Reply 41 of 221, by clueless1
- Rank
- l33t
Testing is a difficult, thankless job. 😉
The more I learn, the more I realize how much I don't know.
OPL3 FM vs. Roland MT-32 vs. General MIDI DOS Game Comparison
Let's benchmark our systems with cache disabled
DOS PCI Graphics Card Benchmarks
Reply 42 of 221, by James-F
- Rank
- Oldbie
It seems the 3D and EQ only work with the 719 version, but it is actually a good thing because the 3D is horrendous and destroys the stereo image, and the EQ is only a deviation form what the developers wanted you to hear.
Carried away eh 😀 , ironically even the good SB16 I own (CT2230) is no match to the cheap common Yamaha YMF71x card.
The B16 is a buggy mess, I do not recommend it to anything, DOS or Windows.
The Yamaha YMF71x is indeed almost THE perfect card for DOS gaming.
I'll be testing the Aztech's as soon as they arrive, hope they up to par with the YMF but with a good Lowpass filter and a functional ADPCM.
Reply 43 of 221, by PhilsComputerLab
- Rank
- l33t++
Yup on a PCI 486 with a fast AMD 133 MHz and Tseng ET6000. That machine does 47 FPS in Doom, now that is what I call smooth 😀 A real race horse. Will test tomorrow if the POD 83 can do better.
Also gave that DOS MIDI player a go and listened to some Last Ninja 2 MIDI files. They are meant for Yamaha XG, but sound fine on General MIDI.
I love how quiet the Yamaha card is. I was listening with headphones, but through an external mixer. I read your comment about impedance, plus on the mixer I can adjust the volume comfortably.
Reply 44 of 221, by James-F
- Rank
- Oldbie
wrote:I was listening with headphones, but through an external mixer. I read your comment about impedance, plus on the mixer I can adjust the volume comfortably.
This is really important to use a proper headphones output, and not plugging the headphones straight to line-out.
A Mixer, Headphones amplifier, Stereo system with proper headphones out should do the trick.
Same when recording the line-out, always use the line-in for recording not the MIC input.
Mic input has a different impedance designed for Microphones obviously.
Unfortunately this is not common knowledge and I post it at every opportunity.
Reply 45 of 221, by PhilsComputerLab
Reply 46 of 221, by firage
- Rank
- Oldbie
wrote:ironically even the good SB16 I own (CT2230) is no match to the cheap common Yamaha YMF71x card.
The B16 is a buggy mess, I do not recommend it to anything, DOS or Windows.
The Yamaha YMF71x is indeed almost THE perfect card for DOS gaming.
Call me when they beat the SB Pro's SNR. 😜
Reply 47 of 221, by James-F
- Rank
- Oldbie
Which one, SB16, YMF, Aztech?
All three SB16 I own have excellent SNR of 75db, the YMF is on par with that also.
Noise is the least of the SB16 problems, that's for sure.
Reply 48 of 221, by Cloudschatze
- Rank
- Oldbie
I'm not sure that the SB Master Volume behavior of the YMF chipset is a technically a "bug," given that the behavior is explicitly described in the YMF715 datasheet:
Since the Master Volume behavior of Creative's CT1345 isn't well-described in their SDK literature, it might be worth checking a Sound Blaster Pro or Pro 2 card to see how the the values set by the Creative mixer utility correspond to the actual 0x22 mixer register values.
Also, for all the talk about the YMF71x cards lacking the SB Pro's output filter, the datasheet also describes a working register control bit for this, as well as pins on the chip to connect the external filtration component(s). It's probably worth looking into this further, as the YMF71x cards may have the filter default to an "off" setting (in contrast to the SBPro), or certain implementations may simply use external filter components that don't result in a 3.2kHz rolloff, or omit them entirely.
Reply 49 of 221, by firage
- Rank
- Oldbie
I saw both the SB Pro and YMF71x hover around 70 dB SNR, but maybe they're better than that. Their DAC distorts less and there's the MPU, but would expect some more improvement from the old tech.
Reply 50 of 221, by clueless1
- Rank
- l33t
Does the SNR correspond with the amount of hiss heard at high volume? I ask because it sounds like the YMF71x has a similar SNR to the SB16, yet I hear way more hiss at volume on my CT2800 than on the YMF71x.
The more I learn, the more I realize how much I don't know.
OPL3 FM vs. Roland MT-32 vs. General MIDI DOS Game Comparison
Let's benchmark our systems with cache disabled
DOS PCI Graphics Card Benchmarks
Reply 51 of 221, by James-F
- Rank
- Oldbie
wrote:I'm not sure that the SB Master Volume behavior of the YMF chipset is a technically a "bug," given that the behavior is explicitly described in the YMF715 datasheet:
Wow, thanks for that!
Apparently I had the wrong datasheet, now I found the right one.
You're right, if it is written in the datasheet it is technically not a bug.
As for the Lowpass filter, I see "Left/Right SBDAC filter" as the only thing that matches that in the datasheet, but the datasheet doesn't specify anything about the lowpass filter.
