You guys need to see this first - It's booted and running!
On a 2MB file test repeated for a few iterations I'm getting anywhere from 123 to 170KB/sec data rates. The readback of that file runs between 321 to 342KB/sec. This is with DOS 3.3 and a Nec V20 running at 4.77Mhz - nothing too fancy. The hard drive is an old 20GB Maxtor.
Compared to my first gen XT-IDE running on the same machine, this is 2 to 4 times faster. The memory mapping was definitely the way to go ...
XT-IDE on PCjr
-
AnonymousFreak
- Posts: 4
- Joined: Sat Sep 03, 2011 9:43 am
Re: XT-IDE on PCjr
De-lurk mode engaged.
Unfortunately, I have few skills to contributed other than testing. (And at the moment, not much money to use to assist, either.)
I have basic engineering knowledge, and basic 'software development' process knowledge, but no hands-on experience in either.
Unfortunately, I have few skills to contributed other than testing. (And at the moment, not much money to use to assist, either.)
I have basic engineering knowledge, and basic 'software development' process knowledge, but no hands-on experience in either.
Re: XT-IDE on PCjr
Ok, I was reviewing potential changes from Rev.P2 (prototype) to Rev A. I seem to recall I was the only one championing the 2.5" PATA drive idea. I've since cooled way down on that. Especially as I just bought a lot of 2GB DoMs for all my 8-bit machines soon to have new hard drive boards. So even though the pinning of that connector was one of the reasons it took 2 prototype runs to get things right (and they still aren't), what do people think about dropping it?
Pros for keeping it:
- Offers a cable-less way to plug a real hard drive in and mount it direct to the card (like a hard card).
- Many may already have old laptop drives from upgrades. Beats buying a DoM if the old drive is free
Cons against it:
- Drives up board cost as there is an extra 6 square inches solely to provide support holes for mounting the drive
- Can only use one or the other. The 40-pin header, if installed, prevents it's use.
I'm leaning on dropping it and cutting down the board size some. The 40-pin header would be moved left a bit but still exist to the right of the support posts.
Few other changes also on the slate:
- Adding a second current limiting resistor so the on-board activity LED and the header can be use simultaneously if so desired.
- Adding a 2 pin header to optionally jumper power over to the key pin on the 40 pin IDE header. This would allow for DoMs or CF adapters that can draw supply power from there.
- Double check with DorkBot on exact drill size used for side car through holes. Absolutely need to get this perfect on the first spin at a different board house the first try.
And some BOM changes:
- Changing the second RAM part to a 128KB capacity since UMB support isn't possible w/o MB mods. Would save $1
- Increasing the resistance value of RA1 to dim the POST display some and limit current/heat
Pros for keeping it:
- Offers a cable-less way to plug a real hard drive in and mount it direct to the card (like a hard card).
- Many may already have old laptop drives from upgrades. Beats buying a DoM if the old drive is free
Cons against it:
- Drives up board cost as there is an extra 6 square inches solely to provide support holes for mounting the drive
- Can only use one or the other. The 40-pin header, if installed, prevents it's use.
I'm leaning on dropping it and cutting down the board size some. The 40-pin header would be moved left a bit but still exist to the right of the support posts.
Few other changes also on the slate:
- Adding a second current limiting resistor so the on-board activity LED and the header can be use simultaneously if so desired.
- Adding a 2 pin header to optionally jumper power over to the key pin on the 40 pin IDE header. This would allow for DoMs or CF adapters that can draw supply power from there.
- Double check with DorkBot on exact drill size used for side car through holes. Absolutely need to get this perfect on the first spin at a different board house the first try.
And some BOM changes:
- Changing the second RAM part to a 128KB capacity since UMB support isn't possible w/o MB mods. Would save $1
- Increasing the resistance value of RA1 to dim the POST display some and limit current/heat
Re: XT-IDE on PCjr
If you're looking for feedback, I'd drop the 2.5" idea. I don't see what's wrong with a CF + adapter; they're both cheaper and more plentiful than DOMs. Just my $0.02.
You're all insane and trying to steal my magic bag!
Re: XT-IDE on PCjr
Right now I'm using the 40 pin header with a cable to an external drive. Eventually that drive will be in a powered enclosure with cooling; they get hot without some sort of cooling. Any 2.5" IDE drive can be put on the standard 40 pin cable with a $2 adapter. I don't see a 2.5" drive for internal use - they are rugged for laptop use but the sidecar is still pretty small and it will get warm in there.
The only part I'd consider using internally is a DOM. A DOM still uses the same header as a cable connection would, and that is true for both 40 and 44 pin versions. Given that you can get DOMs in both 40 and 44 pin versions, straight or right angle, I would not do too much to accommodate one. Just assume 40 pins and run the power on the connector as you suggest. (Or do a little pigtail for the standard Molex connector.)
Trixter - DOMs are guaranteed to work as IDE hard drive replacements. They do CHS, whereas many CF cards don't. I don't think we need CHS for this card as it uses LBA mode, but I like the idea that the DOM is black box replacement for a hard drive. And I suspect the wear leveling algorithm and overprovisioning of FLASH is better than that of commodity CF cards.
I don't think we need both connectors. I think the standard 40 pin IDE connector is enough.
