Genesis Audiophiles... please help me rate the audio quality of my systems

Hey,

I've got a few Genesis systems, and I made a simple mod to one of them... but I've never really been able to notice much difference between what's considered "bad" and "good". Because I know which ones are which, and from actually analyzing the file, I have an order that I'd rank these... but from simply listening blindly, I hardly notice any difference. Would some of you audiophile types check out these unmarked recordings and rank the systems in order from good to bad?

http://www.megaupload.com/?d=N2JETQC9

Sorry for the large file (and the wait from megaupload), but encoding them as MP3s did degrade the quality, so I figured WAV was the best bet.

There's 8 wav files in this zip. 1a, 1b, 2a, ..., 4b. Files a and b are different recordings of the same system (so there's 4 different system recordings). They were all recorded using the same hardware and settings. I did this using Sonic 1 (through the Genesis 6-PAK cart)... a is the first level, b is the bonus after the first level.

If anyone knows a game that's better for noticing differences, I'm all ears :fail: .

Thanks,
DogP

Quote:

---

Originally Posted by DogP

I've never really been able to notice much difference between what's considered "bad" and "good".

---

There's one bad model 1 variant and I think two good model 2 variants. Chances are you ended up with one of those. - Trust me it is impossible not to notice the difference between a really good and a really bad one.

No. 2 sounds like a HDG model 1. No. 3 is horribly distorted. Must be a model 2 or a model 1 VA7.

tldr: Go with #2!

Thanks for the reply... it's good to know that there IS actually a audible difference. Would you mind ranking them in order from good to bad (in your opinion)?

DogP

Here's my take on the systems:

1: No distortion, but with a noticeable hiss. Could be a Genesis Model 2 motherboard revision VA0 to VA1.8 with C324G quad op-amps.
2: Clean, clear and heavily filtered. Seems to be a Genesis Model 1 with motherboard revision VA2 through VA7. I'll need to listen to Spring Yard Zone for quantization noise to confirm.
3: Heavy hiss and noticeable distortion. Definitely a VA0 to VA1.8 Genesis Model 2 with LM324 quad op-amps
4: Sounds very similar to system 2. I assume this is a VA3 or VA4 Genesis Model 2 since the volume balance between FM Synthesis and PSG is correct. Again, I need to hear Spring Yard Zone. If there's no audible quantization noise, this system is one of the better Model 2s(I need to hear the music ONLY, so you'll have to record through the sound test).

From best to worst, they rank as follows: 2, 4, 1, 3. System 3 is absolutely horrendous.

One game that really exposes sound flaws in a Genesis is Thunder Force II MD(there's no MD in the American and European versions' title, just so you know). It's a game that has a HORRIBLE hissing noise when played on consoles with bad sound circuits. It's also a PSG-heavy game, so it will expose any volume balance problems between FM Synthesis and PSG.

I agree with Ace: The best one is #2. The worst is #3. There isn't that much of a difference between #4 and #1. If I had to choose I probably would rank #1 over #4, but that could change after some more in-depth comparisons with Streets of Rage and Thunder Force.


If you wanna hear a really, really bad model 2, check out the Slam City footage in AVGN's Sega CD review:

Okay... so I'll out system #3. It's a Gen 2 w/ BA10324AF. This was the one I thought that this sounded the worst as well, though it wasn't appalling to me to listen to.

I can take some more recordings, but unfortunately, I haven't found my box of Genesis games, so all I have laying around are the 6-PAK (Sonic 1, Golden Axe, Streets of Rage, Revenge of Shinobi, Columns, and Super Hang-On), Sonic 2, Sonic & Knuckles, Aladdin, Lion King, Zoom!, Out Run 2019, and Fire Shark. Any of these good for comparison?

For Streets of Rage, any specific sounds/music that's good? And anything specific in Spring Yard Zone?

I really appreciate the help... hopefully listening to Sega music doesn't feel like work to you guys. ;) And don't worry, I won't make you listen to #3 again. :)

DogP

With SOR2 anything with bass (yeah, 90% of the songs) is a good test. The deep bass notes are horrendous on the bad systems. On my model 2's they sound like farts, on the good systems they are clean midbass notes. The opening song at loud volume may be the best test.

Same with Doom on 32X. Genesis 1s sound like a ton of bass, everything else online sounds like a fart metal concert.

Unfortunately I don't have SOR2. I do have Doom on 32X, but no Gen 1 cable, and I was hoping to simply compare raw Genesis to raw Genesis without anything else thrown into the mix. Oh... about the "sound test"... is there a way for me to get to that w/ the 6-PAK?

