Reel to Reel Taperecorder Equalization

Reel to Reel Taperecorder Equalization

An introduction to this somewhat confusing topic

As mentioned in the earlier PR99 blog, I wanted to design and build a universal tape head amplifier. Just for fun and to see what I could squeeze out of the magnetic source, avoiding the internal Revox electronics. Which sounds great, but honestly I still do not get it as it is full with electrolytic capacitors and overall feedback in the signal path 😉

As I was relatively new (again) to this whole topic I needed to get my head wrapped around the equalizations being used. Honestly, if you google you get a lot of fragmented information, but I could not find clear overall information on the EQ and what signal was coming from the tape head it self… So I had to do that my self – here it comes:

First of all I like this smaller overview (below table) of the EQ time constants (and corner frequencies) being used most commonly. There are much more, but I wanted to stay practical. My focus will be on the ones below – except the 1.875 and 30ips – I will not go into detail why these are as they are (like NAB mostly American and CCIR European) as on that topic there is sufficient other reading material on the web

Tape equalizations summary
Tape equalizations summary

How to come from micro seconds to a EQ filter design?

I am using TINA Basic 12 spice simulator for developments like this. The way I work is to define an ideal curve as reference. Goals is of course to have this next to a spice simulation of a real world design later on (with an amplifier and normal parts) and compare the two curves till they match best – tweak values and automatically having the draft for first prototype ready. For the EQ example I am using the “ideal opamps” in the Tina library and put a RC filter in between two op amps with the right (see table above) corner frequencies and adjust gain so they are all on 0dB at 1kHz:

Reproduction EQ curves for NAB and CCIR 7,5 and 15 ips
Reproduction EQ curves for NAB and CCIR @ 7,5 and 15 ips

First ideas for an universal passive EQ filter

Passive Filter for NAB & CCIR

At the right you see a sneak preview of the EQ amplifier passive filter I came up with tweaking in TINA till the curves almost matched – a separate blog will come when the first prototypes have been built. PCBs are on their way

NB: These values are only working with a low impedance source and 330nF couple capacitor and a high impedance input follower stage…

One switch will enable you to basically select between NAB or CCIR and the second one to select which speed (7.5 or 15 ips for CCIR and 3.75 and 7.5 for NAB) – Front panel switch or relays is up to you (and me – I will use relays to keep the signal path short…)

What signal comes from the reproduction tape head?

Round this passive filter you need quite some gain to get an output voltage of let’s say 2V rms for your main amplifiers. For example looking at the CCIR @ 30 Hz curve above, you would need already 30dB to even compensate for the lower output from the tape head versus the 1kHz reference. After that you need gain to get from the tape head 1kHz to desired output level plus some headroom…

So I needed to know what comes out of a tape head. Google? I could not find anything other than loose and incomplete information, so decided to do it my self. using the Audio Precision Audio Analyzer with source input to the PR99 at 0dbU level. Recording the sweep signal (30Hz-20kHz) and connecting the AP input directly to the head with a strong bandpass filter to avoid noise and hum influence. I only needed a rough idea, so this worked fine for me.

Below is the curve and as you see the output is very low, almost like a MD cartridge. I measured 2.2 mV at 1kHz. Also note the 100uV in the low frequencies – This told me I would need at least an overall gain of 20.000 to get to 2Vrms – of course you would most likely want some headroom for low level recordings. So this could be even more like 60.000 to 100.000. As I wanted to use tubes (transistors and opamps are already inside the Revox PR99 😉 ) it is clear this will be 3 stages (!) and as you want low impedance outputs for the filter, it has to be MU Stages like in my Phonodude design. But more about that in the blogs to come 🙂

Revox PR99 MK3 Frequency response - directly at reproduction tape head over tape 0dBu 15 ips
PR99 measured directly at reproduction tape head with 0dBu input @ 15 ips

Next steps

I hope I gave some short and clear information on the equalization topic and how filter and amplifier design requirements will have to look like.

