Very low signal on MIC port

[rp55]

Good grief! RP55 - if what you are saying is any capacitive loading in the input circuit of the tape recorder will affect the impedance, you'd think it would - it is also loaded to ground by a 2200pf capacitor, which has a reactance of 3.6k at 20KHz - but which when combined makes an impedance of 0.96k (if I did the calculation correctly?!),so no real difference actually!! Only really starts making a difference when getting into RF frequencies, and may well be there to stop the pre-amp/amp/osc chip oscillating where it shouldn't! Another similar but different (if you get what I mean ) tape recorder I have has a 1500pf capacitor to ground (but can't see any resistor..maybe internal to the chip?)...as for the output components of the ZX, that is what really shapes the output characteristics, and that is also loaded to ground by a 1k, but a much higher capacitance of 0.047uf, making a combined impedance to ground of 166 ohms at 20KHz and 860 ohms at 2KHz..now here's a variation, and it's fed through a 1Meg resistor in series with a 47pf capacitor (reactance of 169k at 20KHz, 1.69Meg at 2KHz)...which means, if I have worked it out correctly, the ULA's output is attenuated by 7043:1 at 20KHz and 3128:1 at 2KHz... if then loaded by the 1k of the tape input, it would change the ratios to 8291:1 at 20KHz, and 5823:1 at 2KHz....I think

Odd you calculated such a large value ( 1.69Mohm) at 2kHz for the 47pf. Are you sure that's correct? If so perhaps they are correcting for something.... human voice, isn't that primarily in that area? Perhaps the old table top tape decks have a 1-2k boost on them to help pick up human voice? Perhaps this is compensation??

I'll see what I can come up with for calculations tomorrow...
I guess I will use cap reactance = -1 / (2 pi f C), yes... added into the impedance formula? I'll see what I come up with.

nice job... assuming yours are correct!

[PokeMon]

Well - thats the way physicists or maybe encouraged teachers in school will see or explain the world.
Like this:
http://www.electronics-tutorials.ws/fil ... ter_2.html


Electronic engineers normally do calculate with cut off frequencys for a low/high pass filter.
Like this:
http://www.sengpielaudio.com/calculator-RCpad.htm

I prefer the second variant - it's more practicable.

[rp55]

I get that it is somewhat typical to roll off over 20kHz and below 60Hz to remove non-audio components (typical for audio components). But I don't think that's simply what this circuit is doing.

Assuming the initial 1Mohm + 47pF make a low-pass filter, I calculate a 3386Hz corner:
1 / (1000000 * .000000000047 * 6.283) = 3386Hz cutoff for the 1Mohm and 47pF low-pass filter at the front.

So everything passes below 3386Hz.
And double checking gozzo's calculations for these components I concur with his calculations of:
"1Meg resistor in series with a 47pf capacitor (reactance of 169k at 20KHz, 1.69Meg at 2KHz)"
That's what I get for those frequencies also... so we're on the same page there. And that seems to fit with the low-pass corner/cutoff at 3386Hz.

Now the next RC pair (1k and 47nF) I'm not sure about their arrangement as a filter... since they are parallel. Is this a high-pass filter??

(By the way gozzo I don't see a 47uF in there anywhere... I see a 47nF in parallel with the 1k in all the zx81 schems I have. )


If the 1k and 47nF in parallel then made a hi-pass filter (not sure if that arrangement is correct) then that makes for a 3386Hz hi-pass filter. So is this a notch filter at 3386Hz?

1/ (1000 * .000000047 * 6.283) = 3386Hz cutoff for the 1k and 47nF hi-pass filter at the back???

In any case... 3386Hz is awfully low to be cutting off the top at least in terms of typical audio.

