OPL recording problem

Discussion about old PC hardware.

OPL recording problem

Postby Stojke » 2014-2-02 @ 11:51

Hey guys, i have recently started recording OPL3 from my PAS16 via analogue cable into my XFi Xtreme Music sound card.
What i noticed is that the sound level is quite low, around 20%, and i later have to volume maximize the entire track in order to get something loud enough to listen to with out having to set my sound card to maximum sound output level.

I later discovered that i can boost my sound card to 200% (level 100 in windows) and record at around 30% ~ 40% but thats still far away from 100% maximum that i want to get.
So i tried other sound cards, like Yamaha SW1000XG and noticed the same thing there - 20%.

How can i get more sensitivity with out more noise? If i increase sound source sound level i get more noise. With sound maximize i get the desired sound level (still a bit silent) but i also amplify the noise level (which is logical).

I am using an cable with two ferrite beads on each end and have place the cable where there is least possible interference from anything.
I am also using maximum possible sound level on PAS16 with out getting too high noise floor (the lowest possible noise floor for sound level), and thats 80% FM and 50% WAVE.
Ive seen others get very good sound levels with out noise, am i doing something wrong, or is that how recording actually works :) ?

Here is one track recording: MiR - TenChi MuYo Ending Theme SW1000XG
Note | LLSID | "Big boobs are important!"
User avatar
Stojke
l33t
 
Posts: 2080
Joined: 2012-3-23 @ 14:51
Location: Serbia

Re: OPL recording problem

Postby ik777 » 2014-2-02 @ 18:21

Depending on your devices, maybe hard to solve it. Because of they are quite old, it is possible their analog in/out parts are not healthy as your thought.
I've seen OPL3 recording results by digital recording and common method about recorder was.

1. Use a both OPL3 sound and SPDIF SB. Known one is CT3780. (ISA)
2. Yamaha YMF-744 or 754 with Digital Output. (PCI, OS limit)

But Even a Cmi8738 & a onboard audio Line In can make good results.

http://banti.egloos.com/viewer/5790749

He used CMI8738 card for console recording, but oneday he found out 8738's OPL enabling in Windows 7 and quite pleased. So he made youtube video linking it to onboard audio. He used just volume maximized in recording SW until sounds doesn't clipping. The good thing was he was first owner of soundcard and didn't full it or unpowered it much times. Its condition was quite good.

(edit) One of my card is not healthy so it noises though I unused internal amps. listen my samples.
Attachments
mef-d3dOPL3.ogg
WS32MEF - Grabbag (SAM9233 emulated OPL3)
(2.74 MiB) Downloaded 35 times
User avatar
ik777
Member
 
Posts: 126
Joined: 2012-7-08 @ 13:02
Location: Seoul, Korea

Re: OPL recording problem

Postby jwt27 » 2014-2-02 @ 18:37

The analog line-in and -out on new sound cards is 2V rms max, while on old sound cards it's usually 1V or 0.707V rms. If the signal-to-noise ratio on your recording card is high enough it shouldn't cause any problems, but on consumer sound cards this is usually not that great. I have the same problem on an X-fi elite pro and have been thinking about purchasing a studio sound card just for this purpose.

Another, cheaper option is to use a line amplifier to boost the signal to 2V.
WANTED - Manuals/drivers for:
  • Tecmar Graphics Master
  • Paradise Autoswitch EGA 350 (EGA1A)
User avatar
jwt27
Oldbie
 
Posts: 1628
Joined: 2011-8-15 @ 02:19
Location: Fryslân

Re: OPL recording problem

Postby elianda » 2014-2-02 @ 19:00

How is the problem with the different sampling rates of OPL and typical digital out with 44.1 or 48 kHz solved?
Retronn.de - Vintage Hardware Gallery, Drivers, Guides, HQ Videos.
Youtube Channel
FTP Server - Driver Archive and more
DVI2PCIe alignment and 2D image quality measurement tool
User avatar
elianda
l33t
 
Posts: 2225
Joined: 2006-4-21 @ 16:56
Location: Halle / Germany

Re: OPL recording problem

Postby keropi » 2014-2-02 @ 19:05

ah, that's why I have to use max volume when recording from the midi modules.... that line amp looks a nice option, I wonder where one can get one
User avatar
keropi
l33t++
 
Posts: 6570
Joined: 2003-9-08 @ 06:45
Location: Greece

Re: OPL recording problem

Postby Stojke » 2014-2-02 @ 21:18

ik777 wrote:Depending on your devices, maybe hard to solve it. Because of they are quite old, it is possible their analog in/out parts are not healthy as your thought.
I've seen OPL3 recording results by digital recording and common method about recorder was.

