macdonald@mtwain.dec.com (Paul MacDonald - CUP/ML - 223-3439) (04/30/88)
Whenever (like everytime) I do a warm reboot, my hard disk makes a really harsh buzz-like noise for about two seconds. This only occurs when I use the CTRL-Left Amiga-Right Amiga key sequence. Powering off and then on again to boot the machine does not result in this buzzing. Any thoughts? Paul
erd@tut.cis.ohio-state.edu (Ethan R. Dicks) (05/01/88)
In article <8804300422.AA18746@decwrl.dec.com> macdonald@mtwain.dec.com (Paul MacDonald - CUP/ML - 223-3439) writes: >Whenever (like everytime) I do a warm reboot, my hard disk makes a really >harsh buzz-like noise for about two seconds. This only occurs when I use >the CTRL-Left Amiga-Right Amiga key sequence. Powering off and then on again >to boot the machine does not result in this buzzing. Any thoughts? > >Paul Well, I have seen plenty of these messages, so I thought I would put in my two bits worth (inflation, ya know ;-) I do not have a standard hard disk configuration, as many of you know. I have a WEDGE, which must be Mounted, and not installed via Binddrivers. My drive is a real Seagate ST-225, and it makes noise too. Most of these messages concerning noise say "my disk drive sounds like a buzzsaw when I execute bindrivers... is it broken?" Paul, you did not say what controller, what hard drive, or how you configure the drive in your startup-sequence, but I will take a stab at guessing your configuration. You probably have a hard disk which automatically _PARKS_ its own heads on power off, but did not answer YES to the question on parking your heads after 3 seconds of inactivity, while running the prep program (if you have an A2090). Your drive buzzes on warm start, (like mine too), because the controller forgets where the heads are when reset, and must get the heads back to a known place. Since on the Amiga, the most likely place for the heads to be is over the root track, and most Amiga owners only have one BIG partition, that's a lotta tracks to cover when pulling the heads to the edge of the disk. The noise you hear is as a result of either the hddisk.device stepping the tracks very fast to 0, or a command sent to the drive itself, telling it to reset the heads (I don't know which, because I don't know much about the hddisk.device) An interesting difference in systems, my drive does not 'buzz' harshly, but quietly (probably due to driver differences), and it does not buzz when the drive is Nounted. My drive resets to do its thing when I try to access it for the first time, when (I think) AmigaDOS puts a lock on the root track. If I Mount the other partitions, they do not show up on the Info device table until I at least cd to the partition. I assume this is also standard behavior for A2090 multiple partitions, but I don't know, since I have not yet met anyone who partitions their A2090 configuration. Any comments? Sorry to be long winded, but I hope this clears up any more questions about why hard drives make obnoxious noises on startup. If I am wrong on any point, I'm sure _some_ kind person will be glad to correct me ;-) -ethan -- Ethan R. Dicks | ###### This signifies that the poster is a member in Specialized Software| ## good sitting of Inertia House: Bodies at rest. 2101 Iuka Ave. | ## Columbus OH 43201 | ###### "You get it, you're closer."
raymund@sci.UUCP (Raymund Galvin) (05/02/88)
In article <12083@tut.cis.ohio-state.edu>, erd@tut.cis.ohio-state.edu (Ethan R. Dicks) writes:
<
< Your drive buzzes on warm start, (like mine too), because the controller
< forgets where the heads are when reset, and must get the heads back to a
< known place. Since on the Amiga, the most likely place for the heads to be is
< over the root track,and most Amiga owners only have one BIG partition, that's
< a lotta tracks to cover when pulling the heads to the edge of the disk. The
< noise you hear is as a result of either the hddisk.device stepping the tracks
< very fast to 0, or a command sent to the drive itself, telling it to reset the
< heads (I don't know which, because I don't know much about the hddisk.device)
<
I doubt it. Two seconds of noise for a simple seek. Give me a break.
These days many drives are rated as having average seek times in the
20-60 millisecond range. An average seek for a drive translates to one
third of a full stroke seek. The absolute worst case seek shouldn't be more
than 3 times the average seek time (usually it will be less). I would
bet that whoever wrote the code that is messing around with the hard
disk (during reboots) is doing something silly. I wish someone
from Commodore would explain what is happening.
