Repairing The Legendary Gravis Ultrasound

[@kunka592]

Hey, I'm actually the seller of this card. Really glad you got it working! I was afraid it might have been the GF1 flaking out as I heard those are a common issue. At least it wasn't too difficult or expensive to repair. Couldn't have asked for a better home for this card. Small world to be recommended this video, haha. I bought it as supposedly working about 9 years ago, but I had similar issues back then so I got some more of those RAM chips to try it with but it was always acting weird so I stashed it for years and then tested it again to see if I could figure out the issue and clearly I could not so I listed it as not working/for parts, and the rest is in this video. :)

[@vswitchzero]

Thanks so much for your comment! Wow that's awesome that you found this video! It is a small world indeed. Thank you again for this card, it was in great condition. There were a few people bidding against me from what I remember :) .. I definitely appreciated the detail and honesty about its working state in the auction description. It was spot on and gave me some hope to get it repaired. I had feared it may be the GF1 chip too initially but was a total fluke that I discovered the flaky crystal. Ian was kind enough to send me a couple more OEM crystals to try after I posted this video. When I get some time I'd like to get the proper frequency one back on. I need to check the crystal capacitors in the circuit as well as it's possible they were contributing. Thanks again!

[@kunka592]

@@vswitchzero Great, hope you can get it figured out with the OEM crystals as well. Your channel is really cool, by the way. Cheers.

[@CMDRSweeper]

Glad that it seems part of my theory as to what had happened to this card and the attempt at repair was correct and that you suspected the memory chips. That was my first theory, but the mass failure wasn't very likely to me at first, but the clock chip was a bit further down the rabbit hole than I suspected :D

[@leonardoliveira]

@@CMDRSweeper This absolutely makes sense because DRAM requires "periodic half accesses" (the refresh cycles) to maintain whatever is stored in it and those accesses are generated/timed by the master clock frequency. Without refresh pulses whatever is written to the DRAM "evaporates" over time and then it comes out as bad RAM on the diagnostics program. By the way great channel you have here. Loving your videos.

[@ianpolpo]

Too bad the new old stock crystal I sent was also flaky, but that's great that you've found a substitute that is much easier to source. It sounds like it works just as well!

[@vswitchzero]

Thanks again for sending that over, Ian! I really appreciate it. I'm guessing they are quite old and even though they've never been used, something internally must degrade over time. It was definitely worth a try. I'm very happy the 19.6608MHz crystal seems to work fine. So far it's been rock solid, and I can't notice any difference in the sound quality or speed. Thanks again!

[@jonas-fr]

@@vswitchzero it could be interesting to carefully open up the defective crystals and film the inside to see what's wrong with them

[@vswitchzero]

@@jonas-fr Thanks for the comment! I'm not sure how easy they are to open, but I am also very curious to see if there is anything inside that looks degraded, corroded etc. I may give it a try if I get some time.

[@VShuricK]

@@vswitchzero this one a bit tricky to open - seems it use a contact weld. So only carefully file out a weld.

[@dolphhandcreme]

@@vswitchzero Try 2 capacitors somewhere in the range of 15-22pf from every leg to GND. Most crystal oscillator circuits i've seen have them present to get a stable signal.

[@cburgess5294]

Crystals are highly dependent on accurate capacitors. By touching the case you were adding capacitance. I would have looked at replacing the capacitors at each leg of the crystal with new ones of the correct value - that would also explain why the new-old-stock crystal showed similar errors.

[@vswitchzero]

Very interesting! I will definitely look into this

[@matthewday7565]

@@vswitchzero Just what I was thinking, one typical oscillator has the crystal from input to output of an inverter, and capacitors from each side to ground, this makes the crystal actually act as the inductor of a capacitive center tap grounded tuned circuit to provide the inverted feedback, it's possible the other crystal just has a more forgiving self capacitance. It also requires the appropriate capacitance to pull the crystal to the specified frequency

[@owenvogelgesang7314]

@@vswitchzero FWIW, this is why probing it with the scope caused you to see nothing. The scope probe has (in this situation) non-negligible capacitance, which causes the crystal to stop oscillating or doing anything

[@jeffm2787]

Good to see someone who also understands what's going on. Honestly I would have just grounded it making the can act as a capacitor in a sense around the internal legs of the crystal which are isolated.

[@garyglitta]

You beat me to it. Probably the crystals are fine and the capacitors are the problem

[@ConfusionDistortion]

At 9:33 the evil from beyond tries to communicate with our world using the GUS as its vessel.

[@daishi5571]

The demon Gravis of Ultrasound spoke!

