To Buffer or Not to Buffer (TF536)... That is the question.

[terriblefire]

i played around with adding an IDE buffer..

tf536_r2.JPG (189.06 KiB) Viewed 4316 times

I've also changed the buffers near the 68K sockets to 5V variants of the 3245 chips... I added a silkscreen note to be careful about this but I may manage to get them done as part of the assembly...

Wondering what to use for the IOR, IOW, CS1, CS2 buffering... i guess i dont need bidirectional and a 74125 seems a waste of chip enables?

[8 Bit Dreams]

Living a dream! ☺
Respect

[Danoo]

PCB artwork, love it, this is the kind of art I would like hanging on the walls at home, but the wife

[terriblefire]

So straw poll.. who wants with buffers and who wants without...

I personally think the buffers over complicate the board and i'd bet the performance will be exactly the same unless you want to put a long IDE cable on there (which 99% dont). But i thought i'd see how complex it was to do the board both ways.

I've mitigated the pain of this a bit by putting all but 2 of the buffer chips on the bottom of the board so that they can be soldered on by the boardhouse.

tf536_r2b.JPG (130.57 KiB) Viewed 4269 times

If i go buffered IDE it will be an initial run of 5, 10 max

[go0se]

I'm tempted to vote no as it seems to go against the stripped down ethos of the board. I run a 150mm cable on the TF330 without any issue, although admittedly I've never done a speed comparison on different cables as it's not that important to me.

I assume some people would want a long cable here in order to pull a CF adapter out of the side expansion slot on the 500/+.

I'll fire up the 536 to do a cable comparison, and leave it to others to present a convincing 'for' argument

[Danoo]

IMHO,
A stand-alone buffer is a better option for those that are looking for a longer IDE cable set-up, if a buffer is even required at all.
As Goose stated in the previous post, "a 150mm cable on the TF330 without any issue".

Realizing that the TF536 is aimed at different Amiga configuration, and there could be some possible issues with longer IDE cable lengths with this card. Why not just add a stand-alone buffer for these cases.
Maybe a few of the fortunate people that have a TF536 could run a few tests on cable length and data through-put, and post some results.

But I am a great believer in the KISS principle, so I would have to vote for no IDE buffer on the card.

[terriblefire]

I'll fire up the 536 to do a cable comparison, and leave it to others to present a convincing 'for' argument

Thanks go0se...

So with the IDE Buffer is a TF536 rev 2... Revision bump happens when i need to change the CPLD pin configuration.

Without the IDE Buffer we have a TF536 Rev 1b.

I have pushed both to my git repo. But i dont want to build and support both of these.

[PaulJ_2.0]

I agree an external buffer would fine.
Einstein once said 'Everything should be made as simple as possible, but not simpler'.

[terriblefire]

I agree an external buffer would fine.
Einstein once said 'Everything should be made as simple as possible, but not simpler'.

My mental block with the external buffers is how to reconstitute the chip select to activate the buffers (and the direction).

I guess the CS line can be reconstructed with CS1 & CS2. But the RW line ?

[PhilC]

No buffers for me thanks, as said previously, it works fine with a 150mm cable.