Wednesday, July 26, 2017

Flippin' Flopped Flip-Flop

There's something about blogging and cafe's isn't there? Well, I am writing this from a cafe in Paris, so this blog now has a certian je ne sais quoi it was lacking before I think. But really, je ne sais quoi. Whilst I'm away from home it's a good time to catch up on progress, of which there has been plenty since the last post.

Last time I was basking in the glow from having got one single note to play. Well I wasn't going to stop there.

Striking a Chord.

I decided to buy a few 4-channel opto-couplers from Maplin's and just see what would happen if I got a whole octave working at once.

Well, this was pretty exciting!

The thrill of playing actual chords from the SS-30on a keyboard for the first time in what must be 20 years was quite something for me. I went to bed late but very happy that night, but there was something bothering me at the back of my mind...

 Feeling Jitttery

In the post before last I mentioned that although the master clock on G3 was back in business I had some concerning waveforms. When I looked at the test points I saw that the waves we jittery. Jitter is a term from digital electronics to describe variation in timing between the signal and a clock, so I'm misusing it here but the effect was that the output from the octave divider IC was not stable. My ears told me everything was fine so I didn't immediately know what to do.

Just before connected the K2 board up for my coupler board test above I ran through each octave to decide which to use. It was then that my ears detected a jittery sounding problem with the top octave from K4. No all the keys but about half had a noise that sounded digital and related to the clocks.  After the triumph of playing a chord I came back to investigate this. Initially I was confused because it was coming and going. Some keys which had the problem seemed okay. Then I noticed that the vibrato was effecting the noise so I turned that off and played with the pitch and detune. Now I could make the problem come and go with the detune. Each tone was effected differently so I surmised that this must be related to the interaction between the two oscillators clocks for each tone and that it was specifically effecting the top octave.

Getting busy with the oscilloscope I soon saw that in fact there was no clock at all on one of the G3 board octave dividers. 
Oh no! could this be a dead YM25400? If it was I was in for some serious heartache. Thankfully tracing the input to that IC showed that the problem was further back. So, the jittery signal I'd seen and heard was because one half of G3 wasn't getting the second oscillator.

Divide and Concur

The problem actually stemmed from the flip-flop divider chip on G3 which divides the 500KHz master clock oscillators from G3 and G4 in half, to 250KHz. The octave dividers provide a 1/4 of the input clock output, as well as the octave division for each note. The dividers on G4 provide a 125KHz clock to G2 and the G3 divider 62.5KHz to G1. Hence the input for G3, which must be half of G4, comes from the flip-flop dividers.

Flip-flop - TC4027BP

The TC4027P dual JK flip-flop IC is thus fed inputs directly from the master clock oscillators of G3 and G4.

Master Clock Oscillators from G3 (top) and G4 (bottom)

I could clearly see that whilst the inputs to both of the two flip-flops was okay, the only output was the one with the G4 clock input.

G4 Master Clock Oscillator input (top) and flip-flop output (bottom)

Even after disconnecting the wire running from the output on one side of G3 to the other there was still no signal to speak of. Reducing the scale showed something was there - all but nothing compared to the other output.

Flip-flop input 2 (top) and output 2 (bottom)

The 4027 was half dead.

The fact that this chip is adjacent to the blown transistor I had just replaced and uses the same -15V supply cannot be a coincidence. 

Flip-Flop Swap-Shop

After being slightly disconcerted that none of the main electronics shops sell these devices anymore (I assumed these parts would be around for ever) I found some at a reasonble price on eBay and they were posted the next day.

A few nights later I setteled into my comfort zone (I must have replaced dozens of DIPs back in my test and service tech days) and replaced the part with alacrity.

If you're not familar with changing DIP ICs the trick is to snip off all the legs before you go near the soldering iron. That is unless you have a pot of solder on the heat nearby, and then you can (after protecting the neigbouring parts and being liberal with your flux) place it pin-side down in the bath and pluck the IC out as soon as the solder has melted.

Faulty TC4027BP with legs snipped off.
 Then it's time to deloder each leg in turn. Doing it this way avoids putting additional strain on the pads as you try and get each one free. Ideally you have through-hole plating so that you can desolder from the top, where you pull the legs through. In the case of the SS30 there is no pad at all on the top side for the ICs so I had to do it in two stages: first on the bottom side I heated the solder on each pad and desoldered with wick (suction is liable to pull the pads of); then I used the iron to push each leg through to the top-side. As there is no solder through the hole they usually fall through without too much trouble.

Top-side of G3 board with IC removed

For most I have to do some more desoldering after pushing the leg through and then turn the board over and heat the leg from the top using tweezers to remove the last obstinate couple. Finally I went back to the bottom-side and used the wick again to remove the remaining solder. The main concern with that is to make sure that the legs of the new chip would slide cleanly through. With plated through-holes this can be a real pain to do cleanly.
 Even though I was careful I still lost a part of one of the pads though. This is typical for a pad with no track, as in this case. Because the pin is linked to the adjacent pin it still held enough to solder though.

Bottom-side of G3 with IC pads cleaned.
The new chip was a very close match to the original Toshiba part.

Old part (left) new part (right)

New part soldered in place

Soon everything was back to normal and the jitter was gone as all clocks were present and correct.

Both outputs from the flip-flops

No comments :