[@CaseyFraysier-b7q]

man this takes me back!! I haven't even thought about the GUS in 30-ish years.

[@mistermudpie]

This channel really hits the retro nerd nostalgia spot!

[@UncleAwesomeRetro]

Very nice job repairing this great sound card :) I still have my GUS MAX I had as a kid. I remember I felt I had something luxurious when I could select the "Gravis Ultrasound Max" as the sound card instead of sound blaster. I have also bought the GUS Classic for my 486.

[@Smakificator]

Does that mean you have TWO working GUS? That's awesome

[@UncleAwesomeRetro]

yes :)@@Smakificator

[@SUCRA]

Amazing job identifying that, I've knocked my head around problems of the sort for hours before. Also, great provisory fix with the grounding. It's great that you replaced with another crystal that can be found, that'll serve as a reference to anyone with the same problem. Great video.

[@vswitchzero]

Thanks very much! It was a fun experience getting it fixed :)

[@MaverickStrife]

I really enjoyed watching your journalism from diagnosis to fix My only regret with the video is that after you’d fixed it and upgraded the RAM I was expecting a showcase demonstrating the best of the GUS! It was a bit of an anticlimactic finish 😊 Looking forward to your next repair job

[@vswitchzero]

Thanks very much! I'd very much like to do a deeper dive video on the GUS one of these days. Just need to find the time, which is the hardest part :)

[@eightbit1975]

Excellent! You know at the very beginning of this video and hearing the issue I just had this hunch it was the clock crystal. And it was. What a simple fix and what a relief it was not that GF1. Good job! Another saved GUS!

[@agevenisse3252]

Lol 9:32 sounds like the card was possessed. Great video. Just found your channel. :)

[@vswitchzero]

Thanks! 🙂

[@nmosfet5797]

I have a version 1.0 Gravis Ultrasound, one of the original 10 PCB's that were sent by Gravis to Finland after the Assembly 1992 demoparty, Gravis was looking for programmers to write code for the GUS and I was lucky enough to get one of them. I spent my college days with it. I wonder what it's worth :D

[@vswitchzero]

That's awesome! I'm sure it would be very valuable these days to a collector, that's for sure.

[@arpadmuller7992]

That's a piece of history! Have you ever shared some code, demo, or image of the card somewhere ?

[@nmosfet5797]

Crystals failing to oscillate with IC built-in oscillators is a common issue e.g. with old microcontrollers. Adding load capacitors is often suggested, but often doesn't help. What often helps to make the crystal more compatible is loading the oscillator with a resistive load. Something between 47kOhm and 470kOhm between the crystal pins often helps. Load capacitors decrease the frequency and also the amplitude of oscillation making it weaker, so they might the wrong cure. Try touching the xtal pins with resistor pins, but don't touch it with fingers. Cheers!

[@RetroTinkerer]

If that card is for your 486, then you are working on your 486 build!

[@hilsm9886]

Super film ciekawy i fajnie zmontowany, super że się udało

[@GUCFan]

A very good video that shows that even clock crystals can go bad and when they do, funky stuff starts happening! Like your bad crystal made the GUS sound like a Walkman that's running low on battery

[@withmygoodeyeclosed]

Love your low key channel, the vids are relaxing and it's cool to watch you learn.

[@vswitchzero]

I appreciate that! Thanks so much 🙂👍

[@DJ-Daz]

I still remember the sheer joy and awesome-ness of hearing Doom for the first time on the GUS. I understand why people pay a fortune for them. Nostalgia is so damned expensive.

[@zybch]

I still have my original GUS 'classic'. But I'm fearful of actually using it so I've just ordered a 'PicoGUS' to pop in the P-III retro PC I'm building. Great you could get yours up and running again.

[@vswitchzero]

Nice! I actually just bought a PicoGUS too. Such a cool project that's getting better and better!

[@ozzyp97]

Same, I have an older rev 3.4 which was listed as not being tested for a few years, ended up being just fine though. I'm seriously considering a PicoGUS as well, although personally I tend to think that being scared to use vintage hardware because it might break is a bit of a self defeating argument. What's the value of having working hardware if it's sitting in a box anyway? There's some nuance there of course, if you use your hardware a lot, maybe it's worth saving the real deal for special occasions. It's just hard to quantify, since not being used won't stop the hardware from aging, and the time it spends in storage is time you could've had to actually enjoy what you have.

[@Radek__]

0:08 Gravis Ultrasound was my dream wish from 1992. I finaly bought used one in something between 1994-1995. I remember even now the voice of guy who was speaking "SBOS install" when you load an emulation of sound blaster in dos. such a memories.

