Need help building a driver for "True-IDE" Microdrives

[Guid]

Sorry no success so far. I did compile atapi.dll with PB 4.2, but the driver still doesn't seem to load for some reason. I compiled it using SYSGEN_ATAPI for armv4 (I even tried armv4i).

First I was trying to make my true IDE locked ST1 to load the usual way (using entry point IDE_Init, with the default settings). Then I realized that the driver is not even getting loaded. So I did the following to test it out.

Since atadisk.dll is a subset of atapi.dll, I should able to replace atadisk with atapi. The only difference is atapi should load PCMCIADetectATA instead of DetectATADisk every time I insert a card. To check whether the driver loads or not, I replaced atadisk.dll with atapi.dll in ...\PCMCIA\ATADisk\Dll and in ...\PCMICA\Detect\50\Dll. I also changed subkey ...\PCMCIA\Detect\50\Entry from DetectATADisk to PCMCIADetect. And every time I insert a regular CF card it gives me the unrecognized card error, because the driver doesn't load.

ATAdisk at least gives me the PnP ID for my IDE locked ST1 (SEAGATE-ST1-D8AA).

I'm still researching now, and I found some interesting stuff on microsoft.public.windowsce.embedded. This one here is seemed particular interesting:
http://groups.google.com/group/micro...e5dd9d2ee8e26a
And if what the Microsoft guy is saying is true, we probably can't use wifi when in true IDE mode, because wifi is connected on the PCMCIA bus. And I would guess that WM 2005 won't have native support for true IDE either (even tough it has an improved atapi driver).

Maybe I'm missing something obvious here, or maybe I didn't compile the driver properly. Anyway I'm attaching it, and if you manage to get it to work with even a regular CF card please let us know. Thanks and good luck.

[ravic]

Thanks a lot for your effort. I am loading PB 4.2 tonight. Will try to see if i can get it to work.

[Big_Dawg]

It seems (after some internet detection work, b/c not disclosed by seller) that the 4GB microdrive I picked up on ebay is also a iPod Hitachi white label. I formatted it, but x50v asks for driver. I am not nearly as advanced as ravic any you guys, but I would love to help - not stupid, just not advanced.

[ravic]

Big_Dawg,

I have some mobile devlopment experience and some driver development experience for PCs.

Never developed device drivers for Embedded platform. So i am also new to this game. Still trying to get working environement setup.

However, i saw there are already several versions of ATA/IDE drivers for the Windows CE platform which is the base OS for Windows Mobile.

I am still waiting for my Microdrive to arrive. So here is what i plan on doing. See if you can get it to work.

Download the atapi.zip that guid uploaded recently. Not the first one but the one he uploaded two days ago. Extract and copy that file to \windows folder on PDA.

Then insert the drive. It will ask for the driver. Point to that dll.

No use PHM Registry editor to go into registry.
Look a the HKEY_LOCAL_MACHINE\Drivers area to if there is hitachi entry there. Post the details listed under that key here

We need to create a device entry which looks like this once we know the details about how it is detected

[HKEY_LOCAL_MACHINE\Drivers\BuiltIn\Atapi00]
"Dll"="ATAPI.DLL"
"Prefix"="XXX"
"Ioctl"=dword:4
"DeviceId"=dword:2
"IOBaseAddress"=dword:1F0
"Interrupt"=dword:0E
"FSD"="fatfs.dll"

Not knowing all the details this is how far i could go. Once i have the drive i should have more details.

[UnLoGiC]

This looks interesting!

[ravic]

I finally got my microdrive. And as expected it is not accepted by the Axim X50v.

So now time to find out why. CF is definitely in True IDE mode. I can read in any CF card reader as expected but not in Axim or a CF to PCMCIA adapter.

It surprised me that i am unable to read it PCMCIA, but later found out that PCMCIA does not support True IDE drives. Unfortunately most PDAs including axim implement the CF slot as a variant of PCMCIA configuration.

From what i have read so far a microdrive/CF in True IDE mode might never work in a Axim, even with a new driver from Windows Mobile 5. This is because the IDE drive is tied to a PCMCIA controller which does not understand anything about IDE drives. What WM5 documentation is saying is that it supports IDE drives when connected to a IDE controller. Which will be the case in future PDAs.

However, i have not given up yet.
I am trying to figure why the drive is going to True IDE mode.
The following two things make me believe it is firmware (not hardware) controlling this feature on the drive.

1. The fact that Hitachi and seagate chose CF+ physical design (even though the Ipod and Muvo has no need for CF+ design)
2. Model numbering is identical to the retail drives. Which indicates that the Hardware factory has no way to distinguish between the two until they are packaged later in the process to be True IDE drives.

I am trying to get hold of some generic utility to read from the PCMCIA registers. This will give us more details on what is the drive saying to the controller.

[Big_Dawg]

I have definitely read that it is a "firmware" issue: that the TrueIDE/or CF spec is written to the EPROM in the microdrive. Noone seems to figure out how to rewrite it though.

[darkdestroyer]

If you can figgure out how to rewrite it then you will famous throughout the WWW. Huge companys such as Apple and Rio get HUGE discounts on these drives and sell their devices for even lower than the retailed drives. If we can go "borrow" the new "perpendicular" microdrives that are released, (toshiba?) then we will not only benifit from HUGE amounts of stroage, but it'll be a potential market flood of cheap, high capacity, microdrives, therefore rendering other "retail" versions, obsolete.

[nealzhao]

I has a Magicstor 2.2G microdrive. It can be read in my Axim x50v, but I can not write into it through my PDA. It is weired. I also tried using exdrive.dll (http://www.tjotala.com/hardware/storagebrick/) It did the same thing. By default it using atadisk.dll as the driver.

