CDTV and A500 Harddrive controller

Jope - I fully understand your story and it makes sense in theory, but I do understand Twanger's confusion.

We're trying to send a signal from Motherboard CONFIGOUT to the board1's (DMAC) CONFIGIN, from board1's CONFIGOUT to board2's (in my case GVP harddrive + extra RAM) CONFIGIN and from board2's CONFIGOUT back to the mainboard's CONFIGIN.

There's no special soldering required to do this, because the traces from DMAC are already there (contrary to what some other sources state). You can have a look at the attached (DMAC to Diagnostic 3.jpg) to see for yourself.



However, I think the problem here is that both CONFIGOUT and CONFIGIN at the Diagnostic connector are grounded. So how can we expect a signal to go from ground to ground ????

(

On my Amiga 500's side expansion pin 11 (CFGOUT) is not connected and PIN 11 (CFGIN) is grounded. But the 500 hasn't got DMAC and I suppose any device on the side expansion is the first device.

My guess is that on the CDTV I have to cut the trace between CFGOUT on the diagnostic and ground to make the chain work.

Other attachements:

[Jope]

Judging by the schematics, cfgin is just plain grounded and it seems that cfgout is left floating on pin 5 of the diags connector..

Have you verified all your connections with a multimeter? (you're sure that those pins you outlined in the pics are connected?)

[Twanger]

@ Jope: Thank´s for the abstract again. Yes, I understand what you explained and that was all already clear but:

Originally Posted by jope

the DMAC's config out must be disconnected from it's present destination and be rerouted to your new device's config in. your new device's config out must go to where-ever the DMAC's config out used to go

when you read my post

Originally Posted by twanger

...the CFGOUT goes nowhere else as to the Diagport...

it is not neccessary to do what you said.

-----

CFGOUT is grounded (Yes, I tested it with a multimeter!)

3. Does anyone know where the Autoconfig mechanism is started on the mainboard?


Attached you can find the relevant schematics so we all know what we are talking about (as mentioned there is one mistake in the schematics: CFGOUT is grounded at the diag.port!).

I have just removed part of the Diag connector to have a clear view of the traces underneath.

Both Diagnostic pin 5 and pin 6 are connected to pin 1, but contrary to other sources it looks like pin 1 is NOT GROUNDED.

So it look like DMAC CFGOUT is simply connected to DMAC CFGIN via pin 1 of the diagnostic connector.

If that's the case then currently the AUTOCONFIG signal could go as follows:

Motherboard > connected to DMAC somehow (?) >DMAC CFGOUT > DMAC CFGIN > back to Motherboard (?)

I now guess we should cut the traces between CFGOUT and CFGIN and put our device in between these. Any thoughts ???

[Twanger]

According to my measures pin1 (and thus pin5 and pin6) are grounded...

I just tried the following:
- disconnected the traces of CFGIN and CFGOUT from DMAC to Diag.port
- connected directly CFGIN-DMAC to CFGOUT-A590
- connected directly CFGOUT-DMAC to CFGIN-A590
=> none of the signals are grounded any more
=> result: not working... what now?

maybe the problem is that it is NOT possible to put the device (A590) before the DMAC; but how put it in the chain behind DMAC? And where is the Autoconfig signal started and where does it end?

[Jope]

Originally Posted by Twanger

- disconnected the traces of CFGIN and CFGOUT from DMAC to Diag.port
- connected directly CFGIN-DMAC to CFGOUT-A590
- connected directly CFGOUT-DMAC to CFGIN-A590
=> none of the signals are grounded any more
=> result: not working... what now?

Hmm. Now you created some sort of autoconfig loop with both the DMACs signalling eachother that they are free to autoconfig themselves (although I'm not sure that they'll autoconfig themselves again once they've secured their address range)

What you should do (and what I've been trying to say all along):
motherboard (GND?) -> CFGIN-DMAC
CFGOUT-DMAC -> CFGIN-A590
CFGOUT-A590 -> GND

It's a chain, not a loop.

However I was holding back my reply because I wanted to check my hardware manual to refresh my memory on how the autoconfig gets started.. It could well be that the first board is just plain grounded and thus starts autoconfiguring immediately, but it's been some time since I last read that book.. I'll try to get to the storage room today and check it out.

