Building a Dundovic tape playback preamplifier

SeleniumFalcon Posts: 3,402
edited November 2022 in 2 Channel Audio
I've always wanted to build a magnetic tape playback circuit designed by Joe Dundovic, who was one of the founding members of the Nortronics tape head company. Here is a photo of Mr. Dundovic (in the center) handing the last tape head manufactured to John French who now runs JRF Magnetic Sciences, supplier of professional record and playback heads.

Post edited by SeleniumFalcon on


  • The first step was to convert the schematic drawing to a printed circuit board design. Here is the original schematic:


    It is a two stage amplifier that uses high gain low noise SN5089 NPN type transistors to provide the necessary gain and keep noise to a minimum. Capacitor C1 provides DC blocking to prevent the bias of Q1 from reaching the playback head and possibly magnetizing it. R2 and C2 form a very high frequency filter preventing stray high frequencies from being amplified by Q1. C3 and R5 couple high frequency negative feedback from collector of Q2 to emitter of Q 1 to shape the NAB EQ.
  • I prototyped the circuit so I could see if it actually worked.

  • I've made printed circuit boards (PCB) in the past myself, using a laborious process of laying down dots and connecting lines on a piece of transparent plastic. Then a photosensitive board material is exposed to bright light with the lined and dotted transparent plastic on top to prevent exposure. Finally the board is placed in a tank of acid to eat away all the copper except the parts still coated by the photosensitive pathways. Then hole drilling completes the task, messy and time consuming.
    These days things are easier, I contacted a person in the Ukraine who turned Joe's schematic into a Gerber file and even supplied a rendering of what the PCB would look like.

  • Then a company located in Hong Kong turned the Gerber file into 10 perfectly made PCBs for around $15.00 including shipping.

  • Clipdat
    Clipdat Posts: 12,489

    I always love your photos, so artsy and vivid. That BOKEH!

    You'll have to share your camera & lens setup. Sorry for the off topic post.
  • ken brydson
    ken brydson Posts: 8,632
    The clips got @mhardy6647 ' s attention fo sho....
  • pitdogg2
    pitdogg2 Posts: 24,359
    The clips got @mhardy6647 ' s attention fo sho....

    He's chopping wood as we speak!
  • Thanks, I try and make the photos as interesting as I can. I know interest in tape recorders is pretty limited on CP so I appreciate anybody who tunes in to my threads. I use a Nikon D750 camera with a 60mm lens and try and tell a bit of a story with setting things up on my work bench. Really the camera does all the work.
  • I love these threads. Looking forward to the next installment!
    SystemLuxman L-590AXII Integrated Amplifier|Sonus Faber Cremona Loudspeakers|PS Audio Directream Jr|Sansui TU-9900 Tuner|TEAC A-6100 RtR|Nakamichi RX-202 Cassette
  • mhardy6647
    mhardy6647 Posts: 32,783
    Thanks, I try and make the photos as interesting as I can. I know interest in tape recorders is pretty limited on CP so I appreciate anybody who tunes in to my threads. I use a Nikon D750 camera with a 60mm lens and try and tell a bit of a story with setting things up on my work bench. Really the camera does all the work.

    perhaps limited... but not vanishingly small. :)
  • tonyp063
    tonyp063 Posts: 1,037
    I can't speak for others, but I do enjoy your restoration threads.

    Reminds me how to do a good solder joint too. :smile:
    (yes. I know. Practice)
  • I decided that the best connection method for testing the circuit would be terminal posts, allowing for clip lead attachment. So I needed to drill out the necessary holes.

  • Once done I could use the Keystone terminal kit to punch the ends of the terminals to the underside of the boards.

  • The boards are ready for parts populating.

  • erniejade
    erniejade Posts: 6,267
    Nice work!!!! What a great project.
    Musical Fidelity Tri-Vista 300, Audioquest Thunderbird Zero Speaker Cable, Tyler Highland H2, Audioquest Thunderbird Interconnect, Innuos Zen MK3 W4S recovery, Revolution Audio Labs USB & Ethernet, Border Patrol SE-I, Audioquest Niagara 5000 & Thunder, Cullen Crossover II PC's.
  • It's important to find the correct orientation for capacitors in order to minimize sensitivity to noise fields. Since a capacitor is made by wrapping the conductor and dielectric in a roll the goal is to have the lead connection that is connected to the outermost layer attached to the low impedance point of the circuit.
    I find this by connecting the leads of the capacitor to an oscilloscope set at the most sensitive setting. Then hold the capacitor in my fingers creating an induced hum field, look at the level on the 'scope, reverse the leads and look at the change in the level. The capacitor lead that gives the lowest display is connected to the outer layer and I place a red dot.

