VFD display problem (PFAFF 1473 CD) (FIXED!)

[inse]

As per photo, the driver is able to address separate segments and separate digits, I would tend more to suspect that the uC decoding the display data has got corrupted (degraded (E)PROM)
Or is receiving garbage from the mainboard.
Which would be hard to fix in both cases unless you are willing to reverse engineer the firmware.

[max.wwwang]

As per photo, the driver is able to address separate segments and separate digits, I would tend more to suspect that the uC decoding the display data has got corrupted (degraded (E)PROM)
Or is receiving garbage from the mainboard.
Which would be hard to fix in both cases unless you are willing to reverse engineer the firmware.

That's likely, because the RAM/ROM chips are close to the area affected by battery leak. That's worrying. But for now, at least I would like to confirm the display board is working just fine, or otherwise. Reverse engineering the firmware can be a possibility (if ever possible), only a bit distant for now.

[max.wwwang]

Just to digress, as I said earlier, this machine was (beautifully) made in West Germany.

[max.wwwang]

Thinking of reverse engineering the firmware again. It's almost impossible, unless I purchase another working one and clone its content, or its dump is available somewhere, because I only know what's shown is not right (because not making sense) but don't know what should be shown!

Anyway, have I now catched the tail of a bad guy? I'm probing the clock signal going into NS65118 (pin 21, coming from pin 11 of one of the two uCs, MAB8031, which TxD/P3.1 serial output), which is obviously critical, regardless whether the input data is correct or is garbage. My question here is, why does it look so messy? I think it should be very close to a square waveform? Thoughts?

The scope setting is 0.5us/div and 1V/div. So the clock signal is about 1MHz with p2p voltage 4V-ish.

[max.wwwang]

Caught another guy that is certainly bad, the closest to the battery leak affected area, that is the T74LS03B1 quad 2-input NAND gate. When probing its 1/2/3 pins, I got all constant HI (5V), which is surely wrong - with 1(1A) and 2(1B) both HI, 3(1Y) as NAND(1A, 1B) surely should be LOW (0V)!

This is an easy fix. Hopefully this is all that's wrong! Anyway, eventually I've caught something! 

My question in the above post remains to be answered.

[inse]

Regrading the clock signal:
Did you compensate the probe capacitance?
Make sure to use a good, short ground connection for the probe.
The signal doesn’t look nice but might be acceptable for the receiver depending on threshold and setup/hold timing.
The supply voltage quality at the uC might also be worth checking.
What is the function of the suspicious NAND gate?
I had a look at the mainboard photo and was wondering why they used two microcontrollers.
The LS03 is the one close to the RAM chip?
Probably used to interface it to the 8051/8031 data bus.

[max.wwwang]

Regrading the clock signal:
Did you compensate the probe capacitance?
Make sure to use a good, short ground connection for the probe.

No, I didn't make any compensation. What is it about? 
That's probably the reason of the signal looking strange; the ground lead is long and through another wire. 

The signal doesn’t look nice but might be acceptable for the receiver depending on threshold and setup/hold timing.

True.

The supply voltage quality at the uC might also be worth checking.

Very clean 5V DC.

What is the function of the suspicious NAND gate?
I had a look at the mainboard photo and was wondering who they used two microcontrollers.

Not easy to tell. Tracking the trace is very difficult. Many turn sides via through holes under the chips, as if the designer made an effort to make it difficult for reverse engineering.
Yes, it's very intriguing that two uC's of the same family, one 8051 the other 8031, are used side by side here.

The LS03 is the one close to the RAM chip?

Yes, another photo added. It's the top left dirty one.

[inse]

Probe compensation matches the probe impedance to the scope input impedance.
Decent scopes have a square wave signal available at the front panel and the probe has a tiny trimmer to adjust it to make the signal look as good as possible like a square wave.
I notice the fine traces, make sure none of them has been interrupted by the corrosion you mentioned.

[max.wwwang]

Probe compensation matches the probe impedance to the scope input impedance.
Decent scopes have a square wave signal available at the front panel and the probe has a tiny trimmer to adjust it to make the signal look as good as possible like a square wave.
I notice the fine traces, make sure none of them has been interrupted by the corrosion you mentioned.

My scope does come with a test signal terminal and the probe has a small trim screw for adjusting the overshoot of the test square waveform.
Yes, interruption of traces by corrosion was my concern. I have closely inspected but have not noticed one.

[max.wwwang]

More investigation when waiting for the part to arrive (it will take a good time). Fortunately, this is much easier than having to deal with the chips under the display module.

