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

[max.wwwang]

By redrawing the symbol and re-arranging ..., it becomes much clearer. The bottom two gates (U31B & D) are simply an RS latch. The only difference is that the "prohibited condition" for a normal RS latch is not a problem here because there is no complement of it as another output (so no logical problem). I've included the truth table on the LHS of it.

So signals from 4 sources come in and become the 4-bit preset number (c0) of the counter (counting up), which, once preset, will count up until it's full (reaching 16, i.e. after 16-c0 clock pulses), when carry-out will be set (and all the bits reset). Since there is a bar on the top of C_OUT, which I think means it's the invert of C_OUT, so at that moment C_OUT' (I'm using "'" for a bar) will become L.

The way the magic initial count c0 is set is really difficult to decipher. Just to make this a little closer, I've figured out at least what the input bits are. For this I need to start from the genesis of the clock signal.

It starts from a square wave generator comprising a UJT, the frequency of which can be adjusted at run time through human interface (pedal action). This pulse signal, along with an ENABLE signal (as I understand it), feeds into a 4040 12-bit counter --- only 7 bits are used as a series of pulse signals for output. The lowest bit Q0 has 1/2 of the frequency of the (enabled) input signal (because it flips, from L to H or H to L, once in one cycle of the input). Next one Q1 has 1/2 of the frequency of Q0, and so on, until Q6 (the 7th bit).

These 7 pulses signals go through a series of gates (AND, OR, etc.). When reaching the 4 input ports (bits) of 4516, I've figured out the formulas for all of them (noted on the top). I've also visualised these series of signals and the result of the formulas, as well as the value of the 4-bit number, in Excel. So you can get an idea of --- suppose RESET is L, i.e. at the moment of presetting the initial count --- the likelihood of c0 being non-zero (highlighted in green) and its value (be it zero or not). (Also shown is the result of c0 for another 4516, 2-3C.)

Whichever of these c0 values is set (when the CLOCK signal is H), it will count up upon every clock pulse, and --- as soon as the count reads 16 sends out a carry-out signal (inverted) --- unless RESET becomes H, at which point it will start over again.

(After this 4040, the Q1 signal feeds into the LCK pin of other logic chips, so I'm now naming it "CLOCK".)

While this seems closer, it is still a huge maze to navigate.

[max.wwwang]

While doing this vault-cracking exercise, another progress (or rather realisation) has been made.

Last time when a bad guy was caught --- that was a shorted hall sensor --- there was a big oversight of me. 

It was a "peripheral" fault. Indeed, it would be, if it was only the sensor not working. The fact is, the sensor is not only not working, it has brought down the whole +11V power rail. One beautiful thing is that, the 11V power source is able to handle short circuit of its loads without any drama. But I should have realised this was not a peripheral fault anymore.

So, now I suspect it was the ONLY fault, because the machine once worked well, but only without any sign of life (except for the motor, which is fed directly from the mains) after that. There was even no response to the manual back-tack (BT) switch or the knee switch, which normally would respond regardless of the running status of needle. With the +11V rail down, it's highly likely they are down, because it's unlikely the states of them do not go through ANY of the gates (which all rely on the +11V rail).

Having realised this, simulation of closing these two switches is done (with hall sensor module disconnected, i.e. +11 V rail up), using dummy loads in lieu of the solenoids. +37V was detected for both scenarios (this is different from when the machine was down). So, very likely the machine will be back and running if I put everything together, and will be in full working order as soon as the faulty hall sensor is replaced.

Of course, the lurking crack fault of one of the boards (now without a trace) will probably still be a problem at certain point.

I'm now not hasting to put it back though, when it's wieldy and conveniently laid out on the desk --- so I can do a bit more probing and safe-cracking (it's still a looong way to go). Fingers crossed, hopefully this will not cause any new problems!

[max.wwwang]

Another part of the big puzzle I've partly figured out but hope I can get any input from here, is a part of the synchroniser, which I believe is the head unit speed sensor. Along with the three hall sensors (one is gone), there are other three wires going to a small board close to the centre (a hole). In side the hole, rotating with the needle mechanism, is a donut-shaped magnet. In the static condition (nothing moving), the resistance measurements between the wires resemble two 0.5k Ohm resistors between Black/Brown and Brown/Grey. These are drawn in the circuit (lower half). In whatever way, I believe the rotation of the magnet will induce a (net) potential in way of where a battery is shown. This induced potential will then be sensed by the V comparator and, depending on whether the speed is too high or too low, pulses are output on the top right corner through the gates.

My question ---

Does anyone have an idea of what exactly this speed sensing mechanism is (and what the small PCB is doing), assuming my assumptions are correct?

(In the photo "Magnets", the other magnet than the donut-shaped one, is of course the one that works with the hall sensors.)

[max.wwwang]

There is another thing about this circuit that baffles me. I've drawn up the timing diagram of the signals (outputs and transistor states). Among the signals, only N2-6B-4-sync and N2-6B-11-sync are visible to external. If my timing diagram is correct, their responses to the too high or too low speed are the same --- either rising to high (-4) or falling to low (-11), then return (to low/high). So how can the user of the signals be able to tell whether the motor needs to be driven faster or braked down?

[max.wwwang]

Refer to message #127 and photos there, finally an update. After receiving the replacement hall sensor through snail mail from AliExpress and having it replaced, I put things together and had a test. Everything works now!

[max.wwwang]

From this 1993 NZ code of practice (NEW ZEALAND ELECTRICAL CODE OF PRACTICE For POWER SYSTEMS EARTHING) now still available on WorkSafe website (I have not got into the detail of it yet), and a post from what appears to be a NZ hardcore expert (TT Earthing Systems - Interest by New Zealand ), it appears the system in NZ is (or was) MEN (Multiple Earthed Neutral), but was considering a transition to the TT system. From what you are saying, it seems like this transition has been done at certain point?

No, NZ and Australia are still using the MEN (Multiple Earthed Neutral) system, which -- according to the Electricity (Safety) Regulations 2010 -- "is a New Zealand variant of the internationally defined TNC system". The standard AS/NZS 3000:2007 Electrical installations (known as the Australian-New Zealand Wiring Rules) also includes a great deal of detail of the MEN system.