[@danhaworth6967]

I've honestly had so much fun watching your videos (been on a bit of a binge the last few days lol). I've learned SO much! I have to say though, this is the first time I've watched one of your videos and the very moment I heard the midi playback go distorted and slow I was like "THATS A TIMING ISSUE! I Bet the oscillator is duff!" and literally pointed to that very component on the board!! lmao I wouldn't say I'm exactly clueless when it comes to electronics, but I wouldn't have had a clue three days ago before watching a load of your videos so thanks for the huge amount of education boost :) You have a new subscriber!

[@vswitchzero]

Haha awesome! Really glad you’ve been enjoying the videos. Thanks so much for your kind words. Comments like this really keep me going, it is very much appreciated! 😁👍

[@Shmbler]

The "touchy finger" helped me repair a Voodoo1 with an open 0R resistor on TMU mem RAS line, and an A500 board with some broken traces under the Denise socket. Nice to see how your fingers do the diagnosis just as well ;-) IMHO if you've got ongoing audio or visual problems then touching stuff is always worth a try.

[@vswitchzero]

Haha yep, the finger test has served me well several times now! 👍

[@eizomonitor6003]

I Had this in my 386 DX40 system. I loved it. Unfortunately someone stole it from me. One of build made in the first 100. Good old days, when i went to a demo party and i have shared it with coder to finish the job to debut the demo on the very same night to the audience.

[@GamePlayShare]

I think i had similar problem with ESS card. Gonna try later to replace the oscilator

[@SobieRobie]

Thanks for the video. I have several different GUSes stored in my basement. I wonder do they still work...

[@JimLeonard]

In owning several GUSes over the years, including launch day, I've never seen (heard!) that kind of failure before. That's bananas. One thing to note: While the difference in your replacement crystal is very small, it will likely affect some amiga MOD songs played in modplayers that support the GUS, specifically any song that uses a looping sample (like a drum pattern) that was calibrated to loop on the beat right as the sample ended. Those might have audible beating ("judder") when played. But a slightly slow GUS is better than no GUS at all :-)

[@SianaGearz]

You know this might not be the case. The current crystal is 19.6608; original crystal is 19.7568, which is a sub 0.5% difference. You can kind of expect any of the system crystals to be out of sync with the soundcard crystal by about that order of magnitude if you aren't lucky. For this reason it's considered bad manners to rely on the separate clocks running in sync, and the only timing source for the sound/music routine should always be the soundcard, most game/demo/audio routine programmers know that very well. There's a separate timer subsystem on the GUS chip for this reason, as you can't just time the audio in batches during mixdown, you need actual events in soundcard timebase. The timer has an approximately 12.5kHz resolution i believe. Worth testing for sure to ascertain compatibility!

[@JimLeonard]

@@SianaGearz I wasn't saying the mods will play incorrectly due to bad timing, but rather because the slower crystal means the samples will be played 0.5% slower (that's one of the reasons I called the failure "bananas", I'd never heard a failure like that before). Since the original timing of when to fire off the samples will remain the same, the slower sample playback won't have a chance to get to the end of the sample before it's retriggered. You may think 0.5% pitch difference isn't noticeable, but it can be if percussion sample loops are involved.

[@SianaGearz]

@@JimLeonard the timer peripheral on the GUS runs in lock sync with the sample clock, since there is no other source of timing. If the sample clock is half a percnt slow, so is the overall timebase. The sampled sound routine is generally the source of audio and music related timing in games and music players, and often the master source of timing that all other time sources are calibrated against. Some disparity between clock domains is common and needs to be accounted for by software. Master interrupt controller on the mainboard being 1% too fast is not anyhow out of the ordinary! RTC is usually pretty good but it can't be used for fine timing.

[@JimLeonard]

@@SianaGearz Wasn't aware of that; I agree that any mod player that used the Gus timer would work fine, since the timer and the playback would be in lockstep no matter the frequency. Of course, assuming all mod players that supported the GUS used it is wrong :-) I've seen a lot of player source code, and they almost universally use the system timer for the 50 hz timebase.

[@vswitchzero]

Thanks for your comments, Jim! Very interesting stuff. I haven't had a chance to try out any MOD players yet, but its definitely on my list of things to do. That and some demoscene stuff as well. So far, it seems to be doing well in all of my gaming and MIDI tests. Ian offered to send me a couple more ecliptek 19.7568MHz crystals, so I'm hoping I'll get lucky and one of the proper ones will work properly.

