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

[max.wwwang]

Another repair project. This PFAFF sewing machine (1473 CD) seems working, except that the VFD display shows all 'A', which is incorrect (the photo captured not showing all 'A' is due to the refresh rate). The VFD module consists of 16 characters, 12 segments each. This is the behaviour when powered up without any human input. There is another display mode by pressing a button ("display"). In this mode, it's better - showing what's supposed to be shown except for the first 4 characters - again all 'A's, which is incorrect.

The VFD module is on a separate daughterboard along with buttons and LEDs (it's invisible, but most likely there is a driver IC under it). This board communicates with the motherboard through a 12-wire flat cable.

After a close look, there are residuals of battery leak around the battery holder, causing corrosion (left-hand side of the motherboard). Broken traces of pins due to corrosion were not found. But it was likely the oxides might short-circuit (to some degree) adjacent traces or pins. I gave the affected area a careful clean-up with vinegar, IPA, and flowing water, then dried it thoroughly. However magic did not happen - no improvement.

What do you think most likely the culprit might be? Thanks.

[Edit: By the way, this machine is beautifully designed and made (in West Germany). My hat's off to PFAFF (not existing since 1999)!]

[cruff]

The solder connections of the VFD itself look messy, have you cleaned those up and reflowed them?

[max.wwwang]

Thanks. I checked closely. Despite looking messy, it must be residuals of original flux, without any signs of defect. So I didn't reflow them.

From the condition of the thing, the problem probably is not on the daughterboard. Everything looks tidy except the mess caused by battery leakage.

My guess is that these messy-looking solder points were all manually done in the factory. These are not signs of someone attempting to repair something after it leaving the factory.

[max.wwwang]

There are a few 22uF 6.3V electrolyte caps across the 5V rail and presumably GND, for logic chips. One next step might be to replace all of these (but I need to order new ones). Are they likely to be the cause of this behaviour though? The 5V rail is okay.

Since there are only 12 wires between the two boards (which include at least GND, filament voltage, VFD grid voltage), clearly there is a VFD controller under it, which is invisible (and I don't intend to get into it by risking breaking it). I will need to look into the common VFD controller to find out the communication protocols between it and the MCU(s) on the main board, and how this kind of problem can possibly happen. I suspect (assume, or hope) the display side is okay.

Also, another look at the back of the small board shows that there are clearly three chips under the VFD module, two 16-pins and one 40-pin. At least one of these (probably 16-pin) must be a driver, the big one might be another MCU?

Is there any 16-pin VFD driver IC popping up in your mind?

[cruff]

The VFD can be removed using a desoldering tool with care, it's obviously a tedious task but reasonably safe. The only VFD that I've replaced had on-board driver chips and used a command/data interface to interface to the remainder of the system. I'm not familiar with the types of high voltage driver chips used for VFDs.

If your device uses an off-VFD driver chip, then higher voltages will obviously be present and contamination on the circuit board could potentially cause this symptom. Easy enough to clean up the board and see if it makes a difference.

[inse]

At first, I would check the signals on the interface cable if there is any sign of life when you would expect the display to change.
If so, the suspicion narrows down to the actual display control.
Without removing the VFD tube you won’t be able to inspect what is hidden beneath.
The communication interface can be anything from 4/8 bit parallel to SPI or even UART.

[max.wwwang]

The VFD can be removed using a desoldering tool with care, it's obviously a tedious task but reasonably safe.

This would be the last thing I would like to do and only if absolutely necessary. After all the machine seems to be working well, only with a display problem.

If your device uses an off-VFD driver chip, then higher voltages will obviously be present and contamination on the circuit board could potentially cause this symptom. Easy enough to clean up the board and see if it makes a difference.

There is 32V DC voltage present on the board, supplied through the black wire of the interface cable. On the photo of the back of the board, it's the thin trace on the left edge of the board. It looks like the designer discreetly minimised its interference with other signals.

At first, I would check the signals on the interface cable if there is any sign of life when you would expect the display to change.
If so, the suspicion narrows down to the actual display control.
Without removing the VFD tube you won’t be able to inspect what is hidden beneath.
The communication interface can be anything from 4/8 bit parallel to SPI or even UART.

Good point. Having had another look at the interface cable, there are several wires the functions of which I now know. Starting from the Black (1): 1: 32V DC (this should be for the VFD anode); 2: 5V - this is probably for one or more logic IC on this board (under the VFD); 3: GND; 7: NC; 11/12: 4.5V AC for filament. There are only 6 wires left for signals.

