0

I am doing a speaker driver circuit. When I connect the oscilloscope probe to the output (emitters of npn/pnp complementaries) to where also the speaker is connected; the sound of speaker and the waveforms on the oscilloscope is satisfying. However, when I disconnect the osc probe and leave the speaker alone, than the sound gets much more noisy.

Why does that happen so?

muyustan
  • 2,046
  • 20
  • 54
  • 1
    Stray leakage of audio noise from SMPS is shunt by probe earth ground to 0Vdc. Either reduce CM leakage or connect 10nF RF cap to earth ground... or wire jumper to earth ground like scope ... but in some cases may causes ground loop hum with other power supplies – Tony Stewart EE75 Dec 25 '19 at 02:04
  • @TonyStewartSunnyskyguyEE75 wire to earth ground? Then speaker will always be at gnd level ? – muyustan Dec 25 '19 at 03:27
  • What do you think? – Tony Stewart EE75 Dec 25 '19 at 03:28
  • @TonyStewartSunnyskyguyEE75 did you suggest connecting output to gnd or didn't I understand you? What is the point if the speaker(load) is shorted. – muyustan Dec 25 '19 at 03:32
  • Hmm err no I did not say that. Where did you ground scope probe? 0Vdc? – Tony Stewart EE75 Dec 25 '19 at 03:36
  • At the common ground of the circuit, which should be at 0Vdc w.r.t other voltages – muyustan Dec 25 '19 at 03:40
  • It seems your SMPS has too much stray CM noise that interferes – Tony Stewart EE75 Dec 25 '19 at 03:43
  • 2
    An amplifier might oscillate (often at a frequency above the audible range). What you hear can sound like noise. Perhaps grounding it via 'scope ground kills the oscillation...and the noise. That'd be frustrating, but promising because it shouldn't take much more work to kill the oscillation completely. – glen_geek Dec 25 '19 at 03:59
  • @muyustan Are you still working on that prior schematic I saw, and suggested a minor change? Or is this something different? – jonk Dec 25 '19 at 04:48
  • @jonk I did not forget you, I have done changes you mentioned and it is a lot better now. I was just waiting to get a final result and then notify you. So, yes still the same design. – muyustan Dec 25 '19 at 05:01
  • _"when I disconnect the osc probe **and leave the speaker alone**, than the sound gets much more noisy."_ - do you mean with the speaker disconnected? Can you show us the waveform? Please show a schematic of your amplifier. – Bruce Abbott Dec 25 '19 at 05:58
  • @BruceAbbott no, I mean speaker is always connected but I disconnect the osc probe. https://electronics.stackexchange.com/q/472923/225442 please see this for the schematic. I'll also add it into this question. – muyustan Dec 25 '19 at 06:34
  • @muyustan I don't think your Zoebel network is doing much. Not with a capacitor so small. I conclude it's not causing trouble and we can leave it there or remove it and it won't make any difference. I still haven't given your huge circuit much attention. But after a quick rescan I would be worried about still having gain at high frequencies. You should add a zero in the global NFB, I think, to stomp down on that. Take a 330 Ohm resistor, put it in series with about 220 pF and place that series-pair in parallel to your \$R_{13}\$ resistor in your earlier posted diagram. Report back. – jonk Dec 25 '19 at 06:46
  • @muyustan Oh, heck. why waste time. Make that closer to a \$1\:\text{nF}\$ cap. – jonk Dec 25 '19 at 06:55
  • I remember it now. The amp is probably oscillating due to too much phase shift in the output stage. Also there is no bias on the transistors so you will get crossover distortion, possibility with instability in that region. Since it goes quiet when you put the scope probe on it, have you tried adding a small capacitor to emulate the probe? – Bruce Abbott Dec 25 '19 at 06:55
  • @jonk you have helped me a lot! It really smoothed the sound. Was the idea behind it to direct higher freq oscillations which are "noise" to that newly added parallel low(compared to R13) impedance path? – muyustan Dec 25 '19 at 07:05
  • @BruceAbbott I have thought about it, I have added 1uF capacitors to all power rails between +9/gnd , -9/gnd etc. I am not sure but it looks like it helped a little bit. – muyustan Dec 25 '19 at 07:05
  • 2
    @muyustan It's really, really simple. This bypasses \$R_{13}\$ when the frequencies are higher, destroying the gain at higher frequencies by feeding back a *lot* more signal. It forces a nice roll-off as the frequency increases. You probably had HF oscillations via parasitics where the phase angle was sufficient to become positive feedback. This killed the problem by feeding back still more of the output at HF as NFB through a controlled path. (The same technique can deal with stray inductance in some types of resistors -- not unlike what the Zoebel does for a speaker.) it's a simple paste-on. – jonk Dec 25 '19 at 08:32
  • 2
