Yesterday, I managed to get an EWS64XL sound card to work with an Abit KT7A (KT133A chipset) mobo. Today, I removed the Abit KT7A mobo, and replaced it with a Soltek SL-75KAV mobo. This is also a one ISA slot KT133A chipset based mobo. The only difference between the two is that the Soltek has onboard audio.
The problem is that the EWS64XL sound card appears to be completely dead. It's strange - you can't even hear any hiss noise using a cheap pair of headphones. I've tried plugging these working headphones in to both the EWS's front bay headphone port, and also in to the "out port" on the back of the sound card. I've tried the following to get this to work:
Disabled onboard sound in the mobo's BIOS.
I have set the legacy IRQ and DMAs correctly for the EWS in the mobo's BIOS.
I uninstalled then reinstalled everything inside Win98: VIA chipset drivers, (+ also the graphics card driver), Breese latency patch, EWS driver, EWS applications.
The System Control Panel shows no problems whatsoever! The EWS devices are there, with no conflicts. Everything looks correct - just like it looked when I used the EWS sound card with the Abit mobo. When I test both "basic audio sound effects" and also MIDI music, I get absolute silence.
The EWA64 XL is a diva when it comes to system resources. It refuses to share any resources with other cards. If any other component dares to lay claim on some resource configured for the EWS64 XL, it clams up and stands in the corner, pouting. 🙄
I'm not sure if defining legacy resources for the card in the BIOS is a good idea. These settings are meant to be used for old non-PnP cards, to announce that the respective resources are set with jumpers and can not be changed by PnP mechanisms.
Don't let diagnostic outputs from PnP tools or the Windows device manager deceive you. They often report everything being OK, when in fact the EWS64 is not cooperating.
Look out for common shared resources, like AdLib ports, printer IRQs or ACPI stuff.
Thanks a lot for the info.
Unfortunately, on this occasion I give up. After about 30 Windows 98 reboots and 6 hours later, I've had enough. I don't know what the problem is, so I'll remove the EWS64XL, and test an AWE32 or something just to ensure that the ISA slot is working. Then I'll retest the EWS64XL on the Abit mobo, just to ensure that this sound card is still working OK...
I know a bit more about this problem now. I removed the Soltek mobo, and replaced it with the Abit mobo. I did this in order to retest the EWS sound card - to ensure that it still worked OK. Oh no! It doesn't work anymore - or so it seems. When I run dxdiag.exe and test the music, I get no sound - but if I simply wait for about 5 seconds, suddenly the music can be heard! The sound coming from the EWS sound card is significantly delayed. I tried reinstalling the Breese latency patch, but that didn't fix the problem.
I wonder why the Abit mobo system is now delaying the EWS sound? Before I removed this mobo, to be replaced by the Soltek mobo, it worked OK - there were no sound delays. Also, it's possible that the Soltek mobo system was "malfunctioning" in the same way - if I had simply waited about 5 seconds, I may have heard something from the EWS sound card.
Edit: BTW, all games do not produce any sound. So, this is a fundamental problem.
I don't know if this resolves your problem, but concerning the EWS64 I can give you some general advice:
- As 5u3 said reserving legacy IRQ/DMA adresses is not a good idea for this card; you can try, but especially with windows it might produce more problems. But you can experiment with the auto/manual switch in your mobos bios as long as you keep the adresses on pnp. Try to reset the system/pnp data, reboot and reinstall the card. There is also a small dos tool that resets the card so it has a new pnp ID (default: C:\ews64\EEPROM\EWS64NEU.EXE. It might be nescessary before that you remove the drivers manually with a tool called "Remove.EXE" somewhere in the EWS64 dir). This means that the bios recognizes the card as another EWS64. Of course you have to reinstall the complete card than.
There is an in-depth guide here, which was very helpfull to me:
- As you can see the above guide is mainly written for musicians, so here's what I have to add:
If you have a multiboot system (i.e. DOS 6.22 and Win98) try to configure the card in DOS! Don't forget to load the patch bank and mixer settings and initialize the card correctly as otherwise it won't work in dos!
Also check that you have only one game port in your system, that means that the one of the board or on the EWS is deactivated.
- Sometimes the Crystal codec causes problems. See the infamous SBPro mute bug and its solution (by locutus) here (in this famous long thread started by you ):
- But the EWS64XL does also work completely without the codec. This is for no use in real dos (as you get no sound there) but dramatically increases the sound quality in Windows; music and directX sound should still work in Win98 so this might be a good way to test the card: Simply deactivate the WSS I/O Adress in the config utility (dos or windows, I allways prefer dos to configure the card) and reboot (better you have deactivated the codec / SBpro and FM devices in the win device manager before, I suppose) and than test the card.
