Need some help fixing CPC6128 not booting

[arnoldemu]

http://info-coach.fr/atari/hardware/fd-hard/cd-8002-1.pdf

all in here.

mfm.
double density.
ibm system 34 floppy disc format.

the information near the end shows the gap, id and data fields.

it also describes the mfm encoding too.

[RobertM]

Hi all, I am stuck again.

I downloaded CPCDiskXP in the hope that I could format some 3.5" disks in a format that the CPC can use.

I don't have a Floppy Port on my main board so I am using a USB floppy. Unfortunately it looks like CPCDiskXP can't format a USB floppy, it can only format a Floppy connected to the main board. Is that correct???

This is frustrating lol, If I had CP/M on a 3.5" in the correct format then I could just use the CPC to format disks.

Just to get that first disk @#$%&!

OK, It looks like I have to put together another PC and load XP and use a main board connected Floppy Drive Just to get **something** that will work on the CPC with a 3.5" drive.

OH Please ... is there any other way ?

I haven't done the hardware on the CPC yet. I am going to connect a 3.5" 1.44MB Floppy drive to the internal floppy connector. The 3.5" I have doesn't use 12 Volts so the existing 5 Volt power supply should run it. (5V - 3A). The floppy drive uses less than one AMP.

I tried the DOS Format switches with the UBS Floppy drive and the Floppy disk hole covered over (Double Density), All I get is 'Format not supported'.

[Bryce]

Don't you know anyone with a PC with internal floppy that you could use once? I use ManageDSK to format CPC floppies on the PC, but I don't think it works on USB drives either.
There are also several other CPC users in Australia, maybe one of them can send you CP/M on the 3.5in floppy?

Bryce.

[RobertM]

Hi Bryce,
              I have installed a floppy into the CPC case to work as drive A.

I followed a picture you have here -
Adaptor Cable for Internal HxC
and it didn't work.

The three signals that seem to be transposed in the picture are ...
Main Board - Floppy Drive
NC         ->   /DriveB
Drive0    ->  MotorA
/Index    ->  /Index

NOT transposed is -
/MotorOn   -> /MotorB

I looked up how a 'standard' 3.5" floppy works and found that by default they are hard wired as DriveB

So I just transposed the last two wires - /Index to give -

/MotorOn     ->   /MotorB
NC               ->   /DriveA
/Drive0        ->   /DriveB
empty          ->  gap
/Index          ->  /Index

Now it seems to work .. it lights when I type |A the motor runs and I can hear it seeking. I don't yet have a formatted disk so I get an error.

I also saw this picture on another page but I can't find it now.

I also bridged the /RDY signal to ground on the Main board connector as new 3.5" use this signal for DiskChange instead.

I wanted to do it this way because when I need to replace the drive I can get *any* standard PC 3.5" and just plug it in without needing to worry about jumpers.

I don't have an Amstrad monitor so I am just using a 5 Volt 3 Amp 'Wall Wart' and NO 12 Volt supply. Newer 3.5" floppy drives don't use 12 Volts anyway.

The CPC6128 is 1.7 Amp and the Floppy Drive is 0.96 Amp so the 3 amp supply is fine.

One other thing that I might mention is that the CPC6128 Floppy Power connector uses RED for 12 Volts and Orange for 5 Volts. On my CPC they had the same colours / positions on the connector as a ATX supply meaning they were transposed when compared to a normal 3.5" floppy. I just swapped the pins in the connector at the end of the floppy power lead. Oops - except that orange is yellow (Sorry I am colour blind).

This would be an issue for people using an amstrad monitor and have 12 Volts connected to the CPC as the 12 Volts would be going into the floppies 5 Volt pin. I am hoping that this info might be helpful for someone.

The confusions started for me when I read HXC / Internal adaptor. I read this to mean HXC and Internal Floppy Drive adaptor cable. It now looks to mean HXC running on internal floppy connector adaptor cable. Or something like that.

Now I will put a PC together with a 3.5" floppy and XP and do some formatting / disk dumps.

[Bryce]

Hi Robert,
       the cable I posted will only work if you have changed the 3.5in floppy from DS1 (standard for PC floppies) to DS0. This is usually a small solder jumper on the floppy PCB. There's also usually a RDY jumper on the floppy PCB so that it sends the RDY signal at the right times instead of permanently connecting it to GND which causes problems with some software.

Bryce.

