• Please review our updated Terms and Rules here

WD1004-27X MFM or RLL controller

R

RetroSpector78

Experienced Member
Joined
Dec 30, 2018
Messages
153
Location
Belgum
Hi, I've got an 8 bit WD1004-27X hard disk controller.

According to the documentation :

The WD1004-27X and WD1004A-27X boards work with any hard disk drive that has the ST506/412 drive interface and no more than 1024 cylinders and 16 heads.
The Seagate 238 hard disk drive unit is such a drive.
Look in your drive manual to verify that you have an ST506/412 RLL drive or check with your dealer.
This board does not work with RLL drives. If you have an MFM drive you need either the WDXT–GEN2 or WD1004A–WX1 board.
Click to expand...

So I am a bit confused if this controller is in fact MFM or RLL :)

  1. ST506/412 : could mean both MFM and RLL
  2. Seagate 238 : Seems to indicate RLL
  3. verify that you have an ST506/412 RLL drive : Seems to indicate RLL
  4. This board does not work with RLL drives : typo ?
  5. If you have an MFM drive you need either the WDXT–GEN2 or WD1004A–WX1 board : Seems to indicate RLL

Now, I tried formatting my 21MB MFM drive with it (and with some other cards), and here are the results

  • WD1004-27X (RLL?) : 21.344.256 bytes of total disk space and 28.672 bytes of bad sectors
  • WD1002 MFM: 21.204.992 bytes of total disk space and 0 bytes bad sectors
  • ST11 MFM : 21.274.624 bytes of total disk space and 0 bytes bad sectors
  • Adaptec ACB-2072 RLL : 31.203.328 bytes of total disk space and 28.672 bytes of bad sectors

Now if the WD1004-27X is indeed RLL, I would have expected the total disk space to be around 31MB, as during the LLF, I specified 616 cylinders and 4 heads.
On the other hand, I also notice the WD1004-27X reporting the same amount of bad sectors as the ACB-2072.

Leading me to my questions :)

  1. Is the WD1004-27X MFM or RLL ?
  2. If the WD1004-27X it is RLL, did it format it using 17 sectors / track, leaving the "other" 10MB unformatted ? And if so, what is the point of encoding with RLL and only using 17 sectors / track ?
  3. Are the bad sectors using the WD1004-27X and the Adaptec ACB-2072 the result of using RLL encoding on an MFM drive ?
 
g4ugm said:
Is it this board?

https://stason.org/TULARC/pc/hard-d...GITAL-CORPORATION-2-RLL-ST506-412-drives.html

how are the links set?

Also if the original drive is MFM I would expect errors when formatting RLL
Click to expand...

It does seems to be that one...
So I assume the documentation has a typo when it says "This board does not work with RLL drives".

W17/18/19/20 are all open, so it is configured for : 32MB / 4 / 615 (but I configured the drive manually in the LLF debug routing by specifying the heads / cylinders.
W25 and W26 are also open, so I guess it is using 17 sectors per track.

I didn't give any errors during the LLF and the ms-dos format (except for the bad sectors).

But given this seems to be an RLL controller I'm not going to be using it with my MFM hard drive anymore.

But I am curious what happens when you opt for 17 sectors / track in RLL encoding. I read that RLL drives are more demanding and need a higher quality surface to accommodate for the RLL encoding. Does this solely relate to the fact that you typically need to squeeze 26 sectors / track using RLL ?
And does that requirement go away when you opt for 17 sectors / track in RLL encoding ? (making it possible to use an MFM drive on an RLL controller)
 
RetroSpector78 said:
It does seems to be that one...
So I assume the documentation has a typo when it says "This board does not work with RLL drives".

W17/18/19/20 are all open, so it is configured for : 32MB / 4 / 615 (but I configured the drive manually in the LLF debug routing by specifying the heads / cylinders.
W25 and W26 are also open, so I guess it is using 17 sectors per track.

I didn't give any errors during the LLF and the ms-dos format (except for the bad sectors).

But given this seems to be an RLL controller I'm not going to be using it with my MFM hard drive anymore.

But I am curious what happens when you opt for 17 sectors / track in RLL encoding. I read that RLL drives are more demanding and need a higher quality surface to accommodate for the RLL encoding. Does this solely relate to the fact that you typically need to squeeze 26 sectors / track using RLL ?
And does that requirement go away when you opt for 17 sectors / track in RLL encoding ? (making it possible to use an MFM drive on an RLL controller)
Click to expand...

Yes, RLL encoding requires a higher bit-rate so that is why you can get 26 sectors/track and why its more demanding. When you go for 17 sectors the bit rate is the same, you can just spread the sectors out a bit more. I guess this means you could shuffle the spacings to avoid bad spots on the drive, but I don't know if controllers actually do this....

.. and yes I would stick with MFM on an MFM drive....
... a bit more info here

https://redhill.net.au/d/10.php
 
On WD controllers, when the model number ends with an “X”, I’ve usually found that means it’s an RLL controller.
 
It's an RLL controller.