Thanks,
DogP

It's the typical Sonic the Hedgehog sound test code: UP, DOWN, LEFT, RIGHT at the title screen, then hold A, B and C before pressing START when you hear the ring noise after inputting UP, DOWN, LEFT and RIGHT. Spring Yard Zone is sound code 85.

Cool... I dunno why I couldn't find that code anywhere. Anyway, here's the remaining 3 systems on SOR (Round 8 for 30 seconds, then Round 1 for 30 seconds) (file a), as well as Spring Yard Zone (file b). I feel helpless listening to this stuff... every time I think I hear something bad, I go and listen to it on the others and it's there too. :P

http://www.megaupload.com/?d=8XRR4IV0

Thanks again!
DogP

#3 is the HDG model 1. It provides by far the most clarity and range (see Fighting in the Street, the second half of the recording). There's not much of a difference between #1 and #2. I'd say #1 sounds *slightly* brighter and a tiny bit cleaner, so I'd rank that one over the other, but the difference is barely noticeable with those damn Sony headphones I am using right now.

tldr: #3 is the best, #1 is better than #2

You beat me to it, Christuserloeser. System 1 is definitely better than system 2, which has the typical hissing noise from crappy Genesis Model 2s and the VA7 Genesis Model 1, but with no distortion.

Oddly enough, system 1 sounds considerably more muffled than either my VA3 or VA4 Genesis Model 2s, and this particular system can only be a VA3 or VA4 Genesis Model 2 because the sound is hiss-free and quantization noise is barely audible. I think I'll record the audio coming out of my VA3 and VA4 Genesis Model 2s for comparison purposes.

Thanks a ton guys! That confirms what I thought from looking at the spectrum of the audio. Here's the answer to what each one was:

First set:
1=Gen 2 V1.8 w/ C324G
2=HD Gen 1
3=Gen 2 V1.8 w/ BA10324AF
4=Modded V1.8 Gen 2

Second set:
1=Modded V1.8 Gen 2
2=Gen 2 V1.8 w/ C324G
3=HD Gen 1

Basically, I thought the BA10324AF didn't sound very good, but was intrigued that the same system with just a different op amp (C324G) could sound better (and the LM324 could sound even worse). So, while ordering other stuff from Digikey, I looked for a direct replacement op amp with the best performance (especially GBWP). I came across the TLC274BIDR ( http://search.digikey.com/scripts/Dk...=296-1312-1-ND ). So, I swapped each BA10324AF op amp for a TLC274BIDR. The BA10324AF is actually a dual-supply op amp (+/- voltages), but the Genesis has - connected to GND, so the TLC274BIDR (single-supply) is a direct replacement.

The op amp replacement was the only thing changed on the system (#3 from the first set), and I personally thought it looked better than the C324G, but really needed confirmation from someone who really noticed the differences audibly. It's nice to know that simple mod (~$5 shipped, and ~10 minutes of time) greatly improves the quality of a BA10324AF system and likely a LM324 system as well. It sounds like you're saying that it's better than the C324G as well, though not to the same extent as the others.

Here's a pic of the original:
Attachment 2861

And replaced:
Attachment 2862

DogP

And here is the spectrum of the signals, from the Sonic 1 bonus level... in order that you guys ranked bad to good:

BA10324AF:
Attachment 2868
I noticed the lack of crispness, and lots of distortion.

C324G:
Attachment 2866
More crisp, but still distorted.

TLC274BIDR:
Attachment 2867
Looks a lot more crisp... closer to the HD Gen 1, but not as crisp in the lows, and weaker on the highs.

HD Gen 1:
Attachment 2869
Well... what can I say... this is what everything compares to.

The other thing I noticed is in quiet spots, all Gen 2s had digital noise, especially on the left channel (like at the end of the bonus, and before the title screen). I wonder if that's coming out of the chip on the audio line, or something like bad decoupling of the amp. Of course I'm sure it's coming out during non-quiet times as well, but it's really noticable when it should be quiet. I also did a recording of quiet time to get a base noise level... the Gen 2 had some low frequency noise (<1k or so), as well as the extra left channel noise. The Gen 1 also had a little noise, but it was distributed very evenly over the entire spectrum.

Edit: Well, it looks like it converted my PNGs to JPGs and shrunk them... if anyone's interested, I can post them somewhere else in higher res.

DogP

Quote:

---

Originally Posted by sheath

Same with Doom on 32X. Genesis 1s sound like a ton of bass, everything else online sounds like a fart metal concert.

