Brother industrial sewing machine repair / reverse engineering (repair success)

[tautech]

Also, to make this small schematic professional , I guess the +11V rail should use a different symbol to indicate it's not an input from an external PS but rather the output of the +37V source. What's the appropriate way in EDA?

Rev 2 was fine with 2 Gnds, just fine. While you think you've made it neater with Rev3 it's actually harder to follow. 

For simplicity to follow use NET labels whenever necessary in order not to cross any wiring.

+11V is a power NET and it don't matter a damn if it's devived locally or remotely. You might add a text box saying "From Sheet 1" or something.
If you want it better designated use a different symbol for the supply and load parts of power NETS.
Your Kicad surely has different symbol options other than the arrow which might better indicate data/analog signal flow.
Altium, the one I use uses a T with the NET voltage above it but we do have choices in even the selection of power NETS. Signal path NETS are different and like power NETS can be named.


Which brings me to another point I forgot to make earlier....try to lay out a schematic like this:
Signal flow from left to right. Positive NETS at the top and GND NETS at the bottom.....although they can be placed within the design.
It's also useful to drop a text label or 3 amongst schematics with a single or 2 word description of that part of the circuit. Eg, PSU, Reg, Timer, Vref and so on. There's no need to go overboard but later it can be useful for debugging.


Do carry on...........

[max.wwwang]

Rev 2 was fine with 2 Gnds, just fine. While you think you've made it neater with Rev3 it's actually harder to follow. 

I'm afraid I have to agree. It seems tighter/neater but clarity is lost. That's a good point, clarity and easy-to-follow has higher priority. Here we go, rev 4, which I believe is even better than rev 2.
Clarity is more important than formal neatness.

For simplicity to follow use NET labels whenever necessary in order not to cross any wiring.

True, this is closely related to clarity because crossing but not connected lines increase cognitive load --- use NET labels to minimise crossing lines.

+11V is a power NET and it don't matter a damn if it's devived locally or remotely. You might add a text box saying "From Sheet 1" or something.
If you want it better designated use a different symbol for the supply and load parts of power NETS.
Your Kicad surely has different symbol options other than the arrow which might better indicate data/analog signal flow.
Altium, the one I use uses a T with the NET voltage above it but we do have choices in even the selection of power NETS. Signal path NETS are different and like power NETS can be named.

The point here is --- include right amount of comment/annotation.

Which brings me to another point I forgot to make earlier....try to lay out a schematic like this:
Signal flow from left to right. Positive NETS at the top and GND NETS at the bottom.....although they can be placed within the design.
It's also useful to drop a text label or 3 amongst schematics with a single or 2 word description of that part of the circuit. Eg, PSU, Reg, Timer, Vref and so on. There's no need to go overboard but later it can be useful for debugging.

Signal/power go from left to right, top to bottom --- be intuitive.

Do carry on...........

Will do. 

[max.wwwang]

Known broken parts can be ordered early because it take a long time to arrive. So the broken hall sensor is taken out of the plastic housing. It has such marking on it: N3006 (first line) / SK046 (second line).

Now one (peripheral) obstacle to overcome --- I could not find any place online where it is available. 

But this should not be an unclimbable mountain. I only need to find one that has a TO-92 form factor, the same polarity, and accepts high supply voltage (11V plus some margin).

[max.wwwang]

Was wondering where the square wave signal comes from. I think I got the answer now. It involves a UJT (uni junction transistor), which I didn't hear of before. It's at the top left corner of the schematic. This part of the signal generator is almost identical to figure 8-c in the datasheet, which is different from the more commonly seen (that is 8-a), but is specifically recommended for this transistor.

Its output signal goes to an inverter, as one stage of buffer, then through another AND gate (effectively another stage of buffer). Then there is a stable square wave on check point 2-CP2. Pot R302 is for the adjustment of the frequency.

I think the purpose of R306 and R307 (RHS of the inverter) is to set a mid-point (about 6V) for the signal before the analogue to digital conversion.

Some questions for those who are experienced and have interest in analysing (or appreciating) this circuit.

1. What's the purpose of R303 (area B1, RHS of 2-CP2)? I know it does not hurt. But is there any significance here? There is no such thing deep in the web of the gates.

2. What's the magic going on here in the way the A798 chip (top right corner highlighted) is wired up? This IC consists of two PNP transistors; they are hooked up in an intriguing way.

Will do a bit more poking around and see if all is working.

[james_s]

UJTs are cool, they were once common, used for flashing lamps and generating tones and such back in the 70s. You don't really see them much these days.

[tautech]

UJTs are cool, they were once common, used for flashing lamps and generating tones and such back in the 70s. You don't really see them much these days.

