Research IT Labratory

Contact telephone:
(+359) 32 659 757

Reel-to-Reel Project @LAB2104

   Engineers from the Laboratory of Embedded Systems (LAB2104) at Technical University – branch of Plovdiv,  started a new project for developing of contemporary Reel-to-Reel tape deck. The project includes partnership of some companies, like HWGroup (which have experience in development of smart MIDI mixers for Harman Kardon) and Technocity (one of the biggest new vinyls online shop in Bulgaria). The intention of the project was at the middle of 2015.


   The intention of this project is to build a technology for manufacturing of contemporary reel-to-reel tape player (tape deck). Why analog reel-to-reel tape player? Why not cassette deck? Maybe you know that the sound profile of analog tape sound is little different than the synthetic digital one. This fact is based on signal processing – the analog sound recording is real time process compared to the digital processing, where the sampling, quantization and dithering makes more jitter effects on sound. This is based also on digital filtering technology where old samples are summed to the current sample of the signal and this effect reduces the focus.

   Not long ago most of the analog magnetic tape manufacturers stopped their products, but now there are 2 or 3 active fabrics, which renew their manufacturing process. That is why it is reasonable to start manufacturing of new high-end tape player. Most of cost effective tapes are ¼ inch, so the better configuration of the tape player is for quarter inch. Also the 4 track players have significant reverse side crosstalk, so the use of 2 tracks configuration is the only one potential for the better sound quality. The goal is to use minimal signal path on record and playback stages on 2 tracks, ¼ inch tape. The potential signal to noise level floor without using noise reduction systems on NagraMaster time-constant is about 70dB. Using contemporary tapes like SM900, which can be recorded on +10db peak, and then the common S/N ratio will be about 80-82dB for pure analog source (without noise reduction system). There are no new manufacturing of good audio cassettes.

   Current production of audio cassettes is based on type I (normal tape, normal bias on 120 μS), so that kind of tape is not applicable for good final results. Good cassette tapes are double layer chrome ones, like TDK SA-X (type II) or metal position (type IV), but these cassettes are stopped from manufacturing long ago and there is no intention for resumption.

   Using precise mechanism for mechanics and microcontrollers for tape path will give ability to overcome the wow & flutter effects. The control of the machine is based on graphic TFT touch screen display, where all the current parameters will be listed. For example the current status of bias current, the EQ setting for recording amplifier, the EQ setting for the playback amplifier – NAB or IEC (switchable), the torques that can be read from the motors and etc. The final idea is to make a contemporary reel-to-reel tape reproduces, using state of the art components to get the maximum from the recording media.


   There are several methods to amplify the low signals from the playback audio head. Typical the signals from the head are between 0.8mV and 2mV. Thus every component after the head is very important for the signal quality. One of the good approaches is to bypass the first coupling capacitor. For example that kind of approach is used on most Revox tape players. The bad thing here is that the DC component of the bias current from the first transistor magnetizes the playback head and increases playback tape noise. There are some good bipolar transistors that can be used in the first stage of the amplifier. Both have significant current noise, which produces more noise voltage for higher source impedances. The typical impedance of the playback head is high enough, so the best solution is to be used good FET transistors in the first stage. Also the first stage puts the main signal amplification. The common used voltage feedback amplifiers have reduced frequency response compared to the current feedback ones. The goals in this project:

  • No input capacitor (direct DC signal from the head);
  • No input bias current by using low noise FET;
  • Direct time constant formation on the current feedback configuration;
  • Head loss compensation after the first amplification stage;
  • Cascode configuration to reduce the Miller effect;
  • All DC signal path from the head to the line-out (no capacitors in signal path);
  • Bipolar power supply of the first stage to eliminate current loops problems;


  Tape deck will support both speeds 19cm/s & 38cm/s. The heads are for half-track configuration (2 tracks not 4 tracks). Due different standards the machine will support IEC & NAB (70uS, 50uS, 35uS for 19cm/s and 50/35uS/25uS/13.5uS-NagraMaster for 38cm/s). The low frequency time constant will be switchable (3180uS / ∞). Some modern tapes like SM468 can be forced on 25uS or even NagraMaster EQ. Some old EE tapes can be played back for compatibility.

