Creative Sound Blaster 16s with incorrectly capped TEA2025 amplifiers

[UraBahn]

I've made a bit of a discovery regarding the old Sound Blaster 16 sound cards which use TEA2025 amplifiers. Seems Creative, the penny-pinching shonks they are, wasn't entirely following the datasheet for the TEA2025 and substituted improper capacitors in a few places. This resulted in much of the Sound Blaster 16 line sounding muddy and muffled, a lot worse than they could and should sound.

According to NatSemi's TEA2025 datasheet, the two 1.0uf input capacitors and 470uf 16v output capacitors should be bi-polar, not polarized. Creative, of course, used plain old polarized caps in those places, which had the effect of making the sound darker and muddier. Replacing the 1.0uf input caps with relatively cheap Nichicon VP bi-polarized caps resulted in a much clearer and brighter high-range. (Replacing them with huge 1.0uf metal-film caps brightened it even more, but that may be a touch overkill. Besides the caps look awkward jutting from the board like that. But then I thrive on awkwardness so...) Here are the metal-films on my CT2230: https://www.dropbox.com/s/4ty4aip56kkgzx9/HPIM1095.JPG

In addition, Creative often used 47uf caps on the TEA2025's feedback lines, where the datasheet specified to use 100ufs. Replacing them with 100uf's resulted in a notable increase in bass. (According to the datasheet, using smaller caps increases the low-pass cutoff.)

Last but not least, replacing the old 100uf SVR cap with a 220uf cap results in a generous improvement in overall sound quality.

Here are some before and after comparisons. The before was on a stock CT2770 with the caps Creative selected.

Here's how most SB16's with the TEA2025 amp circuits typically sound:
https://www.dropbox.com/s/6cavo6mopb...put%20caps.wav

And here is how a SB16 (CT2910) with the correct capacitors sounds:
https://www.dropbox.com/s/b1d5fnopqj...n%20inputs.wav

Quite a difference, no? No audio-phoolery here, it's just simply following the datasheet.

And here is how my CT2230 SB16 sounds using 1.0uf polypropylene film caps on the inputs:
https://www.dropbox.com/s/ugqhzo0kzq...zed%20caps.wav

Your results may vary depending on the caps you use and the design of the sound board of course. I tried to compare the results using SB16s which all had very similarly built TEA2025 amplifier sections. Also bear in mind that the TEA2025, being a tape-player amplifier, is not by any means a high-fidelity device, so the best you can hope to attain is a generally listenable sound quality.

So... yeah, you don't need any fancy shmancy SILMIC, Black Gates or anything like that when reworking a SB16 for better quality. All you need is the TEA2025B datasheet, a soldering gun and some good quality bi-polarized caps. (Yes, I used polarized SILMIC on the output of that CT2230, back before I studied the TEA2025B schematics in detail. But that will soon change. The two SILMICs are socketed to allow for easy, solder-free cap replacement.)

Oh yeah, here's the TEA2025B datasheet.

[lti]

Re: Creative Sound Blaster 16s with incorrectly capped TEA2025 amplifiers

I never noticed the bipolar caps on the output before. Actually, it looks like every cap except the power filter cap (connected from the power input on pin 16 to ground) is bipolar.

I haven't had any problems with my headphone amp using a TEA2025B and polarized caps.

[stj]

Re: Creative Sound Blaster 16s with incorrectly capped TEA2025 amplifiers

i dont think the pedophiles - sorry, audiophiles
would give a damn, they would start with a turtle-beach or some other stupidly expensive card.

btw, good luck finding a 470uf bi-polar - they dont exist.
atleast not unless they look like a coke-can!

[R_J]

Re: Creative Sound Blaster 16s with incorrectly capped TEA2025 amplifiers

Here is utc's data sheet..., I also checked a few audio amp schematics nad, H-K, bose etc. they all use polarized caps in the audio path.
looks like that copy from ST was faxed across the country a dozen times

[UraBahn]

Re: Creative Sound Blaster 16s with incorrectly capped TEA2025 amplifiers

"i dont think the pedophiles - sorry, audiophiles would give a damn, they would start with a turtle-beach or some other stupidly expensive card."
Well, this article is for the rest of us who went the SoundBlaster route back in the 90s. With the exception of the SILMIC (which were purely out of morbid curiousity than any kind of audio-phoolery BS) everything I am recommending here is just typical, plain-jane stuff you can order from digi-key or, in the case you are really fortunate to have a GOOD one nearby, ratshack. (I was stunned to see mine had a pair of 1.0uf metal-film caps in stock...)