1. Use a both OPL3 sound and SPDIF SB. Known one is CT3780. (ISA)
2. Yamaha YMF-744 or 754 with Digital Output. (PCI, OS limit)

But Even a Cmi8738 & a onboard audio Line In can make good results.

http://banti.egloos.com/viewer/5790749

He used CMI8738 card for console recording, but oneday he found out 8738's OPL enabling in Windows 7 and quite pleased. So he made youtube video linking it to onboard audio. He used just volume maximized in recording SW until sounds doesn't clipping. The good thing was he was first owner of soundcard and didn't full it or unpowered it much times. Its condition was quite good.

(edit) One of my card is not healthy so it noises though I unused internal amps. listen my samples.


Well, the level when listening to the card it self is loud, its as loud as if i played the recording back at 100% volume on my pc.
The noise is that static buzzing that can be heard if the card is set to too loud, i tried to eliminate it as much as possible and i am recording on levels that produce the least amount of noise (no matter how much i reduce volume the noise level is still the same).
I could amplify a bit more, noise isn't really noticeable when the song is playing.

The only card with SPDIF and OPL3 is Sound Blaster AWE64 Gold, but the OPL on it doesn't sound very nice :)

jwt27 wrote:The analog line-in and -out on new sound cards is 2V rms max, while on old sound cards it's usually 1V or 0.707V rms. If the signal-to-noise ratio on your recording card is high enough it shouldn't cause any problems, but on consumer sound cards this is usually not that great. I have the same problem on an X-fi elite pro and have been thinking about purchasing a studio sound card just for this purpose.

Another, cheaper option is to use a line amplifier to boost the signal to 2V.


Oh i see, i didn't know that, that could also be possible. But the same happens of SW1000XG which is from 1998.
I also thought the line in isn't sensitive enough.
I also noticed that SW1000XG produces a warmer recording than X-Fi Xtreme Music.

I think that Volume Maximize is doing a great job, even though its increasing the noise level too. But what i really wonder is, does the same happen even if the line in is also amplified?
Note | LLSID | "Big boobs are important!"
User avatar
Stojke
l33t
 
Posts: 2080
Joined: 2012-3-23 @ 14:51
Location: Serbia

Re: OPL recording problem

Postby jwt27 » 2014-2-02 @ 23:29

keropi wrote:ah, that's why I have to use max volume when recording from the midi modules.... that line amp looks a nice option, I wonder where one can get one


If you don't mind some soldering, you can easily make a 2x gain amplifier using two op-amps and a few passive components. For a readily available option, most mixers can be used as line-level amplifier.


elianda wrote:How is the problem with the different sampling rates of OPL and typical digital out with 44.1 or 48 kHz solved?


In the case of an YMF754, I think the OPL3 runs on 44.1k sampling rate, like the YMF721 (OPL4).


Stojke wrote:Oh i see, i didn't know that, that could also be possible. But the same happens of SW1000XG which is from 1998.
I also thought the line in isn't sensitive enough.
I also noticed that SW1000XG produces a warmer recording than X-Fi Xtreme Music.

I think that Volume Maximize is doing a great job, even though its increasing the noise level too. But what i really wonder is, does the same happen even if the line in is also amplified?


By "volume maximize", I presume you mean normalize? In that case, you will amplify both the source noise and the noise from your recording card. A line amplifier would amplify only the source noise, but also add a small (negligible) amount of noise itself.

