PDA

View Full Version : Flaw identified in Sanyo MBC-55x motherboard



BradN
May 23rd, 2012, 12:42 PM
Hi all, I have discovered the cause of Sanyo MBC-55x boards that experience disk slow-downs and may eventually lose their ability to access the floppy drives: Part of the circuitry that generates a fake time delay in providing a "head loaded" signal to the floppy controller in response to the FDC's "load the head now" signal becomes nonfunctional, and the floppy controller never thinks the drive is ready.

Symptom: Floppy accesses don't work at all or take a long time to start each time the disk spins up. Possibly this appears or gets worse as the system warms up.

Cause: Capacitor C9 is installed with the wrong polarity - even the board markings are wrong for the circuit. Tantalum capacitors like the one used are sensitive to reverse connection and fail instantly at higher voltages, but at lower voltage, apparently behave like this with slow damage accumulating on the capacitor.

Verification: Check the datasheet for the 74LS221N chip that the capacitor is connected to - the datasheet describes the correct way to connect the capacitor. I have attached a scan of the relevant Sams PhotoFact page that details this part of the circuitry so you can find the pins to measure. If you intend on doing hobby work with one of these machines, buy a copy of this technical manual from ebay - it's that good.

Resolution: Replace capacitor with a new tantalum or electrolytic 47uF, installing it with the correct polarity. Important - be sure the capacitor was installed wrong first as they may have fixed this in late revisions. I am going to do some tests to see whether it is workable to simply reverse the existing (damaged) capacitor - tantalums are known for some "self healing" properties but I don't know if it is effective once it gets to this point.

Additional: Also attached is an oscilloscope trace of the output of multivibrator circuit to floppy controller (pin 4 of U91 in yellow), and the capacitor timing node (pin 15 of U91 in blue) on a WORKING motherboard. On a non-working board, the blue line starts to rise, and levels out before reaching the trigger voltage. This signal is captured during the start of a floppy operation where the drive is spinning up. You don't need an oscilloscope to verify this - a voltmeter is enough to check that pin 4 on U91 is stuck low while the machine is hung up.

Side note, I never thought I'd find a design problem like this on this board. It blows my mind with the attention to detail they used on it.

BradN
May 23rd, 2012, 02:07 PM
Update: The system appears to function correctly without C9 installed - the delay doesn't appear necessary with Sanyo BIOS 1.00 for onboard video or my own dsBIOS. For 100% compatibility it is probably best to install a new capacitor as this provides an extra delay before disk access is allowed and might help drives that don't spin up as fast be functional. It is possible Sanyo may have changed their disk code in later BIOS and there could be compatibility problems then, but I kinda doubt it, as I don't think they did much work on the BIOS at all after it was written except when they absolutely needed to (ie, to support CGA video boards and bigger RAM expansions).

Note: I haven't yet tested write operations that make the disk spin up (usually the OS performs a read first), but I will try to do this soon to validate removing the capacitor as a fully functional workaround.

I guess, if you find yourself in this situation without proper soldering skills and need to get the system working, you can just break capacitor C9 off the board and you're back in business, but otherwise I'd discourage this as it's a sloppy repair, but it is a better approach than trying to desolder without knowing how and messing things up worse.

Attached is a trace without the capacitor installed (note the timescale difference). The time delay is, from a software perspective, completely bypassed, and the system is maybe even a bit more responsive on initial disk access after spin-up. Disk reads are protected by CRC so it is unlikely to experience data loss if the drive doesn't spin up fast enough.

Edit: Clarification on blue line below - it is not the same signal as from the above trace - the signal from the above trace (U91 pin 15) is almost completely flat without the capacitor. I had thought I connected the wrong signal (because there was absolutely nothing there) and went to a different one (the one below is showing pin 14), thinking that is what I connected originally, but it wasn't.

Old Computers
November 20th, 2012, 01:43 PM
I performed this repair by reversing the capacitor. So far the disk drives have reasonable access times. I just need to wait and see what will happen.

Thanks for posting this or I would have never guessed at what could be wrong.

BradN
November 20th, 2012, 05:38 PM
No problem! I had seen mention of this issue before with no resolution. Then I bought a machine from eBay that turned out to have this problem and after probing some of the signals, traced it down.

Glad to have helped! It'd be a shame to see one of these machines get junked for something so simple. They're like the galapagos tortoise of PC's.

Old Computers
November 21st, 2012, 06:10 AM
Reversing the capacitor seems to have been fine. The delay did not manifest itself after an hour.

This is one of my favorite machines because of how unusual it is compared to other PCs at the time.