Line level of the Amiga 600 (and possibly other models)

[ugrnm]

Hello,

I'm going through some maintenance of my Amiga 600, with a special focus on the audio. I'm getting to the point where everything sounds great, no more cracks, hums, CPU activity leak weirdness, etc.

However, while plugging the Amiga on a mixer I noticed that the audio levels were quite low. By low I mean that they are probably perfectly fine for an Amiga, but that in comparison to other gears like synths, it is significantly lower. I want to mix in the sound output of the Amiga with other music gears, and ideally it would be great to not have to boost the signal too much on the line channel where the Amiga is plugged, to avoid adding to the noise floor (and also to give me more room to increase/decrease the sound of the Amiga in relation to other channels).

What followed was a rabbit hole into the world of line level. And in particular this graph found on Wikipedia:



As a result I'm trying to figure out how the Amiga was designed, I would assume for consumer line level? Anyone knows or ever measured this?

I'm also wondering if the audio circuit could be modded to so-called pro line level or at least boost it a little bit. This is where I'm also a bit stuck because when I look on the Amiga 600 schematics of the 1.3 (the model that I have), I see that I could switch R334 and R335 for the right channel (and do the same for left)? I'm suggesting switching because I don't have smd resistors, so that could be a lazy hack.



What I noticed is that later revisions (1.5) have the same voltage divider as the A1200, where the ground resistor of 390 was replaced with a 2K resistor, which makes me also wonder, does it mean they realized eventually that the line level was too low, or that the circuitry before the final output had changed and became too high? (I did not compare the full diff between 1.3 and 1.6 audio circuitry so this is just speculation).

Ultimately, could I also simply get rid of R335? Even.... both R335 and R334?...

Any advice would be very much appreciated

[Stedy]

The attached file is a measured output from a rev 2 A600.

The peak to peak is 1.69V, the RMS is 0.596V.
Doing the sums, it equates to -2.25dB below Pro reference level.

The second attachment shows the audio levels either side of the AC coupling capacitor, C324 or C334, before the potential dividers you mentioned.


Hope this helps explain it all.

[Mick]

I think it's R321/R331 that control the amount of gain, when I built my A600 I used the same values on the audio circuit from the A1200 Rev 2 schematics as it seems earlier boards were prone to either low audio levels or distortion due to being too high. Not quite as technical answer as Stedy is able to provide but it seemed to me Commodore weren't entirely sure what was best audio levels either until maybe late on.

This might be informative:
[ Show youtube player ]

[r.cade]

Very good find and information. Some people might actually like the hot output and distortion, so you never know. It also might not be as bad when hooked up to other output devices like a stereo or mixer.

[ugrnm]

Originally Posted by Stedy

Hope this helps explain it all.

Thanks! This is very useful, I 'm gonna measure mine this week and see what are the differences. What did you use to generate a 1kHz sine wave from the Amiga?

Originally Posted by Mick

I think it's R321/R331 that control the amount of gain, when I built my A600 I used the same values on the audio circuit from the A1200 Rev 2 schematics as it seems earlier boards were prone to either low audio levels or distortion due to being too high. Not quite as technical answer as Stedy is able to provide but it seemed to me Commodore weren't entirely sure what was best audio levels either until maybe late on.

This might be informative:
[ Show youtube player ]

Yes it seems to be that several iterations took place, even for the A1200. I also see that my Amiga 600 1.3 had even a third value for both R331 and C331, so it seems that the 2 changes that occurred through time were both at the opamp input gain and at the last voltage divider before the RCA out.

It's interesting that the peak-to-peak value of the A1200 that is perceived as too loud and distorted is in fact very close to a pro line level. Also it sounds distorted on the TV but on the scope the signal looks fine, so I would guess it's the TV that was designed for consumer line level?

Originally Posted by r.cade

Very good find and information. Some people might actually like the hot output and distortion, so you never know. It also might not be as bad when hooked up to other output devices like a stereo or mixer.

I do believe the problematic A1200 on the video would sound perfectly fine on a mixer or high-end sound card input, or amp.

I know what to do next I think:
- compare my values with Stedy's, and also just to get an idea of what goes out of an A600 1.3
- compare the audio circuit schematics of different generations the A600 and A1200
- report here

[ugrnm]

Oh and, with the loud A1200 signal dipping into the negatives, I'm wondering if this why sometimes people have recommend to put bi-polar caps there...