[@BrassicGamer]

Very lucky repair (as in trivial to perform) but also shows the importance of a) cleaning b) a proper physical inspection (preferably with magnification and c) live testing. Having a scope for things like this is obviously very advantageous because you have diagnostic certainty over signals. Really need to get one. Really happy to see this GUS repair and in good hands!!

[@therealjammit]

Grounding a crystal case is normally done in a circuit. You can pull a higher frequency crystal to a lower frequency with a really small capacitor. You could try using a "gimmick" capacitor to pull a higher frequency crystal lower. It's simply two insulated wires twisted together but isolated form each other. The harder you twist the wires together the lower the frequency. Solder the gimmick to the two crystal pins. You can Google "gimmick capacitor" to see one and yes, they are as simple as they look.

[@micv2]

A very good video, more about GUS please. Best regards 👍

[@O.Shawabkeh]

Pretty nice work, informative and telling it as-it without hype. There was one new, card only v3.7, listed on eBay for $400. It lasted quite a while before the seller discounted it to 375 and it was sold.

[@vswitchzero]

Thanks for your comment! Yeah about $400 USD/$500 CAD is what I've been seeing working ones go for lately. There are quite a few showing up on eBay for really high 1K+ prices. Even though they are collectors items and in high demand, I don't think there are many people who would pay that much.

[@darknase]

Haven't touched my card for more than 20 years, at least I know what to expect and where to look if it doesn't run. Fun fact mine had, at the time, the PAL chip go bad, had it send to CA for warranty.

[@Smakificator]

Hello vswichzero! Sorry for being late to the party on this video! Totally would have wanted to be onboard at broadcast time! I'm a GUS Max owner, with card working as expected in 2023. There are a few questions about how these work when they are working properly that I might be able to help with.

[@Torzelan]

This was great, so glad you got it back in order after an interesting journey! Had to laugh out loud the first time it glitched out, wasn't quite prepared for that sudden turn from lovely music to agonized voices from the fiery abyss 😂 To me who who is a relative super-amateur when it comes to electronics that whole "works if grounded" thing seems very bizarre. I feel like I stubbornly would try to keep the original and "make it work" in as clean and stable way as possible when it is that close / works under the right conditions.

[@vswitchzero]

Thanks for your comment! Haha so true - the glitched MIDI sound is just horrible. I had to subject myself to that mild form of torture for way too long while troubleshooting. When I couldn't find the right crystal, I was very tempted to solder a short ground cable to the crystal can. But yeah, it really shouldn't need to be grounded and I still don't know exactly why that works. Interestingly, even a length of wire attached to it that isn't grounded causes it to start working too.

[@SianaGearz]

I've seen many circuits where the crystal can is specifically grounded or at least very very slightly capacitively coupled to ground. There can be a pour underneath, there can be a blob of solder on the edge of the can (risky, crystals do not like heating), there can be a bit of mask removed so the can will just about touch the pour, there can be a jumper wire hugging the can and clamping it to the board, and also grounding it... might all be useful to keep the crystal stable and for the driver circuit not to go down the wrong overtone, i suppose. Though it appears Advanced Gravis decided to do exactly none of these things. Normally a ground pour is recommended. But i don't know if it's relevant to the fault that this particular family or batch of crystals has developed. I too would be tempted to ground it and leave it like that until it gets worse, what's the worst that can happen anyway? Seems pretty safe, even if it doesn't fix the underlying fault.

[@vswitchzero]

Interesting! Yeah, I've seen quite a few crystals soldered to a ground pad on the PCB or clamped with a bit of wire etc. I had assumed this was just to mechanically hold it down since they are usually mounted horizontally on expansion cards but after this whole experience I can see it must have other benefits as well :)

[@SilverBullet93GT]

legendary North Fraser Way, Burnaby BC ftw!!

[@disgruntledgoat]

I've used Euroquartz before now for custom crystals. Not the cheapest in the world but they have low MOQ and will make basically anything you like.

[@vswitchzero]

Very interesting, thanks for the tip!

[@andrasszabo7386]

I have a laptop designed in Hungary,1987, that accepts 2 cards like this one. I have bought a Sound Blaster 16 for it already. Now all I have to do is to get the laptop's PSU recapped and ready to have stereo sound.

[@kaseyboles30]

I remember this thing. Was a nice upgrade back in the day. I wonder what ever happened to mine.

[@baghdadiabdellatif1581]

Great work 👌

[@OscarSommerbo]

I bought a GUS in 93 with my own money. I read up and understood the potential for the wavetable stuff, but sadly almost no games supported it. So to me, it became a slightly flaky SB emulated in software. But I never had any huge issue with the emulation. Some time around 95, into the trash the card went, mainly because of cheap SB clones and the disappearance of the 16 bit ISA slots.