On your other changes: Yes to all of it. The POST display was useful for debugging, but it's going to be hidden when the shell goes on. That makes me sad. : - (
Is it possible to have an option for an external drive LED? I'd be willing to drill a hole in my sidecar to see what it is doing.
What's the current power draw on the card? I don't want people cooking their little 33 watt power supplies. I have a power sidecar ahead of mine just to provide the extra margin, and to power the LPT sidecar.
Mike
The only part I'd consider using internally is a DOM. A DOM still uses the same header as a cable connection would, and that is true for both 40 and 44 pin versions. Given that you can get DOMs in both 40 and 44 pin versions, straight or right angle, I would not do too much to accommodate one. Just assume 40 pins and run the power on the connector as you suggest. (Or do a little pigtail for the standard Molex connector.)
Trixter - DOMs are guaranteed to work as IDE hard drive replacements. They do CHS, whereas many CF cards don't. I don't think we need CHS for this card as it uses LBA mode, but I like the idea that the DOM is black box replacement for a hard drive. And I suspect the wear leveling algorithm and overprovisioning of FLASH is better than that of commodity CF cards.
I don't think we need both connectors. I think the standard 40 pin IDE connector is enough.
On your other changes: Yes to all of it. The POST display was useful for debugging, but it's going to be hidden when the shell goes on. That makes me sad. : - (
Is it possible to have an option for an external drive LED? I'd be willing to drill a hole in my sidecar to see what it is doing.
What's the current power draw on the card? I don't want people cooking their little 33 watt power supplies. I have a power sidecar ahead of mine just to provide the extra margin, and to power the LPT sidecar.
Mike
Re: XT-IDE on PCjr
OK, the 44 pin is out for Rev A.
There is a header to bring the LED out to the front. I planned on drilling a hole in the front of my side car for a larger green LED. Then hot gluing it in place. ATM though the board will work with both LEDs, but it's not ideal. If you mismatch forward drops, you'll shorten the life of one. It's best to desolder the one on board if you plan on using the header (P2 boards). Rev A will have an additional resistor so you can use both without worry.
You can also turn off the POST display via an I/O port write to conserve power; maybe as the last thing in the BIOS? Yes it's a shame, but it is useful when something goes wrong.
The Rev.P1 boards have a jumper for a current shunt on the supply. This weekend I'll put a P1 back in and recheck it. I seem to recall it only drawing ~140 mA w/o the hard drive. Which is about 700 mW. CMOS is nice.
There is a header to bring the LED out to the front. I planned on drilling a hole in the front of my side car for a larger green LED. Then hot gluing it in place. ATM though the board will work with both LEDs, but it's not ideal. If you mismatch forward drops, you'll shorten the life of one. It's best to desolder the one on board if you plan on using the header (P2 boards). Rev A will have an additional resistor so you can use both without worry.
You can also turn off the POST display via an I/O port write to conserve power; maybe as the last thing in the BIOS? Yes it's a shame, but it is useful when something goes wrong.
The Rev.P1 boards have a jumper for a current shunt on the supply. This weekend I'll put a P1 back in and recheck it. I seem to recall it only drawing ~140 mA w/o the hard drive. Which is about 700 mW. CMOS is nice.
Re: XT-IDE on PCjr
Quick question: Is this fascinating project going to morph into some sort of kit people can buy at some point?
Re: XT-IDE on PCjr
Yes, I'm going to order boards within the week. Will take 4+ weeks to arrive after that however. I'll be selling boards, connectors, and pre-programmed PLDs, and the rest of the parts kit ala-cart for at or near cost + shipping. I'm not sure exactly what those figures will be atm. Likley ~25 for boards, ~10 for new connectors, ~15 for pre-programmed PLDs, ~25-30 for rest of the parts kit (1/3 of that the RTC!), 5 for flat rate shipping. Unfortunately that totals up to about $85 for a full board parts minus hard drive. I wish it cost less. But the reality of it is you just can't build something like this in low quantity much cheaper. I'm going to comparison shop some this week and see if I can get a few dollars knocked off the parts kit and boards. Depending on the demand and pricing based on this estimate, I might do something like $100 for an assembled/soldered board tested and shipped.
Of course you also need the shell of an existing side-car. And the connector if you don't want to buy a $10 new one. They cost $10.25 from State and they are the only people left who apparently still have a lot of them.
Of course you also need the shell of an existing side-car. And the connector if you don't want to buy a $10 new one. They cost $10.25 from State and they are the only people left who apparently still have a lot of them.
Re: XT-IDE on PCjr
I'm willing to pay even more, up to $150 for an assembled and tested board. I mention this in case your time is not compensated fairly enough. The last thing I want is for someone to feel bitter about their hobby project.
You're all insane and trying to steal my magic bag!
Re: XT-IDE on PCjr
I didn't find the assembly very difficult, and I don't have great equipment here. The sockets are fairly friendly to soldering, but there are a lot of pins.
The pricing is going to turn off the occasional user, but I'm also expecting that once the word gets out that they will sell out because the IDE adapter is just such a vast improvement. And unlike prior solutions, it is designed to build into a sidecar so it looks well integrated.
The pricing is going to turn off the occasional user, but I'm also expecting that once the word gets out that they will sell out because the IDE adapter is just such a vast improvement. And unlike prior solutions, it is designed to build into a sidecar so it looks well integrated.