Ray Galvinnic@dworld.UUCP (Nic Bernstein) (05/03/88)
In article <12083@tut.cis.ohio-state.edu> erd@tut.cis.ohio-state.edu (Ethan R. Dicks) writes: >for the first time, when (I think) AmigaDOS puts a lock on the root track. If >I Mount the other partitions, they do not show up on the Info device table >until I at least cd to the partition. I assume this is also standard behavior >for A2090 multiple partitions, but I don't know, since I have not yet met >anyone who partitions their A2090 configuration. Any comments? This is standard AmigaDos behaviour, but only vague reference is made to it in the manuals. This is very similar to the ram: initialization of putting the line "DIR RAM: >NIL:" in your startup-sequence so that an icon appears for it. The system does not seem to really mount the device until some command is run which requires a lock on that device. For a hard disk CD is probably the best command to use since it has no output, and generates the least amount of disk activity. >Ethan R. Dicks -- "You can't spend your history!" Nic Bernstein Melinda Briggerty Discovery World Museum "... but you can sell it!" 818 W. Wisconsin av. Me Milwaukee, WI 53233 ____________________________________________________________________________ {uunet|uwmcsd1|gryphon}!marque{!introl}!dworld!nic ____________________________________________________________________________
atheybey@PTT.LCS.MIT.EDU (05/03/88)
Repository: PTT Originating-Client: flower Raymund Galvin writes: :In article <12083@tut.cis.ohio-state.edu>, erd@tut.cis.ohio-state.edu (Ethan R. Dicks) writes: :< :< Your drive buzzes on warm start, (like mine too), because the controller :< forgets where the heads are when reset, and must get the heads back to a :< known place. Since on the Amiga, the most likely place for the heads to be is :< over the root track,and most Amiga owners only have one BIG partition, that's :< a lotta tracks to cover when pulling the heads to the edge of the disk. The :< noise you hear is as a result of either the hddisk.device stepping the tracks :< very fast to 0, or a command sent to the drive itself, telling it to reset the :< heads (I don't know which, because I don't know much about the hddisk.device) :< : :I doubt it. Two seconds of noise for a simple seek. Give me a break. :These days many drives are rated as having average seek times in the :20-60 millisecond range. An average seek for a drive translates to one :third of a full stroke seek. The absolute worst case seek shouldn't be more :than 3 times the average seek time (usually it will be less). I would :bet that whoever wrote the code that is messing around with the hard :disk (during reboots) is doing something silly. I wish someone :from Commodore would explain what is happening. : I believe that Ethan is correct except for the word "fast". When the hard drive is reset (or powered up), it does not know the track that the heads are on. It therefore steps the heads toward track 0 *slowly*--10s of milliseconds *per track*--until it notices that the heads have hit the physical stop at track zero. If it did this at full speed, it would quickly destroy the heads by bashing them against the stop. I think that this is a function of the drive, not the controller. I have heard the same noise on hard drives on IBM PCs as well. It is entirely normal, and is certainly not the result of any particular hard disk code on the Amiga. Andrew Heybey atheybey@ptt.lcs.mit.edu
hedley@cbmvax.UUCP (Hedley Davis) (05/03/88)
In article <20570@sci.UUCP> raymund@sci.UUCP (Raymund Galvin) writes: >In article <12083@tut.cis.ohio-state.edu>, erd@tut.cis.ohio-state.edu (Ethan R. Dicks) writes: >< >< Your drive buzzes on warm start, (like mine too), because the controller >< forgets where the heads are when reset, and must get the heads back to a >< known place. Since on the Amiga, the most likely place for the heads to be is >< over the root track,and most Amiga owners only have one BIG partition, that's >< a lotta tracks to cover when pulling the heads to the edge of the disk. The >< noise you hear is as a result of either the hddisk.device stepping the tracks >< very fast to 0, or a command sent to the drive itself, telling it to reset the >< heads (I don't know which, because I don't know much about the hddisk.device) >< > >I doubt it. Two seconds of noise for a simple seek. Give me a break. >These days many drives are rated as having average seek times in the >20-60 millisecond range. An average seek for a drive translates to one >third of a full stroke seek. The absolute worst case seek shouldn't be more >than 3 times the average seek time (usually it will be less). I would >bet that whoever wrote the code that is messing around with the hard >disk (during reboots) is doing something silly. I wish someone >from Commodore would explain what is happening. > >Ray Galvin How about this: The system boots, and the driver comes up and says 'Is the disk at track zero ?'. The disk isn't, nor does the disk even know where it s own head is. The veiw of the world here is a single bit which says 'Is the head over track zero ?'. Now, we could tell the disk to seek a few hundred tracks over, but then if the disk was on track 5 , it would come rushing back towards track zero, and WHAM, hit the head stop with incredible velocity. This would not be a good thing. So, the when the system boots, and the head is not over track zero, it backs the head up ONE track and checks again. This takes quite a while ( like a a couple of seconds ). Figure 20 ms seek time, and 100 tracks would yeild 2 seconds of noise. Real numbers may vary, but the general result is the same. Hedley