[ikonsgr]

I just finished a logic mapping of all cpu pins between the working and the faulty cpc's. I found differences for the above pins:
 

PIN	GOOD	BAD
3	all	all	FLASHING RED
4	GY	all	FLASHING RED YEL
5	all	all	FLASHING RED
12	all	YG
16	YR	G
20	YR	R
23	YR	R
29	GY	G

So, as it seems there are differences mostly on some pins of the address bus where instead of having constant light of all 3 leds,i got flashing (with a period of ~10Hz) red and red yellow leds. So, does this mean the problem lies to os/fw rom chip or any of the other larger chips (8255, AY Sound chip etc) might have problem too?
Btw, robertm, as i have a rather big expecrience using 3.5" floppy drives with a cpc, i can tell you for sure that almost ANY 3.5" 1.44mb IDE floppy disk drive for pc, can be directly connected using a ribbon cable! Only thing you need to do is to connect pins 1 and 2 for the "ready" signal! Of course you can also put switches for making the external drive "A" or "B" (pins 23,24) and choosing side of disk (pins 3,4)!I'm making such ribbon cables for some years now, and i have tried dozens of different floppy drives, up until know all seem to work just fine! ;-)

[RobertM]

Your tests results don't seem right.

Can you verify them.

Pin 29 (GND) should NEVER have a yellow light. A yellow light here means you either have a problem with how the probe is grounded to 0 Volts or you have an open circuit Via. On the bottom of the circuit board under the Z80 is one long vertical track. The via at the end opposite pin 29 is where the broken via would be. A dirty Z80 socket can do the same.

None of the other results can have any meaning while you have a yellow on Pin 29 (GND).

[RobertM]

I just finished a logic mapping of all cpu pins between the working and the faulty cpc's. I found differences for the above pins:
SIGNAL    PIN       BAD   
A13          3           all    FLASHING RED
A14          4           all    FLASHING RED YEL
A15          5           all    FLASHING RED
D2           12         YG   
/INT         16         G   
/IORQ      20         R   
/BUSAK    23         R   
GND         29         G

So, as it seems there are differences mostly on some pins of the address bus where instead of having constant light of all 3 leds,i got flashing (with a period of ~10Hz) red and red yellow leds. So, does this mean the problem lies to os/fw rom chip or any of the other larger chips (8255, AY Sound chip etc) might have problem too?
Btw, robertm, as i have a rather big expecrience using 3.5" floppy drives with a cpc, i can tell you for sure that almost ANY 3.5" 1.44mb IDE floppy disk drive for pc, can be directly connected using a ribbon cable! Only thing you need to do is to connect pins 1 and 2 for the "ready" signal! Of course you can also put switches for making the external drive "A" or "B" (pins 23,24) and choosing side of disk (pins 3,4)!I'm making such ribbon cables for some years now, and i have tried dozens of different floppy drives, up until know all seem to work just fine! ;-)

Did you compare the pins to the "expected results" I listed before?

These results show that R110 needs to be replaced but that makes no sense as you don't have a Red on Pin 25 (BUSRQ) which means the CPU is faulty and you have replaced that so it makes no sense.

This is a nonsensical result. 

[RobertM]

Bump

Address bus Pins 1 to 5 and pins 30 to 40 Constant pulse both red and green

Data bus Pins 7 to 10 and pins 12 to 15 Constant pulse both red and green

Pin 6  Clock   Constant pulse both red and green
Pin 11 Vcc     No pulse red
Pin 16 /INT    Constant pulse red
Pin 17 /NMI    No pulse red
Pin 18 /HALT   No pulse red
Pin 19 /MREQ   Constant pulse mostly red
Pin 20 /IORQ   Constant pulse mostly red
Pin 21 /RD     Constant pulse mostly red
Pin 22 /WR     Constant pulse mostly red
Pin 23 /BUSACK No pulse red
Pin 24 /WAIT   No pulse red
Pin 25 /BUSREQ No pulse red
Pin 26 /RESET  No pulse red
Pin 27 /M1     Constant pulse mostly red
Pin 28 /RFSH   Constant pulse mostly red
Pin 29 GND     No pulse green

[Bryce]

Doing this type of debugging with a logic probe is a futile exercise that's not going to solve anything. Without properly reading the signals with a scope or logic analyser is just going to go in circles.

Bryce.

[Munchausen]

You can get a decent logic analyser for under £10 on ebay: Hobby Components USB 24M 8CH 24MHz Logic Analyser & Test Hook Clips | eBay
Hobby Components USB 24M 8CH 24MHz Logic Analyser & Test Hook Clips | eBay


Most of them are seleae logic clones so you can use the seleae software. I bought from that actual seller.