But before I talk you into buying equipment you might not even need, have you already tried to minimize ground loops and other interference?
WANTED - Manuals/drivers for:
  • Tecmar Graphics Master
  • Paradise Autoswitch EGA 350 (EGA1A)
User avatar
jwt27
Oldbie
 
Posts: 1628
Joined: 2011-8-15 @ 02:19
Location: Fryslân

Re: OPL recording problem

Postby ik777 » 2014-2-03 @ 06:33

Stojke wrote:The only card with SPDIF and OPL3 is Sound Blaster AWE64 Gold, but the OPL on it doesn't sound very nice :)


Ghehe. What is this? :happy:

Image
Attachments
CT3780.jpg
Ct3780
(380.36 KiB) Not downloaded yet
User avatar
ik777
Member
 
Posts: 126
Joined: 2012-7-08 @ 13:02
Location: Seoul, Korea

Re: OPL recording problem

Postby Stojke » 2014-2-03 @ 07:25

Sorry, i mean the only card i have with SPDIF and OPL3 is AWE64Gold :lol:
Thanks for showing me that card :D
I will try to find one.

jwt27 wrote:By "volume maximize", I presume you mean normalize? In that case, you will amplify both the source noise and the noise from your recording card. A line amplifier would amplify only the source noise, but also add a small (negligible) amount of noise itself.

But before I talk you into buying equipment you might not even need, have you already tried to minimize ground loops and other interference?


I mean software increasing the possible maximum :)
Here is an example of that [L + 15.97dB , R + 14.74dB] : MiR - TenChi MuYo Volume Maximized
Signal reaches ~50% by doing that.
By my understanding that amplifies both sound and noise.

I have reduced the possibility of any kind of external noise when recording.
Note | LLSID | "Big boobs are important!"
User avatar
Stojke
l33t
 
Posts: 2080
Joined: 2012-3-23 @ 14:51
Location: Serbia

Re: OPL recording problem

Postby Jepael » 2014-2-03 @ 14:04

While the AWE32 in the picture does have digital output and OPL, is it certain you get the OPL sound out from the SPDIF pins?
As far as I can tell, the system is similar than in one of my SB cards. The OPL chip (U13) has digital output going at some rate to separate OPL DAC (U17) from which the output is in analog form.
If it does output OPL3 sounds on SPDIF output, either the sound is digitally passed through the card (original sampling rate unknown to me) or then DAC analog output is again sampled to digital.

Indeed some later OPL3 implementations like OPL3-SA3 (YMF715) or OPL4 (YMF278) have 44.1kHz output, but as they have different master clock to get the 44100Hz rate, they also have different internal timing compared to the original OPL3 (YMF262) chip. I don't know how they internally do the sampling rate conversion, but the timers run at a bit different speed and the pitch is calculated a bit differently, but also I have not tried if they sound very different.

Many sound cards also resample or reclock everything they receive and might even have digital volume control, so it is not certain if you can record digitally exactly the data you receive, unless you have a professional/studio grade sound card that can lock to the digital input stream. Still, it should work with less noise than analog connections.
Jepael
Oldbie
 
Posts: 1195
Joined: 2005-6-15 @ 19:28
Location: Finland

Re: OPL recording problem

Postby Cloudschatze » 2014-2-03 @ 15:58

Jepael wrote:While the AWE32 in the picture does have digital output and OPL, is it certain you get the OPL sound out from the SPDIF pins?
As far as I can tell, the system is similar than in one of my SB cards. The OPL chip (U13) has digital output going at some rate to separate OPL DAC (U17) from which the output is in analog form.

No, the CT3780 is similar to other CT1747/EMU8000-bearing cards in that the (digital) OPL signal passes through the EMU8000, with the output from the EMU8000 then going to both the S/PDIF header and to the TDA DAC (U17).

I've theorized about the pitch shifting being applied to the OPL signal through the EMU8000 in this thread, if you're interested:
http://queststudios.com/smf/index.php/t ... l#msg33801

I haven't personally done any comparisons to see if there are noticeable differences in the 44.1kHz OPL output via the EMU8000, versus the output rate of the YMF262/YAC512 pairing, but everything I've posted in the "Technically Impressive FM" thread is a S/PDIF capture, if some enterprising individual wants to compare....
User avatar
Cloudschatze
Oldbie
 
Posts: 1007
Joined: 2005-6-16 @ 14:32

Re: OPL recording problem

Postby ik777 » 2014-2-03 @ 16:06

Jepael wrote:While the AWE32 in the picture does have digital output and OPL, is it certain you get the OPL sound out from the SPDIF pins?