Standard, (ips) High freqyency EQ, μS Low frequency EQ, μS
IEC, 7 ½ 70
NAB, 7 ½ 50 3180
NAB, 7 ½” EE tapes 35 3180
IEC, 15″ 35
NAB, 15″ 50 3180
NagraMaster, 15” US 13.5
NagraMaster, 15” EU 13.5 3180
Broadcast, 15” 25


   Input impedance is important for the sound quality. Once it must be low to ensure low input thermal noise. But this reduces the down frequency response due low decoupling capacitors. The other requirement is the direct connection without decoupling capacitors. This design implements very high input impedance, direct signal connection without using electrolyte capacitors. Design uses input FET to reduce the input noise voltage and one follower with current mirror to reduce the distortions. The stage acts as impedance converter.


   The mechanism will use four direct drive 3-phase BLDC motors. Two of the motors will be used for the tape drive (take up motor and supply motor). Active electronic tension arms will support both motors. The main processor can read the tension of every lifter. The tension for FF/REW/PLAY will be selected from the control menu of the front panel to support different tape widths (for best performance on 50/35 or even 26 and 18 micron tapes). That way the machine will support standard play, long play, double play and triple play tapes. The main processor for the tape drive will control the tension in real-time. Also it will measure the rotating frequency and tape movement. This will protect tape from damage on fast forward and rewind. On play and stop the torque of the motors will be soft changed for best tape transport. Inverter blocks will drive the motors individually by command from the tape transport processor.

   Two impedance/idle rollers are placed between the tape guides on the capstans and the tension lifters. These impedance rollers provide tape speed stability and reduced wow & flutter. When the tape is fast-forwarded, the contact of the idle rollers will be removed from the tape.

   The other two motors will be used to drive the capstans. The main idea is to use closed loop tape between the two asymmetrical capstans. The speed of every capstan will be quartz stabilized by a microcontroller with error prediction algorithm. The main goal is to get possible low reading for wow & flutter (let say < 0.018% WRMS for 7 1/2ips). The diameter of the two capstans is different, so each flywheel has different rotation speed; thereby avoiding the aforesaid occurrence and stabilizing wow and flutter. As the double capstan system always create a constant and stable holdback tension between two capstans will not be affected by any external condition such as irregular take-up and supply torques, irregular loading of tape, undesirable mechanism vibration and etc., thus assuring the superior wow and flutter characteristics. This asymmetrical capstan method is used in some high performance cassette decks but newer used in reel-to-reel tape decks.


   The control of the deck will use touch screen resistive digitizer. All buttons and parameters will be graphically visible using high resolution TFT display. The reading of the meters, real-time counters, bias, level and EQ parameters will be listed on this touch screen display. Motorized lifter will be available to control the bias, level, EQ, balance and input level for every channel. A standard MIDI in/out interface is used for interconnection with external remote mixer. Ethernet connector will be used for MIDI over TCP/IP, so the control of the machine and all of the working parameters can be done remote using LAN and mobile Android application (working on Smart Phones, Tablets or PC, Notebooks, etc.).

   Level meters will operate as graphical lines in true RMS VU mode, also PEAK mode with or without peak hold function. One of the modes will support both VU and Peak visualization. The reading will be switchable referred to decibels on high range (-40db to +15db) and precise range (-10db to +12db) scale. Also the scale will support reading by flux values in nWb/m referred to ANSI or DIN method. Reference level of 0dB equals to flux of 320 nWb/m.


   Four point automatic or manual calibration is based in this project. The first thing in tape sense is to set the bias level for the left and right channel. Then the system sets the tape sensitivity for the left and right channel. Follows the process of setting the middle frequency compensation for 7 kHz and for the high frequency 15 kHz. Finally the system checks and tunes the tape sensitivity for the low frequency (400 Hz). All the parameters can be set independently in manual mode from the motorized console of the tape deck.  Some benefits of this approach:

  • Electronic volume control for left & right by photo resistor;
  • Electronic bias control for left and right by separate transformer;
  • Switched EQ by relay (16 steps for every channel, mid & high freq);
  • Switched time constants (70/50/35/25/13.5uS);
  • Recording phase compensation;



One Comment - Leave a Comment

Leave a comment