I've only seen them in 2 applications, a now elderly electric fence controller (1 pulse/s) and in a SCR phase control in a 180A single phase welder.
The welder used a pair and a RE schematic is attached.

BTW, Max, that schematic was drawn using MS Office Visio and saved as a Doc in which one of its libraries was quite useful for electronics.

[max.wwwang]

UJTs are cool, they were once common, used for flashing lamps and generating tones and such back in the 70s. You don't really see them much these days.

True, it seems so easy to use, and has plenty of applications. But strangely, it's said to be obsolete now!

[max.wwwang]

BTW, Max, that schematic was drawn using MS Office Visio and saved as a Doc in which one of its libraries was quite useful for electronics.

Thanks. That's one of its common uses --- phase control.

[tautech]

BTW, Max, that schematic was drawn using MS Office Visio and saved as a Doc in which one of its libraries was quite useful for electronics.

Thanks. That's one of its common uses --- phase control.

Very interesting thing to get the SCR's balanced and when I was taught by a EE Prof how to float a scope. 
BTW, that welder was west Auckland designed and built.

Triacs are much simpler to use with a controller IC and another welder I fixed some years back used an Alternistor, a specialised form of Triac.

[max.wwwang]

Very interesting thing to get the SCR's balanced and when I was taught by a EE Prof how to float a scope. 
BTW, that welder was west Auckland designed and built.

Triacs are much simpler to use with a controller IC and another welder I fixed some years back used an Alternistor, a specialised form of Triac.

Will have a closer look and think. Alternistor is again new to me. How many new things not known to me are out there?! 

[tautech]

 
Chill out Max, we can't know everything ! I certainly don't but knowing there is more to learn is what's so exciting about electronics.

[james_s]

I've only seen them in 2 applications, a now elderly electric fence controller (1 pulse/s) and in a SCR phase control in a 180A single phase welder.
The welder used a pair and a RE schematic is attached.

BTW, Max, that schematic was drawn using MS Office Visio and saved as a Doc in which one of its libraries was quite useful for electronics.

I have a flashing red beacon that came off of an old forklift that uses one. I've seen them in a handful of old hobby circuits too, I think the 555 took over a lot of their applications when that came out.

[max.wwwang]

… I think the 555 took over a lot of their applications when that came out.

That probably explains it. I’ve seen a lot of talk about 555 ages ago.

[tautech]

I've only seen them in 2 applications, a now elderly electric fence controller (1 pulse/s) and in a SCR phase control in a 180A single phase welder.
The welder used a pair and a RE schematic is attached.

BTW, Max, that schematic was drawn using MS Office Visio and saved as a Doc in which one of its libraries was quite useful for electronics.

I have a flashing red beacon that came off of an old forklift that uses one. I've seen them in a handful of old hobby circuits too, I think the 555 took over a lot of their applications when that came out.

The welder we brought new that had UJT's was an 80's design but required pulse transformers for the phase control circuitry isolation and sufficient drive via the pulse transformer to drive the SCR gates.
Dunno but guessing 555 would not have the drive to do that.

[TimFox]

I think the 2N2646 unijunction and its equivalents are easier to interface to an SCR than would be a 555, but the 555 is far more flexible for general-purpose timing and oscillation.
I believe that when the 555 came out, some scoffed that it was superfluous since you could do the same thing with a handful of components (instead of one).

[tautech]

I think the 2N2646 unijunction and its equivalents are easier to interface to an SCR than would be a 555, but the 555 is far more flexible for general-purpose timing and oscillation.
I believe that when the 555 came out, some scoffed that it was superfluous since you could do the same thing with a handful of components (instead of one).

Yes at one time pulse transformers were quite common however Triacs removed much need for them.

A pair of reversed bias SCR's and their phase control Gate waveforms are a wonderful thing to see.
That welder ran so much nicer when the SCR phase control was properly balanced.

[max.wwwang]

Yes at one time pulse transformers were quite common however Triacs removed much need for them.

A pair of reversed bias SCR's and their phase control Gate waveforms are a wonderful thing to see.
That welder ran so much nicer when the SCR phase control was properly balanced.

As far as I know, TRIACS allow current both ways whereas SCR only one way. But don’t quite get why the need for transformers is removed by them?

True, probing with a scope and actually seeing what’s going on out there is fascinating.

On your earlier point on balance of two UJTs, sure, even if it was not nicer from the welder perspective, it still feels better with a balanced and symmetrical wave form above and below 0V.

[tautech]

Yes at one time pulse transformers were quite common however Triacs removed much need for them.