[Big_Dawg]

I has a Magicstor 2.2G microdrive. It can be read in my Axim x50v, but I can not write into it through my PDA. It is weired. I also tried using exdrive.dll (http://www.tjotala.com/hardware/storagebrick/) It did the same thing. By default it using atadisk.dll as the driver.

Yeah but that's a "true" Compact Flash compatible drive. Not quite the same thing as the problem we are working on.

[Big_Dawg]

After reviewing more and more docs about CF/CFII and Microdrives, I think I know how Hitachi and Seagate (I've been fooling around with a 5GB ST1 drive from my Rio Carbon in addition to the 4GB Hitachi I bought on eBay) are "crippling" these drives. I suspect enabling them would be very easy if you knew the exact method that they were disabling CF mode. This is because they are disabling them PHYSICALLY, not in the firmware. It seems that one way to disable CF mode is to "ground" pin 9 of the connector. I found a connector diagram, inserted a needle into pin 9, and measured the contact and resistance to the metal part of the casing to the pin. After I was sure I found the right pin (none of the other pins seemed to have any electrical contact) I discovered approx. ~1.45K ohm resistance. I thought that it might have been directly shorted, which would mean a bit of solder (circuit trace) somewhere connecting that pin, but the 1.45K Ohm measurement suggests to me that there is a circuit board somewhere with an (optional) 1.45KOhm surface mount resistor in place; alternately, there may be an open solder contact (or one of the holes I see under the adhesive label) which if electrically closed (w/a bit of solder)provides an electrical bypass to the resistor built into the Microdrive. This way, Hitachi/Seagate would only have to CMC 1) one solder point to differentiate OEM(MP3 Player) drives from Retail(CF spec ones) or 2) solder in one resistor to achieve the same objective. Come to think of it, if they are smart, it is probably the solder bypass method (1). Cheaper and faster, over thousands of drives. :exc:

>_< so it will be quite an expensive adventure to pop open the microdrives to check out the circuit boards of two nearly identical drives. OEM and Retail ST1s.
Probably VERY unwise and fireball prone but grab a old pc power supply... +5v to case GND to pin 9 BWAHAHAH take that resistor... Okay might melt your whole drive so dismantling it is 100x less dangerous in comparison.

So are you able to compare GND to Pin9 on a retail drive in contrast to the OEM? 1.45k ohm doesnt seem logical though, its much easier to bridge a connection over attaching a resistor.

[UnLoGiC]

Sounds like you could be on to something, the question is how do you test this theory without damaging any of you microdrives or phisicaly open them?

[star882]

I've opened some PCMCIA cards before. The key is to work slowly and carefully.
The CF cards seem to be harder to take apart. But I'm sure that it's possible with the right tools. Practice with an obsolete or broken card first.
Also make sure that the metal heatspreaders are actually ground. The easiest way is to insert it into a CF reader (unplugged) and measure between the heatspreaders and the outside of the USB connector.
Finally, maybe the MP3 CF cards are designed for low power and/or low cost and therefore use totally different hardware. Could the controller chip be different? Could the memory devices be low power/low cost rather than high speed? For that matter, maybe it's not even a CF electrical interface but some other type of interface. (Unlikely since the cards work in card readers.) Do normal CF cards work in the MP3 players?
BTW, if you do take apart the cards, please take good high resolution picures of the circuitry and I'll try to help. An easy way to get good pictures is to put the opened card onto a good flatbed scanner, scan at the highest possible resolution, and then crop the image.

[ravic]

I got three drives now.

1. Hitachi extracted from an IPOD (IDE Locked)
2. Seagate 5GB without any firmware mark (Works great in CF)
3. Seagate 5GB with FW 3.04 on the back (Does not work in CF)

In the mean time i also aquired a IDE to CF card adapter. It is an adapter to connect a CF card directly to a 40 pin IDE interface normally found on PCs.

I connected these drives to my PC via IDE. All three drives are detected properly by BIOS and are shown as IDE drives in XP.

I also got a PCMCIA adapter which directly connects a CF card to PCMCIA bus (similar to that in pocketps).
I tried the same three drives in a PC Card slot with a PCMCIA adapter. If it is just a driver issue on a Axim X50, then all the IDE drives should be detected fine in a pcmcia slot, since XP implements ide drivers)

Here are the pcmcia results.

1. Hitachi is detected as an unknown card
2. Seagter with no FW markings detected as a valid drive and shown in explorer
3. Seagter with FW 3.04 is detected as valid drive with ST650211CF model but quickly disappers from the detected the drive list.

Conclusions so far:

1. Hitachi seems to be hardwired at Pin 9 since it is not detected as a valid drive in CF slot or a pcmcia slot.
2. Seagate with FW 3.04 on the other hand is actually detected as an ATA drive in PCMCIA but something is then blocking access. This almost seems like it is implemented in firmware

What i need help on:

1. The seagate that works has FW 3.03. So we need someway to extract firmware out of this and put it on the other seagate.
2. I have a IDE adapter so i can directly connect them to IDE bus.
3. IDE drives support "ATA Commands" to update firmware but could not find a generic utility which allows making some ATA commands.
4. May be we will have batter luck with using linux to extract firmware

There is some one who opened a seagate microdrive. But he opened it on the wrong side (bottom side). Circuit is on top.
I read that the actual drive itself is sealed in another enclosure. So not an issue with dust if we open from top. I am not sure how the circuit is connected to drive. Probably by a flex cable.

Here is a link to his pictures
seagate microdrive ripped apart