Listen up C3 owners (and potential buyers)

Ahle2

A couple of weeks ago when I first got my C3, I couldn't wait to give the C3 a thorough test run.
While I was VERY impressed by all the features and the small size of it, something caught my attention while trying different sound drivers for it. Everything sounded muffled.
I had a little look in the PDF manual and found the values for the RC filter that is used to integrate the audio signal before output. I calculated the -3db cutoff frequency to be just 1.6kHz; which is very low.

Listen to the attached mp3 file and then vote for if you think that parallax has to do a redesign of the audio filter in the C3.

Description:
- 3 tunes are played; each with a different sound driver
- The sound is recorded straight from the C3 to a reference studio audio-card in 44.1 kHz
- First, each tune is played with a cutoff frequency of ~20kHz
- Then, the same tune is played with the normal C3 cutoff frequency of ~1.6 kHz
- The cutoff frequency changes after the beep

Tunes and driver information

Tune 1: "Mega Twins"
Driver: AYcog by Ahle2
Mixing freq: 125kHz
Origin: Atari ST register dump

Tune 2: "Funk Junk"
Driver: SIDcog by Ahle2
Mixing freq: 31kHz
Origin: C64 register dump

Tune 3: "Dire Straits - Money for nothing"
Driver: Wav-player by Raymond Allen
Mixing freq: 44.1kHz
Origin: My private collection.

/ Ahle2

Coley

A bit of a no-brainer really

Oldbitcollector (Jeff)

I think the fix is in the works...

Yes, It's a good idea to address this.

OBC

Dr_Acula

I calculated the -3db cutoff frequency to be just 1.6kHz; which is very low.

Yes, that is low. Even a telephone goes to 3khz.

Is it not just a simple matter of changing the R and C values?

AndreL

Yes, the R or the C, like I said, I used what was on the HYDRA I believe since all my demos that did sound worked best on that. But, this is something on the next rev we can adjust the filter to make it sound better if enough people don't like the way it sounds.

Andre'

Cluso99

It should be a simple matter to identify the resistor or capacitor in a picture. It should be a fairly simple mod until the next rev is available.
To increase the C just put another cap (value tbd) across the first.

potatohead

I didn't vote above. Personally, I think just doing a revision is the right thing to do. Get one sounding good, and just cycle it into the pipe. IMHO, the low cutoff may not be a issue for as many people as one would think. Many simple sounds will work just fine.

If there is to be a revision, I would personally be fairly liberal about higher frequencies leaving the board. It's pretty easy to filter that externally.

Ahle2

Parallax,
What is your input on this?

/Ahle2

AndreL

Well, I don't know what Parallax will have as an opinion on this, but like I said, if enough people don't like the way it sounds then we switch a resistor and cap, simple. But, I need to hear the sounds players people are using and then not "break" other sound players -- so, someone that is really into this, put together a number of plug and play demos that I can compile on a C3 and I will listen and then if they sound bad, make adjustments to make them all sound as good as possible -- In fact, Ahle2, why don't you tweak the filter yourself and try a couple values, for the cap, values that we have are .1, .33 and 1uF and for the resistor, anything you see on the C3 is game or any other common value. So, obviously the best thing is to simply change only R and C, you don't need to change both, and actually, don't make the cap smaller then it stores less charge, so you can make the cap bigger or make the R bigger or smaller. Do a few experiments see what you like and sounds the best for all the sound drivers you are using and then post the values, I will give them a try and we can change them next rev if they really help.

Andre'

Cluso99

I think a standard openly published circuit would help us all a lot. Ale??

Pin-----R1-----x-----C2-----Out
               |
               C1
               |
              Gnd

Phil Pilgrim (PhiPi)

Cluso99,

Try putting your schematic in a code block to preserve spacing.

-Phil

Ahle2

@AndreL
"Ahle2, why don't you tweak the filter yourself and try a couple values"
I already have.

The values I used for this comparison were 1k resistor and 7.5 nF Capacitator.
This gives a -3dB cutoff of 21.2 kHz.
As I've said before, I don't think even a driver with low mixing frequency (like 22 kHz or lower) will "suffer" from a cutoff frequency of ~20 kHz.
I rather listen to some audible aliasing distortion of such drivers than getting all the luscious overtones taken away from my demanding ears.

This reminds me of discussions on the net about the SNES audio filter, which is also vey low.
The low cutoff makes the SNES sound bad IMHO.
On the Amiga you could set a bit in a register to toggle on and off a 5 kHz lowpass filter. The reason for this filter was to get rid of audible aliasing distortion when low quality samples were played back.
The filter was rarely or never enabled, because most people (like myself) thought it sounded bad EVEN when low quality samples were played back.

/Ahle2

Phil Pilgrim (PhiPi)

I suppose the ultimate solution (for future consideration) would be to interpose a switched-capacitor LP filter in the audio stream, clocked from a Prop pin. Then the user would have complete control over the cutoff frequency. Such filters can be obtained as small as a TSSOP-8 and are not frightfully expensive.

-Phil

Cluso99

Ahle2: How does the following sound...
1. 680R & 10nF (0.01uF) f=23.4KHz
2. 750R & 10nF (0.01uF) f=22.2KHz
3. 820R & 10nF (0.01uF) f=19.4KHz

Reason is most likely 10nF is in the toolbox and resistors are easier to get. 680R & 820R easier too.

What DC blocking cap are you using?

Ding-Batty

Cluso99, I thought 100nf == 0.1 uF, not 0.01 uF?

I hope these options work -- I think the easiest modification of the C3 board is to simply solder another resistor on top of the existing one, to lower its value. That's easier than removing the original one...

Cluso99

Thanks - I was suffering heat exhaustion (36C here and no a/c) That's my excuse for a stupid set of typos.

Ahle2

When it comes to low frequency voltage signals (like audio), the impedance isn't very important. So any values will go.
In "utopia" all voltage amplifying output stages has got 0 ohm impedance and all input stages has got infinite impedance.
This is to preserve the the voltage level as much as possible over the voltage divider that forms by the output/input impedance.

However, in reality this is not the case.
The standard for audio signals is more or less:
-Input stages has got an impedance of 10kΩ or more
-Output stages has got an impedance of 150Ω or less

I have found out by experimenting with different values, that the sound will get clearer by lowering the resistor from 1kΩ to 100Ω (and of course change the cap to preserve 20kHz cutoff)
I am not quite sure why this is the case though, but I think it has something to do with the duty mode and current flow through the capacitator.

@Cluso99
The answer to you question is: as long as the cutoff isn't noticeable low those values will do.

Ahle2

Bump!

Parallax have you seen this poll?

Ahle2

Bump!!!!!!!!

porkotron

I got my C3 this week and before I read this post I thought because the muffled sound that it was already broken.
I'mvery interested in ways to make the sound better because my first project is going to be audio related.

pik33

Measured. Patched. No more muffled sound. Used 680R