A5000 Gone for good? :0

[nicf82]

No, the only floppy flashes I have seen have been after initiating the POST using the POSTBox. On startup, nothing. Is that significant?

[baz4096]

How confident are you in your PCF8583? Is it a known good part?

[nicf82]

Ha yeah bit of a sore subject, I changed the original, but it didnt fix anything, so when I cam back to it recently I changed it again in case I had broken it putting it in. I felt that one overheat as I was working on it so I put a fourth one in which is brand new and I was VERY careful so I am pretty confident. Also I am seeing data on the I2C pins on startup so seems to be doing its job.

[baz4096]

Which brings up back to trying to do a CMOS reset. Have you tried power on + R rather than DELETE ?

[nicf82]

No actually I dont think I have, let me give that a go...

[nicf82]

No luck with R either, unfortunatley

[baz4096]

Check traces from keyboard connector to IOEB, and make sure you've not adjusted LK1, 2, 4, 5 by mistake? Check there's signals happening when you're pressing keys on the keyboard.

[nicf82]

Thanks Baz I'll give that a go and let you know how it goes

I have tried probing the kin/kout earlier and saw no activity when pressing the keys.

[nicf82]

I have checked continuity of everything keyboard related, all seems OK and jumpers all seem to be in the right place

When powered on KOUT (at LK2) is HIGH, and KIN (at LK1) is LOW but there is a single HIGH bit immediately after power on. This can be observed as HIGH and a single LOW bit after the IC5 inverter at LK5 also.

The LEDs flash on power on, but they don't work after that. Typing on the keyboard has no effect on either KIN or KOUT. But would you expect to see activity on KIN regardless of KOUT working? Or does the keyboard need to receive anything from the computer to initialise it before it will blindly transmit any keystrokes? I am not sure if it needs 2 way functioning comms before it will work at all?

I am thinking maybe IC5 could be partly broken?


Cheers!
Nic

[IanS]

The LEDs flash on power on, but they don't work after that. Typing on the keyboard has no effect on either KIN or KOUT. But would you expect to see activity on KIN regardless of KOUT working? Or does the keyboard need to receive anything from the computer to initialise it before it will blindly transmit any keystrokes? I am not sure if it needs 2 way functioning comms before it will work at all?

Both ends need to be working to see more.

(from the A540 service manual, but all the keyboards use the same protocol)

[nicf82]

Thanks Ian that's very useful!

[IanS]

So the POST output earlier shows a PASS, so the only thing is the SRAM error, which should just need a reset (with a working keyboard).

Do you have a display connected, do you see any of the colured screens, Blue/purple etc. With a postbox and a PASS, you should see a green screen for quite a while whilst it is flashing the pass code on the floopy drive light

[nicf82]

I have not been using a monitor, so I just went and grabbed one to try it out. Seems when I am powering on the machine normally, I just get a blank screen.

Only when I manually initiate the POST sequence using the POSTbox do I get the red/blue/quick red/green sequence coming up. This also tallies with the fact that I don't see floppy flashes on a normal boot up , they only happen when using the POST box. Maybe it is for some reason not booting to POST??

[IanS]

The post runs everytime, but if you get a pass the colured screens don't show for long. Many LCD screens to respond quick enough to show them.

The long display of the green pass screen only happens with the POSTbox.

In earlier threads you had an A4000, have you tried that keyboard on your A5000, or are you sharing one keyboard between both?

[nicf82]

I am using the A4000 keyboard, but I have also tried another Archimedes keyboard, both of them work on other machines.

I am actually using the CRT from the A4000, but there is absolutely nothing showing without using the POSTbox, this is a vid of me power cycling the board a few times and finally initiating the postbox which you can see run on the screen in the background.

https://cms.carboninternet.net/wp-conte ... G_0800.mov

Cheers!
Nic

[IanS]

Unplug the postbox, once you get a pass, you don't need it, but it will prevent the machine from booting properly (unless you run the python script everytime)

[nicf82]

OK! Let me give it a try with just the monitor

[nicf82]

So I do see the red/blue on boot (without the postbox plugged in), but I don't see the green or any floppy flashes. After that just a black screen unfortunately

[IanS]

How much of a green screen or floppy flashes do you see on your A4000 when it's working?

