Introduction

I said I will talk about LED drivers next. I was lying. I realized that, while I designed the tape read amplifier, I actually never wrote about it. So in this chapter, I'm going to describe that part of the computer.

Variations on a theme

If you look back my previous discussion about the way digital recording in tapes worked, you'll see that we need to detect right small pulses on the read head, corresponding to magnetization domain changes as the tape passes in front of the head. We don't actually care about the polarity of the pulse, only its existence.

This is a very similar problem to the core sense amplifier. Maybe the only true difference is that the amplitude of the pulse is much much lower, probably in the sub-mV range. That, and the frequency range we need to support: here we're looking at audio signal frequencies, so something below 20kHz. Make it 40kHz to account for double-speed playback.

Without further ado, let's look at the schematic:

It's almost identical to the core sense amplifier with different settings of course. The gain is set to 30,000, but that's highly misleading: at these very high gain numbers, my calculations are highly inaccurate. Instead, you should think of it this way: the gain is set such that a +/-1mV input signal clearly triggers the output throughout the whole temperature range, but +/-0.5mV never does.

One thing I've added was a voltage divider between the minimum detectors (Q9 and Q10) and the threshold stage (Q7). This is needed to ensure reliable triggering. You see the output of the diff-amp (OA or OB) never goes below 0V, thus the sense_out node will always sit at least one Vbe above that. It can go higher of course, but can't get lower. This puts the input right at the threshold of the comparator stage. By dividing it by two, I buy some margin at the cost of increased temperature dependence. I'm going to think a little more about this part of the circuit, I might come up with something better. I don't really like this arrangement. Still, it seems to work, so how much do I care?

Finally, the input looks a bit different, it being a tape head. I actually modeled what I think my tape head is, exciting it through a transformer-coupled current source. This should be similar to what the tape is doing to it. The amplitude of the current source can be adjusted to measure the expected output voltage.

None of this deviates enough from the core sense amplifier; the circuit topology is essentially the same. Consequently, it makes total sense to design a combined PCB for both, giving me one less module design to worry about.

One thing I should mention is this: the diff amp is maxed out on gain. further lowering the gain resistors don't actually result in higher gain. I can't really explain it, but nevertheless, that's the case. So, if the input is significantly below 1mV, I will need to do something else. I haven't seen any head specs that would be lower than 1mV, in fact the closest head that I have specs for claims 9mV (peak-to-peak) output, so I'm not terribly worried. Nevertheless, if that's the case I will probably add an extra common-emitter stage at the end (after all audio bandwidths are perfectly within reach for such a stage), before the digital conversion (in place of the resister divider R10/R11). Another benefit is that since we're looking at a tape head input, there's no DC component to the signal. In other words, AC coupling capacitors can be used in the signal chain, if needed (so far I didn't) which makes stage biasing easier.

Conclusion

This is a surprisingly short write-up. There is one thing, I have to think about more: the resistor divider and why I didn't need that for the core sense amplifier. Something doesn't quite add up there. I will also have to do experiments to figure out what input levels this amplifier needs to deal with before committing to the final design.

Still, with a little luck, the current circuit will hold and I can have a combined PCB for two functions: the core and the tape sense amplifiers.

Next time around, I promise to really get around to talk about the LED drivers in the my optical ROM - which might also be a starting point for the tape write amplifiers.