captainbeefheart Posted January 29, 2022 Share Posted January 29, 2022 There are many forum members swearing by taking the two 1N3996 Zener diodes out of the circuit, just unbolt one of them from the bracket and let float to remove from circuit. I was intrigued so I felt like taking some distortion measurements to see what is going on. In theory these diodes should not be doing anything unless you reach a peak voltage of 5.737v to where they would start to conduct and clamp the signal there, nothing higher can pass. In terms of RMS power it works out to a maximum of 2 watts going into the tweeters. Testing confirmed that the diodes indeed do clamp the signal exactly at 5.737v. BUT there was an unexpected outcome to taking distortion readings below this value. Voltage readings are taken at the tweeter output across the diodes. They are peak voltages unless otherwise stated. What I found was surprising. I didn't see any difference at 2.8v, both with and without diodes had identical distortion profiles. It's when you start to reach about 4v where things start go awry for the circuit with the diodes. As soon as you near 4v distortion starts to quickly build up with increase in amplitude. The pictures below are the difference between the network with and without the diodes at 5v. The zeners are clearly starting to do something at 4v, I am guessing it's the very beginning of their 'knee' where very little current is starting to pass and the diodes are somewhere in their dynamic impedance range, datasheet shows 300uA they are 100 ohms and linearly decreases impedance vs increase in current, at 600mA the diode is .2 ohms. Regardless of the mechanism they just don't clamp at 5.737v, clearly current is passing at lower peak voltages, I am seeing 4v is where the difference is happening at. Now you might say to yourself that's still a lot of power. 4v peak is about 2.8v RMS, or 1 watt power. But you must remember dynamic range in music. If you are hovering around say 1 watt at the input of the crossover network, 20db dynamic headroom, a heavy crash cymbal may push a peak transient over the 4v for a very short period of time. 20db is 10x, so for 4v that's only around 400mV. I may be completely wrong but I can completely see even though we know the average power into the tweeter will be low, there must some signals passing over the 4v. Members have made it quite clear that taking these diodes out have impacted the sound positively, so they must be misbehaving during even modest play levels. My engineering mind was doubtful, the diodes are suppose to clamp at 2 watts and up, certainly must not be heard at in home listening levels, well it turns out that distortion is increasing rapidly at anything over just 1 watt or 4v peak into the tweeter. You may not be averaging 1 watt into the tweeter but you may hit it with dynamic music. If for whatever reason you are blasting away and you happen to reach the clamping voltage peak, distortion grossly increases, at the point of clipping you are already at 5% THD and it just gets worse from there. Clearly in home listening is reaching the point where these diodes are creating distortion and effecting the sound quality. You can see the higher harmonics jump right up at 15kHz and 20kHz. 7 3 Quote Link to comment Share on other sites More sharing options...
Curious_George Posted January 29, 2022 Share Posted January 29, 2022 Finally, some visual proof of what people claim to have been hearing. Very good work Captain! 3 Quote Link to comment Share on other sites More sharing options...
svberger Posted January 29, 2022 Share Posted January 29, 2022 1 hour ago, captainbeefheart said: There are many forum members swearing by taking the two 1N3996 Zener diodes out of the circuit, just unbolt one of them from the bracket and let float to remove from circuit. I was intrigued so I felt like taking some distortion measurements to see what is going on. In theory these diodes should not be doing anything unless you reach a peak voltage of 5.737v to where they would start to conduct and clamp the signal there, nothing higher can pass. In terms of RMS power it works out to a maximum of 2 watts going into the tweeters. Testing confirmed that the diodes indeed do clamp the signal exactly at 5.737v. BUT there was an unexpected outcome to taking distortion readings below this value. Voltage readings are taken at the tweeter output across the diodes. They are peak voltages unless otherwise stated. What I found was surprising. I didn't see any difference at 2.8v, both with and without diodes had identical distortion profiles. It's when you start to reach about 4v where things start go awry for the circuit with the diodes. As soon as you near 4v distortion starts to quickly build up with increase in amplitude. The pictures below are the difference between the network with and without the diodes at 5v. The zeners are clearly starting to do something at 4v, I am guessing it's the very beginning of their 'knee' where very little current is starting to pass and the diodes are somewhere in their dynamic impedance range, datasheet shows 300uA they are 100 ohms and linearly decreases impedance vs increase in current, at 600mA the diode is .2 ohms. Regardless of the mechanism they just don't clamp at 5.737v, clearly current is passing at lower peak voltages, I am seeing 4v is where the difference is happening at. Now you might say to yourself that's still a lot of power. 