"btw, good luck finding a 470uf bi-polar - they dont exist. atleast not unless they look like a coke-can!"
What's wrong with coke can caps? Besides Nichicon makes some 470uf's that are 10mm in diameter and a whopping 22mm tall - a little big *ahem*, but not too bad really. So it's more like a MagLite flashlight. Later Sb16's give just enough room for 10mm caps there for the output decouplers anyway. (The Silmics were 12.5mm, a bit too big...)
http://www.digikey.com/product-detai...081-ND/2539563

UTC made TEA2025Bs are cheap clones of the NatSemi TEA2025B. (Though better than a few no-name Chinese knockoffs I've encountered, *shudder* CD2025) I noticed differences in the sound quality between clones and the real, genuine NatSemi TEA2025B articles. Yes their scanning of the document could have been better, but if I look closely I definitely see how some caps were indicated as being NON-polar. (I think the cloners were just too cheap to care when they marked their schematics...)

In any case the proof is in the sound files. I did not turn up the treble or bass on any of the cards, all of them are using the default, middle settings for treble and bass in Windows 9x.

[Per Hansson]

Re: Creative Sound Blaster 16s with incorrectly capped TEA2025 amplifiers

That's pretty interesting.
I wonder how the card sounded when the original caps where brand new though...

Perhaps they have degraded allot through the years?

[Th3_uN1Qu3]

Re: Creative Sound Blaster 16s with incorrectly capped TEA2025 amplifiers

I'm quite sure it sounded about as cheap. Penny-pinching on coupling caps (especially when you make them smaller than they should) always results in loss of bass.

[lexwalker]

Re: Creative Sound Blaster 16s with incorrectly capped TEA2025 amplifiers

Still have one of these soundcards, and in working condition also. Never bothered to modify it, and didn't bother about that TEA2025 anyway (which requires passive speakers) as I've always used the Line Out instead (using active/amplified speakers)...

[stj]

Re: Creative Sound Blaster 16s with incorrectly capped TEA2025 amplifiers

i had most of the sb16 variants btw, but they are long gone.

i do have a pci 5.1 card - if i find it, i'll look at the caps.

[lti]

Re: Creative Sound Blaster 16s with incorrectly capped TEA2025 amplifiers

National Semiconductor (now part of Texas Instruments) never made the TEA2025.

I don't think the caps on the input are needed unless there is some DC offset on the input signal. The increase in treble caused by replacing those caps is interesting.

My headphone amp actually has polarized 10uF caps there, and the voltage across them is only around 10mV. For some reason, those caps have to be perfectly matched or one channel will have less bass. That could be caused by the tone control circuit. It's a really weird circuit, and the caps might not be the actual cause of the lack of bass.

[UraBahn]

Re: Creative Sound Blaster 16s with incorrectly capped TEA2025 amplifiers

National Semiconductor (now part of Texas Instruments) never made the TEA2025.

Oh yeah, derp. ST made the TEA2025, my bad. (Typed all that whilst under the weather.... glowry be pertussis, with hints of pneumonia and fever!) I'm a little bit better today (thank goodness for azithromycin), so now I can think somewhat clearly.

That's pretty interesting.
I wonder how the card sounded when the original caps where brand new though...

Perhaps they have degraded allot through the years?

Oh, the cheap ELGEN and Wincap capacitors degraded over the years for sure, making most Sound Blasters sound even worse as they age. Of particular importance is the SVR (ripple reject) cap and the two feedback capacitors connected to the TEA. Creative LOVES to pinch pennies on the feedback caps, using 47ufs where the datasheet specifies 100uf. They usually don't under-size the 100uf SVR cap but replacing it with a 220uf anyway gives audibly better results. (The new breed of 220ufs occupy the same space as the old 100uf cap does, so replacing it is a cinch.)

My headphone amp actually has polarized 10uF caps there, and the voltage across them is only around 10mV. For some reason, those caps have to be perfectly matched or one channel will have less bass. That could be caused by the tone control circuit. It's a really weird circuit, and the caps might not be the actual cause of the lack of bass.

It's true that caps alone aren't the entire story behind a circuit's sound, the design of the circuit as a whole affects the total sound quality. That sounds like a very strange sound circuit you've got there, obviously one with razor-thin tolerance margins. It'd be interesting to hear what would happen if one were to use bi-polar 10ufs there. (Hopefully not magic smoke?)