I know that 3(1Y) of the NAND gate chip is erroneous. This also goes to 5(2B) of another gate, so 6(2Y) is ruined, which goes to the RAM chip's (SRM2064) 27(!WE, write enable). It is constant LOW when testing, i.e. write enabled all the time. I guess garbage information is probably written when it should not.

Also interesting is that the RAM's pin 27(!OE, output enable) is constant HI, so output is never enabled.

It comes from 8031's pin 17(!RD/P3.7). This means its content is never read by the 8031 uC.

It's a complex spider net obviously. Perhaps wait and see after the replacement of the NAND gate.

[inse]

Very good circuit debugging!
I wonder what they need the RAM for anyway - is it backed up by the battery?
What kind of information could be stored in there?

WAIT - one idea just came to my mind: I had a look at the 74LS03 dataset and it’s an open collector NAND.
There has to be a pull-up resistor and it needs to be supplied.
Did you put a fresh battery in already?
I propose to desolder the LS03, _test all traces for continuity_ and solder a socket for further investigations.

[max.wwwang]

Very good circuit debugging!
I wonder what they need the RAM for anyway - is it backed up by the battery?
What kind of information could be stored in there?

WAIT - one idea just came to my mind: I had a look at the 74LS03 dataset and it’s an open collector NAND.
There has to be a pull-up resistor and it needs to be supplied.
Did you put a fresh battery in already?
I propose to desolder the LS03, _test all traces for continuity_ and solder a socket for further investigations.

Thanks!

Good question. Yes, this machine requires two AA batteries to keep the memory (that's what caused the mess around the battery compartment), which I now think is probably what is stored in this RAM, though I cannot be sure yet.

As far as I understand, this machine can let you customise your own stitch patterns and stored in its memory for use whenever you want (so far as it's not lost).

Yes, I tried multiple times putting in batteries and taking them out. It should work just fine without the batteries.

This train of thought makes me think even if this one is replaced, the problem of the display might not be fixed? I don't like this thought!

Surely when taking it off, with more spaces exposed, I will have another good check and clean before putting the new one back.

[inse]

If the obvious failure was the battery leakage, I would thoroughly test all the fine traces in the affected area for continuity.(electrically, not optical)
I have seen cases where corrosion went along the leads into components, Idk if this may also have happened at the LS. But the leads don’t look so bad on the photo.
A second failure on the machine is certainly possible but less likely.
We don’t know how the firmware reacts to issues with the RAM.
Don’t give up

[max.wwwang]

If the obvious failure was the battery leakage, I would thoroughly test all the fine traces in the affected area for continuity.(electrically, not optical)
I have seen cases where corrosion went along the leads into components, Idk if this may also have happened at the LS. But the leads don’t look so bad on the photo.
A second failure on the machine is certainly possible but less likely.
We don’t know how the firmware reacts to issues with the RAM.
Don’t give up

In this case, it's almost certain that the corrosive alkaline has managed to get into the LS chip (probably along the pins) and failed a part of it. Fingers crossed and hope it will get well fully after a new NAND gate chip! 

[max.wwwang]

The original NAND gate DIP-14 IC to be replaced is T74LS03B1. I was unable to find the exact IC but found SN74LS03 on Aliexpress. Ordered 5 and arrived very quickly (which was a pleasant surprise). Also to my surprise though, with a quick test before soldering one on, they do not seem to work!  This is really a surprise because I bought many electronic parts from Aliexpress before, and never had such experience.
[Edit: I was wrong. Pull-up resistor is required on the output pin of each NAND gate. The chips are all good.]

Now searching with only "quad 2-in NAND" again, there appear to be many (more than I thought) that all look like good as replacements. From the local store Jarcar alone, I got:

74LS00 Quad 2-Input NAND Gate Schottky IC;
74HC00 Quad 2-Input NAND Gate CMOS IC;
4011 Quad 2-Input NAND Gate CMOS IC;
74LS01 Quad 2-in NAND Gate Low power Schottky IC

These all seem ok as a replacement? The last one looks the closest to me (from the name of it). Can someone please help me confirm? Thanks.

[picburner]

How did you test them?
The 74LS03's are open collectors (outputs), you cannot replace them with the 74LS00's which are Totem Pole if the circuit does not allow for it.

[inse]

Open collector, because they might be used to interface the battery backed up RAM.
Analyze the circuit, maybe a LS00 with a diode can be used as replacement (botch).
As picburner suggested already, did you have a pull-up at the output for your test?
One more tip: solder an IC socket, so you’re free to swap forward and backwards as you wish

[Martinn]

Just to digress, as I said earlier, this machine was (beautifully) made in West Germany.