A pair of reversed bias SCR's and their phase control Gate waveforms are a wonderful thing to see.
That welder ran so much nicer when the SCR phase control was properly balanced.

As far as I know, TRIACS allow current both ways whereas SCR only one way. But don’t quite get why the need for transformers is removed by them?

True, probing with a scope and actually seeing what’s going on out there is fascinating.

On your earlier point on balance of two UJTs, sure, even if it was not nicer from the welder perspective, it still feels better with a balanced and symmetrical wave form above and below 0V.

Unlike a SCR reverse biased pair for AC phase control a Triac gate is zero crossing referenced I believe which negates the need for a pulse transformer.
See these devices and datasheets.
https://www.hestore.hu/prod_getfile.php?id=5680
No current sense required for a welder if the set point is correctly set. (max rated current draw)
A high current Alternistor in substantial airflow on a decent heatsink completes the phase control:
https://ie.rs-online.com/web/p/triacs/1106594

This pair using U2008B was just too easy for a repair I did on a mates 180A welder that had been parked atop some obstacle shoving a grille into a bottom mounted fan stopping it which fried the Alternistor. The previous DIP16 control IC was no longer available so a U2008D was carefully shoehorned into the same PCB footprint with a couple of cut traces and wire links for the welder to give further decades of use.
I'm sure there are even better solutions available today.

[max.wwwang]

Unlike a SCR reverse biased pair for AC phase control a Triac gate is zero crossing referenced I believe which negates the need for a pulse transformer.
See these devices and datasheets.
https://www.hestore.hu/prod_getfile.php?id=5680
No current sense required for a welder if the set point is correctly set. (max rated current draw)
A high current Alternistor in substantial airflow on a decent heatsink completes the phase control:
https://ie.rs-online.com/web/p/triacs/1106594

This pair using U2008B was just too easy for a repair I did on a mates 180A welder that had been parked atop some obstacle shoving a grille into a bottom mounted fan stopping it which fried the Alternistor. The previous DIP16 control IC was no longer available so a U2008D was carefully shoehorned into the same PCB footprint with a couple of cut traces and wire links for the welder to give further decades of use.
I'm sure there are even better solutions available today.

So it seems Alternistor is only another name for "snubberless TRIAC". Good to know. 

Substantial progress has been made on two fronts --- chasing down the culprit (though the battel is not yet conclusive) and deciphering the working of the entire circuit and the machine. The aims now are two-fold: 1) to get the thing back and running; and --- more importantly --- to fully understand how it works. I will not be too disappointed if the first goal ends up a failure now.

I said before this is a beautiful machine. It now looks even more beautiful than it was before. Without an MCU, its program is literally hard wired so is completely transparent. No secret is hidden whatsoever.

Details to follow...

[max.wwwang]

A very strange thing that I've never come across before.

Accidentally I think I found one fault on one board just by holding it in hands, turning around and closely inspecting (because I need to work out the rat nest across the components).

In doing this, I noticed a very light but distinct clicking (metallic) sounds when turning the board, definitely because of the flexing of it due to its weight (this is when it's held in only one hand), and due to the (very light) forces imposed on it by my hands. Because there are no moving parts on the entirely board, there must be a crack somewhere, either in a trace or a component, or somewhere else.

Strange enough, having done whatever I can do (including with a magnifying lens), closely combing through every square mm of each side of the board, I just cannot find where exactly the crack is. By holding one part each time and repeat, I think I located where the problem probably sits --- inside a resistor. There is no sound when it's lightly pressed (or touched) when turning. My explanation is that the sound is either eliminated by pressing it (causing the gap to close) or the sound it largely dampened. That said I'm still unable to observe any relative movement between parts of it (such as between a pin and its body) when bending the board slightly, nothing whatsoever. So I cannot be 100% sure.

What else did I do with it? Tried to measure the resistance of it with too alligator clips on its two ends --- the reading is correct: 6.8k. What's more interesting is that, after doing this, I cannot get the same sound again!

The only explanation I can come up with is: the use of alligator clips on its two ends -- due to the weight of the clips and the wires --- its two ends have been bent slightly closing whatever gap there existed before.

Because this fault is so evasive, if I take it off, by applying heat and force, it's very likely it looks perfectly good.

Not in a haste to take it off yet because first I'm still not very sure about this and second, there is much more test and investigation that need to be done. Even if it's faulty, there are probably more problems somewhere because its role here is still quite peripheral --- it's related to another solenoid not related to the motor.

In the photo, it's the smaller resistor starting with a blue band at LHS of a transistor.

[tautech]

A very strange thing that I've never come across before.

