I am installing new heads on a reel to reel tape recorder, which have much lower impedance than the originals, like 3 vs 250mH. For the playback amps this is mostly a question of level perhaps. But for recording I will probably need a different amp and bias oscillator.
Then a thought occurred to me...a low impedance output amp and an oscillator? Wait, is that not the definition of a class D amp?
I suspect an analog tape recording head amp is a very unpopular notion by now, in this forum and elsewhere -- who could possibly need it? Well people like me, I guess...
A class D amp has a pulse wave output with some high frequency backbone...100-400kHz, perhaps. We then filter this HF with an inductor and, hey presto, audio signal!
What if I applied it to the recording head, which is an inductor after all? I realize the frequency and level of biasing current has to be precisely set and I am not sure that can be done all that easily in a class D amp.
In any case I thought it was an interesting idea. Until someone corrects my erroneous thinking.
Then a thought occurred to me...a low impedance output amp and an oscillator? Wait, is that not the definition of a class D amp?
I suspect an analog tape recording head amp is a very unpopular notion by now, in this forum and elsewhere -- who could possibly need it? Well people like me, I guess...
A class D amp has a pulse wave output with some high frequency backbone...100-400kHz, perhaps. We then filter this HF with an inductor and, hey presto, audio signal!
What if I applied it to the recording head, which is an inductor after all? I realize the frequency and level of biasing current has to be precisely set and I am not sure that can be done all that easily in a class D amp.
In any case I thought it was an interesting idea. Until someone corrects my erroneous thinking.
You won't be happy with the harmonics, the noise, switching artifacts (you don't want those in your recording) and the output filter lowpass. The bias frequency of 70kHz is way too close to the filter. And since the bias is usually adjusted to the tape, the level will change with the frequency (remember the filter?) - not something you want to have while adjusting the bias/level. At all.
You've measured the inductance, not the impedance. I'd suggest you measure the impedance (not the inductance) at the bias frequency of the heads first, then think about the amplifier. And don't forget: The amplifier does not only amplify the bias, it also amplifies the actual music signal, so the amp which is already in use has to work with that too, that means, it already deals with low impedances. You more likely have to adjust or modify the bias circuit. If you are unsure, just look at the donor shematics and compare it with your machine. To re-build that part is probably much easier and safer than to design a class-D amp and eliminate all the drawbacks, that's quite some R&D time you're looking into investing there.
You've measured the inductance, not the impedance. I'd suggest you measure the impedance (not the inductance) at the bias frequency of the heads first, then think about the amplifier. And don't forget: The amplifier does not only amplify the bias, it also amplifies the actual music signal, so the amp which is already in use has to work with that too, that means, it already deals with low impedances. You more likely have to adjust or modify the bias circuit. If you are unsure, just look at the donor shematics and compare it with your machine. To re-build that part is probably much easier and safer than to design a class-D amp and eliminate all the drawbacks, that's quite some R&D time you're looking into investing there.
Many thanks! As I said, it was a nice thought for a few minutes of reverie, nothing more...
You say, "And don't forget: The amplifier does not only amplify the bias, it also amplifies the actual music signal, so the amp which is already in use has to work with that too, that means, it already deals with low impedances."
In most schematics that I have seen, the bias appears to be added right near the output head, so it does not actually go through the amp (except in my infeasible Class D version!). Am I wrong?
You say, "And don't forget: The amplifier does not only amplify the bias, it also amplifies the actual music signal, so the amp which is already in use has to work with that too, that means, it already deals with low impedances."
In most schematics that I have seen, the bias appears to be added right near the output head, so it does not actually go through the amp (except in my infeasible Class D version!). Am I wrong?
Often there are different solutions to do the same thing. If you add a strong signal directly to a weaker one, you have to protect the 'weaker' output circuit. That is much more complicated than to add them in the pre stage, which make it rather unlikely. It's much more likely it's added directly before the last amp stage or it's in fact the circuit for the erase head or something switched for the different tape speeds etc.. If in doubt, look at the shematics.
Class D output can be very clean, some modulators and switching implementations achieve it. However, modulator design dictates an idle oscillation frequency at least 10 times the highest freq of desired audio bandwidth. Consider that just closing a linearizing NFB loop across a sawtooth modulator (to linearize power stage) requires loop crossover frequency to be Fsw/(2*pi) or lower. For self oscillating loops NFB can be used up to Fsw/2 or so.
Many thanks.
Why is the modulation frequency so much higher than Nyquist? Is that because PWM is a bit like Bitstream so a single "bit height" has to be provided with a larger frequency?
10 x 20 or 200kHz is not crazy high as bias. The next matter would be providing some level control to adjust "bias."
All of this can probably be tested quite easily with a low power PWM chip.
Why is the modulation frequency so much higher than Nyquist? Is that because PWM is a bit like Bitstream so a single "bit height" has to be provided with a larger frequency?
10 x 20 or 200kHz is not crazy high as bias. The next matter would be providing some level control to adjust "bias."