[Daedalus]

It's interesting that the peak-to-peak value of the A1200 that is perceived as too loud and distorted is in fact very close to a pro line level. Also it sounds distorted on the TV but on the scope the signal looks fine, so I would guess it's the TV that was designed for consumer line level?

There are some early A1200s that have too much gain at the op amp stage, and on such motherboard you can definitely see the distortion on the 'scope at the output. What's key to note though is that the output adds together the channels at the output, so for my measurements at least, playing a sine wave on a single channel wasn't enough to get clipping. You needed to play the wave on both channels for the output (e.g. channels 0 and 3 or channels 1 and 2), and then with the combined amplitude, you see the clipping.

AmigaTestKit has some useful functions for generating sine waves and turning on and off channels and the audio filter.

[pandy71]

Increase value of R325/335 - they form voltage divider with R324/334.
Sadly to say but Paula has no separate pin for Audio Vref so or they use some internal Vref source or using Vcc as Vref so expected Vpp for Audio signal will be =<5V for sure.
Hope you already replaced R309 with proper ferrite bead, normally i would also separate Paula GND and corresponding Audio GND trough ferrite beads (also add ferrite beads on Audio line outs from Paula to separate relatively slow SR OPAMP from HF).

Technically whole audio subsystem and PCB layout should be changed accordingly to nowadays rules for HiFi Audio.

[Stedy]

Hi,

This topic intrigued me and I had a bit of a slow day.
The schematics on Amigawiki.org do not match the original printed versions from Commodore, I should know, I have paper copies of them.

For A600 REV C/June Bug schematics from 1991, R331 and R321 are 360R, the amigawikie_R1.5 and R2 schematics show 1.5K, as does the A1200 schematics.
My printed schematics of the A1200 rev 1D show 1.5K for R331/R321.

C331 and C321 change from 100nF on the earlier Junebug schematics to 3.9nF on the REV 1.5 A600 and A1200. This should keep the high frequency roll-off of the first stage circuit the same.

The other principal difference is the output potential divider. On the A500/A600/A1000 it is R334=1K and R335=390R. This means the output is 0.28x the input.

For the Amigawiki schematics for A600/A1200, the output is R334=1K, R335=2K, so the final stage output is 0.66x the input.

I have created a simulation in MPLAB Mindi/Simetrix but I don't know for sure what the output stage of Paula is to model it. R331 and R321 are used on inverting amplifiers with Avref at 2.5V. This means Paula must output a signal centred around 2.5V and transitions at most +/-1V.
To calculate the gain, it is Rf/Rin or R331/Rpaula and Rpaula is unknown. I'll do some emperical analysis (trial and error) and see what I come up with.

[Krashan]

Originally Posted by Stedy

I don't know for sure what the output stage of Paula is to model it. R331 and R321 are used on inverting amplifiers with Avref at 2.5V. This means Paula must output a signal centred around 2.5V and transitions at most +/-1V.

Paula DACs have current outputs and the first stage op-amps work in usual current-to-voltage converter configuration.

[Stedy]

Originally Posted by Krashan

Paula DACs have current outputs and the first stage op-amps work in usual current-to-voltage converter configuration.

I should have spotted the Transimpedance amplifier!

I modelled it as an 8mA current source and the simulation has a nice sinewave output with 360 ohms for R331/R321.
Change to 1.5K and the output goes a bit crazy. I can play a bit tomorrow.

[Stedy]

Me again.


This should explain the differences in audio circuits. To adjust the signal level, you need to change the parts highlighted. If it is built with the values shown in the schematics from Amigawiki, it will be too loud and distort.

I checked my rev 2B A600 and it has the values I highlighted as need changing in red. The schematics I have, issued by Commodore, match my A600. It was on this A600 that I obtained the measurements shown previously.

[ugrnm]

Wow, so many replies!

Originally Posted by Daedalus

What's key to note though is that the output adds together the channels at the output, so for my measurements at least, playing a sine wave on a single channel wasn't enough to get clipping.

Of course! Ooops. My bad.

Originally Posted by Daedalus

AmigaTestKit has some useful functions for generating sine waves and turning on and off channels and the audio filter.