Jope > OK, after I've seen for myself where the traces go I have now a clear idea of how I can make your suggested scheme work. It makes sense and I'll give this a go. I was just confused because of all the wrong info on the web about grounded and non-grounded signals.

Twanger > Are you measuring with your device connected to the Diagnostic port or disconnected. If you have connected pin 1 of Diagnostic to pin 1 of your device then it is grounded through your device.

[Jope]

Ok, I took a look at my trusty Amiga Hardware Reference Manual at the section on AUTOCONFIG(TM). Indeed it works just like can be deduced from the schematics and the grounding bit, that I was unsure of, is really like that. I don't know why I hesitated on this, as it's clearly indicated in the schematics that the signals are active low. Sorry.. Then again, I wasn't really working this problem with all my heart as I thought I had already started you guys on the right track. (and also I didn't have that much free time to devote to this)

Autoconfig works in the following manner:
The backplane (motherboard) activates the first card's _CFGIN line and then that card starts listening to the autoconfig memory space. The OS tells the card via this memory space where it will end up or whether it should shut up and not do anything.

Once the card has been told to configure itself, it does so and then it activates it's _CFGOUT line, which is connected to the next card's _CFGIN (like I've mentioned a couple of times now).

If you have a properly designed zorro backplane, the cards don't need to be one after another, as the logic on the backplane can route the signal to the next populated slot (it basically detects whether a card is there or not and then just connects the signal to the next slot). This however does not matter in the A1000/A500/CDTV, as the cards are always adjacent.

This leads me to believe that the last _CFGOUT can indeed be grounded, it doesn't matter as it never gets read by anything and also the chip itself pulls it down.

Twanger, I missed this question before (I just saw it in one of your side sentences while reading through the thread again), but a low signal means that if it's grounded, it means it's active.

If you have a signal, it's active state is normally +5V

If the signalname has a line drawn over it in the schematics or if it has an underscore in front of it (_CFGIN) it means it's a "ground to activate" signal (active low, high == +5, low == signal gnd).

Ok. Now armed with this knowledge that autoconfig is a chain that begins with ground and ends with ground, can you guys test it?

[Twanger]

Aaah, I think now I got it (in theory at least) - THANKS!!!

I think there are generally two options on how to get it work:

1. DeviceXY before DMAC: Ground > DeviceCFGIN; DeviceCFGOUT > DMACCFGIN; DMACCFGOUT NOT grounded (simply not connected to nowhere else as there is no further device needed to be configured).

2. DeviceXY after DMAC: Ground > DMACCFGIN; DMACCFGOUT > DeviceCFGIN; DeviceCFGOUT again not connected.

Remarks:
* All above assumed that there are no further connections which could ground anything that should not be grounded (connections between DMAC and Diag.port).
*A german developer told me not to ground the last CFGOUT (but maybe it also works grounded as Jope said).

[Jope]

2 is better.

If someone said leave it floating, then it'll probably be just as good, but like you saw from the CDTV's DMAC, it works if grounded too.

(the DMAC in the A590 is the same chip as the DMAC in the CDTV)

[Twanger]

@Doelen:
Did you get it running?
I tried both ways but it doesn´t work

But there are still some small things that I could change/try. I´ll keep you informed if it works...

Twanger, not sure if you're still on this board, but no, I never got it running either.
BTW, I agree that option 2 is better but in the case of my GVP it's not really an option since DeviceCFGIN is not connected on my board so that would break the chain (it may be grounded on the board).
There's an A2091, similar to your A590, for sale on e-bay. I might give it a go with that one. If that fails I'm afraid I'll just try to look for a proper SCSI-TV (rare and expensive!).

[Twanger]

Hi doelen,

if you find out anything new it would be great to hear from you!

I´m also going to buy the CDTV-SCSI Module (but I still want to get my idea running!!!).

You can buy a NEW SCSI 2 CDTV module here: http://forum.cd32-allianz.de/posts/list/128.page

Good for me that the producer is from Germany...

Here is another very good Amiga board with also one english speaking board: http://www.a1k.org/forum/
You can find me there...

Twanger, that's very random! I read your message now for the first time and I have just 2 days ago ordered the same SCSI controller from Jochen!
Anyhow, I'll give that other forum a try as well as you suggested.

Cheers,
Roland