[@danielleblanc5923]

Nice video, Quartz crystals are physically very fragile parts. It's basically a quartz disc pinched in 2 places for the electrical connections. It is basically impossible to weld / solder on the crystal as the temperature would cause it to crack. Grounding the case of crystals is common practice, probably for the stability of the oscillation or to remove stray capacitance. Depending on the topology serial resonant or parallel resonant capacitors of a few pF can be used to adjust the fundamental frequency. Also, measuring crystals in circuit is tricky because the probe itself has a built-in capacitance (usually 10pF) that is not negligible and that can destabilise the oscillator.

[@vswitchzero]

Thanks for your comment and for sharing! I hope to revisit this card at some point. I have two more OEM crystals to try and will be looking much more closely at any capacitors in the circuit too.

[@JohnSmith-iu8cj]

Good job!

[@io4439]

Some very helpful tips in the comments to nail down the issue with this card. A short follow up video perhaps?

[@vswitchzero]

Indeed! I've learnt a lot going through the comments on this video. Doing a follow up on this one is definitely on my list of things to get to at some point. Ian sent me some additional OEM crystals to try too.

[@SianaGearz]

With a crystal that's really close you can tune it closer by adjusting the load capacitors in the circuit. You could take a lower crystal and then try reducing the series load capacitor values to nudge it up. Or i suppose you can take the higher crystal, and tack a ceramic capacitor across it to nudge it down. Though i don't expect the adjustment range up to be above 0.05%, and you need a 0.5% up; down you should have more latitude, but you'd need to go whole 12% down, which is a tall order indeed. Programmable MEMS oscillator next step? Then again it might not be needed. 0.5% difference is just on the audibility threshold in direct comparison, so as you concluded, you can't hear it especially if you don't have a reference, you wouldn't be wrong on that; on the other hand, this sort of disparity between different system clocks wouldn't be entirely out of the ordinary so the software may need to account for it anyway by using GUS as the time base.

[@SianaGearz]

I made a brainfart! You need to nudge 1.2% down from 20MHz rather than 12%. This seems plausible, much more plausible... then again your scope doesn't seem to have THAT much precision as to estimating frequency. So probably worth leaving the repair as-is :D

[@vswitchzero]

Thanks very much for this and your other comments! I don't have much experience with crystals and system timing in general so the information you provided is very interesting. I have heard of adjustable/programmable oscillators so that could be a very interesting option. So far, the 19.6608MHz crystal has been working very well in all of my tests, but I still need to try out some mod players and demo scene stuff. Since I couldn't get a reading on the old oscillator without causing it to cut-out, I'd be very curious just how in-spec it was. Considering how flaky it was, I wonder if it has drifted over it's 30+ year life. I suspect it may have. But yeah, my scope really isn't good enough for that level of precision. On a positive note, Ian very kindly offered to send me a couple more of the original ecliptek crystals to try out. Maybe I'll get lucky with the next one. Even if I can't hear a difference, I'll feel better knowing the crystal is to the original spec. If it were any other card, I probably wouldn't care, but I think a GUS deserves the effort :)

[@eightbit1975]

I forgot to ask you.. Where did you obtain the ram test software for the GUS Classic? I have the 4.11 installer as well but it does not have this tester. I would love to test the 1MB onboard mine. If you can let me know it would be greatly appreciated!

[@vswitchzero]

Good question. I believe I downloaded a special "pack" of GUS software from Vogons, but I can't seem to find the thread/link anywhere. From what I recall, it was a ZIP file containing the entire directory structure all ready to go with the latest drivers and tools. I'm pretty sure GUSDRAM was created by someone and not part of the original software installer bundle. I'll keep looking to see if I can find it.

[@MegafOfficial]

Good video and repair, just some observations. Maybe, just perhaps, the crystal might be fine, and the problem is elsewhere, a broken solder joint or trail, or potentially a bad capacitor. Something could be overheating after some time, hence not having issues at first and problems steadily getting worse. Using a slightly slower crystal, makes things run slightly slower and use slightly less power, causing stuff to heat slightly less.

[@vswitchzero]

Thanks for the comment and the tips! A few people have mentioned a bad capacitor possibly being the issue. I'm definitely going to look into this. I got a couple more original crystals from Ian that I'm going to try when I get some time.

