Ahhh the trying to get a 16bit card to work in an 8bit slot ? Looking at the pictur the jumpers seems proper except: Did you jump J1 ?
Have no experience with that particular Trident but have some Oak 16 bit cards that work in 8 bit slots fine if you jumper them proper but are not from 1994 era (the chips on yours tells me they are)
Added: I have a Trident 9000B that does work proper in 8 bit slot, a 512k card dated 1992. Works fine in my XT clone. Not sure about a 5150....

Horun - yep, I've tried all the jumper settings with no change (except for the block of jumpers at the bottom right - you remove any of those and it won't work at all). The card works great, there's just the booting problem, which only happened once I removed my MGA card and put in the VGA one (which has opened up a LOT of gaming options!)

How exactly did you mod the card? Cut trace? Masking tape?

Here are the pictures I took of the mod I did - I've also tried a pullup resistor between the ALE and +5V with no change in the way it boots:

(Yes I know the solder blob looks crummy - I was trying to get the pictures while I was doing it, and it's been cleaned up since!) 😀

And just to show it actually works - and the BIOS screen:

Oh I have no doubt it works, I was just wondering if the method you picked had something to do with it not booting properly. But it looks OK to me so that's not it. When I experimented with ALE on one of my cards I did a soft-mod, used capton masking tape on the gold contact instead of cutting the connection.

Anyway, did you try a manual reset via a switch when it hangs? Does this help any or do you need to wait for the system to warm up a bit? Unless this system has no reset button, being IBM that obviously would never hang...

Dead shorting ALE to +5 is retarded

maxtherabbit wrote on 2022-02-20, 18:15:

Dead shorting ALE to +5 is retarded

I have to agree, maybe using a 470 ohm or 1k resistor be OK, be enough to pull high w/o risking some over current thing.
After looking over 3 - 8bit VGA cards and 3 - 8/16bit VGA cards that I have: 5 of the 6 do not even use the ALE/BALE line. Only the one Trident 9000b does, the others including a Trident 8800c do not.

Deunan - I like the idea of covering the pin, but the trace is already cut, and it shouldn't be much problem to reconnect if I decide to use this card in another system. I will keep this in mind for the next experiment, though!
I did implement a reset button (temporarily) that grounded pwrgood with a push button switch - it had no effect on the VGA boot problem, and the 5150 did NOT like the reset button, so it's gone for now.
As of right now, if I power it up and leave it hung for about 3 minutes, power off for 2 seconds and power back on - it works every time. There was no boot problem with the original monochrome adapter in the system.

maxtherabbit - good to hear - I'll be sure to let the several people that recommended it to me know your thoughts - do you have a better solution for this specific card?

Horun - just edited the original post as I forgot to state - I now have an 820 ohm resistor in place of the direct connect, and its made no difference with the boot hang.

TheRealAnubis wrote on 2022-02-20, 19:12:

Horun - just edited the original post as I forgot to state - I now have an 820 ohm resistor in place of the direct connect, and its made no difference with the boot hang.

Good ! The resistor will help to protect the video card chipset from over current. Right the boot hang must be a 5150 thing. Wonder if it would hang with a different 256k VGA card.

Horun - I only had it straight connected for about 3 days, then I tried the resistor with no change, so I just left the resistor in. I may update the pictures to show the resistor option, but hopefully anyone wanting to do this will just read the posts.. I did the mod, then took the pictures, then added the text and resized everything that night.

On other VGA cards working or not - I'm not sure - since the prices on all this stuff have gone insane, I don't have the spare cash to try too many ones that may not work.. The ones that are listed as working with the 5150 are about 3-4X what I paid for this card. I need to get a better understanding of the ALE and see if maybe some of the pins from the unused (in the 8 bit slot) portion of the card may need connected to something. BUT - since this card has jumper settings to choose 8-bit, probably not. To me it really seems like either the card or the PC is getting ahead of the other component and not getting some signal that it needs - possibly my turning the power back on after about 2-3 off seconds allows the current not to fully dissipate and then it works right? I really wish the reset button would have done it.

I see now that you cut the trace to the ALE pin going to the ISA bus, so I formally rescind my accusations of retardation. Now it's simply "unwise"

maxtherabbit - HA! Yeah, I wouldn't have done that to the 5150! The +5V only connects to the card's ALE.

It's a good practice not to connect inputs directly to VCC or GND, but use a resistor to limit the current. That being said, if the input is a modern CMOS with protection diodes, and it's always an input, and it's connected to the very same power plane that supplies the chip, what exactly could go wrong? Also, the choice of the resistor matters. If people do not understand what it's for and what value it should have I could easly picture someone trying to pull-down a bunch of TTL chips with something over 1k and wonder why it's not doing its job.

Fact is, that chip might already have some of its inputs permanently tied either low or high, to configure something like memory interface or DAC type, and I bet those connections would be directly to VCC/GND on the PCB. So what's one more? And I can't remember what chip(s) that were right now but if I had to I could dig up some datasheets that specifically say NOT to use resistors. So it's not like this is a golden rule that has no exceptions.

I took an educated guess based on the 4.7k resistors for J2, J3, J4. so a 470 to 1k ohm would overrule a 4.7k pullup/pull down if that was also in use for the ALE signal. It could not hurt 😀

added: there are some simple circuits that can be used to make a true reset switch connection to the 8284A and it's res line, using an RC network on the Schmidt trigger input....assuming the reset line goes to the 8088 reset pin as in the 5150's schematic ;p

Deunan - all good advice - I'm fairly new to the early PC scene, so most of my ideas were actually the combined input from online blogs/chats/groups. I was looking for a better reset button, but lots of online information said it was fine.

I read that the ALE was improved in the 5160 and up - what did they do to fix it, and is it something that someone else could re-create without a crazy amount of work? I'd be happier if the card just worked when I plugged it in, and even now, waiting 3 minutes and restarting isn't the end of the world, but it does cut into the time I have available as these 3 minute waits add up! I will say that on one of my reset button forays, the VGA BIOS read CGA and 8 BIT instead of the picture I posted. I wonder if there's a way to trigger those modes directly then? Anyway, I'd love to just flip it on and have it work. I may need to hunt up another VGA card at some point, but this one is all I have and I'm willing to tinker to get it more reliable on boot.

Horun - I'll see if there's something I can find that would do it properly - I recall reading one that mentioned that they got some extra 8284a's because the reset button could easily blow them. This didn't sound like a good enough fix for me to try, but that's just one instance.

Horun - I forgot to ask you - with the VGA cards you have, have any of them been tested in a 5150? I'm looking for examples from people that have actually had them working in their machiens. Also - a new detail from the cold boot problem - if it's been on for a while and being used, I can turn it off and work with the files on the CF card for at least 10 minutes and it will startup first boot.