Elna Stella TSP Air Electronic

Refurbishing the electronics.

(Revision 2, component value information updated)

This story begins when I got a call to say that one of Kate's favourite machines, the Elna Stella TSP Air Electronic, had been knocked over at school and had started to misbehave. The report was that it would continue to sew even after foot pressure had been removed from the pneumatic pedal.  I was going to have to take it apart.

With trepidation I removed the power and control connections then unscrewed the four screws holding the end of the case on. It seemed to be sprung, which made me nervous, fearing a sudden twang! However, when it came off I was agreeably surprised. The board was held to the plastic case end moulding, and connection to the motor and lamp was by four coil-sprung connectors, making disassembly simple and reliable. This machine was designed for ease of maintenance.

Looking in to the case  Board mounted in case end  Board from above showing failed components

I found four components with clear signs of breakdown, two cooked resistors and two bulged capacitors. Both resistors were so discoloured that their values could not be read from the colour coded rings. One was open-circuit, had obviously burned out. The other still held its value at 2.7K Ohms.

It isn't obvious that these component failures had caused the running-on, but they still needed to be replaced before they caused a worse failure. The running-on could have been caused by the speed controller device sticking slightly. As this is a pneumatic bellows device, I was loth to apply any contact cleaner or spray treatment near it, as the bellows will be hard to replace, if not impossible.

I detached the board from the case, drawing a diagram of the connections for later reassembly, and started with component replacement. New discrete components were bought from Maplin Electronics, who didn't have all the values I needed in 2 Watt resistors, so I had to build up the replacement for the 2K7 component from two smaller values. I chose 2 Watt resistors to reduce the likelyhood of them overheating again. Capacitors were easier to source, being fairly standard metallised paper non-polarised components.

Old and new components

Removing the old components was tricky as their leads had been bent over before soldering, and I couldn't find my desoldering pump. I bent the leads back with cutters while heating the solder pads, careful to avoid damage to the board.  I have been reminded that the resistor leads could have been clipped on the component side and the stubs removed.  The capacitor leads were not accessible due to the packaging.

Bending the leads up - not good practice  The board with all the old bits removed

Once all the old components were off, the new ones had to be applied. All three of the old paper capacitors were replaced, as this type does decay with age. The new 0.1 F capacitor needed its leads extending to fit the holes in the circuit board, and the 2K7 resistor was made up by joining a 2K2 and a 470R resistor in series.

Board with replacement components  Refurbished board back in the case

Initial placement of this combination proved to be too close to the motor when the board was reassembled into the case, and it had to be reshaped as below. This is the board as it now is in the machine.

Reworked resistor pair  Reworked board in place

In order to make a decent guess at the burned-out resistor I needed to reverse-engineer the circuit diagram as an original has not yet turned up at any of our sewing-machine contacts.

If you know of one, please get in touch!

Solder-side was compared to component-side and the connections drawn in. Then the net had to be reorganised into something like a recognisable circuit diagram. My effort is shown below.


This looks to be a triac speed control driving a universal motor in DC mode. The triac is comparatively elderly, a Philips BT138 500E, and needs a snubber circuit, and it was the snubber resistor which had burned out.

I've had a message from an Experienced Sewing Machine Guy giving the colours of the snubber resistor on an intact board, orange-white-red-gold, translating to a value of 3K9.  Thanks, Ray!