---

There's not much difference either way (your own recordings are less distorted and almost exactly like fusion, but it still sounds poor to mediocre).
There's perhaps 1 or 2 tracks that sound as good or better than Adlib on the PC version, the rest are worse IMO. (all should have been better by a good margin . . . E1M1 being one of the worst examples doesn't help first impressions either)

Even for the default GEMS midi instruments it's bad, no percussion samples, no remixed arrangements to cater better to the GEMS instruments, just a weak conversion from the PC MIDI. (I don't think there's any use of PSG either)

I'll need to look through the tracks again to remember the 1 or 2 that actually managed better than Adlib.

Would you guys mind listening to this one: http://www.megaupload.com/?d=GBZPFHK7 ... it's the same modified system, but with a couple upgrades. I was trying to reduce the base level of noise (especially digital noise), and looking at the signal I think it cleaned up a little bit, and maybe slightly improved the signal.

I just added a couple bypass caps, and probing the ground, I noticed >50mVpp potential difference between the outer ground pour and ground at the chips... so I also added a ground jumper directly to each chip. I don't think the difference is drastic, and might not even be noticable... but hopefully someone could confirm that, or tell me otherwise.

Of course I probably should have just done TmEE's mod... but I'm way too lazy to solder all that stuff together. I was hoping just to improve the Gen 2 to the best it can be with the least amount of effort.

Thanks,
DogP

Digital noise is introduced in the sound right in the ASIC so there's not much to do to get rid of it. The output of YM is very high impedance and you should buffer the signal right next to the chip which would prevent it from picking up additional noise through the traces on the board. Boosting YM DAC voltage also helps regarding the noise, which is exactly what those solder bridges do on my CCAM install instruction pics. Only problem with that boost is that those shitty opamps will mess the sound up even more as the amplitude of FM signal gets increased, and the opamps are already out of bandwidth, incapable of amping without distortions....

Thanks for that info. Question about boosting the YM DAC voltage though... if that increases the signal levels, doesn't that affect the balance of the audio? I assume your circuit takes care of that, but wouldn't work on the original without more tweaks.

IMO, this sounds pretty good, but could use more highs. Is there a component that can be changed to bump up the filtering frequency.

Have you ever considered making a PCB with your CCAM? It seems like you could make it quite small and clean using all SMT components. Maybe the demand isn't there, but IMO if there was a simple to install "mod chip" style CCAM board, it'd be pretty popular.

BTW, thanks for posting your schematic of the Gen 2 audio section... it makes probing around with the scope much nicer than trying to trace everything myself. :)

Thanks,
DogP

You can change some filter capacitors but this only increases distortion... the opamps are out of bandwidth so there's no improvement without tradeoffs possible.

I have considered making a PCB of CCAM, in fact I have actually made a PCB, but having lack of time and free funds its on hold.

While I'd agree that the stock op amps are out of bandwidth, looking at the spectrum of mine w/ the TLC274BIDR op amps, I think I could go up in freq without too much distortion. The TLC274BIDR has ~4x the GBWP of the BA10324AF and C324G, and I should be able to increase that even more without much trouble. Do you know which caps and values would work well to reduce the filtering of the highs?

Thanks,
DogP

Could you guys give this set a listen? http://www.megaupload.com/?d=S0OPANTT

I can personally hear a difference between this and the others, and to me, this sounds the best out of any of my recordings. Could you guys give me comments on what you think?

Thanks,
DogP

Bump... would someone mind listening to the last one and give me your thoughts?

Thanks,
DogP

Sounds crystal clear and very clean. A lot brighter than even your model 1, no doubt about that.

Quote:

---

Originally Posted by Christuserloeser

Sounds crystal clear and very clean. A lot brighter than even your model 1, no doubt about that.

---

Thanks... that's what I thought as well. For anyone interested, I simply bumped up the low pass filters C21 and C23 from 1000pF to 270pF, which should increase the passband (of that stage) by about 4x. With the original filter, it seemed to start rolling off pretty steeply at ~6k... the Gen 1 seems to slowly roll off through ~10k, and this one seems to start rolling off at ~13k.