Perfect, thanks!

Originally Posted by pandy71

Sadly to say but Paula has no separate pin for Audio Vref so or they use some internal Vref source or using Vcc as Vref so expected Vpp for Audio signal will be =<5V for sure.

But that should be fine for pro line level, no? Vpp of pro line level is <3.5V.

Originally Posted by pandy71

Hope you already replaced R309 with proper ferrite bead, normally i would also separate Paula GND and corresponding Audio GND trough ferrite beads (also add ferrite beads on Audio line outs from Paula to separate relatively slow SR OPAMP from HF).

Hmmm, starting with R309, which ferrite bead value would you recommend then?

Originally Posted by Stedy

To adjust the signal level, you need to change the parts highlighted. If it is built with the values shown in the schematics from Amigawiki, it will be too loud and distort.

I checked my rev 2B A600 and it has the values I highlighted as need changing in red. The schematics I have, issued by Commodore, match my A600. It was on this A600 that I obtained the measurements shown previously.

Thanks for following the rabbit hole I haven't had time yet to measure my A600, will hopefully be able to do that this week-end, and then get into changing values.

[pandy71]

Originally Posted by ugrnm

But that should be fine for pro line level, no? Vpp of pro line level is <3.5V.

Issue is not in level but in inability to isolate analog circuitry from digital part as they share same power line - unless Paula has embedded some fancy circuitry to provide true reference current/voltage to DAC's then you will have lot of unwanted sounds from digital circuitry on analog signal out lines.

Originally Posted by ugrnm

Hmmm, starting with R309, which ferrite bead value would you recommend then?

Something like https://eu.mouser.com/c/passive-comp...t%7C~Impedance

Highest possible impedance, maximum DC resistance bellow 1 Ohm, not sure about accurate Paula power consumption but i think ferrite bead should allow current not less than 500mA, ferrite bead tailored for rather low/standard frequencies...
Anyway even randomly selected ferrite bead will be probably better than resistor (strange that Amiga guys didn't selected ferrite bead instead 1 Ohm resistors - this can be explained by relatively new idea of ferrite bead and perhaps underestimation for signal quality expectations) - there is famous application note from linear technology (now owned by analog devices) https://www.analog.com/media/en/tech...tes/an101f.pdf - last page says everything important about ferrite beads - "megahurts to minihurts converter" this very good description for this small component capabilities.

Also avoid MLCC capacitors as most of them is piezoelectric and may produce audible interference's.

[ugrnm]

Originally Posted by Stedy

Me again.

Attachment 74022
I checked my rev 2B A600 and it has the values I highlighted as need changing in red. The schematics I have, issued by Commodore, match my A600. It was on this A600 that I obtained the measurements shown previously.

So basically the audio circuit on your 2B A600 is the same as the audio circuit of my 1.3 A600? Below is the full schematics found there http://www.devili.iki.fi/mirrors/4x4.hopto.org/ and that matches the components on my machine.



Also does that mean that the schematics on Amigawiki are wrong, or that *some* A600 models had a for while the same audio circuit as the A1200?


Going to measure the voltages now.

[ugrnm]

Some measurements with a scope, using the 500Hz sine wave from AmigaTestKit. No clipping, offset, weirdness was visible.

At the exit of C334:

• Vpp 1/R + 2/R 5.64-5.68
• Vpp 1/R 3.12-3.16
• Vpp 2/R 3.08-3.12

At the exit of C324:

• Vpp 0/L + 3/L 5.84-5.88
• Vpp 0/L 3.04-3.08
• Vpp 3/L 3.24-3.28

Not sure if it's worth investigating but 0/L and 3/L are not as balanced as 1/R and 2/R. I should probably not loose sleep over it...

At the exit of CN3

• Vpp 1/R + 2/R 1.66-1.68
• Vpp 1/R 1.00-1.02
• Vpp 2/R 1.00-1.02

At the exit of CN4

• Vpp 0/L + 3/L 1.74-1.76
• Vpp 0/L 0.98-1.00
• Vpp 3/L 1.04-1.06

So as far as I understand my stock A600 1.3, and yours Stedy, are a bit louder than the standard for consumer line level (0.894).

And if my math is correct, I would just have to replace R335 and R325 with 1.5K resistors or maybe 1.4K to be on the safe side? Vpp of pro line level is 3.472.