The pins go through a SMD capacitors to ground on my board.
Reply 52 of 221, by Cloudschatze
- Rank
- Oldbie
wrote:As for the Lowpass filter, I see "Left/Right SBDAC filter" as the only thing that matches that in the datasheet, but the datasheet doesn't specify anything about the lowpass filter.
The pins go through a SMD capacitors to ground on my board.
Yeah, Yamaha's documentation isn't as complete as would be desired either, so this is largely conjectural. Sound conjecture though, I think. 😀
If you want to try some tests with your card, I suspect Creative's early SBPro mixer can be used with the YMF71x chipset to set the output filter operation, otherwise, debug can be used:
debug
-o 224 0e
-i 225 (reads the current 0x0e register value, so we can determine the YMF71x default setting)
-q
debug
-o 224 0e
-o 225 00 (filter on)
-q
debug
-o 224 0e
-o 225 20 (filter off)
-q
Reply 53 of 221, by firage
- Rank
- Oldbie
wrote:Does the SNR correspond with the amount of hiss heard at high volume? I ask because it sounds like the YMF71x has a similar SNR to the SB16, yet I hear way more hiss at volume on my CT2800 than on the YMF71x.
Yeah, that's right, the base level hiss under OPL or DAC output. I'm half joking, but still... Pretty surprised if a Vibra card is noisier. Have you compared line-out vs. line-out or just the amp outputs?
Reply 54 of 221, by James-F
- Rank
- Oldbie
0x0E reads 11.
After writing it still reads 11.
sbp-set.exe (SBPro mixer) has the /DNFI:on/off which controlls the 0X0E as per the SB programming guide.
It doesn't work.
I guess Register 0x0E is not programmable on the YMF71x?
Reply 55 of 221, by Cloudschatze
- Rank
- Oldbie
wrote:I guess Register 0x0E is not programmable on the YMF71x?
It may just be that 0x0E bit 5, which controls the output filter, isn't programmable, despite the datasheet not suggesting otherwise. 0x11 is certainly an interesting value to read back from the register though, as bits 0 and 4 don't correspond to anything in any documentation that I've seen. Do you have an SB Pro or Pro 2 that you can check as well? If not, I can check later, when I have a chance. I'll also test my YMF715-based AudioTrix 3D-XG, to see if its behavior/implementation is any different.
Reply 56 of 221, by James-F
- Rank
- Oldbie
What about the D numbers?
The Stereo Switch is also on 0x0E but on D1, while the Output Filter is on D5.
With the debug commands we failed to specify to which D to read/write.
Reply 57 of 221, by Cloudschatze
- Rank
- Oldbie
wrote:What about the D numbers?
Per the SB documentation, bits 1 and 5 are "off" by default, meaning the stereo bit is set to mono, and the output filter is on. This corresponds to the 0x00 value sent via debug (00000000). Leaving the stereo bit set to mono, but flipping bit 5 to disable the output filter, corresponds to the 0x20 byte (00100000).
Reply 58 of 221, by James-F
- Rank
- Oldbie
Alright, so Register 0x0E on D1 (Stereo Switch) and D5 (Output Filter) are writable but D0 and D4 are always 1.
It doesn't matter, writing 22 returns 33 when queried so it should work just like the SBPro but it always adds D0 and D4 to the sum.
D5 Output Filter, it is by default on 0 so the filter should be enabled as in the SBPro.
The "L/R SBDAC filter" output pins on the microchip trace to a capacitor then ground, it may be that the capacitor is small or the D5 switch doesn't do anything.
I'll try to piggyback another cap on top of the existing one, but I'll measure the existing one before.
I definitely don't want to destroy my YMF card just for experimentation.
EDIT:
The caps on SBFLTR are 250pF, on the small side I assume.
Reply 59 of 221, by clueless1
- Rank
- l33t
wrote:wrote:Does the SNR correspond with the amount of hiss heard at high volume? I ask because it sounds like the YMF71x has a similar SNR to the SB16, yet I hear way more hiss at volume on my CT2800 than on the YMF71x.
Yeah, that's right, the base level hiss under OPL or DAC output. I'm half joking, but still... Pretty surprised if a Vibra card is noisier. Have you compared line-out vs. line-out or just the amp outputs?
Actually, the YMF is jumpered so that line out is amplified, which I've heard reduces sound quality, but I have not experienced that. It's quiet as can be. No audible hiss when cranking up to max usable volume. The CT2800 is on line out. Both line outs go to an audio switcher which connects to a single set of powered desktop speakers. When I turn up the volume on the speakers and play the same game music on both cards and switch back and forth with the audio switcher, the hiss is noticeable on the CT2800 but not on the YMF.
The more I learn, the more I realize how much I don't know.
OPL3 FM vs. Roland MT-32 vs. General MIDI DOS Game Comparison
Let's benchmark our systems with cache disabled
DOS PCI Graphics Card Benchmarks