The confusing mixer panel and its settings can also cause trouble, so try to experiment with that too. AFAIR the last mixer settings a stored every time you shut down the machine. Anyway - with the above constellation (codec deactivated) you should only need the following sliders: CD/SYN2; IN-2 (switch ADM position "A")Out-1; Out-2, Rec.
Problem solved!
Thanks very much to Silent Loon and 5u3, and to everyone else on Vogons who offers excellent advice! 😀
I decided to banish the forces of unspeakable evil (I removed the VIA KT133A mobo), and replace it with a different chipset based board, an AMD 751 Irongate based mobo: Gigabyte GA-7IXE4. The installation and subsequent testing worked immediately. Inside "pure" DOS 7.1, I was able to do this:
ewsinit B <dream 94b file> F
I could then hear funky tunes inside DOS games. I could also do this inside the Windows 98 "dos box" too. The dxdiag testing worked fine: all sound and music could be heard on all of the tests, and there were no sound delays. Please note that I do not have a wavetable daughterboard installed inside the "break out" box, so some tests were omitted. I don't think I will use a wavetable DB, because you need another IRQ for this, and that would "max out" all of my mobo's resources! And I only have 1 graphics card and the EWS sound card in it. This thing eats mobo resources for breakfast! 😉
BTW: You can also free the LPT port if you don't use a printer there; and of course the two serial ports COM1/2 = + 3 IRQ.
As your "new" board should have 2 ISA Slots you might be able to run the AWE32 alongside peacefully, which would give you additional SB16 and AWE support.
EDIT: Concerning the initialization of the card for dos - you can also load a specific mixer setting in your autoexec.bat, one that you have previously saved in Win98. this feature allows you to use the effects also in dos, even for the FM synth.
i.e.: C:\EWS64\EWSINIT.EXE F V B C:\EWS64\SOUNDSET\GSSBK320.94B M C:\EWS64\SETTINGS\DEFAULT.TTM
Well the problem is not solved at all, you switched the mobo but you dont know that the cause was.
I am a bit curious about this behavior, because I had a identical problem.
EWS64XL in DOS playing MIDI from Dream or Wavetable and after 5 to 10 seconds somehow the silence breaks and music plays. There is no delay with the music, but with the out-enable. Interesting enough is, that it also enables digital playback etc.
If I play digital, the output does not get enabled.
Sometimes it stays silent even when playing midi.
If output is enabled, I drop back to dos cmdline and restart the midi player it takes 5 to 10 secs again...
I think it has nothing to do with resources, because I got two identical system (at least as identical as it can get).
The other one also utilizes a EWS64XL, GUS, GA586HX Mainboard. Cards on same resources (EEPROM and setting), bios same setting, bios same version.
I made a boot disk that even runs the identical software for initalization...
In one PC the EWS64XL works flawless in the other one the 'output silent' problem. (Besides the silent output the card seems to work flawless too).
I removed the front panel and RAM, still silent.
I asked myself, why I noticed this problem now, because the PC has been running some time in this config yet. Well if the problem was limited to half a second before, I might not have noticed it. I moved the (silent) EWS64XL to another system, booted the check disk and it worked instantly. I havent exchanged it against the other one - never touch a running system 😉.
My guess is, there might be something wrong with the voltages on the ISA port. Maybe the power supply ages and some voltage is not correct. I got currently no multimeter here. Though the remaining system runs (replaced the EWSL64XL with a Maxi Sound 64).
I am pretty sure this effect has nothing to do with misconfiguration. It is something else...
Oh and btw. there are ways to minimize resource usage and with some knowledge it is even possible to use same resources on different ISA cards reliably.
there is no 10 MB SF on the disk, but this disk was a quick hack and it simply skips loading a SF. At the end, it tries to play digital.
Resources are:
WSS 104, I5, D1, D3
FM 388
SB 220
GP off
Control 108
MPU401 330/9
Dream 300
Digital Control 110
Have you swapped system ram for a check? Ram causes a wide range of weird stuff.
Unlikely, both Systems are memtest86'ed. On one I also got NT4 Pro installed and didn't encountered a bluescreen in the last years. (A BSOD is typical for RAM problems).
Switching RAM isn't as easy as it sounds. This usually means removing the power supply in an AT Midi Tower.
Same Bios options selected, even where you don't think it matters? (Like Year 2k bug or other oddballs)
Yes, except the HDD ofcourse. On first order there are resource conflicts and timing to think of. I didn't wanted to rip off the remaining PC, so on PCI side they are a bit different.
One has a Riva128 + Voodoo1, though since there is no IRQ assigned to the VGA card they use memory locations where no ISA can be. In this machine is also a HighPoint UDMA controller with 2 HDDs. I have to check if this one maps ISA resources. Though I think it doesn't, it brings its own BIOS.
The other PC has a Voodoo3, same applies as for the Riva128.
