php1800 sidecar disk controller no worky

[Ed in SoDak]

My dad picked this up in the mid-80's. I don't think it worked then and sure doesn't now. Any schematics or common failures in these? The controller chip might be suspect as a common disk controller problem, so why should the sidecar version be any different? It has voltage, but the +5 is pulled down to 4.87 when connected to the circuit and when it is connected and powered, the -5 line goes a bit more negative compared to its disconnected reading. If that's a clue.

 

This one is serial 1528133 and LTA0381. So early '81. I know how fragile the thin ribbon wires between the boards are, so I'm reluctant to do my usual disassemble and then give it my best baleful stare.

 

The drive motor spins at powerup, but no light, head movement or anything other than a console error report. To try it, I just moved the outboard drives' 1&2 cable from my PE box to the sidecar and plugged it into the console port.

 

I don't have a real use for it, but figure it's worth more working, if it's a fairly simple fix, such as locating and installing a new controller chip, assuming it's not unobtainium.

 

Thanks in advance!

 

-Ed

Edited March 29, 2018 by Ed in SoDak

[+Ksarul]

There's a schematic for it (and a bunch of other sidecars/PEB cards) on Mainbyte's Site. Do remember that this controller only works with single-sided drives, IIRC.

[+FarmerPotato]

I just tested two working PHP1800 sidecar disk controllers with original TI external (single sided) floppy drive.

 

What model# of drive do you have in the P-Box?

[Ed in SoDak]

Thanks for the replies! I looked on Mianbyte, but only saw pics of the outside. Guess I missed seeing the schem.

 

I knew about the SSSD-only, but I was just trying to get the LED access to light, I didn't even have a disk inserted except for one test. The drives are DSSD from a PC clone, but I do have original TI SSSD I can drag out if it has to be that.

 

I gotta make a delivery of a couple tylosaur jaws, but I'll check into it more later today. Thanks again, it'll be neat if we can get it going!

-Ed

[+Ksarul]

The link I sent you to on Mainbyte's site is a list of all of the schematics he has for download. . .

[RickyDean]

You know, I have two of the side car disk controllers, one intact and accounted for, the other somewhere in storage, taken apart back in the 90's,, due to the fact in didn't work either. If they use the 1771 chip, why aren't they able to see a double sided disk? Can this be changed with a rom upgrade or something? also what about 80 tracks?

[+mizapf]

According to the schematics on Mainbyte, the SIDSEL line (pin 32) is pulled up to +5V and cannot be controlled. On the standard TI DSSD controller, the SIDSEL line is controlled by a CRU bit, not by the FD1771. The reason is that the FD1771 cannot control the head selection; the sector header (cyl,head,sect,size) is specified to have zeros for the head. The WD177x controllers can optionally verify the head field, but do not have a SIDSEL line either. The HDC9234 of the HFDC does have that line, but it is anyway tailored to control MFM hard disk drives as well, so it is advantageous to know that there is more than one head.

 

In short, it seems impossible to me to make the PHP1800 double-sided, because the circuitry is missing that is needed to control the SIDSEL line; a change in the DSR is clearly not sufficient without a hardware mod.

 

The 1771 specs define on page 13 the track number to be a value from 00 to 4C, which means 77 tracks. I could not find whether this constraint is enforced or just assumed.

Edited March 29, 2018 by mizapf

[Ed in SoDak]

For my simple tests, I'm guessing there's nothing special about the TI SSSD drive that would make it a requirement. I'm using my regular DSSD drive setup off my PEB, moving the cable from it to the sidecar. I don't even have a disk mounted, I'm just looking for the delay while EXB searches for DSK1.LOAD or I'm trying OLD DSK1.<anyname> to see it the drive is selected and lights up.

 

Since I had a couple new caps in my stash, I changed out C12 and C13 that filter the +/-5v lines. That improved the voltage drop I was seeing in my earlier tests, but still no joy. All along, I've been suspecting the 1771 controller chip, simply because it seems the most likely chip to blame.

 

So now I'm debating dragging my second PEB out of the shed to "borrow" the 1771 chip to see if it gets this sidecar working. But it's way up in the loft and not easy to get to. I'm also unsure how easy it is come across a replacement chip (haven't looked yet) and if it's worth buying one on a guess it might help.

-Ed

[Stuart]

What command are you typing at the console, and what error code is it returning?

[Ed in SoDak]

Upon choosing XB, it comes to the cursor with no delay nor action from the drives. Entering OLD DSK1.L or the same for DSK2, it gives an error code 50. "The drive specified could not be found with OLD."