phils@tekigm2.TEK.COM (Philip E Staub) (05/04/88)
In article <20570@sci.UUCP> raymund@sci.UUCP (Raymund Galvin) writes: >In article <12083@tut.cis.ohio-state.edu>, erd@tut.cis.ohio-state.edu (Ethan R. Dicks) writes: >< >< Your drive buzzes on warm start, (like mine too), because the controller >< forgets where the heads are when reset, and must get the heads back to a >< known place. Since on the Amiga, the most likely place for the heads to be is >< > >I doubt it. Two seconds of noise for a simple seek. Give me a break. >These days many drives are rated as having average seek times in the >20-60 millisecond range. An average seek for a drive translates to one Don't be so hasty to reject the explanation. What you are referring to by the 20-60 millisecond range is the *seek* time. What Ethan was referring to is a completely separate function known as "recalibration", which does not use a standard seek. This function literally starts and stops the motion of the head for each track it crosses, resulting in *much* longer times to get across the multiple hundreds of tracks on a hard disk. Imagine the time it would take to do a seek from track 0 to track 599 if you had a 600 track floppy, and you're more in the range of the time it takes to do a recalibrate. >Ray Galvin Phil -- ------------------------------------------------------------------------------ Phil Staub "I do NOT approve. I merely said I UNDERSTAND." tektronix!tekigm2!phils - Spock phils@tekigm2.TEK.COM
cmcmanis%pepper@Sun.COM (Chuck McManis) (05/05/88)
In article <3718@cbmvax.UUCP> hedley@cbmvax.UUCP (Hedley Davis) writes:
->How about this:
->
-> The system boots, and the driver comes up and says 'Is the disk
->at track zero ?'. The disk isn't, nor does the disk even know where it
->s own head is. The veiw of the world here is a single bit which says
->'Is the head over track zero ?'. Now, we could tell the disk to seek
->a few hundred tracks over, but then if the disk was on track 5 , it
->would come rushing back towards track zero, and WHAM, hit the head stop
->with incredible velocity. This would not be a good thing.
No offense Hedley but that is what you get for programming a hard disk
like a floppy. How about this :
The system boots and the driver comes up and says "I wonder what cylinder
the disk is on?" So it issues a READ-ID command which causes the heads to
read the first sector header they happen to come upon and return the cylinder
number. Remembering this, it tells the drive to seek to the position of the
root block. No noise, no fuss. SCSI drives are even easier because they
have no idea what the heck a cylinder is. You just say "Go to logical block
207" or whatever and they do. Anyway, it could be made to work better.
Your explanation does explain the noise though...
--Chuck McManis
uucp: {anywhere}!sun!cmcmanis BIX: cmcmanis ARPAnet: cmcmanis@sun.com
These opinions are my own and no one elses, but you knew that didn't you.pds@quintus.UUCP (Peter Schachte) (05/05/88)
In article <3718@cbmvax.UUCP>, hedley@cbmvax.UUCP (Hedley Davis) writes: > So, the when the system boots, and the head is not over track > zero, it backs the head up ONE track and checks again. Thanks for this explanation. It is all clear now. But one question: wouldn't sending the head to track 0 BEFORE booting or shutting down cure this problem? So why not fix the old three-finger-nerve-pinch reboot code to seek to track 0 before actually rebooting? Surely a short program could be run (manually) before powering down? -- -Peter Schachte pds@quintus.uucp ...!sun!quintus!pds
hedley@cbmvax.UUCP (Hedley Davis) (05/05/88)
In article <51959@sun.uucp> cmcmanis@sun.UUCP (Chuck McManis) writes: >In article <3718@cbmvax.UUCP> hedley@cbmvax.UUCP (Hedley Davis) writes: >->How about this: >-> (deleted) > >No offense Hedley but that is what you get for programming a hard disk >like a floppy. How about this : > (deleted) > >--Chuck McManis >uucp: {anywhere}!sun!cmcmanis BIX: cmcmanis ARPAnet: cmcmanis@sun.com >These opinions are my own and no one elses, but you knew that didn't you. > Hey, I didn't do it ! I'm just passing on the explanation for the noise as it was explained to me by Jeff Boyer who did the hardware design. The person who could really provide the authoritative software responce would be Bart. Hedley
andy@cbmvax.UUCP (Andy Finkel) (05/05/88)
In article <927@sandino.quintus.UUCP> pds@quintus.UUCP (Peter Schachte) writes: >Thanks for this explanation. It is all clear now. But one question: >wouldn't sending the head to track 0 BEFORE booting or shutting down >cure this problem? So why not fix the old three-finger-nerve-pinch Except that track 0 is not the usual place to park heads. Usually, when you park them, you try to avoid having them right over critical data. A park program is pretty easy...but I wouldn't advise parking at track 0. -- andy finkel {ihnp4|seismo|allegra}!cbmvax!andy Commodore-Amiga, Inc. "C combines the power of assembly language with the flexibility of assembly language." Any expressed opinions are mine; but feel free to share. I disclaim all responsibilities, all shapes, all sizes, all colors.