  • Reversing the lead connections.

  • As you can see the label that is printed on the capacitor can't be relied upon for consistent orientation.

  • Since this is going to be a fairly high gain preamplifier it is important to place the transistors with the highest gain in position Q1 and also it is important to try and match all four of the transistors for gain. To do this I use a really helpful little device that determines the characteristics of the transistors and measures their Hfe or DC gain.

  • I bought 10 transistors and this is how they checked out.

  • I wanted to examine the performance of transistors made by Central Semiconductor and the same older version made by Motorola. The best way to do this is with a transistor curve tracer which shows how the device operates under dynamic conditions. This is the 2N5089 made by Central and the slight loop is caused by a small amount of collector to base capacitance. This causes a displacement current to add and subtract from the transistor's current creating the loop.


    The amount of capacitance can be roughly measured by paralleling small value capacitors between the base and the collector until the trace doubles in width. I did this with a few small value capacitors and found the amount to be around 200pF. This amount of looping can be especially detrimental in common emitter amplifier designs which the Dundovic design is.
  • Next I tried looking at the same model transistor made by the Motorola company, here is the trace:


    The amount of looping about the same, but the slight rounding of the top of the third step shows a lack of linearity when compared to the Central version.
  • My edition of Thomas Transistor Substitution showed that two other transistors could be considered as replacements, the European BC349C and the American 2N3904 also made by Motorola. I found some of the 2N3904 and this is the trace:


    This shows a bit less looping and a more ideal linearity stepping, although this transistor has considerably less gain than the 2N5089. I decided to use transistor mounting sockets on the circuit board to allow me to quickly swap transistors and see how the performance changes.
  • I had to wait a bit to get the transistor mounting sockets, but they arrived and after some creative pin bending I was able to get them installed. Here is a photo of the completed circuit board:


    So I don't make a parts mistake I make up a list of everything needed and check off when a part has been installed.
  • I made a small wooden bracket to support the high frequency trim resistor and the output volume control. Then I connected a signal source, added the DC power supply and connected the circuit board to my meter and began a frequency response check looking for the necessary NAB playback response. At first there was a problem, the circuit was behaving like a flat linear amplifier with no frequency shelving. This had me puzzled for a day, it was as though there was no feedback from Q2 to Q1 and the amount of gain was way too low. The circuit was working, just not as intended. At first I thought maybe I had not wired the 10k potentiometer correctly and I spent some time making sure I understood how the pot was supposed to be in the circuit. Then I started making measurements between Q2's collector and Q1's emitter and that's when I found the problem. When I inserted the transistors I used the orientation graphic on the circuit board to get the emitter, base and collector correct.

  • The error is that this orientation is backwards, it places the emitter in the collector's position and vice versa. Once the transistors were flipped around all was well. In fact the NAB accuracy seems to be really on the mark and the adjustable high frequency response should be great to calibrate to the deck's characteristics.

  • JayCee
    JayCee Posts: 1,472

    So I don't make a parts mistake I make up a list of everything needed and check off when a part has been installed.

    I've made many check boxes, but, 3.........D........check boxes? Your attention to detail is just as interesting as your projects. B) You must be the long lost twin of Doc Brown. Just as talented but polar opposites.

    Speakers: Polk1.2tl's (Uber Mods) Pre/Amp/DAC: PS Audio BHK Signature & 250, DirectStream Cables/IC's: MIT S1Bi-Wire/S1 Balanced +Avel Lindberg 1000VA "Dreadnought" Power Conditioner: PS Audio P15 Power Plant Power Cords: Core Power Technologies Gold, DH Labs Power Plus DIY w/Neotech NC-P301 & P311ends Streaming: Roon ROCK on wifi'd NUC, TP-Link WAP, & Uptone EtherREGEN, AfterDark, Emperor Double Crown Clock, Black Modernize LPS, PS Audio AirLens⟿Ω☯☥☮⟿🔊♩♪♫♬♩♪♫♬♩♪♫♬
  • I got in the 3D box habit when I worked for Polk's CS. When I fist started every task I had to do for a customer was written down in a spiral notebook. In those days before direct computer entry I had to write their name, address, telephone number and what part they needed along with credit card information. I would put a box beside the entry so I would know if I'd done that item and the 3D aspect just grew out of mentally thinking about what had to be done. It was all part of the process of not forgetting anything that I had to do, I guess.