You probably have it configured to do 17 sectors, which it is also capable of. Check jumpers W25 and W26.
 
A common misunderstanding is the since RLL involves a higher bit-rate for the decoded data, that it requires more high-frequency headroom on the drive. That's not so--and if you have access to one, a spectrum analyzer on the raw data lines will verify this. However, it does require more exact timing from the drive, so that most plated-media drives are capable of (2,7) RLL and some better oxide-coated ones are also. But media with large particle sizer or thicker oxide coatings generally don't have the ability to reproduce the smaller timing margins required.

That being said, my advice to those with RLL controllers and some non-RLL-certified drive, is "try it--you might be surprised". I've still got a Quantum Q540 30MB drive that formats to 45MB without issues, for example--as well as a couple of large Atasi/Priam 100MB ones that behave similarly.
 
Stone said:
It's an RLL controller.

You probably have it configured to do 17 sectors, which it is also capable of. Check jumpers W25 and W26.
Click to expand...

That was indeed the case. I assumed that RLL always used 26 sectors / track.

If RLL drives are optimised to facilitate 26 sectors / track (and thus having an increased capacity of 50% over their MFM counterparts), why would one opt to configure an RRL setup with 17 sectors / track ?

The typo in the documentation + the fact that the size wasn't increased by 50% led me to believe this was an MFM controller. Should always RTFM twice :)
 
Chuck(G) said:
That being said, my advice to those with RLL controllers and some non-RLL-certified drive, is "try it--you might be surprised". I've still got a Quantum Q540 30MB drive that formats to 45MB without issues, for example--as well as a couple of large Atasi/Priam 100MB ones that behave similarly.
Click to expand...

Would running a Sprinrite "Level 4 - Locate Surface Defects" be a good way to validate that initially after a LLF ?
 
Stone said:
Even Norton Utilities v4.5 DT (Disk Test) should be sufficient as well as being much simpler to run.
Click to expand...

I did a LLF with the drive using the RLL controller, installed NU 4.5 and did a DT (Disk Test) and NDD (Norton Disk Doctor) and no issues were reported.

I don't know if it's important to make sure there is a good amount of data on the drive because I only installed ms-dos and norton utilities.
I do want to note that before I did the LLF there was lots of data on the drive and I did ran NDD and it did see some bad tracks.

The drive itself also has a defect list with 4 entries, I somehow would have expected the disk test to detect those as well. (or is a more sophisticated tool like Spinrite required for that ?)
 
It's entirely possible for no tool to detect the defects shown on the attached list because they're given as bit offsets on an unformatted track. If the defect falls between sectors (i.e. in the gap area), it won't be detected by the usual suspects.
 
You really do want to test the drive with spinrite. I've seen too many times where sectors with nothing but zeros test perfectly with tools like Norton Utils, but totally crap out when actual data is put in them. Spinrite's pattern testing catches that kind of problem.
 
RetroSpector78 said:
I did a LLF with the drive using the RLL controller, installed NU 4.5 and did a DT (Disk Test) and NDD (Norton Disk Doctor) and no issues were reported.
Click to expand...
For future reference NDD itself would be sufficient as it is actually DT with additional features.

RetroSpector78 said:
I don't know if it's important to make sure there is a good amount of data on the drive because I only installed ms-dos and norton utilities. I do want to note that before I did the LLF there was lots of data on the drive and I did ran NDD and it did see some bad tracks.
Click to expand...
Actual written data is not necessary to successfully weed out the bad sectors which are a format issue.

But SpinRite is always a good idea.
 
SomeGuy said:
You really do want to test the drive with spinrite. I've seen too many times where sectors with nothing but zeros test perfectly with tools like Norton Utils, but totally crap out when actual data is put in them. Spinrite's pattern testing catches that kind of problem.
Click to expand...

I'm doing the level 4 test now in Spinrite 5.0 ... 2h30min in and 2h30min more to go :) Just passed the 50% mark so now the interesting stuff might happen. I'll update here when its done ...

Just looking for the fastest and/or best way to "stress' test the thing and and get some kind of feedback.
Was also thinking it might not be good to test an empty disk so I started duplicating folders like crazy to fill up the disk but figured there should be a better way :)

But it did make me wonder why it would not be possible to just generate some big files using some kind of repeating pattern and fill up the drive and then trying to read (stream) these files back in and verifying the pattern. Are there tools that can do that and wouldn't that also go a long way in terms of verifying the disk ?
 
I believe that I answered that one in #13. Further, depending on the drive electronics, there are certain pathological patterns that you may miss.

Hint: Neither 111111.... nor 000000.... are particularly pathological patterns.
 
So a little under 8 hours of testing and the results are in .... didn't find any defects. Was a bit surprised to see that.
Will try another MFM hard drive today as I hope to see one fail. (as I assume errors related to RLL encoding a regular MFM drive should be easier to spot than the defects found on a defect list. Post #13).

2019-07-29 06.59.09.jpg
 
You must log in or register to reply here.
Back
Top