[PokeMon]

Yes, it is a very simple and clever bandpass filter with frequency of 3.3 kHz /(right as you calculated).
http://www.electronics-tutorials.ws/fil ... ter_4.html


This match to the transferred pulses of 150us which give a cycle time of 300us which results in a frequency of 3.3 kHz.
See cassette pulse documentation:
http://nocash.emubase.de/zxdocs.htm#zx8 ... ttesignals

[rp55]

Right... 300us each of the 3386 cycles = about 1,000,000 microseconds which equals 1 second. Makes sense.

so then the output is very frequency specific (centered at 3386 given this formula for center resonant frequency via low and hi filters):
f = sqrt (fLow * fHi)
f = sqrt (3386 * 3386) which is obviously 3386

so that might explain the "low signal" issue to some degree (depending on how it was measured).

It is also in parallel with the tv output... and it's own reactance (since it is not rf but direct connected?). Odd that the op mentioned the signal dropping completely when disattaching the UHF (tv).

[PokeMon]

so that might explain the "low signal" issue to some degree (depending on how it was measured).

Keep in mind that this is a clever combination which does not only build a bandpass filter with 3.3 kHz but also reduces the signal by factor 1/1000 due to different resistors (1M to 1k) and corresponding capacitors (47p to 47n). In fact this is a combination of reduce the output level and filter it. As Sinclair always did to reduce any not necessary components. So they spared a divider 1M/1k and integrated this in the filter.

It is also in parallel with the tv output... and it's own reactance (since it is not rf but direct connected?). Odd that the op mentioned the signal dropping completely when disattaching the UHF (tv).

The TV output has no influcence on that filter as it is feeded directly from the ULA into the modulator and discoupled via the 1 M resistor. So I would say no influence at all.

[rp55]

Keep in mind that this is a clever combination which does not only build a bandpass filter with 3.3 kHz but also reduces the signal by factor 1/1000 due to different resistors (1M to 1k) and corresponding capacitors (47p to 47n). In fact this is a combination of reduce the output level and filter it. As Sinclair always did to reduce any not necessary components. So they spared a divider 1M/1k and integrated this in the filter.

Interesting. But why also lower the 3386 output as well? Perhaps the tabletop tape recorders of the era did boost around 2k and this aided in that?

The TV output has no influcence on that filter as it is feeded directly from the ULA into the modulator and discoupled via the 1 M resistor. So I would say no influence at all.

Right.. by the diode. I can see that. It's in parallel anyways. Perhaps 100monkeys could verify that's what he saw.

[PokeMon]

Interesting. But why also lower the 3386 output as well?

The output of the ULA has TTL level which is 5Vpp level.
This would be too much to record so they dropped it to 5mV for the sensitive MIC input of cassette recorders of the 80ies (where microphones have been connected directly).

[rp55]

Interesting. But why also lower the 3386 output as well?

The output of the ULA has TTL level which is 5Vpp level.
This would be too much to record so they dropped it to 5mV for the sensitive MIC input of cassette recorders of the 80ies (where microphones have been connected directly).

Has that been measured/verified that the ULA is outputting 5v? I wonder why they wouldn't have at least lowered the amplitude out of the ULA down to <1v (line level) internally in the ULA. I guess I'm not sure what the best voltage is for a tv (as it needs to source that as well, I guess it wouldn't need to be very much and is probably cut way down in the tv encoder). Seems they might have saved components there perhaps (probably too far apart and the decoder was 3rd party)... but still needed the components on the mic filtering side obviously. Doesn't this mic output also serve as the actual audio Output too (I haven't looked at that)?

[rp55]

Good grief! RP55 - if what you are saying is any capacitive loading in the input circuit of the tape recorder will affect the impedance, you'd think it would - it is also loaded to ground by a 2200pf capacitor, which has a reactance of 3.6k at 20KHz - but which when combined makes an impedance of 0.96k (if I did the calculation correctly?!),so no real difference actually!! Only really starts making a difference when getting into RF frequencies, and may well be there to stop the pre-amp/amp/osc chip oscillating where it shouldn't! Another similar but different (if you get what I mean ) tape recorder I have has a 1500pf capacitor to ground (but can't see any resistor..maybe internal to the chip?)...

I'm still trying to comprehend this gozzo (after going back and looking at your obvious genius gozzo!). Give me some time to absorb this please, yes?