[@GabrielZ666]

Nice job! I wonder if there really is any difference in sound between the original crystal and the newer part. I have a GUS Max 2.1 and paid a decent price for it a couple of years ago, next to nothing compared to what it costs today. This retro hobby is getting very expensive! 🥴

[@BlackHoleForge]

Man it sounds like you're Baroque music is possessed.

[@Anamnesia]

We lusted after a GUS back in the day... Ended up with a TB Maui & SB AWE64...

[@CoreyDeWalt]

The failed midi sounds are from nightmares!

[@movax20h]

Great work bringing it to life. However.. As others mentioned, it is probably not a bad crystal but auxiliary capacitors that are bad. They are usually external to the crystal. and need to be rather specific values, to form a part of resonant circuit. I would desolder the crystal, and measure capacitance and resistance of each pin to ground on the board, then trace where these caps are (usually few pF), they should somewhere close to the crystal, and preferably replace.. The new crystal might be working, but if the caps are still there and have high ESR, this could cause issues in the future or a short. However, looking at this board I do not see any obvious places for these caps (C20 is connected to a different net, so it is not it). If you check the rev 2.4 board, there are some, but on 3.7 I do not know where they are. If there are no external load caps, then new repair should be fine without them either. My guess is that the old crystal when it was new was working fine, but due to aging and moisture, it got slightly different driving characteristic.

[@OldSkoolF]

What is the Model# of that snazzy Oscilloscope please? And UGH.... I scrapped all sorts of old ISA/PCI/AGP cards about 5 years ago.. Not thinking they were worth anything... Buffalo Computer Recycling hopefully made out.. PS. Gravis cards were notoriously a PIA to install.

[@vswitchzero]

It’s the fnirsi-5012h. There are lots of hardware clones of the same model made by numerous companies and all are about the same as far as I can tell. It’s not the most accurate and the 100MHz it’s rated for is a stretch, but it’s perfect for my basic hobbyist needs. I think I paid about $120CAD for it.

[@OldSkoolF]

@@vswitchzero thanks... wanted a portable for checking signals when working on the road

[@Coolman6564]

Stupid question... what's the handy portable scope you used here? I've been dying to get one that doesn't cost $1000, lol!

[@vswitchzero]

It’s the FNIRSI 5012H. There are quite a few clones of it on Amazon and can be found for around $100 on sale. My needs are very basic so not sure how it performs compared to higher-end scopes but I’ve been quite happy with it so far 👍

[@MarcKloos]

I have a 3.7 too, which I upgraded to 1Mb of RAM back then. I haven't tried it in years, but if they are worth so much I really should find a motherboard to see if it's still okay!

[@mikebe4530]

Oh a GUS... I remember this card (and the gus max) used in the demomakers vidéos with the their great sound. But personnally i prefer the Guillemot maxi sound 32 wave FX with the daughter card ( a Real Korg wave table for great sound with midi files).

[@KuntalGhosh]

Holy shit i somehow thought it was the cryastal while u were cleaning the sockets. Idk why it made me think the crystal is bad. Most likely it was my experience with these crystals going bad often

[@unperrier5998]

The tolerance on crystals is very tight, typically 0.02% (20 ppm) Using a 19.6608 MHz instead of 19.7568 results in about 0.5% variation, so you're way out of the normal tolerance. But for sound, 0.5% difference in speed is just not perceptible, especially when using wave samples (patches) that last for about a second or so. Great find.

[@arpadmuller7992]

Hey, it's a fantastic video! I'd like to ask, after you fixed that card, have you tried to run the ultrasnd/setup and see if the DMA tests are OK ? I am qurious, because I am facing some very similar issue. DMA tests are fail, but the MIDI tests are sounds well. (AFAIK Midi playback dont need DMA.)

[@vswitchzero]

Thanks very much! No, my card would fail the DMA tests every time. As far as I know, this is just a bug in the software because the card works just fine. Are you seeing any issues with playback at all? Or just the failed test? Thanks for watching!

[@sniperviperman6400]

Used to have the GUS Max for a while with the extra memory slotted in for wave tables. wasn't really happy with it because it wasn't 100% Sound Blaster compatible in the later games.

[@okona1up]

I do have one in a 486 build complemented by an orpheus.