The man who did it.
http://www.xain.pe.kr/zonx/image/clien/ ... esized.JPG
http://www.xain.pe.kr/zonx/image/clien/ ... esized.JPG
http://www.xain.pe.kr/zonx/image/clien/ ... esized.JPG

http://zonx.egloos.com/5761410
You can visit there for asking uncompressed outputs about some his work.
User avatar
ik777
Member
 
Posts: 126
Joined: 2012-7-08 @ 13:02
Location: Seoul, Korea

Re: OPL recording problem

Postby NewRisingSun » 2014-2-03 @ 19:01

I've theorized about the pitch shifting being applied to the OPL signal through the EMU8000 in this thread
Given that the YMF289 (OPL3-L) is perfectly capable of outputting 44.1 kHz without any pitch-shifting, it's more likely that the CT1747 is based on the YMF289 and thus uses no pitch-shifting either.

The YMF289, according to its data sheet, expects a 33.8688 MHz input, which is 3/4 of 45.1584 MHz. The data sheet calls 33.8688 MHz the "third overtone oscillation", indicating that the fundamental frequency is 11.2896 MHz. Wikipedia lists this frequency as common in CD audio because dividing it by 256 yields the common 44.1 kHz. This means that while the OPL3 (YMF262) has a master clock of exactly 14+7/22 Mhz and generates an output sample every 288 cycles, yielding a sampling frequency of exactly 49,715+10/11 Hz, the OPL3-L (YMF289), and likely the CT1747, has a master clock of exactly 33,868,800 Hz and outputs a sample every 768 cycles, yielding a sampling frequency of exactly 44,100 Hz.

The YMF289's data sheet lists no formula for calculating the f-number values that the programmer needs to write into the OPL3-L, but the YMF278 (OPL4)'s data sheet, listing a similar master clock of 33.8688 Mhz and also a 44.1 kHz sampling rate, specifies a master clock divider of 684 for f-number calculation. 33.868.800 divided by 684 of course yields 49,515 + 15/19 Hz, very close to the OPL3's original sampling rate. The error is about 0.4 percent. This indicates that while the OPL-3 uses the same 288 divider to derive the output sample rate and to drive the waveform/phase/envelope/timer, the OPL3-L and OPL4 use a 768 divider to derive the output sample rate but a 684 divider to drive the waveform/phase/envelope/timer for compatibility. Therefore, no pitch-shifting is needed on the output.
NewRisingSun
Oldbie
 
Posts: 798
Joined: 2005-9-02 @ 02:26

Re: OPL recording problem

Postby elianda » 2014-2-03 @ 19:52

Very well, this explains a few things.
Retronn.de - Vintage Hardware Gallery, Drivers, Guides, HQ Videos.
Youtube Channel
FTP Server - Driver Archive and more
DVI2PCIe alignment and 2D image quality measurement tool
User avatar
elianda
l33t
 
Posts: 2225
Joined: 2006-4-21 @ 16:56
Location: Halle / Germany

Re: OPL recording problem

Postby Cloudschatze » 2014-2-03 @ 22:52

NewRisingSun wrote:Given that the YMF289 (OPL3-L) is perfectly capable of outputting 44.1 kHz without any pitch-shifting, it's more likely that the CT1747 is based on the YMF289 and thus uses no pitch-shifting either.

I'd be inclined to agree (and did up until August 10, 2011), but for Dave Rossum having stated in an e-mail that "the OPL3 I2S output is routed to the EMU8000, in which it is sample rate converted and digitally mixed with the other EMU8000 signals."

In the QuestStudios thread, I've assumed use of the YMF262 frequency divisor of 288, given that the CT1747 chips on the non-EMU-bearing CT2230 and CT2290 SB16-variants are tied to 14.31818 crystals.
Last edited by Cloudschatze on 2014-2-04 @ 02:44, edited 3 times in total.
User avatar
Cloudschatze
Oldbie
 
Posts: 1007
Joined: 2005-6-16 @ 14:32

Re: OPL recording problem

Postby keropi » 2014-2-03 @ 23:04

jwt27 wrote:
keropi wrote:ah, that's why I have to use max volume when recording from the midi modules.... that line amp looks a nice option, I wonder where one can get one


If you don't mind some soldering, you can easily make a 2x gain amplifier using two op-amps and a few passive components. For a readily available option, most mixers can be used as line-level amplifier.
[...]