Since PnP OS=Yes no resource except the HighPoint controller gets initialized (as Realtek network adapter).
What is left is the different Upper Memory Layout with different BIOS ROMs. Though I cant see a possible conflict there, when accessing the EWS directly.
I also removed the HDD cable from the HP controller (in that case it doesnt installs its BIOS). This had no effect.
Well the problem is not solved at all, you switched the mobo but you dont know that the cause was.
Hehe, yeah I know. 😊
elianda wrote:I am a bit curious about this behavior, because I had a identical problem.
...
I think it has nothing to do with resources, be […] Show full quote
I am a bit curious about this behavior, because I had a identical problem.
...
I think it has nothing to do with resources, because I got two identical system (at least as identical as it can get).
...
My guess is, there might be something wrong with the voltages on the ISA port. Maybe the power supply ages and some voltage is not correct.
...
I am pretty sure this effect has nothing to do with misconfiguration. It is something else
...
still the silence problem is a strange one.
Your speculation about the voltages on the ISA port is very interesting, and does seem highly plausible. However, I am not 100% certain about it. Here's why:
1. I put the EWS in to an Abit KT7A mobo. After a lot of "weird stuff" (it's a VIA KT133A board! 😉 ), the EWS behaved itself and worked: no sound delays.
2. I put the EWS in to a Soltek KAV75 mobo. (Same layout + chipset as the Abit mobo.) The EWS did not work correctly: sound delays problem.
3. I put the EWS back in to the Abit mobo. The EWS did not work correctly! : sound delays problem. Please note that it worked one day ago on this board!
4. I put the EWS in to a 3rd mobo, AMD chipset based. It worked - no sound delays.
To me, that sounds like some kind of resource and/or configuration issue. However, the "ISA voltage theory" is very interesting too.
Also, I tried an AWE32 CT3900 non-PnP in the ISA slot on the Abit 1 ISA slot mobo, and noticed that the "Windows 98 start up sound" of thunder sounded bad - lots of crackles. Perhaps the ISA slot is defective in some way, or perhaps it is a latency/configuration/resource problem?
Am I wrong, or do you initialize the cards codec (CS4236 I guess it was) with two mixersettings - micha.ttm and the parameters of CS32MIX?
Why? Does CS32MIX not override the former *.ttm settings? Or is the ttm configuration just for loading effects somehow? Does this work?
What happens if you dismount the GUS?
@retro games 100:
I tried an AWE32 CT3900 non-PnP in the ISA slot on the Abit 1 ISA slot mobo, and noticed that the "Windows 98 start up sound" of thunder sounded bad - lots of crackles.
Are you sure that this card is 100% non-pnp? I thought that all the higher version numbers are pnp? I think I have this version myself (somewhere) and it was pnp, but I'm not sure.
Could the described effects (sound delay, crackling) result from a pnp-problem?
Are you sure that this card [ AWE32 CT3900 ] is 100% non-pnp? I thought that all the higher version numbers are pnp? I think I have this version myself (somewhere) and it was pnp, but I'm not sure.
Could the described effects (sound delay, crackling) result from a pnp-problem?
Please see the schematic diagram of this card here -
On this diagram, it shows jumpers for setting the "IO stuff", such as 220 and 330. However, this card doesn't have an IRQ jumper. Does that make it a non-PnP card? I have a vague feeling that there were two CT3900 models. One PnP, the other non-PnP.
Am I wrong, or do you initialize the cards codec (CS4236 I guess it was) with two mixersettings - micha.ttm and the parameters of CS32MIX?
Why? Does CS32MIX not override the former *.ttm settings? Or is the ttm configuration just for loading effects somehow? Does this work?
You are perfectly right.
The only difference is that running cs32mix does solve the "Wavetable analog audio gets muted when initializing SB portion of the CS4236B" bug.
What happens if you dismount the GUS?
If I remember correctly I did this with no effect on the silent EWS64XL.
The AWE CT3900 is one of this semi PnP cards, you set the IO only and initialize the rest with BLASTER= followed by diagnose /s.
Resources are set according to the previously set BLASTER environment.
On non-PnP cards you only call diagnose /s and it sets BLASTER= to the detected values. (Or if you know, which is usually the case, just set BLASTER= yourself).
On PnP cards you have to use ctcm and ctpnp.cfg.
isapnp.exe usually works too, but is not effective with soundcards cause they usually mix initialization code and pnp resource setting code in the same exe file.
I would like to mention another point, why I think it has nothing to do with resources.
As pointed out it gets loud again after some seconds playing MIDI. MIDI itself is just output to Port 330. It can be played without using IRQ, just by sending the data.
It is hard to imagine a kind of misconfiguration where sending midi data to the mpu401 port to the external wavetable triggers some codec muting setting after some rather random time.