When I some years ago bought a new sewing machine, I wanted something robust, purely mechanical without unobtainium chips that would fail on me.
It turned out that Pfaff sold off their Tipmatic production line and they are to date still manufactured in Iran, trade name "Gritzner". I have a tipmatic 1037 which has full cam control, no electronics besides lamp and motor.
Great machine, however in hindsight I really would love to have needle positioning, which of course only the newer electronic machines have.
Reading your post however tells me that decision was not so bad after all.

[max.wwwang]

How did you test them?
The 74LS03's are open collectors (outputs), you cannot replace them with the 74LS00's which are Totem Pole if the circuit does not allow for it.

Thank you. Lucky I asked here! They took all identical to dumb me!  I didn't use any extra components testing the 74LS03's - no pull-up's or pull-downs or anything else. I just plug it on a small breadboard, feeding the 5V and GND rail to the chip's Vcc and GND pins, mock one gates input with Vcc and/or GND (for high or low) and see what I get from its output pin.

So I suppose they are probably ok just not suitable as replacement?

Open collector, because they might be used to interface the battery backed up RAM.
Analyze the circuit, maybe a LS00 with a diode can be used as replacement (botch).
As picburner suggested already, did you have a pull-up at the output for your test?
One more tip: solder an IC socket, so you’re free to swap forward and backwards as you wish

Thanks! IC socket is a great idea, though I need to check if there is enough (head) room for it.

Knowing the above, and getting back to square one (looking for an identical replacement), does anyone know where I can find T74LS03B1 or anything that is equivalent? Does this mean none of the NAND gate chips I listed above is a straightforward equivalent replacement for T74LS03B1? (Datasheet of each can be found under the tab "download".)

Will take time to look closer into the datasheets, and the circuit about the T74LS03B1, and digest the information I got from you guys.

[inse]

The last posting confuses me- just drop in the LS03 you received and see if it works or behaves differently

[max.wwwang]

The last posting confuses me- just drop in the LS03 you received and see if it works or behaves differently

Not like those experienced experts, I tend to do enough 'mental exercises' before action just to avoid ruining things by repeated soldering and desoldering unnecessarily, particularly with things having more than 3 pins. I have not even pulled the bad guy out. 

As pointed out by other people above, obviously I missed an important piece of information about pullup resisters in the datasheet when testing the new chips. I guess they are probably good (based on my experience with Aliexpress). Will test again. It looks like they should be ok as direct replacements based on the TI's reference Understanding and Interpreting Standard-Logic Data Sheets (page 15). They are both 74LS03, where 74 means commercial grade (54 military grade), LS means "Low-Power Schottky Logic" (though the meaning of 03 is still not obvious to me). It was these magic codes that puzzled me.

[inse]

There is a certain level of logic behind the naming scheme of logic ICs as you have already started to understand.
The last digits code the function of the IC, starting from 00 - in the past entire data books used to be common.
So don’t worry, any 74LS03 will work in your sewing machine unless the mainboard is designed to the edge of technology like some kind of supercomputer.
If you are careful while desoldering and don’t use force, everything will be fine

[max.wwwang]

BINGO! FIXED ! after popping in the new chip. 

As usual, abundant care was taken. To avoid stress to the board and the pads/traces, I cut break the legs of the chip with a sharp knife before desoldering them - now a piece of cake! It's not super easy but much easier than you might think. Some cleaning (not the best or the tidiest but reasonably good) ...  I'm satisfied with the outcome.

Thanks heaps for all the help and support along the way, particularly by reminding me of the need for pullup resistors (so I didn't judge the good chips as faulty!)... I apologize to Aliexpress, you rock!

[Edit] Of course, before using the new chip, I test it again - this time with a pull-up resistor on the output pin of a NAND gate. It worked as expected!

[max.wwwang]

Just to digress, as I said earlier, this machine was (beautifully) made in West Germany.

When I some years ago bought a new sewing machine, I wanted something robust, purely mechanical without unobtainium chips that would fail on me.
It turned out that Pfaff sold off their Tipmatic production line and they are to date still manufactured in Iran, trade name "Gritzner". I have a tipmatic 1037 which has full cam control, no electronics besides lamp and motor.
Great machine, however in hindsight I really would love to have needle positioning, which of course only the newer electronic machines have.
Reading your post however tells me that decision was not so bad after all.

Whatever your decision was, if it was a PFAFF, that's the right decision. Yes, as soon as you have tried the electronic needle up/down feature (one of the very first new features compared with the pure mechanical machines that the early electronic machines offered), you feel you cannot live without it. 

[inse]

Congratulations