Audio Handbook.
 
They usually want high input resistance (playback) and high output resistance (record), the higher the better.
Also, recording heads need a high freq bias. I've experimented long ago with recording using pulse-width modulated by a sound signal high freq bias, good results.
I liked EF86 tube in playback preamp with 10 MOhm resistor in the 1'st grid (grounded cathode), however proper mounting is needed for low microphonics...
 
There's a pretty good discussion in chapter 28 of the third edition of the Handbook for Sound Engineers, ed. Ballou, ISBN 0 240 80454 6.

Been meaning to get that one for awhile now, will have to finally order a copy.

There is also some material IIRC about tape head preamps in the National Semiconductor Audio Handbook.

Must have missed that section, will look it up tonight.

They usually want high input resistance (playback)

Thats what I thought, the Multivox uses bipolar input opamps here, I was thinking JFET input opamps seemed more appropriate, but that really was just a hunch.

I've experimented long ago with recording using pulse-width modulated by a sound signal high freq bias, good results.

Not completely sure what you mean here. Do you mean a pulse being modulated by say a high frequency sine wave used as the bias? I guess "sound signal high freq bias" throws me.

I liked EF86 tube in playback preamp with 10 MOhm resistor in the 1'st grid (grounded cathode), however proper mounting is needed for low microphonics...

I am probably going to stay away from tubes for this, but I have thought about it. The case is big enough and I am much more comfortable with tubes, but it is about time I get this transistor stuff down better.

thanks,
adam
 
I'm not sure what you're looking to do but your performance will likely be limited by tape media and head designs, assuming you provide them with proper biases and such.

If you have a schematic and it's a low cost design you can assume there are no spare parts in the design. Look at the schematic and when you are comfortable you understand what "every" thing is there for, upgrade individual circuit blocks.

JE
 
I'm not sure what you're looking to do but your performance will likely be limited by tape media and head designs, assuming you provide them with proper biases and such.

Maybe nothing, I will find out when I am done.

If you have a schematic and it's a low cost design you can assume there are no spare parts in the design. Look at the schematic and when you are comfortable you understand what "every" thing is there for, upgrade individual circuit blocks.

Thats the plan. I will be in the staring at the schematic stage for awhile yet, and have already been there for abit. It is what prompted this thread and will probably prompt more specific ones like it in the near future. At this time, staring at the schematic just brings up alot of questions I can not answer without more knowledge of tape heads.
 
Whatever happened to the idea floating around a couple of decades ago of passing the record audio through a "cube rooter" (not sure how to do that outside of DSP but I'm guessing Analog Devices has a widget) and thus obviating the need for an applied HF bias.
M
 
I think I've seen pre-distortion in addition to HF AC bias but it was usually pretty crude approximations. Perhaps DSP could do that better, but why bother when they could print digital words to magnetic tape.

JR
 
IIRC, before digital recording really took off, one of the hot ideas was to record a high frequency, FM carrier on the tape and (of course) demodulate on playback. I thought it was a good idea when I read about it, but didn't get the chance to hear a demo first-hand
 
there was also the 'myth' of "contour biasing"...

A scam artist or two tried to claim that you could get a compact cassette to exceed the noise performance of a CD... :?

...yeah...

Keith
 
[quote author="adamasd"]

I've experimented long ago with recording using pulse-width modulated by a sound signal high freq bias, good results.

Not completely sure what you mean here. Do you mean a pulse being modulated by say a high frequency sine wave used as the bias? I guess "sound signal high freq bias" throws me.

[/quote]

I generated linear triangles of a bias frequency and applied it to one input of a comparator, to another input of the same comparator I applied the audio signal I recorded. On comparator's output I had pulse-width modulated square signal of bias frequency.


[quote author="NewYorkDave"]IIRC, before digital recording really took off, one of the hot ideas was to record a high frequency, FM carrier on the tape and (of course) demodulate on playback. I thought it was a good idea when I read about it, but didn't get the chance to hear a demo first-hand[/quote]

Tape noises and non-linearities is one problem; another problem is detonation because it is impossible to make the speed perfectly stable.
 
[quote author="Wavebourn"]

Tape noises and non-linearities is one problem; another problem is detonation because it is impossible to make the speed perfectly stable.[/quote]

I was getting detonation in my mustang after I put the supercharger on it, but stopped it with a special ignition that retarded the timing based on amount of boost present.

Wow and flutter is yet another problem in analog tape systems that can be eliminated with digital recording.

JR
 
[quote author="JohnRoberts"][quote author="Wavebourn"]

Tape noises and non-linearities is one problem; another problem is detonation because it is impossible to make the speed perfectly stable.[/quote]

I was getting detonation in my mustang after I put the supercharger on it, but stopped it with a special ignition that retarded the timing based on amount of boost present.
[/quote]

I'd rather put Nissan engine under the hood instead of attempting to do something with that crappy design. I suppose Ford engineers had an order from police to make such engine for teenagers so they can't accelerate faster than Crown Victoria. :cool:
(However, there is an other meaning of "acceleration" word, growing up faster.)

Wow and flutter is yet another problem in analog tape systems that can be eliminated with digital recording.