4v peak is about 2.8v RMS, or 1 watt power. But you must remember dynamic range in music. If you are hovering around say 1 watt at the input of the crossover network, 20db dynamic headroom, a heavy crash cymbal may push a peak transient over the 4v for a very short period of time. 20db is 10x, so for 4v that's only around 400mV. I may be completely wrong but I can completely see even though we know the average power into the tweeter will be low, there must some signals passing over the 4v. Members have made it quite clear that taking these diodes out have impacted the sound positively, so they must be misbehaving during even modest play levels. My engineering mind was doubtful, the diodes are suppose to clamp at 2 watts and up, certainly must not be heard at in home listening levels, well it turns out that distortion is increasing rapidly at anything over just 1 watt or 4v peak into the tweeter. You may not be averaging 1 watt into the tweeter but you may hit it with dynamic music. If for whatever reason you are blasting away and you happen to reach the clamping voltage peak, distortion grossly increases, at the point of clipping you are already at 5% THD and it just gets worse from there. Clearly in home listening is reaching the point where these diodes are creating distortion and effecting the sound quality. You can see the higher harmonics jump right up at 15kHz and 20kHz. That's great! Thanks very much for doing the work. 2 Quote Link to comment Share on other sites More sharing options...
jason str Posted January 29, 2022 Share Posted January 29, 2022 I believe it was the 561 dome lamp bulb Dennis recommended using for tweeter protection, something about fuses not quick enough. Not forum literate enough to dig and confirm my memory. 1 Quote Link to comment Share on other sites More sharing options...
Curious_George Posted January 29, 2022 Share Posted January 29, 2022 1 hour ago, Crankysoldermeister said: ‘“Golden Ear Audiophile hit with a brick "Oww, that hurts! I'm bleeding!" Tin Ear Meter Reader hit with a brick "You can't prove I was hit with a brick! We need to do a double blind ABX test!"’ - DJK I never found that quote funny or even relevant. 1 Quote Link to comment Share on other sites More sharing options...
henry4841 Posted January 29, 2022 Share Posted January 29, 2022 We all appreciate the work you have done. Good work. 2 Quote Link to comment Share on other sites More sharing options...
Peter P. Posted January 29, 2022 Share Posted January 29, 2022 The question is, what did the engineers have in mind when they chose diode protection and the breakdown voltage? Could be they were aware of the trade-offs. Quote Link to comment Share on other sites More sharing options...
svberger Posted January 30, 2022 Share Posted January 30, 2022 27 minutes ago, Peter P. said: The question is, what did the engineers have in mind when chose diode protection and the breakdown voltage? Could be they were aware of the trade-offs. Maybe they were but figured the owners wouldn't be and to protect the influx of blown tweeters this was the easiest and most inexpensive remedy? Quote Link to comment Share on other sites More sharing options...
Curious_George Posted January 30, 2022 Share Posted January 30, 2022 35 minutes ago, Peter P. said: The question is, what did the engineers have in mind when chose diode protection and the breakdown voltage? It was a compromise between blown tweeters and sound quality. It’s a known fact the tweeters were (are) delicate. Perhaps the threshold could have been a bit higher so the distortion would not begin until a higher wattage. Newer technology performs better such as poly switches, but they are not the final answer. I personally prefer the right light bulb. 1 Quote Link to comment Share on other sites More sharing options...
Deang Posted January 30, 2022 Share Posted January 30, 2022 Knowing what distortion sounds like and turning down the volume works pretty good too. 1 Quote Link to comment Share on other sites More sharing options...
svberger Posted January 30, 2022 Share Posted January 30, 2022 15 minutes ago, Crankysoldermeister said: Knowing what distortion sounds like and turning down the volume works pretty good too. Yes but the distortion is occurring at fairly low levels to begin with. Taking out the diode solves that. Quote Link to comment Share on other sites More sharing options...
captainbeefheart Posted January 30, 2022 Author Share Posted January 30, 2022 These are taken right at 5.5v peak just before it starts clamping, sine wave on the scope still looks nice but clearly we are over 2% THD with the zeners in. With them out the distortion remains the same even going past 10v peak. Images are both taken at 5.5v, with and without the diodes. Check out the third harmonic (15kHz) it's only -30db below fundamental. Quote Link to comment Share on other sites More sharing options...