All of this can probably be tested quite easily with a low power PWM chip.
Consider some facts:
- To get top linearity from a PWM modulator, enclosing it in a negative feedback loop is required.
- To get enough phase margin for a NFB loop, the Nyquist frequency must be several times higher than the unity gain frequency of the loop.
- Since a 2nd order lowpass filter is involved, and some carrier attenuation is required, this imposes also that Fsw must be several times higher than the top frequency of audio bandwidth.
- A clean sine carrier residual is only obtained with such a 2nd order LPF, and the >~10 ratio between top frequency of audio bandwidth and switching frequency.
- Additionally, self oscillating modulators allow the highest linearity, and tend to provide constant amplitude carrier residual at output, but oscillation frequency suffers a typical 3:1 drop between idle and threshold of clipping, the cleanest solution does *not* have constant switching frequency.
- On the other hand, constant frequency sawtooth modulators do not provide constant carrier residual amplitude, which tends towards zero at threshold of clipping.
- I did some research about the practical cost and advantages crossover point between a linear solution and a switching mode solution, for voltage voltage regulators and audio amplifiers. The crossover is around 100mA output and/or about 400mW dissipation in regulator or amplifier, at this point a linear solution and a switching mode solution have similar complexity and price, with currently available chips and technologies. The exception is compact size and light weight battery operated equipment where the "crossover" point between linear and switchmode may be lower.
As far as I know, power and current levels involved in a tape head are lower than this "crossover" point. And the two families of class D modulators are unsuitable, one produces constant amplitude residual but not constant frequency and the other produces constant frequency but not constant amplitude.
- To get top linearity from a PWM modulator, enclosing it in a negative feedback loop is required.
- To get enough phase margin for a NFB loop, the Nyquist frequency must be several times higher than the unity gain frequency of the loop.
- Since a 2nd order lowpass filter is involved, and some carrier attenuation is required, this imposes also that Fsw must be several times higher than the top frequency of audio bandwidth.
- A clean sine carrier residual is only obtained with such a 2nd order LPF, and the >~10 ratio between top frequency of audio bandwidth and switching frequency.
- Additionally, self oscillating modulators allow the highest linearity, and tend to provide constant amplitude carrier residual at output, but oscillation frequency suffers a typical 3:1 drop between idle and threshold of clipping, the cleanest solution does *not* have constant switching frequency.
- On the other hand, constant frequency sawtooth modulators do not provide constant carrier residual amplitude, which tends towards zero at threshold of clipping.
- I did some research about the practical cost and advantages crossover point between a linear solution and a switching mode solution, for voltage voltage regulators and audio amplifiers. The crossover is around 100mA output and/or about 400mW dissipation in regulator or amplifier, at this point a linear solution and a switching mode solution have similar complexity and price, with currently available chips and technologies. The exception is compact size and light weight battery operated equipment where the "crossover" point between linear and switchmode may be lower.
As far as I know, power and current levels involved in a tape head are lower than this "crossover" point. And the two families of class D modulators are unsuitable, one produces constant amplitude residual but not constant frequency and the other produces constant frequency but not constant amplitude.
Last edited:
That is wrong. Your used upper frequency isn't the 20kHz of the audio, it's the 70kHz of the bias! So we are talking about 700kHz! And that's not that trivial anymore.
Digalog
I saw somewhere that towards the end of musicassette's reign in the early 1990s, in an effort to improve pre-recorded cassettes, digital masters were used.
The signal, on a sped-up basis was delivered to the record head as "digital" -- I presume that means PWM and replaced the bias also, called Digalog. I have not been able to find much more about it yet.
I saw somewhere that towards the end of musicassette's reign in the early 1990s, in an effort to improve pre-recorded cassettes, digital masters were used.
The signal, on a sped-up basis was delivered to the record head as "digital" -- I presume that means PWM and replaced the bias also, called Digalog. I have not been able to find much more about it yet.
Well, they digitally created the complete, modulated and sped up signal and then used a PWM amplifier and completely different machines and heads for the recording. That means, they included the BIAS in the signal and an amplifier with a very large bandwidth. They did not use the amplifiers triangle signal to create the BIAS. You can ofcourse do the same but without a clean BIAS you won't get any usable recording.
Why don't you go ahead and use a frequency generator and try it before cobbling up such an amp?
Why don't you go ahead and use a frequency generator and try it before cobbling up such an amp?
Yes, pretty much like that. Instead of a square wave you'd have to use a triangle wave form signal generator to get the result of what you're planning. But even at the bad sound quality of the video you can clearly hear the distortion. If that's good enough sound quality for you, go for it!
I wish I were as young and care-free as the experimenter in the video and be able to say, “because it is time to experiment!” with his dirty soup bowl still sitting on a cassette deck as he films his “experiment.” Unfortunately with age many of us learn, “it is no longer time to experiment but do the sure thing.” Tsk.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.