• Please review our updated Terms and Rules here

Booting Pdp-11 from flash USB

Crawford,

This is good. Unfortunately, I have been working with an 8GB stick so far. I need to find a small one. There is a 512MB around here somewhere.

You have figured out how we can share individual images and reassemble them into a stick. Now I'd like to figure out how to build the individual images from stratch, and so I would like study the format of the disk images. I find all this intersting since I learned a lot making the conversion utilities to go to/from PUTR images to Reinhard's RL02 emulator file format.

Lou
 
Lou,

Yeah, I was able to dig up a 64MB and a 128MB stick. I can send you a single image to examine if you want via personal email. I looked at it briefly in HxD. It looked like some header stuff then a bitstream. One of the blogs mentioned the unused portions keeping previously written junk as if it was just skipped over (buffer? Easier to map?)

? ; - )
 
Crawford,

If you haven't e-mailed the single floppy image, don't bother. Using your techniques I looked at them myself.

My next plan of attack is to study the raw image of an RX33 and the exact duplicate in the emulator. I'll take a bootable RT-11 RX33 and use copy /image to copy it to the emulator. Then, I'll use imagedisk to make a raw image of the physical RX33 floppy. Then I'll study both images side-by-side (I found a utilitiy called vbindiff that is good for that.)

Lou
 
You know, a linux box with dd, od, and diff would handle this very easily...
I will try that once I get myself a USB stick. I have a Linux box
that I used to create my SCSI images for my 11/84.
 
Exactly like you'd want it to be

Exactly like you'd want it to be

Well no linux box here and this nut is cracked.

The image files on the stick are exactly how you'd want them to be. They are sector sequential data, starting from track zero, head zero, sector zero, progressing to the end of the track, padding the end of the track to fill up the left over space (if any) then switching heads, then going to the next cylinder. Repeat to the end of the disk. Start over at the next byte with the next disk.

The RX33 track has 15 sectors of 512 bytes for 7680d bytes (1E00h). Each track in the image contains 9216d bytes (18 sectors x 512 bytes) (2400h). So, the first 7680d bytes have the RX33 track data, and the remaining 1536d (600h) bytes are padding. I saw padding of garbage sometimes, and padding of repeating 00h through FFh. It may not matter.

So, it is simple (even for me) to make a little program to convert from emulator RX33 images to PUTR compatable RX33 images and back.

Next question - Is the disk image "spinning" at 300 RPM or 360 RPM? It must be 300 right? Could I put an RX23 (1.44MB 3-1/2") on RQDX3 and format it with ZRQC?? and work with it as if it was an RX33? We will have to try this tomorrow for the hell of it. Totally useless, but interesting. Actually it may not be useless to the young guy who just got his machine but has no RX33s, but has piles of trashpicked RX23s.

Lou
 
You know, a linux box with dd, od, and diff would handle this very easily...

You don't even need a separate linux box. Use CYGWIN (www.cygwin.com) installed on Windows XP/7/8 etc and you get all this stuff (plus a lot of other linux commands if you so desire). I use this all the time as a linux box 'inside' my XP machine.

Don
 
Lou,

Good analysis. Yes it's 300 RPM, and very spot on. One of the bloggers connected a Gotek to a kyroflux device that apparently gives lots of info - kyrofux is intended to capture disk data at the magnetic flux level for preservation purposes.

The bootloader is locked/protected according to chuck(g) so unless that is cracked, no firmware changes short of chip replacement :(
 
Who would have guessed? RX23 does work on RQDX3 and looks like an RX33!

Who would have guessed? RX23 does work on RQDX3 and looks like an RX33!

And so, since the 300RPM 3.5" floppy emulator works on an RQDX3, so does a real 3.5" 1.44MB floppy drive (RX23)!

I have a dead common jumperless Teac FD-235HF as DU1: (the jumperless ones are all hardwired to drive select 1) and the floppy emulator as DU0: on the RQDX3. I booted XXDP from the emulator, ran ZRQC??, and formatted a 3.5" floppy in in the RX23. It also passes the test with no bad blocks. Then I made the real floppy a bootable DUDP. I just booted it and it works!!

How come I never read anywhere that this works? The RQDX3 can tolerate the 300RPM somehow. The RX50 is 300RPM, but it has a 250kHz data rate, while the RX23 and 33 have 500kHz data rates.

Next, I'll examine this weird 3.5" floppy with imagedisk.

OK, I'm editing this post to say that what I saw in imagedisk is what I expected. 15 sectors per track.

I think all the hardware is there on the RQDX3 to have supported 18 sectors per track. I'll have to look at the datasheet for the SMC HDC9224 IC.

Lou
 
Last edited:
Lou,

Ummm, yeah, I've had a 3.5" floppy connected to my RQDX3 for some time, I guess I didn't think there was anything special about that... Maybe I have later ROMs?
 
Lou,

Good analysis. Yes it's 300 RPM, and very spot on. One of the bloggers connected a Gotek to a kyroflux device that apparently gives lots of info - kyrofux is intended to capture disk data at the magnetic flux level for preservation purposes.

The bootloader is locked/protected according to chuck(g) so unless that is cracked, no firmware changes short of chip replacement :(

This is a bit old but I just found that apparently someone has reverse engineered the Gotek (not reading out the firmware though since it is protected) and created a new firmware to be able to handle Amiga sector format: http://cortexamigafloppydrive.wordpress.com/

Sadly, I was unable to find any source for his new firmware, only the hex file, but there is a good schematic on the unit.
 
Last edited:
Mattis,

Yes, I also asked Herve if he would release source, but he's mad that folks are selling his firmware that he provides for free. One of the nicer things Herve did was to write a menu that boots automatically and allows you to chain-load another disk image.

Crawford
 
I just bought two of these drives to play with. What is the largest USB stick that people have been able to get working? I have a 256 meg that does work, formats to 165 floppies. But when I try an 8 gig it fails. :(
 
8008guy,

What format utility failed? Was it one of the utilities from the mini-cd shipped with the drive? If I recall you can just use the drive to format the usb stick by some button presses.

Realize that a drive bigger than 2gb makes no sense as 1000 images * 1.5MB is still less than 2Gb. Having said that, I think I tried a 4gb or 8gb drive without issue.
 
It was the built in format utility that creates the 1000 disks when you push the two buttons that failed. I ordered some 2gig drives from walmart, for $3 each, to test with, they work fine. I couldn't get anything larger than 2gig to create the 1000 virtual floppies.


I used imagedisk to copy a working RT11 bootdisk to my gotek drive. I swapped out my floppy with the gotek. When I try to boot from the gotek I get this:

Trying DU0

Message 11
Non bootable media in the drive

In the previous messages I've seen that others have used the RDQX3 format utility to initialize the virtual disks. Has anyone successfully used imageDisk to create a bootable image on their gotek drive?

len



8008guy,

What format utility failed? Was it one of the utilities from the mini-cd shipped with the drive? If I recall you can just use the drive to format the usb stick by some button presses.

Realize that a drive bigger than 2gb makes no sense as 1000 images * 1.5MB is still less than 2Gb. Having said that, I think I tried a 4gb or 8gb drive without issue.
 
I got my pdp booted from the gotek drive tonight. I made three floppy images on the gotek, only two booted.

1. Formatted using the rqdx3 formatting program, copied files with putr, did not boot.

2. Formatted using rqdx3 and copied image using image disk, booted fine.

3. Formatted with putr and copied files with putr, booted fine.

I'm not sure why the first floppy image didn't boot... The same position, 0, was the same one that failed originally whe. I imaged it with imagedisk.
 
Back
Top