[ugrnm]

Originally Posted by pandy71

Issue is not in level but in inability to isolate analog circuitry from digital part as they share same power line - unless Paula has embedded some fancy circuitry to provide true reference current/voltage to DAC's then you will have lot of unwanted sounds from digital circuitry on analog signal out lines.

Yes you're correct, and there are 2 issues in fact. Signal level is important for me as the A600 will be plugged in a mixer with other synths that are all louder and not necessarily with a master volume control, so I'm trying to bring the Amiga to the same line level as everything else using this so-called pro line level as ref.

I understand this is not going to make the signal more clean, but with everything more or less on the same level, I don't have to mess with extra gain on the mixer, and it also gives me more room to adjust the volume up and down on the Amiga channel.

If the resistor switch I mention above allows me to achieve this, then it will be already a great quality of life improvement

Then issue number 2, is the cherry on top, that is to say, any simple things I could do afterward to help improve the signal quality, like your ferrite bead suggestion.

[Stedy]

Originally Posted by ugrnm

Some measurements with a scope, using the 500Hz sine wave from AmigaTestKit. No clipping, offset, weirdness was visible.

At the exit of C334:

• Vpp 1/R + 2/R 5.64-5.68
• Vpp 1/R 3.12-3.16
• Vpp 2/R 3.08-3.12

At the exit of C324:

• Vpp 0/L + 3/L 5.84-5.88
• Vpp 0/L 3.04-3.08
• Vpp 3/L 3.24-3.28

Not sure if it's worth investigating but 0/L and 3/L are not as balanced as 1/R and 2/R. I should probably not loose sleep over it...

At the exit of CN3

• Vpp 1/R + 2/R 1.66-1.68
• Vpp 1/R 1.00-1.02
• Vpp 2/R 1.00-1.02

At the exit of CN4

• Vpp 0/L + 3/L 1.74-1.76
• Vpp 0/L 0.98-1.00
• Vpp 3/L 1.04-1.06

So as far as I understand my stock A600 1.3, and yours pandy71, are a bit louder than the standard for consumer line level (0.894).

And if my math is correct, I would just have to replace R335 and R325 with 1.5K resistors or maybe 1.4K to be on the safe side? Vpp of pro line level is 3.472.

I wouldn't fiddle with R335/R325 if I were you. A typical measurement accuracy of an oscilloscope is +/- 2.5% +0.2 division, your readings are within the margin of error. If you have a true RMS 5.5 digit or better, DMM, you could get a more accurate reading.

The audio circuit uses 5% tolerance resistors, a typical worst case RSS error could be a +/-8.66% variation in the signal amplitude due to the resistors. alone. I'll spare you the pain of tolerance analysis calculations, what you have now is about right and good enough.

[ugrnm]

What could go wrong if I mess around with R335/R325? Bare in mind I'm very new to electronics so I may not see the bigger picture.

My intention is to make a simple mod to have a higher signal output to plug the Amiga on a mixer where all the other devices are much louder (pro line level vs consumer line level of the graph I posted in the first message). Said differently, I'd like to have the A600 send something around 3.4V. Currently it generates 1.7V.

That's why I thought to achieve this by altering the last voltage divider.

[Stedy]

Originally Posted by ugrnm

What could go wrong if I mess around with R335/R325? Bare in mind I'm very new to electronics so I may not see the bigger picture.

My intention is to make a simple mod to have a higher signal output to plug the Amiga on a mixer where all the other devices are much louder (pro line level vs consumer line level of the graph I posted in the first message). Said differently, I'd like to have the A600 send something around 3.4V. Currently it generates 1.7V.

That's why I thought to achieve this by altering the last voltage divider.

Hi,

My apologies, I've treated this circuit like a typical precision op-amp circuit when I did not need to. To boost the output change R325 and R335 from 390R to either 470R or 510R. This will get you a peak voltage of 0.9408V for 470R resistor or 0.9702V for 510R. You won't get the exact 1.095V reading due to the 5% tolerance of components used on the Amiga.

You need some 1206 510R (or 470R resistors) 1%. Ebay are pricey for a few resistors but if that is all you need, it's not worth using a regular distributor with their postage costs.

Good luck.