    @muyustan The huge clue to me was the capacitance in the oscilloscope probe you applied. It grounded out the HF right at the point where you were taking your feedback path. So it shorted out the HF while you were probing, destroying the positive feedback for the HF, and making things sound right. The fix to the circuit was then obvious. By the way, it was NOT leakage of stray *audio* from the SMPS as Tony mentioned at the outset. But Tony did grasp that it was related to capacitance of the probe. – jonk Dec 25 '19 at 08:37
  • @jonk thanks again, I get it. Now, about your comment on zoebel filter, I was using 68nF despite it is 22 in the schematic. Actually someone from here, I think he was tony, advised 0.1uF, I have found 68nF as closest. What is your opinions about values? – muyustan Dec 25 '19 at 08:46
  • 1
    @muyustan I don't know. It varies a lot and it also depends on if you are separating things out for a tweeter, and so on. But I almost never see anything less than \$1\:\mu\text{F}\$ being calculated. I haven't cared that much, though. I'm no audiophile. So I'm just reflecting the writing of others, not reflecting my own considered opinion after studying the matter, myself. But just now looking on the web, I find [this page](http://www.wavecor.com/html/zobel_networks.html) which has a table of values more in concert with my memory. – jonk Dec 25 '19 at 08:51
  • 1
    @muyustan So. Twice now, I've managed to pin the tail on the donkey, so to speak? Pure luck, I'm sure. I'm only a hobbyist with no more than a high school education and never a single class on so much as DC electronics. There are real engineers here in the group who know this stuff way better than I do. They are trained. Hopefully, they chip in more often and provide better help than I can. – jonk Dec 25 '19 at 08:58
  • 1
    @jonk I am geting its education, so in my case, it is pure **unluck** not to learn these things but instead learn how to integrate complex integrals while you can use any computational device :) – muyustan Dec 25 '19 at 09:05
  • 1
    @muyustan That's a great "funny/sad" remark! :) You provided a good laugh. Thanks! – jonk Dec 25 '19 at 09:07
  • 1
    @muyustan I've enjoyed watching you push forward. (By the way, a JFET is better for electrets.) I would have expected others with degrees to have enjoyed it, too, and then have wanted to nurture it when your interest is there and ready. We owe who we are to those who helped us in our day and although we cannot ever fully make amends for what we've been given, we must try to pay some of that debt back by helping those who show effort. To me, that's most of what really matters. So it's been my pleasure and honor. I wish it were the pleasure of more, though. Best wishes. – jonk Dec 25 '19 at 09:19
  • If you have additional information related to your original question, use the "edit" button to add the information there. Do not create a duplicate question. – Dave Tweed Dec 25 '19 at 13:22
  • @jonk There are engineers here who have not looked so closely at these circuits (too much else to cover in limited time). Such intrepid approach as muyustan has applied is never wasted and never forgotten. – glen_geek Dec 25 '19 at 16:11
  • @glen I've not looked closely at then, either. In fact, I refused to, partly because the OP just dumped one large mess of a schematic into a question. So I limited myself to only a very cursory look at a couple of things. About two minutes at most. I never actually expected to add a helpful comment. But some things are just too obvious. And I got lucky. He'd made an obvious blunder. Later, here in this post which I considered to lack way too much context and so I only hoped it was still the same topic, it was likewise luckily obvious. Pure luck, that way. – jonk Dec 25 '19 at 17:33
  • 1
    @glen But I'm doing several things at once with my comment. In some part, I'm talking to some who don't have the chance at formal education and lose hope because of that. In part, I'm talking to those who have been given, through serendipitous circumstances none of their making, so much more in life. More need to give back generously than do. Partly, I'm just teasing. And a few more things, besides. Nothing personal meant to anyone in particular. – jonk Dec 25 '19 at 17:39
  • @Dave I don't think this question was so much a duplicate question, as it was a poorly formed one. He gave no context. Sure, it was about a similar circuit. But that circuit didn't work. He repaired one problem with it due to a comment I made about it, and then ran into a completely different problem he was wondering about. He should have provided the new, changed schematic and didn't. That was his fault. But it isn't the same question and it was about a very specific circuit. In any case, it is answered for now. – jonk Dec 25 '19 at 17:45

0 Answers0