Cool card! Wouldn't be surprised if this turned out to be the fastest VLB card out there... I'd like to see some benchmark results once it is fixed. 😀

Check for any bent legs on the main ARK chip, if you haven't already.

Check the tantalum caps to see if any are shorted (continuity between the legs of a cap).

Clean the VLB contacts with a pencil eraser.

Check for any tiny scratches or dings in the PCB that may have damaged a trace.

With a very good magnifier check for any broken solder joints, especially on the (populated) RAM sockets. This is very very common with cards of this age. If they've been mishandled at any point there's a good chance one of the legs from a through-hole component will have gotten tugged on enough to break free of the solder. It might be very hard to tell.

I use one of these when inspecting broken solder joints, it works beautifully, though there may be more convenient devices for this.

Poke the various parts of the card while its running to see if the lines change, this could help pinpoint any physical issues. (broken traces, broken solder joints, etc.)

I'd be very very surprised if it wasn't fixable. Basically, anything other than an internal defect in a non-replaceable IC (like the ARK chip) could be fixed. One bad trace or a bad solder joint are actually very easy fixes if you're somewhat experienced with a soldering iron and have the tools.

How do the lines change in different video modes ?

Since it is in text mode I am suspecting the connection to RAMDAC or the RAMDAC itself being the problem.

I'd check the solder joints on the memory sockets.
Also, as far as I know with this card it can work with only the DIP RAM sockets filled. My card has the DIP sockets filled and uses the SOJ sockets to upgrade to 2MB. I suggest filling those and removing the SOJ chips to see if it makes a difference.

Like the Cirrus Logic CL GD542x cards, the second meg of memory on this card is fairly useless. I think it will at least get you true colour in windows at 800x600 though.

At first glance the legs of both the 1000VL and the RAMDAC look good. I'm not sure if I have enough DIP RAM chips of the correct size and speed to populate those sockets. I'll have a rummage through my boxes this evening for those. This card gets prio over anything else I have in any event 😉

I'll check the different modes this evening too, and the tantalum caps (I assumed they wouldn't be the problem as they'd short the whole card out).

Did you try setting the waitstate to 0 or 1 to see if that has effect?

Funniest thing, this was my ARK1000 VLB when I got it:

problem was obvious in my case, bent leg on the chip that shorted:

Hmm, looks very familiar - but all the pins here are straight.

Didn't have as much time to look into this as I hoped. Still not found enough 70ns or better DIPs to try that. Did quickly boot it up to see how other modes looked. Same number of pixels between lines it seems:

Not sure what that means...

I'd definitely look for bad solder joints on the back with a magnifier. Also! check for loose legs on the surface mount components. Don't be afraid to poke at them a little. I've seen a few cards where the legs of surface mount chips must have had been poorly soldered and the leg actually came free from the solder pad easily. It is very hard to see but if you touch it you will see movement.

Replace memory chips just in case.

Cheers, pentiumspeed

I hope you get the card fixed... in terms of performance, it is certainly in DOS the best i've seen.

Come to think of it, I had this very problem a while ago... I changed out the memory chips and the lines went away!

I'm still mixed on this card. Is it really faster than a Trio64?

I have two of these. One just like in this thread (the 2theMax version), and an Octek.

Anonymous Coward wrote:

I'm still mixed on this card. Is it really faster than a Trio64?

I have two of these. One just like in this thread (the 2theMax version), and an Octek.

Depends on your benchmark.

3DBench 1 & 2; both cards equal

PCPBench; mostly equal, though Trio64 edges past slightly @ 8bpp

Doom; -
Ark1000 - 1907 realtics, 39.27 fps
Trio64 - 1923 realtics, 38.84 fps

In reality though, nobodys going to notice the difference. Have you found any difference benchmarking between the 2theMax version and the Octek?

I'm pretty sure my doom results with the Trio64 are *slightly* faster than the ARK1000VL...but it might just be my board (SiS 486 EISA chipsets). One thing is for certain, the Trio64 has a much better picture, is more compatible and has a way better RAMDAC which makes Windows look pretty darn nice.

The only real difference I have noticed between the 2themax (Bali32/Stingray Pro etc.) and the Octek, is that with the Octek I can actually get 0WS working. The setting did not work with the Stringray on any of my 486 boards...however, it did not make ANY difference in the benchmarks. I was pretty sure the jumper on my motherboard is also set to 0WS, but who knows. The Octek also has a very rad yellow and blue sticker.

Update after quite a while - today I received 8x DIP DRAM chips, M5M44256BP-7. These should be compatible with the card, so installed them.

Results pretty disastrous - first I booted with just the two SOJ chips. Behaviour was exactly as previously seen. Then I installed the DIPs for a total of 2MB. This resulted in lines, but also colour shifts (grey background). So RAM had some kind of impact 😀 Finally I removed the two SOJs and... absolutely nothing, screen stayed black during POST and it hung on 85, which I believe is AMI WinBIOS for primary display error.

Checked the DIPs, all legs were seated correctly, but wiggled and pressed them just in case. No change. Added the SOJs again. No change. Removed the DIPs. Still no change, no picture and 85 error. Looks like I seriously borked something 🙁

Unless the keys on the card are wrong, orientation isn't the problem, keying on card and chips is clear and all 8 were correctly lined up. Apart from that... no clue really.

Did you repair this card?
I found your thread because have another strange issue with this card.
It produces a bit poor quality output, text is not so sharp. Couldn't find the problem..

Afraid not, it went onto the pile of "will dig deeper once I have more time and skills".

Have hardly had any time to do serious retrocomputing work these past months, with new job and a lot of studying for electrical engineering course. Tomorrow I'll have last exam until after the summer, so should be a bit more relaxed. I'll take this as a reminder to look into this card again. I now have a USB microscope which might show up soldering issues I missed visially. But no promises and at best this will be #3 or 4 in my priority list, as I also have to get a PC ready to donate to a good cause, then get my new Ryzen system connected to a fanless case and after that there are 1001 other things that need attention too.