do you own that CT1600? if yes please upload a proper picture not a screenshot from your device - it is interesting to see what they did but that picture does not help at all

Yes I own it but the card is in a different country right now so it might take a while. It looks pretty much exactly like a regular 049219 but with an oscillator in a different location to where it is on the r5.

I see... please do not forget to add some good pics when it is possible - maybe this is a hardware mod to make a rev4 to a rev5 , interesting!

There is no sign of it being an after market mod. Maybe its something Creative tried on a few R4's before making the proper change in R5. Anyway something to look out for on R4 cards to see how prevelant it is.

yeah that's what I mean, something Creative did and it must be analyzed 😁

keropi wrote:

.. something Creative did and it must be analyzed 😁

Don't tempt me... 😁

@Ivader
Can you take a picture of the back of the card, especially below the oscillator.

I just have access to a few crappy photos at the moment but it might help for now...

lvader wrote:

What do you guys make of this rev 4 card? Spot the oscillator that isn't meant to be there!

What makes you think it is not meant to be there? Sure, it sticks out, metal can oscillators are huge, but any other reason?

I think I see a trend here, first cards with no crystal or oscillator use the ISA OSC signal (14.318 MHz), but for some reason it may not work so great so first they slab an oscillator there (quick, easy, expensive), and later they replace it with cheaper circuitry where the oscillator is built with a crystal and a 74LS04 TTL inverter chip.

I think nothing else on the board requires 14.318 MHz than the OPL3 chip (YMF262).

I am sure lvader just points out that this is a nice find since we now know there is a rev4 CT1600 with a crystal modification.
If this gets documented maybe I will try and add it to my rev3 card for testing

That raises the obvious question of WHY did Creative decided to add the 14.318 MHz oscillator on board?
I know that SB1.5 and SB2.0 are extremely sensitive to CPU Clock and/or FSB and they don't have an oscillator on board, so I wonder if the older SBPro2 without the oscillator have the same problem.

I am using my rev3 with my 233mmx build - other than OpenCubicPlayer I have not noticed anything else crashing (except from games that need L1 disabled anyways - these bug out no matter the soundcard in use)
Maybe they added it because some mobos had clock issues so they played it safe.

keropi wrote:

Maybe they added it because some mobos had clock issues so they played it safe.

Yeah, maybe some cheap MB manufacturers omitted the oscillator completely, or maybe the MB oscillator signal was too weak and was distorted by the time it reached the card.

My go-to games for OPL sensitivity testing is Prince of Persia 1/2 and Mortal Kombat 2, these games go insane on a Pentium with a YM3812 (OPL2) based cards.

Normally having its own independant clock helps the DAC this things like jitter. Maybe Creative experinced some quality issues with some motherboards and decided better not rely on them.

at least POP1 plays music correctly on my 233mmx/full speed with my rev3 CT1600 - ofcourse it's OPL3 but still there are no issues from the lack of a dedicated oscillator. Mobo is a Shuttle HOT-591p

lvader wrote:

Normally having its own independant clock helps the DAC this things like jitter. Maybe Creative experinced some quality issues with some motherboards and decided better not rely on them.

In this case I really don't think it is for DAC jitter reasons - OPL is already so low cost consumer-grade stuff that it's sound would not improve with a better clock 😀

I think it is more likely that the boards using ISA OSC signal have sometimes trouble receiving it cleanly for some reason (amplitude, noise, ringing, or it might have wrong duty cycle), so that it just does not work.

If someone posts a picture from the underside of such card, we can see if the TTL level ISA OSC goes directly to OPL3 chip.
From the datasheets we'll see the OPL3 clock input pin can't really work with TTL voltage levels, it needs more amplitude.

James-F wrote:

That raises the obvious question of WHY did Creative decided to add the 14.318 MHz oscillator on board?
I know that SB1.5 and SB2.0 are extremely sensitive to CPU Clock and/or FSB and they don't have an oscillator on board, so I wonder if the older SBPro2 without the oscillator have the same problem.

I don't think the missing oscillator would be the reason for CPU/FSB clock sensitivity, because the 14.318 MHz from ISA should be constant, and the cards have circuitry to divide this by four for the OPL2 chip so it ends up getting 50% duty square wave. If the 14.318 MHz from ISA had a different frequency, then also OPL songs would play at wrong pitch as well.

So most likely if changing CPU/FSB clocks makes it work, then it's a sign of too fast I/O cycle length (card would like more wait states even for a single I/O write which is not dependent on CPU but chipset/BIOS/HW, or, CPU can execute more I/O cycles in a burst so it makes the required software delays between OPL chip writes too small so it is theoretically a software issue).

what picture is needed? from a CT1600 that has no dedicated oscillator?

I recently bought the latest Rev.08 SBPro2 and tested it.
I can say it is very quit compared to Rev.06, all else is the same, including the frequency response and filtering behavior.

Rev.06 has some self noise when when all mixer levels are maximized (default) and I can hear the noise go down when I lower the mixer level (not wheel).
The main culprit for noise in Rev.06 is the OPL3 because when I lower only the FM level, the noise goes away.

Rev.08 on the other hand has so little self noise that I can't even measure the noise level when the mixer levels are full or zero.
I can attest that Creative definitely did something right in Rev.07 and Rev.08.

It would be great if the Vogonswiki page for the CT1600 could be updated with details on some of the different revisions!

Definitely.

We also recently proved here on Vogons that there is absolutely no difference between the SBPro1 and SBPro2 in terms of reversed stereo issue, they are absolutely identical besides the dual-OPL2 vs OPL3.
Th reversed stereo issue is actually a NON issue, because game developers that took the time to test with a real SBPro had the stereo correct for any SBPro.
Game developers that did not bother to test with a SBPro card and simply let the DSP handle stereo PCM as is, in that case it was reversed to the SB16 option in that game.
Yet, we have to remember that the SBPro handles stereo PCM in a completely different way than SB16, so blaming the SBPro for reversed stereo is not an excuse for lazy developers.

We also know that OPL2 (YM3812) is extremely sensitive to FSB and/or CPU speed making the SBPro1 (including SB1.5 and 2.0) completely useless in a Pentium.

Moreover the SBPro2 CT1600 (07 and 08), and the two CT2600 revisions (01 and 02), all from 1993-1994 have the same revised PCB layout and are very quiet cards.