I like American terms such as "wow" and "flutter", they are so intuitively understandable, unlike "detonation" that needs understanding of "de" and "tone" parts. :thumb:
 
I generated linear triangles of a bias frequency and applied it to one input of a comparator, to another input of the same comparator I applied the audio signal I recorded. On comparator's output I had pulse-width modulated square signal of bias frequency.

Ok, that is alot clearer. It also cleared up a few other related questions I was trying to sort out in my head. Its all starting to make sense.

Found a couple books on the subject over at Peter Millett's place, no clue how useful they are yet.

http://www.pmillett.com/Books/intro_TapeWire.pdf

http://www.pmillett.com/Books/intro_read_recording.pdf
 
Dave notes:
IRC, before digital recording really took off, one of the hot ideas was to record a high frequency, FM carrier on the tape and (of course) demodulate on playback. I thought it was a good idea when I read about it, but didn't get the chance to hear a demo first-hand

Sure you have.

VHS HiFi.

used an fm modulated carrier on tape.
 
The magnetism on tape is proportional to record head current, not voltage. And our starting point is flat recording. So the basic record amp is a constant current source or an output impedance much higher than the head's inductive impedance at the top of the band. You could drive constant voltage and pre-compensate; that's usually the Hard Way.

The wound playback head is a differentiator. Your basic playback amp is an integrator.

Since it is hard and unnecessary to have infinite DC gain, a Spec tells you to use a pole around 50Hz.

Since there are a dozen reasons treble falls off, a Spec tells you to add a Zero, typically around 3KHz.

The playback head could be wound for any impedance. Inductance changes with turns, capacitance mostly doens't change. In audio heads, a low impedance gives a low voltage and amplifier noise voltage can be an issue. But a high impedance head will have a self-resonant L-C peak and drop. For very wide response, you use the lowest impedance which will overwhelm amplifier noise voltage. For economy design, you wind-up until you see the peak (or until winding costs rise), and shift it to compensate some other droop.

Same thinking applies to phono and guitar pickups, and input transformers. All windings work the same; differences are physical size (capacitance and wire cost), expected response, and cost of amplification.

These will almost never "require" FET impedances. And historically FET noise voltage has been higher than a good clean BJT. Most tape heads are wound to work fine with BJTs.

But with half-good design and semi-modern parts, tape noise will overwhelm all other noise sources. You always have a hiss when tape is stopped, you should have a lot more hiss when tape rolls. So don't bang your head on amplifier noise.

The playback head has an infinite null where gap length equals recorded wavelength. No cure. Your tape head designer picked the gap for the application. This is one reason we throw in a ~3KHz Zero. We may have another +/-3dB tweak in the last half-octave of the band.

At this point you should be able to play a Reference Tape perfectly well aside from LF bumps.

You can check head and preamp response by injecting a small signal across a 1 ohm resistor spliced into the head groundy lead. It should trace your NAB/IEC curve up to some reasonable top-of-band point.

There are many more losses; these are compensated in the record side. For 30ips you might just rise the top octave 6dB and be very happy. For 1.7ips this would give dull voice recording; cassette record channels have huge resonant peaking at the top of the band. Many decks don't fudge below the ~50Hz pole, but you can boost if you want, if you average-out the subsonic noise which ultimately limits linear tape.

You really want to upgrade an existing amp+heads set, NOT work totally from scratch. Even if you will replace all the electronics, note all the signal levels and impedances before you trash it.

For total from-scratch: set up your playback first. Play standard tape, adjust gain, check response and see if you screwed-up. Except for finding a gain which brings standard tape level to standard patchbay level, and verifying that amp-hiss is lower than tape-hiss, this should be trivial.

Record is tough with mystery heads. Inject a signal a decade away from the band ends. 400Hz is habitual for audio. At low levels it will be distorted, ignore that. Bring it up until the peaks flatten. Note the level. Trim the azimuth. Raise frequency a few octaves and re-trim azimuth.

Come back to 400Hz. Back-off a good chunk, 10 or 20dB. Add bias, at least twice your hoped-for HF band-limit. Up to a point, playback level increases (and that low-level kink goes away); after that, it erases signal faster than it averages-out the lo-level kink. Note this level. But now go an octave down from the HF limit (which will be a matter of judgement for a from-scratch design). Again increase bias for maximum playback output, and a little more to make signal fall a bit ("overbias"). The lowest-THD condition will be slightly overbiased, but the maximum bandwidth is with maximum output bias. You pick your compromise. It is easier if you want 15KHz response at 30ips than it you demand 20KHz response at 1.7ips. Oh, and high tones at slow speeds saturate the surface of the tape; we traditionally test at -20dB, but hot rock drum tracks exceed -20dB in the top octave, so again you compromise.

Having found a good bias, run frequency sweep and bump-up your top (and maybe sub-bass) for flat response. For any economic recording speed, this will need a resonant circuit with more than 6dB/oct slope (however such steep slopes are dubious for sound quality).
 
You can find my playback amp (tubes again :wink:) here: http://stiftsbogtrykkeriet.dk/~mcs/Tandberg/index.html

It works well with old Tandberg decks at least... And the input impedance is not 10M, but around 50k, so making something similar with BJTs should be no problem.

My recording amp has never made it beyond PSpice.

Best regards,

Mikkel C. Simonsen
 

Latest posts

Back
Top