[Bryce]

Technically, that is a logic analyser, but it's far from what I would consider "decent". It might just be enough for the job here, although it would be much more complicated and slow compared to using a proper logic analyser and it definitely wouldn't be enough to fault find anything even slightly more complicated. I wouldn't recommend it to anyone who wanted to regularly fault find digital circuits. Of course a real logic analyser is going to set you back a lot more, but it's worth it if you intend making electronics your hobby.

Bryce.

[pelrun]

Actually they're a lot better than you give them credit for! I don't have a Saleae or clone myself (although a friend does), having opted for the OpenBench Logic Sniffer (and more recently a DSLogic from the kickstarter) and those are pretty awesome. Definitely compared to the 'high-end' logic analyzers that end up having fewer features because you have to pay $$$$ for each and every protocol you want to decode.


I've debugged some pretty hairy issues using them, on all sorts of boards (running at far higher clocks than the trusty CPC, at that.) My friend has used his Saleae to good effect debugging some high-performance radio modem hardware, amongst other things.

[Bryce]

But how do you manage something like this:

Monitor a UART signal, trigger on a particular word and show what's on the data and Address bus for the next 5 seconds.

I know how to do this on a proper logic analyser and it's typical of the type of test I might need to do, but I've no idea how I'd do it with just 8 channels unless I repeated the test several times and even then I would be unsure of the results.

Bryce.

[RobertM]

But how do you manage something like this:

Monitor a UART signal, trigger on a particular word and show what's on the data and Address bus for the next 5 seconds.

I know how to do this on a proper logic analyser and it's typical of the type of test I might need to do, but I've no idea how I'd do it with just 8 channels unless I repeated the test several times and even then I would be unsure of the results.

Bryce.

Well the simple answer is 'with FPGA'.

Back in the 80's we didn't have FPGA! and we seemed to manage with our dungy old 20MHz oscilloscopes.

[pelrun]

On the saleae, I wouldn't know (although it's continuous streaming, so you could capture a large chunk of data without triggering, run the UART decoder on the lot, then search for the bit you're interested in.) But yeah, 8 channels isn't enough to capture a couple of wide buses simultaneously. Lucky for me most interfaces have moved from parallel to high-speed serial these days, so you need fewer channels anyway


My OBLS captures to BRAM in the FPGA, so I can capture up to 200MSPS but I'm limited in buffer size - so I need to set up an appropriate serial trigger on the appropriate signal line, set it to capture 99% after the trigger point, then run the capture. But it has 32 channels.


The DSLogic has a separate DRAM buffer, so it can capture a huge amount of data (only 16 channels though ) and do either approach


(and when I get my Novena board, I could roll my own off the FPGA interface...)

[Bryce]

Well the simple answer is 'with FPGA'.

Back in the 80's we didn't have FPGA! and we seemed to manage with our dungy old 20MHz oscilloscopes.

With FPGA??? You're suggesting building your own device for one measurement? Or how will the FPGA suddenly add channels, functions and buffer RAM to an 8 channel analyser?

Back in the 80's I was limited to a 10Mhz analogue 2 channel scope due to finances, but it was a struggle and many things just couldn't be tested. Today you can get a real 24 channel analyser with word / byte / pattern recognition and a decent sized buffer for a few hundred Euros. I am also not limited to pocket money and cutting the neighbours grass to earn money

Off course, as I said earlier, it all boils down to how much you intend to use it and how complicated your target device is. If you intend doing regular fixing / developing then I'd still recommend spending a little more and getting a semi-decent MSO. That way you have an oscilloscope and analyser that will serve you well. Something like this will get you a lot further than those low-cost USB solutions: Hantek MSO5102D 100MHz 2 Channel 1GSa/s Oscilloscope 16CH Analyzer FREE EXPRESS 

Bryce.

[RobertM]

I am also not limited to pocket money and cutting the neighbours grass to earn money

Good for you Bryce.

[Bryce]

Now I just get to cut my own grass and nobody pays me for it

Bryce.

[Munchausen]

Technically, that is a logic analyser, but it's far from what I would consider "decent". It might just be enough for the job here, although it would be much more complicated and slow compared to using a proper logic analyser and it definitely wouldn't be enough to fault find anything even slightly more complicated. I wouldn't recommend it to anyone who wanted to regularly fault find digital circuits. Of course a real logic analyser is going to set you back a lot more, but it's worth it if you intend making electronics your hobby.

Criticising this seems very harsh when we are comparing to a logic probe. It's easy when you have access to very expensive equipment, for example I know the electronics engineer at my previous job had a >£4000 scope, while at home I work with a USB one that cost ~£90, but for doing the odd job I can get away with it. The logic analyser above is better than many that cost more than ten times its price, and for a hobbyist is likely to (a) just about fit your needs and (b) be all you can afford. Electronics has been a hobby of mine for a long time, and I wanted a good scope for almost as long, but eventually I realised that I can get a £90 now or a £400 one "one day". I can't really justify the greater expense when the things I play with generally cost less than ten or fifteen pounds, and I have much greater other expenses to worry about.