raymund@sci.UUCP (Raymund Galvin) (05/06/88)
In article <2897@tekigm2.TEK.COM>, phils@tekigm2.TEK.COM (Philip E Staub) writes: > Don't be so hasty to reject the explanation. What you are referring to by > the 20-60 millisecond range is the *seek* time. What Ethan was referring to > is a completely separate function known as "recalibration", which does not > use a standard seek. This function literally starts and stops the motion of > the head for each track it crosses, resulting in *much* longer times to > get across the multiple hundreds of tracks on a hard disk. Imagine the time > it would take to do a seek from track 0 to track 599 if you had a 600 > track floppy, and you're more in the range of the time it takes to do a > recalibrate. > > > Phil > I didn't mean to reject the explanation. It was, as many people pointed out, probably correct. My real intention was to criticize the behavior of the disk controller/software during reboots. Ok, maybe I was a bit hasty. Let me put it another way: I have a miniscribe hard disk with an ST506 interface that usually makes 2 to 3 seconds of noise during the recalibration process when used with my amiga. The same drive never made any objectionable noise during the recalibration process when used in my PC clone. According to the my drive's manual, the drive will ignore requests to step the heads out past track 0 or in past the head parking cylinder. This means it is not necessary to check for track 0 after every step during a reclibration process. Just do one huge seek and verify that you're at track zero when the seek completes. If some disk drives aren't intelligent enough to handle this, then the amiga's recalibration routine seems to be a reasonable (but noisy) compromise; otherwise, the amiga's drive recalibration routine could be made quieter. Ray Galvin
wayne@utflis.UUCP (Wayne Young) (05/06/88)
In article <3718@cbmvax.UUCP>, hedley@cbmvax.UUCP (Hedley Davis) writes: > So, the when the system boots, and the head is not over track > zero, it backs the head up ONE track and checks again. If the drive forgets which track it's over on a warm boot and consequently has to back up a track at a time to find track 0, then why doesn't it do the same thing on power up? I own a MiniScribe too, and I notice that when I power the sucker on, it makes *one* seek, and there is no buzzsaw sound. Only on a warm boot does it make the buzzsaw sound. What now? wayne -- ----- {uunet,utzoo,decvax,ihnp4,allegra}!utcsri!utflis!wayne
erd@tut.cis.ohio-state.edu (Ethan R. Dicks) (05/08/88)
In article <1080@utflis.UUCP> wayne@flis.toronto.edu.UUCP (Wayne Young) writes: >In article <3718@cbmvax.UUCP>, hedley@cbmvax.UUCP (Hedley Davis) writes: >> So, the when the system boots, and the head is not over track >> zero, it backs the head up ONE track and checks again. >If the drive forgets which track it's over on a warm boot and consequently >has to back up a track at a time to find track 0, then why doesn't it do >the same thing on power up? I own a MiniScribe too, and I notice that when >I power the sucker on, it makes *one* seek, and there is no buzzsaw sound. >Only on a warm boot does it make the buzzsaw sound. >What now? Well, _my_ drive claims in the documentation to PARK on power off (Seagate ST-225) It makes sense that if the drive parks on power off, that the drive would know where it put the heads. On a warm start, the heads could be anywhere. >wayne > {uunet,utzoo,decvax,ihnp4,allegra}!utcsri!utflis!wayne -- Ethan R. Dicks | ###### This signifies that the poster is a member in | ## good sitting of Inertia House: Bodies at rest. This space for rent | ## | ###### "You get it, you're closer."