I hope you are as intrigued as I am right now and interested how a direct tape head amplifier will look like and how it will sound. PCBs are on their way and after my upcoming vacation, I will start building first prototyping and do some first testing & trials. And of course blog about it as promised!

ohh… just one day later and the PCBs arrived ! So no official “update” just added the image below. May be I can find some time before I go on vacation to do first test…

PCBs MU-Stage CCIR NAB Filter - HV PSU-b
The new PCBs for the Tape EQ amplifier

To be cont’d

8 thoughts on “Reel to Reel Taperecorder Equalization

  1. Hello Doede. You have very interesting developments. Thank you for sharing your experience.

  2. Hello Doede,

    Any progress with the tape head amplifier?
    I’m very curious and interested. I’m searching for something like this for my Telefunken M15A.

  3. Re: 4.75cm/sec EQ
    I know it’s slightly off topic, but the subject of pre-emphasis and de-emphasis (REC and PB EQ) is unclear for cassette decks at 4.75 cm/sec, pre-1981. From what feedback I can gather, manufacturers like Teac, Sony, Pioneer, Sansui and others broadly followed the IEC (‘IEC I’) spec (similar to RIAA?), but appeared to have tweaked their machines to suit the limitations of their heads. Compromises were seemingly made with regard to frequency response, noise, and distortion. If I haven’t misread anything, Nakamichi broadly didn’t change their spec that much pre-1981 and after.

    The Transfer Function Y(s) = X(s)*H(s) {H(s) is the Transfer Function in the s-domain} that describes the relationship between input and output is also a bit vague?

    But no such messing about for reel to reel?

  4. Hi Eric,

    thanks for interesting comment. As I am not involved in cassette decks I did not realize this…

    My knowledge is, that reel to reel deck manufacturers held strictly to IEC and NAB. Probably as technology allows for it and also studios were all on reel to reel.

    not sure what you mean with the transfer function?

  5. Re: Transfer Function.
    A transfer function is the mathematical relationship between output and input, it’s usually expressed in the frequency domain – typical written H(s) or similar. So in this case of ‘steady state’ sinusoidal analysis, s=2πf, note that 2πf is usually written as … ω=2πf. And of course we know that f = frequency in Hz.

    Any pre-emphasis and de-emphasis spec will follow their individual transfer function – this gives us the ‘shape’ of the curves.

    There are pivotal transition points in the mathematical expression from which the ‘time constants’ are derived. A typical pre-emphasis may look like this H(s)=(sT1+1)(sT3+1)/(sT2+1).

    The time constant signatures T1, T2, T3 etc are derived by the condition when sT1=1, or sT2=1, or sT3=1. At one of these corner frequencies s (also called ‘transition points’) we have the condition where sT1=1 or 2πf =1/T1.

    Re-arranging gives … f=1/(2π*T1) So now we have the relationship between frequency f and this ‘time constant’ T1. (I can’t write subscripts so I had to write T1, T2 etc)

    Now I’ll leave it to you to calculate the transition frequencies for 3180us, 120us, 70us etc. ;o)

    The arrangement of resistors, capacitors, and sometimes inductors are modelled to give us that pre-emphasis/de-emphasis curve.

    Hope this makes some sense.
    regards.

  6. Hi Eric,

    sorry I had you write this all up. I know this of course. Even so good summary. I just asked why you brought this up in your previous comment and why you thought it was vague?

  7. Re: Cassette EQ.
    I thought it ‘vague’ because obtaining official infomation is difficult – nowhere (so far) have I seen someone like Sony, Teac etc state what they did specifically, ie what their mathematical model was.

    Most folk, like myself , are happy just to recognise visually the EQ curves, but I also wanted to see the s-domain model, if and where it was published. So I guessed it was roughly based in the RIAA (phonograph) model? Some people refer to the pre-1981 model as ‘IEC 1’, and post-1981 model ‘IEC 2’ – but even the post-1981 model is difficult to source.

    No worries. :o)

    Regards.

Leave a Reply

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