DogP

It does not sound ideal but certainly better than stock stuff. On my MD2 VA0/1 models doing what you did would have ended up in significantly distorted sound. If one can get a negative voltage rail created then bumping up the opamp supply would do wonders, +/- 12V would be sweet

Quote:

---

Originally Posted by TmEE

It does not sound ideal but certainly better than stock stuff. On my MD2 VA0/1 models doing what you did would have ended up in significantly distorted sound. If one can get a negative voltage rail created then bumping up the opamp supply would do wonders, +/- 12V would be sweet

---

Well, this is with the replaced op amp, as well as bumping up the LPF. I could gain another 500KHz of GBWP by running it at 10V rather than 5V (which I could do by jumpering to the voltage input, but would need some filtering, and maybe regulation). I'm sure if I searched harder I could just find a better chip as well... maybe something like this: http://search.digikey.com/scripts/Dk...L94T-E/SLCT-ND , which should be pinout compatible, with 10MHz GBWP @ 5V. Single supply op amps typically have less noise than bipolar ones, due to the typically lower input bias current, so there is an advantage to staying w/ single supply, since Sega already did the work to properly couple everything.

BTW, what doesn't sound ideal?

Thanks,
DogP

The model 1 still has the advantage of the cleaner bass & pcm sounds (kick drum in Fighting in the Street, 2nd part of 5a IIRC).

TmEE's Crystal Clear Audio Mod was adjusted to faithfully reproduce the model 1 sound (without the earlier roll off of course) so it remains the superior method.

Still, from what I understand your approach requires less work- and material and would work with most model 2s (aside of the VA0/1?). So that is quite an achievement!

I'd love to ask you for a detailed schematic / manual / guide on what you did. (I have no idea if the information you provided is already sufficient to understand what you did exactly, so sorry if this question is redundant)

Ideal is full audible freq range without noticable distortion. CCAM freq cutoff is around 23KHz

Quote:

---

Originally Posted by Christuserloeser

The model 1 still has the advantage of the cleaner bass & pcm sounds (kick drum in Fighting in the Street, 2nd part of 5a IIRC).

TmEE's Crystal Clear Audio Mod was adjusted to faithfully reproduce the model 1 sound (without the earlier roll off of course) so it remains the superior method.

Still, from what I understand your approach requires less work- and material and would work with most model 2s (aside of the VA0/1?). So that is quite an achievement!

I'd love to ask you for a detailed schematic / manual / guide on what you did. (I have no idea if the information you provided is already sufficient to understand what you did exactly, so sorry if this question is redundant)

---

Yeah, I'll post a quick write up on it later... it really is very simple though (swap two op amps and two capacitors). It should take less than 10 mins for someone familiar with soldering. I have to order a few other things from Digikey, so I'll try to find a better op amp than the TI one that I'm currently using (I found the TI one as a single supply cross for the 324 op amp), and I'll do the writeup when I modify my other VA1.8.

I did notice that the lows on SOR sounded a little different... it didn't sound bad, so I wasn't sure if the Gen 1 just had something causing a beat frequency that was showing up, or if it was really missing something. I wish I had a game w/ a sweep generator to really get a feel for the frequency response, though I guess I could just inject my own signal.

Quote:

---

Originally Posted by TmEE

Ideal is full audible freq range without noticable distortion. CCAM freq cutoff is around 23KHz

---

Yeah, I really liked the look of the spectrum w/ 100pF caps (opened up the top end), but it didn't sound as good IMO... probably running into the BW limit of the op amp. A better op amp should fix that. You're much more familiar with this circuit than I am... if you have any other suggestions, feel free to chime in. Of course I don't mean to step on your toes in any way... the CCAM is quite impressive, but I'm just far too lazy to build that circuit, and if a simple parts swap will make it sound good, I'm all for it. Plus, as I'm sure you know... hacking hardware is fun. :)

DogP

EDIT: Problem solved!

Quote:

---

Originally Posted by chessage

EDIT: Problem solved!

---

I saw your question, but didn't have time to reply... anyway, how does it sound with that chip? I took a quick glance at that datasheet earlier and saw a GBWP of 4.5MHz, but that was at +/-15V... at +5V, I'll bet the GBWP is quite a bit lower.

DogP

Quote:

---

Originally Posted by Christuserloeser

The model 1 still has the advantage of the cleaner bass & pcm sounds (kick drum in Fighting in the Street, 2nd part of 5a IIRC).