Know of any good/tested schematic? :blush:
User avatar
keropi
l33t++
 
Posts: 6570
Joined: 2003-9-08 @ 06:45
Location: Greece

Re: OPL recording problem

Postby NewRisingSun » 2014-2-04 @ 06:49

According to my ID.COM utility, the CT1747's registers are not readable; in that regard, it is more similar to the OPL3 than the OPL3-L. In any case, my explanation of why the OPL3-L and OPL4 can output 44.1 kHz without sample rate conversion or adjustment of f-numbers stands, though it may not apply to most of the Creative cards.

A 14+7/22 MHz-master clocked CT1747 will output at the 49,715+10/11 Hz sampling rate, which the EMU8000 can then convert to 44.1 kHz. Cards without the 14+7/22 MHz oscillator (and which don't use the OSC signal from the bus) that have the CT1747 clocked by the EMU8000 will either have the EMU8000 convert the 33.8688 MHz to 14+7/22 MHz --- a conversion factor of exactly 2.36544 --- or it will have the EMU8000 clock the CT1747 at 33.8688 MHz (or 11.2896 MHz), and the CT1747 in those cards uses the 768 (or 256) and 684 (or 228) combination of internal dividers. I could imagine that the CT1747 can do both depending on the state of another external pin: if a particular pin is high, use 768/256 and 684/228 dividers, if low, use a divider of 288. This would be called something like "input clock select". The EMU8000 would have to perform no sample rate conversion in this case, and the Dave Rossum statement would only apply to the latter case. Check how the other CT1747 pins are connected between cards with and without 14+7/22 MHz inputs (either from an oscillator or the ISA bus' OSC signal).

In any case, while the EMU8000 may do resampling (which E-MU may well call "pitch shifting"), it is extremely unlikely that it does any actual pitch-shifting (with time-stretching), as that would yield way more artifacts than heard in sample recordings.
NewRisingSun
Oldbie
 
Posts: 798
Joined: 2005-9-02 @ 02:26

Re: OPL recording problem

Postby jwt27 » 2014-2-04 @ 19:10

Jepael wrote:[...] OPL3-SA3 (YMF715) or OPL4 (YMF278) have 44.1kHz output [...]


I measured the YMF719 sampling rate a while ago and got about 49516 Hz. I think it's slightly detuned, too (haven't checked thoroughly)


keropi wrote:
jwt27 wrote:
keropi wrote:ah, that's why I have to use max volume when recording from the midi modules.... that line amp looks a nice option, I wonder where one can get one


If you don't mind some soldering, you can easily make a 2x gain amplifier using two op-amps and a few passive components. For a readily available option, most mixers can be used as line-level amplifier.
[...]


Know of any good/tested schematic? :blush:


The standard non-inverting amp configuration should work. See here: http://en.wikipedia.org/wiki/Operationa ... _amplifier
Use R1 = R2 = 1k and you'll have a 2x gain amplifer. Add an RC filter (R = 39k, C = 470n should be okay) on input and output to eliminate any DC offset, and you're done :)
WANTED - Manuals/drivers for:
  • Tecmar Graphics Master
  • Paradise Autoswitch EGA 350 (EGA1A)
User avatar
jwt27
Oldbie
 
Posts: 1628
Joined: 2011-8-15 @ 02:19
Location: Fryslân

Re: OPL recording problem

Postby keropi » 2014-2-04 @ 19:23

^ thanks jwt27 :)
User avatar
keropi
l33t++
 
Posts: 6570
Joined: 2003-9-08 @ 06:45
Location: Greece

Re: OPL recording problem

Postby jwt27 » 2014-2-04 @ 19:34

You're welcome :)
Forgot to add, an easy way to power it is using +12 and -12V found on the wavetable header on your sound card. Make sure to add two 100n decoupling caps to ground, on both op-amp supply voltages, and as close to the chip as possible.
A good audio op-amp is OPA2134: you'll get 2 amps in one IC, low noise, FET input, and they're widely available.
WANTED - Manuals/drivers for:
  • Tecmar Graphics Master
  • Paradise Autoswitch EGA 350 (EGA1A)
User avatar
jwt27
Oldbie
 
Posts: 1628
Joined: 2011-8-15 @ 02:19
Location: Fryslân

Next

Return to General Old Hardware

Who is online

Users browsing this forum: stamasd, Sudos and 3 guests