 

The drives are connected same as I have them to my PEB, they have their own power supply, so I merely moved the drive's cable to the sidecar, which replaced the PEB. No disk was inserted, I don't think that should matter till I at least see the drive LED turn on. So it also doesn't matter what filename I use, since it's not reading either drive to begin looking for a specific file.

 

I guess it could be related to some signal or line the SSSD drive provides which my DSSD drives do not, but that doesn't make sense.

-Ed

[marc.hull]

If XB returns immediatly then that indicates it cannot see the card. Items to check in order of probability.

 

1) drivers and tranceivers 74244 & 74245 on the front of the card.

 

2) CRU decode circuit

 

3) card rom.

 

It's not at all likely the 1771 at this point.

[+mizapf]

If the error code is correct (there are some examples of misleading error codes), the 50 should say that on OLD, the device "DSK1" was not found by the DSR. This should not relate to the hardware (should give a 56 error) but to the DSR search routine. That is, either the ROM does not work, or the circuitry fails to turn on the ROM.

[Ed in SoDak]

Wow, I'm getting lots of help here, thanks! Mark and Mizapf, thanks for pointing out things to check. It's looking like a plug'n'play solution is not in the cards. I'd guess the ROMS are even more unobtainium than the 1771, the only socketed chip. I'd as likely bodge the ribbon wire while replacing soldered-in chips.

 

No big deal, it's been sitting for over three decades now.

 

-Ed

[Ed in SoDak]

How would a guy go about testing the ROMs and the 74244 & 74245 chips Marc mentioned? I have a couple DMMs, one can measure a small range of frequencies. No logic probe, but one might be useful here. More knowledge and a scope is also lacking.

 

Otherwise, about all I got is a shotgun approach of finding and swapping in replacements for the suspects.

 

What about peeking into the sidecar's DSR to see if there's a response?

-Ed

[+Ksarul]

Since you don't have a logic probe, I would probably replace the 244s and the 245 first, as pretty much everything else is behind them on the board.

[+mizapf]

What about peeking into the sidecar's DSR to see if there's a response?

 

Using Extended Basic; you can do it on the command line.

 

 

CALL INIT
CALL LOAD(-16384,2,12,17,0,29,0,4,91)
CALL LOAD(-31804,192,0)
CALL PEEK(16384,A,B,C,D)
PRINT A;B;C;D

 

This should print some values not equal to 0,0,0,0. As a test, leave out the second CALL LOAD line; this should deliver 0,0,0,0.

 

(Explanation of the lines: The first CALL LOAD loads a sequence of machine code bytes at address >C000 that first set R12 to >1100, then set CRU bit 0 to one, then return. This should turn on the card DSR. The second line sets the interrupt hook to the address >C000, which causes the code to be executed. The PEEK line reads some values from >4000 to >4003.)

Edited March 31, 2018 by mizapf

[Stuart]

Also take a look at http://atariage.com/forums/topic/273707-in-need-of-corcomp-fdc-pics/?p=3930964 if you want to do a deeper dive into the DSR.

[Stuart]

 

Using Extended Basic; you can do it on the command line.

 

 

CALL INIT
CALL LOAD(-16384,2,12,17,0,29,0,4,91)
CALL LOAD(-31804,192,0)
CALL PEEK(16384,A,B,C,D)
PRINT A;B;C;D

 

This should print some values not equal to 0,0,0,0. As a test, leave out the second CALL LOAD line; this should deliver 0,0,0,0.

 

(Explanation of the lines: The first CALL LOAD loads a sequence of machine code bytes at address >C000 that first set R12 to >1100, then set CRU bit 0 to one, then return. This should turn on the card DSR. The second line sets the interrupt hook to the address >C000, which causes the code to be executed. The PEEK line reads some values from >4000 to >4003.)

 

That's a nice way of getting the code to run, Michael. When I've played around with this before, I found that the CALL PEEK command switched off the card DSR, but I can't reproduce that now. Re-enabling the card every interrupt would be a neat way around that.

 

Edit: I've found what I done. If you CALL PEEK with the XB cartridge then the card DSR remains enabled. If you CALL PEEK using the E/A cartridge, the card DSR is switched off after the CALL. Strange.

Edited March 31, 2018 by Stuart

[Ed in SoDak]

Thanks Mizapf for the code I can use and the explanations! I did a bit of my own sleuthing today and found some info in the MG Explorer docs. Then a HEXtoDEC converter got me the 16384 location, and the CRU 1100 address, but I was pretty much drawing a blank after that.

 

I have an original (actually purchased!) Explorer disk and manual, but with no working drive, well... I pondered piggybacking the PEB with its controller onto the sidecar, but figured that wouldn't work or just the PEB disk controller would respond. Load Explorer and then unplug the PEB? Too risky.

