Interpreting uPD765 FDC result bytes

[JonB]

Thanks kevin.c


What or wher are the other traces do you think? I have a 4 channel DCO on its way so I should be able to make these measurements, but only if I know where to put the probes..

[kevin.c]

JonB, for the final test of this FDC PLL I would treat it as black box and just check  if it works correctly or not.
Connect the first probe to  output clock TP2 500Khz or 250kHz. Set o-scope synch to first CH ONLY!
Connect second probe to Read Data (30) or Output Read Data (pin 23).
Set floppy to read still one track to get valid and stable Read Data stream. The PLL clock should lock.
Observe on the screen if CH2 Read Data bits are stable in frequency and phase against  CH1 clock pulse.
Of course bits must change, that's the data stream, but it should be clearly visible if they are synchronized or not.

[JonB]

Right. Should get the scope in a couple of weeks, then I can "scope it out,dude"!


Looking forward to it (scope is right complicated though)!

[Bryce]

Is it your first scope? Do you know the basics? If not, there are some really good documents and videos on the net that I'd recommend you read/watch. Scopes can be very deceiving if you don't know how to set it up properly.

Bryce.

[JonB]

Yes, first scope. Cannot wait. Fortunately it has an Auto feature so I should at least be able to see the waveforms, but feel free to recommend online resources. There does not appear to be a "Scopes for Dummies" book (I prefer books to online resources for reference materials).


Ah yes, also I am shopping around for a solder station - any recommendations?

[kevin.c]

.........Scopes can be very deceiving if you don't know how to set it up properly.
Bryce.

It is quite right! When you all set to auto - the scope, specially digital one - will try to show waveforms as clean and stable as possible. In our FDD PLL case, you must recognize if it is stable because of good scope digitalisation or PLL properties. It is not so obvious and needs some experience.

[JonB]

One has to start somewhere...

[Bryce]

The Auto button is useful for your first experiments, but you need to learn what's really happening, so try to move away from using Auto as soon as possible. My recommendataion is to experiment with known signals and learn how the different adjustments affect what is displayed.
A good example of why Auto is not good: If I measure a clock signal of 4Mhz which is full of noise at 60Mhz (which is the problem and what you want to find in this example). Hitting the Auto button will attempt to show a clean 4Mhz signal. The horizontal and vertical will be set correctly, but it will also turn on the 20Mhz bandwidth limit which hides the 60Mhz noise and hence you don't see the problem. If you were manually looking for this problem, you'd have left the bandwidth limit high and probably turned up the persistance value too.
I don't know of any "physical" books about scopes, but I can put together a list of links to pdf files which you could read (or print if you really want to).
As far as irons are concerned:
- Buy one that allows you to set the temperature and displays it too.
- Don't buy one with a handle / tip big enough to weld a ship together. The tip should be a chisel tip and about 2mm wide at the end.
- Don't buy a combi version that includes a rework station - If you need both, then buy two separate units.
- Buy a known brand so that you can be sure you can still buy new tips in the future.

Personally I prefer Ersa, but Weller, Hakko and many other companies do good irons too.

Bryce.

Edit: For anyone just starting out with there first scope, here's a whole collection of documents worth reading. Although they are from Tektronix, the information is valid for any scope: Oscilloscope Tutorial | Tektronix

[JonB]

Hi Bryce


Sage advice as always, thanks!


On the Rigol I read you use auto to get to a known state. The display is all over the place otherwise. But I will find out once I have it.


I ordered a Hakko FX-888D in "CBeebies" livery (blue & yellow).

[Bryce]

I've never used Hakko personally, but I'm told they are good irons.

Auto button doesn't really bring you to a "Known state" unless you know what state the scope got to. When you get used to the scope you'll find that it's better to know exactly what you are watching rather than what the scope thinks you want to watch.

Bryce.

[JonB]

OK, I have it at last. Took six weeks, because they are so popular. Damned fine 'scope it is too!


I have measured the monoflop at TP1 as shown in the manual and it is showing a pulse width of 3uS. It is supposed to 1uS or as near as possible. Adjusting the trimmer 4713 brings it down, but only to 2.77, however it is an uneven waveform with the gap between falling and rising roughly 33% the gap between rising and falling. So I am thinking that the smaller gap is the one I need to calibrate, and sure enough I can get it to 1uS. Hurrah.

[JonB]

But... It is still not working, unable to read disks. So I'm wondering, what should I check next? I do have the timing diagrams for the FDC and I will take a look.

[Bryce]

Yes, it's a negative pulse, not a positive. The "gap" you're talking about IS the pulse and it looks fine.

Bryce.

