exxos's Atari, Amiga & retro forum

I'm looking at the audio output stages on a Falcon030 and noticed something while comparing the factory design with the C-LAB/Cubase audio mods. Maybe there is some knowledge here...

So the audio mods break down into 3 parts:

1) line level output (reduce gain on 'output')
2) line level input (reduce gain at mic socket)
3) remove bass boost

Lets ignore (2) & (3) for now - I'm mainly interested in checking the audio output voltages with (1).

The Falcon I'm looking at seems to have a high output level at the stereo headphone socket. It works but it overloads headphones etc. if volume is set more than about 40% and it would completely blow up anything expecting line-level. I'm trying to confirm if that's normal or not.

So I started investigating the audio output stage, using the description of the line-level mods to see how the output gain is being modified. I bodge in the suggested 10k resistor at U21 pin 2 and see.... no change whatsoever.

Hmm..... rabbithole developing!

Studying the schematic, the suggested 10k resistor for line-level output is applied to the signal-input pin of one of the four LF/LM 347 amps in U21. It appears to take HLI/HRI stereo outputs from the CODEC, mix them to mono using what looks like an opamp add configuration and the gain-adjusted output goes to.... the SPEAKER and AUDIO nets.

The SPEAKER net (internal speaker, not present) is obviously not that interesting as it goes nowhere except to a speaker-specific amp - but the AUDIO net seems to go to the TV modulator and the VIDEO OUT connector as a mono signal. It does not go anywhere near the headphone socket....

The tapped CODEC HLI/HRI signals continue on to a couple of completely independent audio output amps in U22, U23, one for left one for right. These go nowhere except the stereo headphone jack.

My only conclusion from this is that the 'line level out' mod can't affect the stereo output on the Falcon. It only affects the speaker and the video output.

This makes no sense to me whatsoever. Does anyone have any idea whats going on here and what the intent was with the 10k resistor at U21? It's not a bias voltage for the stereo circuit - its a gain-setting preamp for two unrelated mono circuits.

o__O

Is there more than one version of the line-level output mod? I have seen the same version in two places now and they say the same thing.

My best guess is that the original information has got corrupted over the years to interpret a mono-only part of the mod as the 'line-level out' mod, while the 'remove bass boost' mod might actually be doing the stereo line-level out mod as well as bass-boost... Still, I don't understand why anyone would want to mess with the audio output gain at the video connector?

Completely aside from this, it looks like the mono preamp in U21 has pin #1 (mono audio out) sitting at 8.25V all the time. It is AC-coupled to the AUDIO and SPEAKER nets but it still doesn't seem right. That's basically pegged - the opamp's rail voltage is 9V.

I think the 8.25V might be correct actually.

The stereo-to-mono mixed audio from CODEC input is fed into pin #2 (inverting input) and the reference voltage is set at pin #3 (noninverting input).

The reference voltage comes directly from the CODEC on a net called REFBYP which suggests a bypass condition.

So I think the CODEC can adjust or switch off the mono circuit via REFBYP. The pegged output is maybe normal when the mono output devices are turned off.


I adjusted the input level using a potentiometer instead of the 10K bodge and the output does act like the signals are being subtracted so U21 is probably ok. It's just completely irrelevant to getting line-level output at the headphone jack.

Some context, showing the headphone L/R amps and the mono speaker amp are completely separate.

The headphone L/R amps get their input directly from the CODEC.

The 10K resistor mod only affects the bottom circuit and the VIDEO connector...

So the headphone jack / stereo out amp gain is actually being set by:

LEFT: R132/R131
RIGHT: R125/R124

I see no mention of those in the audio line-level mods - It only mentions R123, R130 in the 'remove bass boost' mod.

I suppose that achieves the same although they are not equal on the reference schematic so it's a bit inconvenient to set them correctly. L/R use 15k,16k respectively.

[edit] the schematic is a bad scan - so if I twist my eyes just enough, the 16k might be 15k but it certainly looks like it says 16k.


The audio mods are described here:

https://www.holmerup.com/linemod.html

I suppose if you do 'all the mods' at the same time you'll get line-level output.

If you ONLY want line-level output though (or line-level in+out but leaving bass-boost alone), this description is not going to help you at all :)

Ok here's my remix:

LOWER THE STEREO OUTPUT AMPLIFIER GAIN:
Change R123, R130 from 10kohm to 15kohm

LOWER THE MONO SPEAKER & MONITOR/VIDEO OUTPUT AMPLIFIER GAIN:
Put a 10kohm resistor between pin 2 of U21 (@ R119) and AGND (@ R111).

REMOVE BASSBOOST:
Change C58, C59 from 100pF to 1000pF
Change R59, R68 from 100kohm to 4.7kohm
Change C95, C90 from 0.033uF to 1kohm resistor

LOWER THE STEREO INPUT GAIN TO -10DB:
Remove R56, R57, R90, R92, C21, C81
Change R66, R87 from 100kohm to 1.8kohm
Put a 470pF capacitor in parallel with R67, R88
Put a 100pF capacitor in parallel with R66, R87

Looks interesting - I'd love to do the audio mods to my Falcon at some point, so this will be a good thread to refer back to.

Thanks @dml

There was a modification, something to do with grounding on a transistor, but I remember having this conversation a while ago and believe that half the Falcons had the mod incorrectly done because the ground wire was literally linking to the same ground like 5 mm away from the transistor.. Which made absolutely no sense.. But I cannot find the topic at the moment where it was talked about...

Maybe @Steve remembers ?

EDIT:

Found it viewtopic.php?p=90440#p90440

I was also wondering about this - the nets seem to be designed such that AGND (audio ground) and GND (general, common ground) are independent grounds, isolated from each other. Which sorta makes sense for an audio output stage if you want completely isolated grounds.

I would then expect the two grounds to be joined by a capacitor so there is no DC path between them but they are still 'joined' for the signal path.

However they seem to be DC-connected on two Falcons I checked. Maybe that's an oversight or maybe there's a mistake that causes those grounds to be joined.

It's a rabbithole I sort of sensed was near - and slowly backed away from. :P I think it would be difficult to figure out where the two nets are actually joined on the board, especially if it is a mistake.

The field service guide does show AGND connected to 'earth' symbol via some blurry thing with some blurry text over it. The RevG schematic does not have that 'earth' connection. So.... shrug.