Audio filter mod

[Mick]

I'm going to remove the op amp and regulator board anyway, I was just interested to see if there would be any difference but I don't think Commodore skimped at all really. I'll see if I can lower the volume and if so will record from the beginning to get an idea of noise.

BTW Stedy, do you have an A600 shielding with those boards you received? I plan to sell my A600 to make back some money but mine seems to have gone AWOL.

[Mick]

Capture volume is set at 100% so the clipping must be down to Commodore? unless there's something I'm doing wrong.

I've remove the bodge board and replaced the op amp back to TL084.

https://ufile.io/f/b9ga0

https://postimg.cc/5HrKPzF2


I'm not really sure why they drift out of alignment over time? the time between powering on which is the spike at the very beginning and the intro sound is the same but then it drifts out after...

[pandy71]

Originally Posted by Stedy

Hi,

You don't need a faster op-amp, quite the reverse. For an op amp maximum slew rate is 2*pi*V*fs, where V is the peak voltage (2.2V) and Fs is 20KHz (Audio range) . Multiply this together and you get 251327 volt/second. Divide by 1E6 to get V/us and you get 0.25v/us, the antique 741 Op Amp could handle Amiga audio (0.5v/us) and the TL084 has a slew rate of 20V/us.

Paula has a current mode output, hence the I/V converter in the first amplifier circuit, I guess around 8mA into 360 ohms creates the output voltage. These are nice, linear D/A converters, they can work fast, not needed in this application. All this talk of PWM is nonsense, this is not a PCM output, as used on more modern sound cards, it's older technology.

The audio circuit, in the Amiga, in simulations, shows the cut-off of the I/V converter around 27 KHz but the filter circuit, does add gain around 1MHz due to the 3900pF capacitors in the filter adding a zero in the feedback circuit, increasing the gain at 1Mhz then tailing of at 10MHz, the gain bandwidth of the TL084. So frequencies around 1Mhz will be amplified slightly to -6db instead of the -34dB they should have been rather than attenuated by.
Never use a faster op-amp than you need as the higher frequencies can bite you in the bum.

This work nicely for sinewave but issue is in volume regulation method used by Paula - you have sinewave (lets assume sine as sampled waveform to simplify discussion) multiplied by PWM (square wave with duty variable from 1/64 to 63/64) - so sampled sinewave is additionally sampled by 3.58MHz/64 square wave - you need fast OPAMP to deal with complex product or you get distortions dependable at OPAMP slew rate (large signal response).
That's why lot of audio DAC's use or passive I/V network or fast OPAMP's.
To get "14" bit you need address this.

[Mick]

It might be worth someone who knows what they're doing looking into reducing the gain some more? I have A1200 2B spec and it was almost constantly in the red zone slamming 0dB and probably beyond, I think Audacity recommend input levels to be between -3 and -6dB.

The problem is if I change the feedback resistor value it might change what's needed as the feedback capacitor as from what I've read it is a stabiliser and it's beyond me to measure what's best, or maybe I can just tweak the voltage divider near to the RCA's?

Also, I'm going to guess the alignment issue was down to slight difference in floppy drive reading time.

[pandy71]

Perhaps you can adjust input mixer level on your capture board?

[Stedy]

Originally Posted by Mick

It might be worth someone who knows what they're doing looking into reducing the gain some more? I have A1200 2B spec and it was almost constantly in the red zone slamming 0dB and probably beyond, I think Audacity recommend input levels to be between -3 and -6dB.

The problem is if I change the feedback resistor value it might change what's needed as the feedback capacitor as from what I've read it is a stabiliser and it's beyond me to measure what's best, or maybe I can just tweak the voltage divider near to the RCA's?

Also, I'm going to guess the alignment issue was down to slight difference in floppy drive reading time.

What values do you have for R325/R335 on your A1200?
Mine's packed for storage so can't check it but they should be 390R-510R, not 2K as shown on schematics at AmigaWiki. This will give a true line level output.

To truly answer the output levels from Oaula, I'd need the spec of the chip, it's one of the few custom chips I don't have the spec for. I assumed 8mA, a common value for the I/V converter and it worked in simulations which matched reality.

I don't think I have A600 metalwork, like I said, all my Amiga gear is packed away for a few months, whilst building a new workshop

@pandy71, do you have any idea how filter and op-amps work?
You're spouting a lot of rubbish. A filter at the I/V converter, will attenuate higher frequencies above 27KHz, the slew rate will easily handle the required changes in output over a 20KHz range, any talk of 3.58MHz is irrelevant, the circuit works at ~20 KHz due to the filter.

I've spent the last 18 years specialising in precision analogue and power electronics, kinda know what I'm doing (mostly).

[pandy71]

Originally Posted by Stedy

@pandy71, do you have any idea how filter and op-amps work?
You're spouting a lot of rubbish. A filter at the I/V converter, will attenuate higher frequencies above 27KHz, the slew rate will easily handle the required changes in output over a 20KHz range, any talk of 3.58MHz is irrelevant, the circuit works at ~20 KHz due to the filter.

I've spent the last 18 years specialising in precision analogue and power electronics, kinda know what I'm doing (mostly).

Good for you Steady - don't get me wrong but I/V converter in Paula is active not passive - PWM volume control is not 20kHz sinewave for sure.
If you do passive I/V and low pass before OPAMP then i agree with all you wrote.

[Mick]

Originally Posted by pandy71

Perhaps you can adjust input mixer level on your capture board?

I just used line in on the sound card so I would assume it doesn't amplify the input signal?

Originally Posted by Stedy

What values do you have for R325/R335 on your A1200?
Mine's packed for storage so can't check it but they should be 390R-510R, not 2K as shown on schematics at AmigaWiki. This will give a true line level output.

360 ohms. I built the whole sound stage the same as the A1200 2B schematics and also checked with motherboard photos to make sure that's what Commodore actually used.

[pandy71]

Originally Posted by Mick

I just used line in on the sound card so I would assume it doesn't amplify the input signal?

Amplify - probably not, but attenuate probably yes so you can reduce level by 50% (-6dB from 0dBFS) or more to prevent clipping if Amiga level is so high (usually supported maximum input level in consumer devices is around 2.504 Vrms)

[Mick]

Originally Posted by pandy71

Amplify - probably not, but attenuate probably yes so you can reduce level by 50% (6dBFS) or more to prevent clipping if Amiga level is so high (usually supported maximum input level in consumer devices is around 2.504 Vrms)

It's too late run tests again now, well it's not but I'm just too demoralised now. I just thought that if Amiga levels are still too high (even after all of the messing around Commodore themselves did) then it might be better to bring them down to modern day recommended output levels, assuming that's what Audacity's recommendation is. If the Amiga output is clipping in Audacity then wouldn't it be reasonable to assume it may also be clipping when connected to other devices?

[pandy71]

Originally Posted by Mick

If the Amiga output is clipping in Audacity then wouldn't it be reasonable to assume it may also be clipping when connected to other devices?

It looks like clipped - but perhaps everything is fine (source dependent) - best test will be some sinewave with level for example -6.0206dBFS so easily maximum output level can be calculated (basic multimeter can be used to measure level if your sinewave frequency will be like 50 or 60Hz - depends on country).