[@RetroTinkerer]

If I stop purchasing parts that are such a big deal to let them go I would had saved a little for that Orpheus. 🤣

[@Crack84baby]

I remember getting a chip puller when o was a kid. Wow was a game changer haha 😂😂

[@Cruz0e]

I'm sure I got one of these and the Voodo 5 graphic card too somewhere in a box :) I may have to dig it out... if they become this pricy :)

[@RealGengarTV]

I'd like to see if you play a A4 Piano note with the card if a chromatic tuner tuned to 440 Hz to see if it reports the correct tuning

[@vswitchzero]

Thanks for the tip! I'd definitely like to revisit this card soon. I received two more OEM crystals from Ian to try and would like to do some recordings with the 19.66, 20MHz and OEM crystals for comparison. Just need to find the time, which is the problem currently :)

[@RealGengarTV]

@@vswitchzero if the card still play all samples at the intended tuning, i don't see a reason to change the crystal TBH unless you have a huge urge to keep it stock

[@RealGengarTV]

Either way, I'm glad to see a follow-up video 😄

[@Otakunopodcast]

I burst out laughing when at 9:18 the music turned from a happy classical piece into some sort of grotesque eldritch horror abomination.

[@SJ-co6nk]

"I'm afraid your friends will find this gravis ultrasound is fully armed and operational!"

[@jefftiller5035]

sounds just like i remember even with the different crystal .

[@MichaelVx]

that is why these crystals having a ground connected to the can for 100% stable working and the only way for these old crystals to work fine is the ground with the can

[@gmergmer2606]

is it possible to make a card with a pci or pcie connector for new computers? As an example of a name - pcigus or pciegus (translated from Google)

[@vswitchzero]

There are some "bridge" adapters that allow AGP cards to work on PCI-e, but I have never seen anything like that for ISA cards. I'd imagine it may be possible to bridge an ISA card to PCI, but old DOS games need direct hardware access to the sound card via IRQ/DMA channels etc, so I'm not sure how well it would work. Would be very interesting to see though!

[@Stefan_Payne]

PRO TIP: use SBOS and play a bit Wing Commander 2 with the GUS. I guarantee, its interesting!

[@skoog7091]

I bought one as a teenager in 1993 and it costed about 147$ with 1mb memory. Retro hardware has become ridiculously high priced!

[@binface9]

9:35 i love it

[@dominikschutz6300]

I wonder how different frequencies affect the card... Great repair :)

[@vswitchzero]

I wonder the same. I actually got a couple more OEM crystals from Ian to try. I may actually do some before/after recordings and may even try the 20MHz crystal too for comparison if I get some time.

[@dominikschutz6300]

@@vswitchzero I saw a video where someone connected a frequency generator to a gameboy to run it faster. Would be interesting to see if the pitch changes or remains the same, so if higher clocked crystals can be used as replacement :)

[@mrbuddyboi5054]

The Gravis Ultrasound sounded haunted

[@wasteandglory]

i think i have a gravis ultrasound version in my shed somewhere laying around i know its an isa, thats why i stoped using it.. your going t pickup the mailcosts if you want it. last time i used it in a cyrixdx40 it worked but thats ... 30y ago?

[@vswitchzero]

Thanks very much for the offer! But since I already have one (and possibly a GUS a clone on the way), I'll have to pass. I'm sure there are many GUS enthusiasts out there who would love to have it though. You should consider putting it on ebay or a local classified site. They are definitely worth a fair bit these days and I'm sure it'd sell really quickly.

[@johanhansson9292]

I saw one thing you did wrong.. the IC extractor, you pulled with lifting force to extract the chip.. instead you should press the extractor together with your hand and the tool force the chip out..

[@vswitchzero]

Thanks for the tip! I will definitely try that next time. It was a bit scary the way it suddenly came loose. To be honest, I usually just pry them out with a very small screwdriver :)

[@johanhansson9292]

Don't worry, I did the same in the start, you can squeeze them together in your hand not taking out a chip, and you will see how the hooks go up in the tool. Btw... I bought about 5 or 6 of them untill I realized that there is a great differences in quality between the tools sold, i end up with one that was more then five times expensive the the one you find on AliExpress etc .. worth every dollar extra.. Keep up with your channel, we are in the same hobby 😉

[@makuuhne]

I don't think that crystal frequency matters at all. If I'm correct GUS never plays bit exact same anyway. Commands to GF1 are sent by io port in system time and synthesis plays at in its own clock. So sent commands can actuate on sample or another. Also I bet no-one can hear such minimal pitch shift.

[@andreavico6198]

Usually crystal are cacitive loaded (maybe the two yellow cap?).Weaker crystall can soffer for wrong loading. Maybe the correct vintage one are not so strong like the modern one and a sligly drifter capacitor can stop it to work properly.

[@vswitchzero]

Thanks for the tip! someone else mentioned this to me as well and will definitely take a look at the capacitors near the crystal. Ian sent me a couple more of the original crystals to try, so I'll be working on this card again (if I can find some time).