[JonB]

JonB, for the final test of this FDC PLL I would treat it as black box and just check  if it works correctly or not.
Connect the first probe to  output clock TP2 500Khz or 250kHz. Set o-scope synch to first CH ONLY!
Connect second probe to Read Data (30) or Output Read Data (pin 23).
Set floppy to read still one track to get valid and stable Read Data stream. The PLL clock should lock.
Observe on the screen if CH2 Read Data bits are stable in frequency and phase against  CH1 clock pulse.
Of course bits must change, that's the data stream, but it should be clearly visible if they are synchronized or not.

So, if I trigger on the 500khz test point (Ch.1, Yellow) and try to see the read data (pin 23 on the FDC chip; Ch. 2, cyan) all I am getting is a narrow pulse (~120nS) that aligns with the 500Khz wave forms. Can't see any actual data. Aren't these supposed to change with 0 / 1 ? And I see this even when it's not reading (shown below).

[Bryce]

What Data seperator does the controller use? Have you read its input?

Bryce.

[JonB]

Sorry Bryce, I don't understand what you mean by that. It's a uPD765 controller.


In looking at the timing diagram, I am not sure where the measurement points are that correspond to some of the timing signals shown. I assume RDD is pin 23 of the FDC, but if that is the case, all I am getting is a stream of 1s. Could you help me identify the test points for the signals READ DATA(MFM), SHAPED DATA, VFO CLK/2, No.5 U2, No. 9 U2?

[JonB]

I'm looking at !RD on pin 2. According to the timing chart it goes active low during the read operation. All I can see is noise right now, even when reading. Does that tell you anything?


..other than I don't yet know how to use my new 'scope, ha ha!

[JonB]

Ah silly me, !RD is for transferring to the system data bus, not grabbing stuff from the drive read pin. Duhh..

[Bryce]

The signal from the disk drive isn't fed directly to the µPD765, it has to go through a Data seperator circuit/chip first. What chip is used in your case (the CPC for example uses an FDC9216 or later an SED9420 to do this). What IC is used in this case and what signal are you getting at the input to this chip?

Bryce.

[JonB]

..and RDD (pin 30 in the Shugart drive interface) is shown on the PLL schematic. So...


Yes, it's showing data and I can see it getting fed into the PLL circuit bottom left.


The other input is the SYNC pulse which looks like a sawtooth at 25-27Mhz. Disconnecting the drive makes it bigger; I'm thinking it is noise.
SYNC is also shown on the PLL circuit coming in from the right hand side. Without it, nothing of the RD signal makes it past the first gate of 7414.


Sure enough, there is nothing coming out but a sawtooth. So should I conclude that the issue is with the SYNC line?

[JonB]

The signal from the disk drive isn't fed directly to the µPD765, it has to go through a Data seperator circuit/chip first. What chip is used in your case (the CPC for example uses an FDC9216 or later an SED9420 to do this). What IC is used in this case and what signal are you getting at the input to this chip?

Bryce.

Not sure - it should be on the FDC schematic. I think it is that discrete logic bottom left which is marrying up SYNC with RD pulses. As you can see I can't detect anything on the SYNC line but some 27Mhz sawtooth, so I think it may be not connected properly. Going to check it for continuity.

[JonB]

More and more confusing. FDC Pin 24 isn't reading anything and is showing a logic low with a probe, even when attempting to read. I think this indicates an issue with the FDC. Anyone care to comment?

[JonB]

More info from an application note for the FDC that explains how to build an interface:

In the Read mode, the VCO Sync signal goes true after the R/W head has been loaded and the head settling time (head load time) has elapsed. (See bottom of Figure 7.) The VCO Sync goes false between the 10 Field and the Data Field; and after the Data Field. This is done to blank out discontinuities which appear in these gaps when the Write current is turned on and off. When VCO Sync is true, valid data is coming off the diskette and the data recovery circuit does not have to be configured to accommodate any abnormal conditions.

So it won't be high all the time.


Additional info: Head load is showing pulses (very sloooow ones). Since this is related to the VCO Sync I thoiught I'd better check it, but it looks OK. Again, though, there is no activity on the VCO line.

[Bryce]

To avoid all this back and forth, can you patch together a schematic of the complete system, from the drive head up to where it's outputted as 8-bit data. That way I/we can see exactly what's happening and tell you what you should be expacting where.

Bryce.

[JonB]

OK Bryce, it'll take a short while, though. The schematics in the service manual are a little fragmented.


Do you really want it from the drive heads? The drives are Teac F55 and I know they work. I used one to read the P2000C's boot and application disks on my PC with 22disk144.exe with no problems (once I'd found the correct parameters for the P2000C's format of course).


Bearing this in mind I propose to start it from the 34 pin connector (standard Shugart) for now. If we need to go upstream of that I can dig out an FD55 service manual. Sounds good?


Cheers
JonB


Oh, by the way, do you think that the FDC may be suspect given that it is not outputting a VCO signal? (I shall also check for a short to ground somewhere..!)