 

This takes me back to locating which 4116 RAM was dead on a Timex 1500. My first approach was piggybacking some spare 4116s I had from a TI99. When one of those proved to be shorted, or I crossed-up some pins while testing; which burned my finger and also smoked a transistor, that method was out! Next up was looking for shorts and such with my DMM. I found some discrepancies comparing the chips, but nothing I could write home about. So I ordered a shitpot of 4116s and also the various transistors, ready to swap in all of 'em if needed, plus sockets.

 

To finally fix that little beast, my buddy Ian from Canada and I were bashing our heads till I had a little brainstorm. We added a 16k RAMpack (which I'd learned on the Timex 1500, gets mapped in before the internal RAM). We loaded Ian's nifty memory-check program into it, then investigated the internal 16K. The bad chip was found by noting which digit of the R/W test was failing. This spared me from removing all eight chips and got it working with only one RAM chip needed.

 

This got me to wondering about a similar software approach of looking into the ROMs or DSR on the TI's sidecar.

 

On the shotgun side of shooting in some parts, good advice from Ksarul to start early in the chain and those chips should be available. Not so sure about the ROMS, if we end up there.

-Ed

Edited March 31, 2018 by Ed in SoDak

[marc.hull]

Buy a cheap logic probe for 20 bucks and download the data sheets.

 

ROMs are pretty bullet proof compared to the front line TTL. Not saying it couldn't be them but low in probability at this point.

[Ed in SoDak]

Yeah, I really should add a logic probe to the arsenal. A lot of my test equipment is for servicing tubes or testing transistors. I never did move on to those new-fangled ICs. Other than swapping in new ones for the guessed-at suspects, I'm typically clueless. I even bought a 300-in-one electronics trainer with a breadboard to try learning more, but never got too serious. I should drag it back out for another go.

 

My old electronics magazines published dozens of plans for probes. Most are pretty simple.

 

I found the 244/245 chips at Newark for under 2 bucks, but $5.50 shipping. On the plus side, they are actual TI-branded.

 

-Ed

[Ed in SoDak]

Digi-key has the same TI chips with less shipping since they'll ship small orders 1st class. I've used them before, so will order my chips from there.

 

I tried the XB code Mizapf posted above, using SXB.

 

CALL INIT
CALL LOAD(-16384,2,12,17,0,29,0,4,91)
CALL LOAD(-31804,192,0)
CALL PEEK(16384,A,B,C,D)
PRINT A;B;C;D

 

The CALL INIT caused a Syntax Error. So I left it out. On two attempts, it locked up upon entering the -31804 line. Leaving it out gave 0,0,0,0 for the peek command. Just the 16k console RAM, if that matters.

 

I repeated it with the PEB instead of the sidecar. This time it worked and gave 170,2,0,0. So I'm guessing it needs the 32k? I suppose I could pull the PEB's disk controller and insert the sidecar to try again with 32k card in play. Not tonight, though.

 

While dinking around, I had the sidecar on for awhile, the transformer was warmer than I might expect, but not hot by any means. The regulator was cool and no ICs felt warm at all. I did notice the 244/245 pins were oddly discolored a sort of dark gray, like they had overheated at some point, but the original fuse is intact. The other IC pins were shiny like you'd expect.

 

Dug out a couple logic probe circuits from the magazines. One uses just a 2N222, a couple LEDs and a few other parts good for TTL and CMOS. Can't get much simpler or cheaper than that! Another uses a 74L04 Hex Inverter IC, which might be harder to find. I'll keep looking for other circuits.

 

-Ed

Edited April 2, 2018 by Ed in SoDak

[+mizapf]

Ah, ok, you don't have a memory expansion. This is required for my approach, sorry.

[Stuart]

You can also do some testing using the MiniMem cartridge, if you have one of those. I finally got round to putting some notes together for this: http://www.stuartconner.me.uk/ti/ti.htm#checking_disk_controller_card_dsr.

[Ed in SoDak]

With the decision to just replace the 244 and 245, the testing and logic probe is more for my own education, though it would be nice to verify if the problem is with those two chips.

 

My PEB has TI 32k, disk controller and RS232. I've been messing around with only the sidecar controller attached, so no 32k. It's worth a shot to connect both the sidecar and the PEB, but with the PEB's controller removed. That should work. I see the sidecar does not have an access LED, so disk activity is all I have to go by.

 

I also have the Minimem, so I'll look into Stuart's method of using that cart.

 

Somewhere I have another sidecar. I'm pretty sure it's the 32k, but could be the RS232, it's been awhile since I looked at it. It didn't seem to work either as I recall and has been stored for decades, like the sidecar controller was until recently, when I came across it and got curious about it.

 

Thanks again guys! Maybe I'll learn something along the way!

-Ed