Accidentally I think I found one fault on one board just by holding it in hands, turning around and closely inspecting (because I need to work out the rat nest across the components).

In doing this, I noticed a very light but distinct clicking (metallic) sounds when turning the board, definitely because of the flexing of it due to its weight (this is when it's held in only one hand), and due to the (very light) forces imposed on it by my hands. Because there are no moving parts on the entirely board, there must be a crack somewhere, either in a trace or a component, or somewhere else.

Yup, an old trick when assessing a faulty PCB is to bend and flex it while listening for the telltale clicks or a crack which while you've identified there is one, finding it can be an entirely different matter !

We could of course go over it with a DMM and while components are in circuit and can give misleading measurements any that measure higher than bands/markings suggest must first be removed and checked.

I had a PCB some years back from a sorta modern welder that had been carelessly dropped/tipped in a workshop and then refused to work.
Initial diagnosis was easy, a small PCB mount transformer that powered the control PCB despite being cable/zip tied down had dragged a PCB pin from its housing breaking the primary winding deep in the potted transformer case a repair was not possible however it was a common voltage, pinout and construction that another was sourced from RS for a straightforward replacement resulting in a successful repair......well so I thought !
When it came back I found a cracked solder joint on a SMD passive and a reflow remedied it again......so I thought !
Returned again with an intermittent.....goes sometimes, sometimes not.... and based on previous finding it was obvious the PCB had suffered severe flexing when the initial damage was done to the transformer that some some substantial Sherlock Holmes'ing was required to beat it.

This PCB fair crackled and clicked when flexed and reflowing several joints shut it up some but never completely fixed it.
Even careful study with a good magnifying headset couldn't identify further cracks but it still clicked and cracked when flexed.
As it was only some 150x150mm but mostly SMD I decided to reflow each and every solder joint by hand to near MIL spec...you know perfect fillets and all that.

Never saw it back again ! 

[max.wwwang]

Ok, so I was not alone! With some little experience before, now I got a good appreciation of the magnitude of impact of the seemingly very trivial forces and fluctuation of forces, such as  due to weight, heat, vibration, handling etc etc.

Yes, there was definitely a good deal of Sherlock Holmes'ing going on. I was reluctant to take it straight off not only because I don't want to break things that were not faulty but also really want to clench the dirty hands when they are misbehaving! I was irritated if not insulted.

[max.wwwang]

With strenuous effort, I think I've transformed all the PCBs into schematics, except for one module, which can be treated as a simple black box --- its behaviour is well defined.

It seems more strenuous effort is needed to detangle this huge net. There appear no obvious winning strategies other than re-arranging, transforming, labelling signals with meaningful names, isolating parts of the circuity that are relatively independent or have relatively clear functions, keeping notes, etc etc.

Take this part as an example, although it's already a hugely reduced/simplified application of the chip (with Aout/Bout/Cout/Dout all NC), it's still insanely complex --- perhaps too complex to decipher by human brutal force.

The datasheet of TC4516BP is very vague to me (perhaps crystal clear to experienced). But at least I've figured it out (the working of the chip, not this circuit!) by staring into the timing diagram.

[srb1954]

The datasheet of TC4516BP is very vague to me (perhaps crystal clear to experienced). But at least I've figured it out (the working of the chip, not this circuit!) by staring into the timing diagram.

It will be easier to understand the operation of the chip in this circuit if you redraw the TC4516 symbol to put its inputs on the LHS and outputs on the RHS. Group the signals together with their associated control functions wherever possible. e.g. the PRE/EN signal is associated with the parallel inputs Ain - Din.

If you still can't decipher the operation of the chip you may get a better description or diagram from another manufacturer's data sheet - the 4516 is a standard CMOS chip available from many other manufacturers apart from Toshiba.

[max.wwwang]

The datasheet of TC4516BP is very vague to me (perhaps crystal clear to experienced). But at least I've figured it out (the working of the chip, not this circuit!) by staring into the timing diagram.

It will be easier to understand the operation of the chip in this circuit if you redraw the TC4516 symbol to put its inputs on the LHS and outputs on the RHS. Group the signals together with their associated control functions wherever possible. e.g. the PRE/EN signal is associated with the parallel inputs Ain - Din.

If you still can't decipher the operation of the chip you may get a better description or diagram from another manufacturer's data sheet - the 4516 is a standard CMOS chip available from many other manufacturers apart from Toshiba.

Thanks. The difficulty at first is that it's not explicitly stated D/C/B/A are formed into a 4-bit counter. But this was figured out by carefully examining the timing diagram. 

Yep, redrawing the symbol more organisedly is a good idea!