---

Model 1 doesn't have cleaner bass, just heavier filtering. :p
Some things sound "wrong" without that filtering as they relied on it, other things sound fine (or better) without filtering.
I think most/all of Tiido's stuff is intended to be best with super clean/clear sound without filtering. (closest thing short of modding would probably be the CD-X/Multimega's audio -not from the headphone jack, but the AV port)

Of course, if you have no internal filtering, you can always use treble/bass filter/boost controls on a TV and/or stereo/surround receiver/amp to adjust things as desired. (including on a per-game basis, which I often do)

On another note, it's bassy stuff (or lower frequency sounds in general) that seem to suffer less from bad (or on the edge) amps in MDs, as well as "good" models with distortion from aging components.
Loud, high-pitched sounds are the absolute worst, though loud bassy sounds are also susceptible. (PCM, especially for speech is generally bad too as there tends to be some high pitched and loud areas -speech has an extremely wide dynamic volume and frequency range range in most cases, so can be especially problematic)

Heavy filtering also seems to minimize those problems . . . or mask them rather. (make everything all muffled) That works OK for the on-the edge, acceptable amps in some VA0/1.x model 2s (forget the actual amp models), but the really bad ones are pretty much a lost cause without modding. (albeit, they aren't as bad as the Tectoy MD or some similar clones . . . or the separate emulation related issues with Atgames stuff)
The super heavy filtering on some crappy Adlib/SB compatibles didn't really address those problems either, though I think those tended to get even worse than the really bad MD/Genesis examples. (probably not as bad as some modern clones)

Quote:

---

TmEE's Crystal Clear Audio Mod was adjusted to faithfully reproduce the model 1 sound (without the earlier roll off of course) so it remains the superior method.

---

That's just added filtering if you want it. ;) You still avoid the bugs and quantization noise issues of "good" model 1s . . . and also avoid the hissy headphone amp. (albeit, in a recent comparison I did with some friends 'and my own MD/Genesis systems, it seems that the mono amp is often no better than the headphone amp in model 1s -in several cases, mono showed more audible distortion than the headphone amp, though in some others it was better than the headphone amp -several systems shows audible, though moderate, distortion as well, similar to what bad models do, but far more subtle -not sure if that's related to aging caps or what not, though I know at least 1 has some technical problems -pops in the headphone amp at low volume and crashes the CD)




On another note: have there been any successful mods completed (with instructions) for improving model 1 sound output?
The main issue would probably be bypassing the less than ideal headphone amp for a proper stereo amp set-up. (playing the the filtering levels would be another area of interest)

No, I am pretty sure the bass sounds are cleaner on a model 1 than on any model 2 I heard so far. And model 2s do sound the worst with deep bass sounds. It's more like farts, really. But yeah, higher tones are awful too. Then again I am not sure why we are even debating this: ANYTHING is awful with a model 2. =P

Quote:

---

Originally Posted by Christuserloeser

No, I am pretty sure the bass sounds are cleaner on a model 1 than on any model 2 I heard so far. And model 2s do sound the worst with deep bass sounds. It's more like farts, really. But yeah, higher tones are awful too. Then again I am not sure why we are even debating this: ANYTHING is awful with a model 2. =P

---

A lot of FM stuff is inherently "farty" even under the best circumstances. :p (albeit I'd call it more grating than "farty" in the worst cases)

I think you and I are thinking of different types of bassy sounds too (bass guitar-ish sounds are among the most noticeably distorted). OTOH, it's the bass stuff on the borderline model 2s with OK amps that generally sounds fine. (and it's also the bassy stuff that doesn't degrade much from heavy filtering -being in the low frequency range and thus not cut off)

I'm still not sure what you mean about clearer bass. The CCAM recordings I've heard have very clear bass, but obviously much stronger treble . . . so in that sense the bass is "clearer" as it has less to compete with. :p (as far as noticing more distortion in bass sounds, I haven't noticed that)



On another note, here's a good (or bad) example of what poor Adlib/SB cards can do:
http://www.youtube.com/watch?v=MK_OwAaltLw


vs how it's supposed to sound:
http://www.crossfire-designs.de/inde...dcards/&page=6
http://www.crossfire-designs.de/down...ib_tempest.mp3


this one isn't too bad either:
http://www.youtube.com/watch?v=lQkQo2h3S6k

Looking at the recording of the Model 1, it appears that it has a sort of "bass boost". Where higher frequencies are equal in amplitude between the Model 1 and Model 2, the Model 1 has a higher amplitude of lower frequencies. I added a bass boost "filter" on my model 2 recording, and it sounded better IMO (though still not identical to the Model 1).

DogP

Do you normalize volumes between your recordings ?

I set the PC recording volume to good levels for the Model 2, then adjusted the Model 1 headphone output to match that (and now leave all volume controls untouched).

DogP

Quote:

---

Originally Posted by chessage

EDIT: Problem solved!

---

The second issue has been solved!