exxos's Atari, Amiga & retro forum

exxos wrote: 22 Dec 2018 10:39 But 3 different chips from 3 different manufactures all doing all manner of various crazy.. Obviously I am doing something wrong or killing them somehow.

I assume you already checked trivial things, such as accidentally assuming a wrong pin-out? I have used MOSFET switches in the past (would need to look up the part number, though) and I didn't find them "temperamental".

I've worked from the data sheets in all cases.

So single mosfets isn't going to be pretty...


So I have found a dual mosfet in a slightly smaller package and some SMT resistor networks to keep sizes down...

Will see how that layout goes later....


EDIT:

Its a lot smaller and that could work..

I got sidetracked looking for IC's once again..

I found a NXP part GTL2003 so ordered one yesterday and arrived today.. and this one works!!


3.3V as expected on the "left side" of the chip..


5V on the right side of the chip..



If either side goes low, the other side goes low as well.


So it's very possible I will be able to use this part instead of single mosfets :D :D


EDIT:

Interestingly, The 3.3V side doesn't seem to actually need a pull up resistor, and it still outputs 3.3V... So I need to look into that more, but that would save needing all the pull up resistors!


EDIT2:

exxos wrote: 28 Dec 2018 15:21I found a NXP part GTL2003 so ordered one yesterday and arrived today.. and this one works!!

Great!

Interestingly, The 3.3V side doesn't seem to actually need a pull up resistor, and it still outputs 3.3V... So I need to look into that more, but that would save needing all the pull up resistors!

I don't think you necessarily need pull ups on either side. This is how this type of switches works. You only might need pull ups on the high voltage side (5v), and only if you want a full voltage swing. Without pull ups it would work, but it would never reach 5v.

However, for most if not all of these signals you already have 5V pull ups on the motherboard. Most pull ups are not too strong, you know better than me, they are designed just to avoid the signal ever be floating. You might replace them for stronger pull ups. But if you consider this important, IMHO you should better use a transceiver that actively drives all the way to 5V.

ijor wrote: 29 Dec 2018 12:04 However, for most if not all of these signals you already have 5V pull ups on the motherboard. Most pull ups are not too strong, you know better than me, they are designed just to avoid the signal ever be floating. You might replace them for stronger pull ups. But if you consider this important, IMHO you should better use a transceiver that actively drives all the way to 5V.

It does drive to 5V. Though that is done on the motherboard pull-ups anyway. The 3.3V side doesn't need the pull-ups, so I hope I can tweak my original IDT layout to use the NXP chips without much fuss.

I've been looking at the GTL2003 chip footprint (TSSOP20) , and with my current PCB design its just not going to fit :( Because they are 8bit devices, 2 are needed each side of the FPGA... there is only just enough room for the 16bit chip..

I have been talking to Icky about this earlier today as well. How it looks, it will have to have the FPGA "above" the blitter socket.. but this does mean the board will be solder in , not "plug in". Its not going to fit all machines, but compromises & choices have to be made and that looks like the only way forward now.

As mentioned by Exxos, we have been discussing the fit of these Blitter boards.

It should fit both orientations of the bottom right Blitter socket location on most ST motherboards and miss the shielding but we won't fully know until we get one of these built.

Of course this version will not fit motherboards with SMT Blitter locations.


All That Glitters - Here is an image of the Blitter board. There still is a lot to do but we are making progress.

I know I've asked before but why hot choose the combined blitter,glue,mmu you have on the STE?

Is it just a matter of knowing the pin out etc? Was especially thinking this when doing the Mongrel board. One chip instead of 3 SMT ones.

Damn, somebody had another idea, I'm going to have to build one of these too! Takes mental note to buy a mainframe to maintain complex list of 'projects in flight'.