[UraBahn]

Re: Creative Sound Blaster 16s with incorrectly capped TEA2025 amplifiers

OK, sorry for resurrecting this old thread, but I've made a few further discoveries here: ST (repeat, NOT NatSemi... :P) specified .22uf non-polarized caps on the TEA2025's inputs, so what I did is I replaced the 1 uf film caps I was using with .22 uf film caps. The end result was a very noticeable increase in treble, allowing my CT1770 to finally almost match my modded CT2230 in the high-range. The bass did decrease slightly, but this was offset by the improvements made elsewhere (eg. the 47uf feedbacks being replaced with 100ufs proper) and I still get a very decent bass output.

Also I've installed 470uf bi-polars for the output decouplers, and yes, there was a noticeable improvement in sound quality, though nothing too dramatic. A quick look at the recorded waveform in Audacity, though, reveals that the negative swing on the output waveforms is now much less distorted looking. (I do understand that polarized caps, when given a negative charge, react in a non-linear way...)

So yeah I'm on the way to a better sounding SB16, just still have a few adjustments to do here and there. A few sound cards were maimed in the process (such is the nature of 'science') but all is for a good cause.

[chozo4]

Re: Creative Sound Blaster 16s with incorrectly capped TEA2025 amplifiers

So out of curiosity, what kind of results did you get with recapping the bootstrap capacitors? Have you given raising their capacitance a try to see the potential differences? I only ask as I noticed a marked increase in clarity when replacing them with 16v/100uf Chemicon PXA polymers and curious if an increase in capacitance is worth looking into on them at some point.

[UraBahn]

Re: Creative Sound Blaster 16s with incorrectly capped TEA2025 amplifiers

I've replaced the bootstraps with 100uf bi-polars, as the datasheet's diagrams (see 'Stereo Audio Circuit' in the ST TEA2025B datasheet) specifies to use bi-polars there. Haven't heard a lot of difference though. (I surmise the bi-polar application is to prevent distortions at some point.)

Considerable improvement in the treble range for both amplified and direct line-out, however, comes from replacing the small 1.0uf caps near the CT1745 chip with 1.0uF metalized polypropylene-film caps (requires some ingenuity to fit them due to their much larger size though!) Replacing all the 4.7uf and 10uf caps near the chip helps too, though you'll probably have to use regular electrolytics there since you're not likely to come across film caps of those values that will fit.

Afterward you'll begin hearing the flaws in Windows 95's startup sound much better, revealing it as the hissy 8-bit sample that it is. You'll also hear Sierra's Police Quest VGA kick out some serious bass: https://www.dropbox.com/s/9olev2ykbj...0CT2230%29.ogg

[lti]

Re: Creative Sound Blaster 16s with incorrectly capped TEA2025 amplifiers

I don't see why bipolar output caps would help. I really don't think the output of a TEA2025 can ever go negative. I don't even know how close to 0V it can get.

[chozo4]

Re: Creative Sound Blaster 16s with incorrectly capped TEA2025 amplifiers

I don't see why bipolar output caps would help. I really don't think the output of a TEA2025 can ever go negative. I don't even know how close to 0V it can get.

Thinking it's more so because of the AC going through it being attenuated more in one direction than the other.

[mockingbird]

Re: Creative Sound Blaster 16s with incorrectly capped TEA2025 amplifiers

Just thought I'd show you my card, and point out that there are subtle differences between them, your having a later datecode. They're both CT2230, but different revisions it seems.



(See attachment for higher-res version)

[Th3_uN1Qu3]

Re: Creative Sound Blaster 16s with incorrectly capped TEA2025 amplifiers

Is that a SB16? I have one too round here, and i believe i've some pics of it somewhere... If not, i'll just take a couple more. I also have an AWE64 ISA, that one has the TEA1517 amplifier on it.

[mockingbird]

Re: Creative Sound Blaster 16s with incorrectly capped TEA2025 amplifiers

Yup. Please do take some pics, I'm interested in seeing yours. Are you considering re-capping with the "mods" (More like circuit corrections) posted here?

[Dans2530]

Re: Creative Sound Blaster 16s with incorrectly capped TEA2025 amplifiers

So what you are telling me is that Doom would have sounded MORE awesome!? I had a 486 running this card through a maranz amp with some beefy speakers and it sounded good then.
Good on you for fixing something a massive company thought was "good enough" or couldn't be f'ed fixing it. If I still had that gear I would do your mods and test again.
Top work mate!