Probed some of these 6 wires with a scope, but didn't get a clue yet.

Not I think the problem is very likely with those chips underneath. It's unlikely for the detail of display segments to be determined by the MCU(s) on the main board and signal transferred through this 12-pin cable. Very unlikely. And it's very likely the 40-pin IC is an MCU for the VFD (and potentially for the buttons and LEDs). Now my guess is that the two 16-pin's are logic ICs. The culprit is probably something around them or one of these two ICs themselves.

It looks like now I have a hard decision to make, whether to desolder the VFD module, risking making things worse. (This would not be a hard one if I had a desoldering station that can suck when the head is on heating, such as the one used by Mr Carlson's Lab - this desoldering job would be a piece of cake!)

But among the things that should be tried before this move, I will replace all the 22uF 6.3V electrolyte caps (and another 1uF 63V) first.

[picburner]

The DIP-40 under the VFD is most likely the vfd driver, it could be an SN75518 or HV518 or similar. If it had partially failed it could very well highlight the defect shown in the pictures.

[max.wwwang]

The DIP-40 under the VFD is most likely the vfd driver, it could be an SN75518 or HV518 or similar. If it had partially failed it could very well highlight the defect shown in the pictures.

Thank you. This is a plausible theory. Both of these drivers have a 32-bit register, which looks like something that suits this VFD module very well. What the other DIP-16 ICs (or their function) do you think are likely to be?

I'm now considering buying a desoldering station for this job because I really don't want to ruin this machine (and if possible get it fully funcitoning), which I value highly. I saw once in Mr Carlson's repair videos a nice kit, with which this job is just easy as. Does anyone have a recommendation on this?

[inse]

You don’t need to buy a desoldering station for just one job.
A good quality desoldering pump, fresh (leaded) solder and some extra flux will do the job if supported by a decent soldering iron.

[max.wwwang]

You don’t need to buy a desoldering station for just one job.
A good quality desoldering pump, fresh (leaded) solder and some extra flux will do the job if supported by a decent soldering iron.

Not just for one job of course. It looks like only for one, but the fact is that often one job triggers the idea of buying some tool that has been on the wish list for long and just been waiting for a proper/final justification, and surely will be useful later on. Another thing is that I’m hopelessly pedantic!

[inse]

Thats an acceptable answer

[max.wwwang]

Have done a bit more trace tracking and got this table for the 12-wire interface cable. The DIP-40 on the display board is almost certainly a VFD driver, such as SN65518 or HV518.

[max.wwwang]

All electrolytic capacitors are replaced (except the big ones in the power supply stage). The old ones do not look too bad (or not bad at all, which is surprising). Unfortunately, no miracle happened. The same.

The next step is probably to get a proper desoldering tool and tackle the bigass VFD module. One problem with that though, is that when it is off, there is no display available as a handy indicator of what the problem is, or when it's back working properly, without soldering it back. So I guess before doing the big job, I need to do a bit more in understanding what voltages it should give on some critical pins/pads, if it is working properly.

[sarahMCML]

Above the green connecting strip at the bottom of the first pic, some way up is a thick trace. Just above this, in the middle are a pair of solder joints, one of which looks to be barely, if at all, soldered. There's another similar poor joint over to its left which needs re-doing as well!

[max.wwwang]

Above the green connecting strip at the bottom of the first pic, some way up is a thick trace. Just above this, in the middle are a pair of solder joints, one of which looks to be barely, if at all, soldered. There's another similar poor joint over to its left which needs re-doing as well!

Thanks for your spotting. I will have another close look at these, but very likely they are good. These messy-looking joints I strongly believe were done manually at the factory. There are the LEDs that stand up and fit with the right height in the holes in the face panel. This machine was made in West Germany so before 1989. Since they are only LEDs, even if they were out, normally it would not affect the rest of the circuit.

One obvious thing to take note of before taking off the VFD module is perhaps the voltages on its 32 pins in the middle. If replacement of any of the chips under it is necessary and when it’s done, I can check if there is any change to these voltages. If there are, it’s likely fixed and I can proceed with soldering the VFD back.

[Edit]I had another look at these soldering points of the LEDs. Despite looking messy in the photo, they are perfectly fine (just because they were done manually with an abundance of natural rosin as flux. Its residue is not harmful.)

[picburner]

If you have a scope you can check the problem with the help of the SN75518 datasheet.
I reported on your pic the position of the main pins and their function (at least what I can understand based on the pic).