More to the point, for this problem the above logic analyser is the debugging option with the minimum possible outlay, and is still almost infinitely better than a logic probe. I don't think it would be useful to suggest that someone goes and spend hundreds on a logic analyser or scope to fix a CPC, but £10 might be ok. The other option of course (which I'd probably have gone for by now) is to send it to you

[RobertM]

A digital oscilloscope would be a marvelous tool indeed, but at the moment i can't afford for something like that. The logic tester on the other side, seems more interesting, and if i could make all those checks you mention, then i might get one!

It's not so easy to fix something like a CPC on budget. Mine is working now. I have all the original disks working with a 3.5" drive installed into my CPC6128.

This was achieved with a multi meter, logic tester and some greatly appreciated advice from Bryce.

For me seeing the very low frequency of the /INT pin of the CPU gave an indication that I had a RAM issue.

I am sorry to be such a 'stick in the mud' about test equipment. I am simply trying to assist ikonsgr in a manner that is suitable to the budget.

[Bryce]

@Munchausen: I wasn't really criticising the analyser, rather just pointing out that it is only suitable for people who don't intend doing more than one repair or very occassional testing. I have seen many people "climbing the testing equipment ladder" and being frustrated that they didn't buy proper equipment in the first place. They spend €20 on a cheap analyser, €100 on a cheap scope and the same on soldering iron, DMM etc. then quickly realise that all that money was more or less wasted because they quickly "outgrow" those devices. Electronics definitely isn't a cheap hobby, but if you think you are going to be doing more of it, then it's worth dimensioning your equipment (and budget) for a higher level. A scope that you can use for years is well worth a few hundred, but spending that for one repair would be ridiculous, so a €10 analyser will hopefully be just enough to get it fixed.

Bryce.

P.s. I've no problem fixing stuff for you (or anyone else here), so send it over if you've given up. It's always good to see a CPC come back to life

[pelrun]

Off course, as I said earlier, it all boils down to how much you intend to use it and how complicated your target device is. If you intend doing regular fixing / developing then I'd still recommend spending a little more and getting a semi-decent MSO. That way you have an oscilloscope and analyser that will serve you well. Something like this will get you a lot further than those low-cost USB solutions:

I'm 100% behind paying for a decent scope or not getting one at all; those dinky $100 handheld lcd ones are pretty much worthless. Either spend $100 on a good second hand analogue scope, or $400-$500+ on one of the low-end Rigols at a bare minimum.


An analyser is a bit different; it's possible to get a very capable fpga based usb logic analyser for under $100 these days - much of the power comes from the software, where it's possible to roll your own protocol decoder in code. I definitely preferred it over the 16 channel analyser built into the expensive tektronix MSO I used at work - it was clunky, and I had no licenses to unlock the protocols it could decode. But the scope functions were top-notch.

[Bryce]

The main difference I find is when you are fault-finding. If you are developing something you usually set up the equipment to measure a specific situation and that's easily done on both a real analyser and a USB software controlled analyser, but when you are probing around a faulty board, testing many things in different ways, then a real analyser is much faster and more comfortable to use.

Bryce.

Edit: Just looking at the offer for  the USB analyser on ebay again I noticed a "deal killer" in my opinion: The input impedance is 100K!! That means that connecting this to a circuit has quite a large influence on how the circuit reacts. This would probably be ok for the CPC, but definitely not for anything else. An analyser should have at least a 1Mohm impendance. They also don't mention the voltage range it works in, which is a bit worrying. Is there a page where all the specs are listed?

[gerald]

Edit: Just looking at the offer for  the USB analyser on ebay again I noticed a "deal killer" in my opinion: The input impedance is 100K!! That means that connecting this to a circuit has quite a large influence on how the circuit reacts. This would probably be ok for the CPC, but definitely not for anything else. An analyser should have at least a 1Mohm impendance. They also don't mention the voltage range it works in, which is a bit worrying. Is there a page where all the specs are listed?

100k input inpedance is rather standard for a logic analyser, and you will find this on Agilent logic analyser, which are not known for being cheap. It can go down to 50k/20k on some probes, an this is just from DC point of view.
1M or even 10M is more common on scope, where you may significantly disturb the signal your looking at because the source impedance is high. This is usually not the case for digital signals, and in there you use active probes.