Covox Sound Master Reverse Engineering / Replica

Discussion about old sound cards, MIDI devices and sound related accessories.

Re: Covox Sound Master Reverse Engineering / Replica

Postby moturimi1 » 2019-4-16 @ 20:56

moturimi1 wrote:Is this my Sound Mater Original that you thankfully repaired (due to my terrible desoldering skills)

Fagear wrote:Yes it is your board. I've repaired pads and traces around GAL, soldered in new fresh socket and put in a GAL with my firmware. Your *.JED did not work.


I am so impressed by all of your skills.
I would have loved to follow your Youtube videos, but as these are in Russian and there are no English subtitles you do not get the attention you deserve!

Fagear wrote:Prototype also works, but has no R2R DAC yet and has some issues with output level.


I would like to order one of your designed boards in the future: the FMonster!
moturimi1
Member
 
Posts: 159
Joined: 2014-6-01 @ 16:45

Re: Covox Sound Master Reverse Engineering / Replica

Postby Fagear » 2019-4-24 @ 20:36

moturimi1 wrote:I would have loved to follow your Youtube videos, but as these are in Russian and there are no English subtitles

Well... I barely have time to shoot and edit my videos and translating takes a lot of time, so that's out of the question. On top of that, I don't write scripts, so it's impossible to just dump my script into Google translate and bodge auto-translated subtitles on each video. I have to transcribe all my speech into text first. Which also will take a ton of time, so it is not possible.
If I have to do something that will allow auto-translated (by Google) subtitles and/or community translated ones - let me know. I'm all for translations, but I just don't have time to do them myself.

moturimi1 wrote:I would like to order one of your designed boards in the future: the FMonster!

I think it will be a long time from now... :roll:

Now on the topic.
I've made some measurements and tested original board alongside with my replica.
Not good news for original board... As I've described some time ago, layout of the board is terrible. Measurements show noise (more like rumble and crackle) level is about -35 dB. :exclamation: When anything is active in the PC (CPU, RAM, etc.) noise level rises to -15 dB. :dead: You can notice this noise in the video above even when music is playing.
Also I've discovered that output of AY (about 1.0...1.5 Vpp) gets divided by 100! Right after AY output there are 100k:1k dividers for each channel right before "mixer" on a group of resistors. So... signal gets drown in noise and after that all resulting garbage gets heavily amplified in the final stage on LM386. Result is pretty awful. :depressed: Insufficient voltage filtering on +12V (used to power opamps) and complete lack of such filtering on +5V rail (that powers AY directly) makes matters worse.
CSM_orig_noise.png
CSM original: idle noise

After some thinking, cutting of traces and testing I've done the following:
- replaced crude "RC-filter" on the +12V rail with 78L09 linear +9V regulator;
- disconnected power supply of AY and 4066 switch from digital +5V, added 78L05 regulator from +9V rail and powered AY from it;
- temporary replaced 100k resistors in dividers with 2.2k ones.
The last one will have to be ajusted later, because signal from R2R DAC in the same mixer on the original board gets through 6.2k:1k divider (much less attenuation), but transistor amplifier on the original is powered from +5V and on my replica it is powered from +9V. So i have to level out differences between AY and DAC channels later (when I'll make my own R2R ladder).

All that gave me noise level about -60 dB. Not perfect yet, but pretty good. :cool:
CSMR_mod_noise.png
CSM replica, power modified: idle noise

After that I've made recordings of original board and of my replica in the game Bad Blood and also I've recorded DuoSID in the same game on the same machine. DuoSID get lower than -70 dB of noise, which is great.
But when I accidentally put SID and AY together on the timeline... I've discovered something new, something that nobody didn't hear before.
You see, music for SID and for AY in Bad Blood is pretty simple and "lazy", but the same melody is written differently. And when SID and AY play together... they rock! :cool:
That's how this short video was born: https://youtu.be/1fpP_-TrkMY
CSM replica with AY-3-8910 (which sound very different from AY8930 in the original, it plays in higher octave) and DuoSID with SwinSID.
User avatar
Fagear
Member
 
Posts: 268
Joined: 2014-7-10 @ 22:27
Location: Russia, Moscow

Re: Covox Sound Master Reverse Engineering / Replica

Postby shock__ » 2019-4-25 @ 21:49

Automatic transscription + translation by youtube/google works incredibly well to get a good clue of what is going on in videos that are in russian.
If you have indepth questions you can always ask here I guess :P
User avatar
shock__
Oldbie
 
Posts: 994
Joined: 2010-12-22 @ 01:53
Location: Berlin, Germany


Re: Covox Sound Master Reverse Engineering / Replica

Postby Fagear » 2019-5-25 @ 23:02

I've assembled a couple of R2R DACs. Resistors are not those ones that should be there, but it will be helpful for comparison with proper resistors.

IMG_8932.jpg
R2R DACs
IMG_8941.jpg
R2R DACs
IMG_8944.jpg
R2R DAC + CSM
IMG_8947.jpg
R2R DAC + CSM
User avatar
Fagear
Member
 
Posts: 268
Joined: 2014-7-10 @ 22:27
Location: Russia, Moscow

Re: Covox Sound Master Reverse Engineering / Replica

Postby Fagear » 2019-10-01 @ 23:24

An update.
I've revised and updated GAL replacement logic and redid SMD layout from scratch. :dead:

csm_sch_2019-10-01 22.28.19.png
CSM GAL replacement @2019-10-01

I've managed to remove a couple of ICs from the GAL replacement section and re-structure logic utilization between different parts of the board.

Also, at the moment I beleave that no software was using "4-bit volume control" feature, so I've removed diode-resistor DAC from AY's port A and corresponding LM13600 opamp.
I've left in place "mono switch" on output, but reconstructed analog output section a bit.

AY8930 (and YM2149) have internal frequency divider, that was used on the original board, giving 14.318M (from ISA) /4 (on D-triggers) = 3,58M /2 (on internal divider) = 1.78975M for AY.
But "the original" AY-3-8910 doesn't have that. So, I've decided to switch off internal divider and place an external one on another D-trigger.

Next, I've thought "what the heck, AY-chips of all kinds were used in many different machines, what frequency it was run in those?".
I've did some research and came up with the list:
  • 2.000 MHz = Atari ST
  • 1.7898 MHz = CSM
  • 1.7734 MHz = ZX Spectrum
  • 1.75 MHz = Pentagon 128k
  • 1.000 MHz = Amstrad CPC
Feel free to correct me if I'm wrong.

So... Keeping in mind FMonster project and its clock generation block I've managed to implement all those frequencies in new board. :lol:

Also I'm thinking about removing joystick ports from SMD-variant of the board. But I'll keep those in "proper" TH replica. :confused:

On the screenshot below there are no:
  • PCM circuitry
  • analog mixing/filtering stage
  • clock generation unit

AY_SMD_2019-10-02_top.png
CSM @2019-10-02

White thick line denotes fully completed "AY-only" block (no PCM, no DMA, no joysticks) ready to be put into FMonster.
User avatar
Fagear
Member
 
Posts: 268
Joined: 2014-7-10 @ 22:27
Location: Russia, Moscow

Previous

Return to Sound

Who is online

Users browsing this forum: No registered users and 0 guests