The extended green screen and "error" code flashes only happen for a pass when the postbox is used. Your machine has no errors that need flashing.

Press DEL, power on, wait, do you see the bref colour screens? Do you see a red border screen. If not, there is something wrong with the keyboard input.

Did I say left go of DEL? no, tnen don't (You can't press DEL too early or for too lomg, err on the side of caution, wait lomger)

[nicf82]

Ah ok, I see so the "SRAM Err xx" is not considered an error worth flashing, if I understand you correctly?

I am seeing the brief red/blue but then no I am not getting the red border screen, so like you say I am pretty convinced there is a problem talking to the keyboard. CMOS not being reset coupled with no activity on the KIN/KOUT, it must be a problem I think.

Tomorrow I will test the KIN/KOUT on the A4000 to see what a proper handshake between the keyboard and IOC looks like then compare to what I am seeing here.

Cheers!
Nic

[IanS]

The SRAM error just means junk in the cmos, you can get that just having an invalid econet station id. It's not enough of an error to cause the red screen.

[nicf82]

I pulled out the A4000 and measured the KIN and KOUT lines straight after power up. This first thing that happens is that KIN goes high for a second or two, then back LOW. The handshake then happens which I have taken some screenshots of, posting here for info, I will use them to evaluate what is happening on the A5000.

A4000 Keyboard circuit

Magenta is KOUT measured at R213, Yellow is KIN measured at R63

Edit - I have just re-read this, and actually Magenta/KIN is R213 and Yellow/KOUT is R63!

Complete handshake

First exchange

First exchange zoomed

Second exchange

[nicf82]

I have probed the A5000 to compare the keyboard handshake and it certainly looks different.

Again KIN at LK1 is Yellow, and KOUT at LK2 is Magenta.

All activity

KIN activity zoomed

It looks to me like the keyboard is trying to initiate the handshake on KIN a lot earlier, but is not able to even hit 5v?

Then the long period where KOUT goes HIGH is strange because I saw something similar happening on KIN on the A4000 (no, it was KOUT on the A4000 as well), but before the handshake data

Maybe the keyboard it trying to communicate too early?

[IanS]

Again KIN at LK1 is Yellow,

And is the signal the same after IC5, LK5, and the kin pin on the IOC?

[nicf82]

Yes but inverted, it looks like the inverter has not got up to full voltage by this point either?


Here yellow is LK1, and magenta is pin 52 at the IOC:

Zoomed out

Zoomed in

[IanS]

That looks like the power supply is still coming up, It only makes sense to be looking at these signals after the reset signal.

Can you monitor the reset signal and kin? You could even measure it afer pressing reset on the keyboard as that will reset the arc and the keyboard controller.

[nicf82]

Sure! So here is the RST and KIN together, showing power on, and also after holding the reset button for approx 2s.

RST and KIN

It is quite hard to see, but about 2s after the RST goes high, there is a spike of activity on KIN, here is one of them zoomed in:

RST and KIN, zoomed on the KIN spike

I also took a shot of KIN and KOUT together and it looks to me like the KOUT queries the keyboard, the keyboard tries to reply on KIN but the reply is just poor quality (not square or sustained for long enough):

KOUT and KIN together

Compare that to the same interation as it happens on the working A4000, at the same scale, in fact it just looks a lot cleaner in both directions:

KOUT and KIN together on A4000

Could it be the problem? I am wondering with my limited knowledge if replacing any capacitors might clean up the signal?

Cheers!
Nic

[IanS]

Looks like you do have two way comms, it's worth looking a bit further along.

Blue is kout
Yellow in kin
First pic is the initial exchange looks very similar to yours. The second is the later exchanges.
The top row shows the zoomed out view showing both transactions.

Traces taken on an A3000 as it has the keyboard encoder on-board so is easy to probe. My signals are inverted as they are taken at the IOC end.

Your traces look a little noisey, with a lot of ringing, do you have a good earth connection to the scope?

[nicf82]

Yeah it looks like the same initial handshake there doesn’t it, I will check if the earth is ok and get another look at it tomorrow.

Looking at the second transaction, it looks the same as my A4000, but I could not see anything happening further along on the A5000 though but I will make sure of that.

[nicf82]

I managed to get a better ground and the signal looks a lot cleaner

But that's it, there is no second exchange later on