Help debugging Merlin's Signal Generator

[tristanc]

I'm having some issues with my build of Merlin's Sine-wave generator (details here: http://www.valvewizard.co.uk/siggen.html and discussion here: http://www.diystompboxes.com/smfforum/index.php?topic=83930.20). Schematic: http://s81.photobucket.com/user/merlinblencowe/media/Sine%20Wave%20Generator/siggenschem.jpg.html?sort=3&o=6



I'm getting a clipped wave generated... Any ideas on what might be causing the below, or where to start looking?

The voltages on the two ICs are (from pin 1 to 8):

U1
7.3V
3.4V
4V
Gnd
4V
4V
~
7.88V

U2
~
~
~
Gnd
4V
4V
4V
7.88V

The waves on U1:7 and U2:1-3 are given below. The output signal is in blue, the probed pin is in red.

U1:7:

U2:1:

U2:2:

U2:3:


Thanks!

[anotherjim]

It's probably the relationship between LED brightness and LDR resistance that's out of whack. Clipping because the LDR isn't going low enough to keep the amplitude down as signal level increases.
It could be the wrong choice of LDR or LED or a more subtle fault, like a leaky or reversed capacitor.

[tristanc]

Thanks Jim. I had assumed the LED / LDR combo was fine as Merlin's description seems to indicate there would be no oscillation if this was out of whack. But you could be right - that could explain the lovely smooth clipping seen at U2:3 and U1:7.

Or it could be that the top-most electolytic cap in the picture above is round the wrong way compared to below...

http://s81.photobucket.com/user/merlinblencowe/media/Sine%20Wave%20Generator/CIMG7130b.jpg.html

[merlinb]

Check that the LED isn't the wrong way around or that the leads are touching inside the heatshrink. If your LDR has too high resistance then you can reduce R5 until you get a clean sinewave, although this seems like an unlikely problem to me. I think the LED is the culprit -it's not turning on, or is not bright enough or something.

[Rob Strand]

Make sure you have the LED and diodes around the right way.

Sometimes oscillators with level feedback are tricky and frustrating to get working.

If there is a problem with the output level stabilization (D1 through to the LDR), which is a common problem, then you might be better off trying to increase the value of R6 (and/or increase R9 and/or decrease R7).

A guaranteed way to force it to work is to reduce R2, you will need very fine adjustments, so use a pot.  While ultimately you might *have* to use this method it is not a good approach unless you can prove the level control circuit (D1 through to the LDR) is actually working first.

Forgot to mention, sometimes the oscillator will work better at low or high frequency so play with the frequency pot to see if it changes things.

[tristanc]

Merlin, Rob, thanks for the pointers.

This will test my desoldering skills to the limit! I, stupidly, assumed it was working fine based on hearing a nice tone coming out of the amp. So went ahead and soldered the 'on' LED which basically anchors the PCB in the enclosure due to the panel mount used...

Lesson learnt.

[PRR]

> stupidly, assumed it was working fine based on hearing a nice tone coming out of the amp.

Not "stupid".

Audio-workers use sine-waves because it simplifies the math and hides certain problems we tend to study other ways.

But a pure sine wave is very nice, even "too nice". Total lack of harmonics is not natural, and is a narrow slice of wide-band audio.

As a "does this amplifier work?" tool, your "ugly" bent wave may be quite useful. It has musical overtones. As you turn tone-knobs, the sound will change.

It is no good for THD analysis, but you don't have a THD meter.

It is awkward for reading the clipping point of an amplifier, since it is already semi-clipped. However it is squinty to read clipping on a sine because of the gentle curve over the tips. A triangle wave is much better for that. (And easier to generate clean and vari-frequency.)

Fix it just because. But you just might want a switch to go back to "im-pure sine" as a general bench tester.

[tristanc]

Thanks for the kind words.

This was very much a learning experience for me - having jumped in to amps (with extensive help from Merlin along the way!) I bypassed the more intricate effects-building step and PCBs altogether.

The secondary motivation was to test a PCB fabrication house (ragworm.eu) and also my new soldering iron; another learning experience, going from a cheap 40W all-in-one iron to a temperature-controlled station. What a difference!

You'll laugh, but the whole exercise is redundant anyhow - I managed to salvage a soon-to-be-disposed Thandar signal generator (to join my similarly-acquired Kenwood scope) from the uni here in Oxford. But where's the fun in that? I'll persevere with getting this build working.

[PRR]

> soon-to-be-disposed Thandar signal generator

That's a good score.

Un-bolt the BNC jacks and put in 1/4" jacks (if you can). BNC is fussy, and unreliable if you don't keep GOOD connectors of proper impedance. 1/4" is fine for any audio.

While it has Sine, you will find the top of the sine always has a point or a flat-spot (often still with a point). They get it by beating on a triangle wave. That sharp corner never goes away.

For 99% of audio work, that is fine. You learn what your Fun-Gen output looks like, and you know when your test-object is not reproducing exactly. The overtones are not large enough to mess-up EQ measurements.

But that does give you a reason to trace-out Merlin's pure Sine generator and make it right. See/hear what a PURE sine sounds like. Maybe Oxford will throw-out a THD meter or spectrum analyzer next.

[Rob Strand]

You should find a service manual for the Thandar signal generator.   These things come in handy.   It's good to look at what's under the hood in these things.

[tristanc]

Check that the LED isn't the wrong way around or that the leads are touching inside the heatshrink. ... I think the LED is the culprit -it's not turning on, or is not bright enough or something.

I think you were right! Finally got round to bug fixing this - removed the LED / LDR (badly - pad lifted), patched in replacements and got a lovely sine wave. It is most likely I had the LED back to front - a case of following the, slightly obscured, silk screen rather than checking with the schematic.

Thanks again for the help. On to build the tone burst generator!