If it was indeed the DIP-40 that was faulty (and I think so) and you have no experience with a desoldering station, I advise you to gain experience on another faulty PCB first.
Try desoldering many components without breaking them or damaging the PCB before moving on to desoldering a delicate glass VFD!

[inse]

I would not be so much afraid of the VFD, it has long legs, so thermal stress shouldn’t be a problem.
If all pins are properly desoldered and are moving free in their holes, the display should be easily removable.
It will come off with that plastic holder, just make sure to unclip all clips, sometimes plastic feet will be deformed by wave soldering.
Edit: should not be a problem as the display is hand soldered, obviously.

[max.wwwang]

Edit: ... the display is hand soldered, obviously.

I believe this is 100% true, the same for the stand-off LED's.

[max.wwwang]

If you have a scope you can check the problem with the help of the SN75518 datasheet.
I reported on your pic the position of the main pins and their function (at least what I can understand based on the pic).

If it was indeed the DIP-40 that was faulty (and I think so) and you have no experience with a desoldering station, I advise you to gain experience on another faulty PCB first.
Try desoldering many components without breaking them or damaging the PCB before moving on to desoldering a delicate glass VFD!

That's sensible advice, and was exactly my worry -- I didn't want to make it worse! After all the machine is now working except only the display (or one of the two display modes - or a little bit more than that), and it's a beautiful machine!

Yes, I need to look a bit closer at what's going on around the DIP-40, which is almost certainly an SN75518 or similar. Only if I'm sure that the problem is indeed with it, or the DIP-16's, should I start to remove the delicate display module.
[And obviously, I need to source a replacement chip before doing that. It does not seem easily available now (the DIP version)!]

My soldering skill is not too bad, desoldering not as good.

[inse]

Once the VFD is removed you could solder in a socket terminal strip or a IC socket split in half for debugging.

[jchw4]

That's sensible advice, and was exactly my worry -- I didn't want to make it worse! After all the machine is now working except only the display (or one of the two display modes - or a little bit more than that), and it's a beautiful machine!

My soldering skill is not too bad, desoldering not as good.

I would strongly recommend buying a dedicated tool for that purpose.
Hakko FR-401  FR-301 is a bit expensive but it's worth having. If you damage the glass the cost of repair will easily outweight the price of the pump.

There are many alternatives too, you just need to have this kind of tool even if you use it for this project only and sell later.

[max.wwwang]

Once the VFD is removed you could solder in a socket terminal strip or a IC socket split in half for debugging.

Good idea. Thanks.

I would strongly recommend buying a dedicated tool for that purpose.
Hakko FR-401 is a bit expensive but it's worth having. If you damage the glass the cost of repair will easily outweight the price of the pump.

Very true.

There are many alternatives too, you just need to have this kind of tool even if you use it for this project only and sell later.

Thanks. I was considering Hakko 301 and was not aware of the FR401. Yes, buy one, use for one project and sell is also a possibility. Just keep your mind open!

[jchw4]

Thanks. I was considering Hakko 301 and was not aware of the FR401. Yes, buy one, use for one project and sell is also a possibility. Just keep your mind open!

Ouch! You are right. 301. Or any other soldering pump.

[max.wwwang]

Had a little bit more gazing at the thing and got the connectivity of the VFD and the (presumably) SN65518 (mounted upside down). VFD pin # counting from left to right, ignoring the outmost pair. Segment pinout is yet to be figured out.

One observation. There are 16 letters in the display, each with 17 segments. Based on my understanding of the workings of VFD so far, there will need to be at least 33 pins, other than the 2 for filament, if we want to make full use of the segments. There are only 32 pins in the middle. The answer I have is that probably one segment is never used (which is strange)? Another possibility is that, there are two segments that are only visually separate but logically as one combined (i.e. can only be on or off at the same time).

Also probed some legs of the VFD with a scope. Shown is for the 4th leg (for one segment, which is unknown yet). As I said, there are two display modes, let's say mode 1 and 2. 1 is when you don't do anything whereas 2 is when the "display button" is pressed and held. In mode 1 all letters shown are "A"; in mode 2 what's shown is in photo. Another two photos of the scope for the two modes, of leg 4. The scope setting is 0.2ms/div and 1V/div.

Does this give any clue of what's going on and what's going wrong? Any insights are appreciated.

My guess was, since all letters are "A" in mode 1, the segment lines do not even need to have pulses - they can stay either constant high or low. My guess was apparently wrong, because the scope does not give me a horizontal straight line.