[@telekom1t244]

@@vswitchzero Hi, you can also try to build a very simple test circuit to measure the driving level of the crystal. You can then compare the original bad GUS crystal vs the original replacements and vs the new ones.

[@yatikarton2735]

Hey new here I love your content and I can relate with many points you make haha. I am more Into retro consoles than Pc's but I have my fare share of Pc's too haha. I also need to do a bit of saudering on my stuff and I am kinda new to this. I also don't have any good equipment so I wanted to ask you what you use. Thanks

[@vswitchzero]

Thanks very much! I really like Hakko soldering equipment. I have an 888d iron and a FR301 desoldering tool. It's quite expensive (especially the FR301), but the quality is excellent. If you decide to purchase Hakko gear, just be careful where you buy it from as there are many fakes/clones out there. Places like Digikey/mouser are a safe bet.

[@MrVITO5911]

Hi, I found in my basement a Nesxa 80386SX motherboard rev 0.5. Is it worth something?

[@vswitchzero]

Not familiar with that specific board but yes, just about anything from that era has some value these days 👍

[@jirja3192]

9:18 that was demonic...

[@andreaginepro1767]

ahhh the beauty of the isa bus :_)

[@amindlost]

I''m still kicking my ass for casually throwing out my Voodoo3 2000 PCI, AWE64 Gold, and Gravis Ultrasound (with a full 1MB of RAM!) during a hardware purge about 15 years ago. :(

[@Shawn_White]

If the card gives you problems again you might want to check the decoupling capacitors around the crystal.

[@Zerbey]

Giving away my GUS is one of my deepest regrets, one day I’ll get another.

[@modlabs]

What scope model you are using?

[@vswitchzero]

It’s the Fnirsi 5012H. Looks like it’s sold under numerous brands on Amazon. My needs are very basic so I can’t really comment on its performance compared to higher end scopes but I’ve been happy with it for the price 👍

[@modlabs]

@@vswitchzero i'm already bought a microscope almost like yours - Tomlov 602, and i will definitely buy oscilloscope like yours. Thanks.

[@worldgate989]

The one I have is older i guess because it has ram sockets on the tail end.

[@kenabi]

reminds me i need to find drivers for my pro audio spectrum.

[@ruthlessadmin]

My initial thought was demon possession, but then I realized it was probably a clock crystal. Always rewarding to turn out correct :D

[@anzelmasmatutis2500]

I amazed HOW repair depends on some random find, this time by touching...

[@seekerrsa8275]

Maybe Gravis ran out of 19.66 crystals and just stuck a 19.4 one in there they had laying around lol!

[@hi-friaudioman]

As soon as i heard that awful audio i though "man, how much do you wanna bet that crystal is bad" Also believe it or not grounding crystals is a common affair. Sometimes theres too much interference on that frequency and the can has to be grounded.

[@mattruddick8919]

Put a 47pf capacitor from the crystal oscillator pin to ground

[@mattruddick8919]

Or extend the leads on the crystal

[@ThePred2009]

aren't the creative soundblaster cards way better?

[@vswitchzero]

Thanks for your comment. All depends on what was being done with the card. The Gravis Ultrasound had some very clear strengths over the sound blaster (32 hardware channel mixing, wavetable MIDI etc).. but when it comes to gaming and compatibility, the Sound Blaster is usually a better choice. The GUS’s lack of hardware fm synthesis was an unfortunate weakness. That’s why many enthusiasts try to run both in the same system for the best of both worlds 🙂

[@framebuffer.10]

they said to not touch the components while the computer was running 😬 😅

[@vswitchzero]

Haha very true! Sometimes deseparate times call for desparate measures! I don't have a grounding bracelet or anything, but I do touch the power supply casing frequently when handling hardware just in case.

[@BrassicGamer]

Haha, try diagnosing an intermittent CRT fault without live running! Sometimes you just gotta do the risk assessment and get in there.

[@jeffm2787]

Just ground the can on the crystal and be done with it. A couple picofarad cap across the crystal would probably also fix it.

[@awilliams1701]

LMAO it sounds like it's possessed.

[@PROSTO4Tabal]

who expect at the end with 1mb ram to listen some more music ?

[@MrFlooperke]

I have the knockoff altrasound :)

[@mirabilis]

Perhaps the crystal capacitors are bad and out of spec.

[@vswitchzero]

Yeah that's definitely a possibility and something I will be taking a look at when I revisit this card again. Ian was kind enough to send me a couple more OEM crystals. I'll probably desolder and test the associated tantalum caps.

[@vad1mcso244]

song0.psm is a tracker module