kids these days

Yesterday afternoon, trying unsuccessfully to rewire an SC70 analog switch. :(

WDNNS SC70 analog parts.

Cheers

Phil Hobbs

Some of my favorite cmos parts come in the dreadful US8 package. Like the fast triple buffers and a couple of 1ns parts.

When I got the science faculty workshop at Nijmegen to let me use an ECLinPS part in 1996, the specialised mounting gear was a lot smaller and appreciably cheaper. Nobody found either tool set hard to use. I used the Groatmoor gadget myself. The technicians in the science faculty workshop were a bit territorial about their tools, so I didn't get to play with the more modern gear.

If John Larkin is still trying to make do with fine-tipped soldering iron (which can work) he's nuts.

Surface mount packages offer much lower inductance connections. I exploited that back in 1979, but with a t-packaged PNP transistor (the

Ghiggino, K.P., Phillips, D., and Sloman, A.W. "Nanosecond pulse stretcher",Journal of Physics E: Scientific Instruments, 12, 686-687 (1979).

I should have used an emitter-coupled monostable, but the circuit was always a get-it-working bodge, even if my co-authors insisted on publishing it.

Yesterday - and the day before that - and the day before that - and the day before that.......

I've got 180 op-amps to mount on 7 perforated cards (which are horribly prone to shorts between the tracks) - to test out an idea based on analogue computing.

Hand-wiring 180 op-amps isn't going to test yours idea so much as test your wiring skills.

The last time I was involved with hand-wired boards was at EMI Central Research (1976 - 79) where we had a bunch of skilled wiremen, who were still horrifyingly slow. It was quicker to use a skilled draftsman to create a printed circuit layout and get it made, and loaded, only by about a week, but still faster.

And the board - once made - kept on working for longer.

Not all printed circuit layout drafts-persons are equally skilled. We did check the layout before we had it made. Once we were given a layout created for an EMI subsiduary - SE Labs - and decided that it wouldn't work, and handed-edited the layout into one that would (and did) work.

We warned SE labs, but they'd already had 210 board made, and ended up having to hack every last one of their boards to make them work.

Now that we've got printed circuit layout programs (try KiCad - it's free), and fast turn-around printed circuit shops, designing directly onto a printed circuit board makes even more sense, and it's a whole lot faster.

What is the idea you are testing, if I might ask?

It says that 99% solder, and 1% do not. Hmm. If they even know that eevblog exists, they probably are well acquainted with soldering. Self-selected crowd.

A good follow-up questions would be what percentage can program in at least one computer language (like C) or hardware description language (like VHDL).

Joe

Analysing the left and right outputs of a stereo gramophone cartridge into vertical and horizontal vectors. When playing a mono record thare should, in theory, be no vertical component. With a crackly 78, the lumps and holes on one groove wall drive the stylus up and down the slope of the opposite grove wall, creating a vertical component which corresponds to the unwanted crackles in the horizontal signal. (The vertical signal cannot just be subtracted, as that simply regenerates the left or right channel complete with noise)

I built a system many years ago that used a FET to blank out the noise - its gate was controlled from a full-wave rectified vertical signal. It left 'holes' in the signal but by splitting the frequency range into two bands and carefully controlling the various amplitudes and time constants, the intermodulation products could be minimised.

The idea I am testing is to split up the vertical and horizontal audio spectra into ten bands, each an octave wide. Within each band, comparing the vertical and horizontal amplitudes will give the angle of stylus movement. Signals greater than a certain angle from horizontal can be deleted. A gain control in the vertical channel determines the relative amplitude of the two signals and hence the angle at which 'cutoff' occurs. There will also have to be an automated variable threshold control to prevent false synchronous operation caused by the 'pinch effect'.

A CMOS switch is used to momentarily interrupt the signal in each octave band, but a tuned filter is allowed to go on 'ringing' to fill the gap with an approximate replica of the signal utnil the switch can close again. The signal is fed into a state-variable filter through the feedback components (not the usual arrangement); when the CMOS switch opens, it alters the impedances in the feedback circuit and removes some of the damping so as to allow ringing.

The splitting filters give phase changes in the signals which would cause cancellation notches at the overlap frequencies when the signals are re-combined. This is overcome by inverting the signal in alternate bands (similar to a loudspeaker crossover network). The effect of the 'ringing' tuned circuits is an unknown factor but I suspect it will cause an additional phase change that removes the need for alternating polarities. I have allowed for both possibilities in the design.

The mechanical arrangements of stylus size and tracking geometry are critical for correct operation, so I need to monitor the vectors with an oscilloscope. If the idea works, it will be a lot simpler to operate than the system I have been using until now

I check at every step that the stage I am building is working as expected, before going on to the next stage. Power line shorts are the worst and they may not necessarily be in the area of the board you have just been working on (they occur where the solder splashes have landed).