Curious_George Posted January 30, 2022 Share Posted January 30, 2022 55 minutes ago, Curious_George said: It was a compromise between blown tweeters and sound quality. It’s a known fact the tweeters were (are) delicate. Perhaps the threshold could have been a bit higher so the distortion would not begin until a higher wattage. Newer technology performs better such as poly switches, but they are not the final answer. I personally prefer the right light bulb. I assume the tweeter in the below diagram is a K77? 1 Quote Link to comment Share on other sites More sharing options...
captainbeefheart Posted January 30, 2022 Author Share Posted January 30, 2022 I wanted to see exactly where it reaches 1% THD and it's 4.5v peak. To put things into perspective the input at the crossover network is 9v peak. That's 5 watts RMS power into the Network for 1% THD at the tweeter caused from the diodes. Quote Link to comment Share on other sites More sharing options...
garyrc Posted January 30, 2022 Share Posted January 30, 2022 5 hours ago, captainbeefheart said: If you are hovering around say 1 watt at the input of the crossover network, 20db dynamic headroom, a heavy crash cymbal may push a peak transient over the 4v for a very short period of time. 20db is 10x ... Questions: Is 10 dB 10 times, when dealing in watts? I realize you were looking at volts at that point. On what speakers did you do these measurements? Khorns? The overall signal, up to maybe 2K Hz might hover around 1 watt (95 dB at 13 feet away, in my ~ 4,300 cu. ft. room with Khorns) but would even a cymbal crash produce a peak of 1 watt in tweeter range, above about 1 octave below crossover (2.25K Hz -- 36 dB down in the AK4 & AK5 crossovers) into my Khorns? The figure we often hear is that "the treble" is 20 dB down in orchestral music (electronic and other "non-acoustical" music excluded). Wouldn't that exactly offset the 20 dB dynamic headroom quoted above? Subjectively, when I disconnected the woofer in my old JBL 030 c.1960 CE, cymbal crashes through the tweeter only (Xover @ 2,500 Hz) sounded like someone emphatically making a "sheeeesch" sound with their mouth (but a little more sfz). I measured it at about 92 dB @ 3 feet. O.K., is "the treble" a full 20 dB down? Maybe, maybe not. Let's look at some possibilities: If the treble above tweeter crossover is 20 dB below the overall level of 95 dB =1 watt (Khorn in my room) overall level for the full orchestra, that would be somewhere around 0.0052 watts into the tweeter, when the orchestra is churning out additive and interactive combined fundamentals and lower and medium harmonics (below tweeter Xover) of 1 watt, or 95 dB. To play with the big kids, at THX/Dolby full scale for peaks (20 dB dynamic headroom over their arbitrary "typical" or "nominal" SPL of 85 dB) of 105 dB at the listening position for the full orchestral sound (not counting frequencies that would go to a subwoofer), this would be 10 watts (105 dB peaks) for the music as a whole, and still only 95 dB (1 watt) for the tweeter. If the treble above tweeter crossover is only 10 dB below the overall level, the tweeter might get 10 watts, the zener diode would clamp down, and distortion would result. 5. Can we substitute another component for the zener diode that would protect the tweeter, but not introduce distortion? Quote Link to comment Share on other sites More sharing options...
captainbeefheart Posted January 30, 2022 Author Share Posted January 30, 2022 Dynamic range I view in terms of amplitude, I think most do. Where watts is a product of voltage and current, a measure of power. For a 20db increase in amplitude to reach the 4.5v peak we are seeing 1% THD at, -20db down is -10x or 450mV amplitude. In terms of power that's a change of 13mW to 1.29w. Maybe we can get a clue looking empirically at power ratings of the drivers and deduce a reasonable answer. 100 watt woofer and a 3 watt tweeter. Let's change these power terms to amplitude. 100 watts into 8 ohms is 39.59v peak 3 watts into 8 ohms is 6.85v That puts us -15db down I think that is ballpark to say treble or tweeter content will be -15db lower vs low frequency content. I think I'll wire up a microphone soon and go from tapping on a cymbal lightly to full crash like a drummer would do and record it to see the dynamic range in amplitude. This information could probably just googled. I'll check that first, google 'dynamic range recording of cymbals' or something like that. Clearly people are hearing a difference so they must be hitting peaks up past the 1 watt into the tweeter to get the audible distortion. 1 Quote Link to comment Share on other sites More sharing options...
garyrc Posted January 30, 2022 Share Posted January 30, 2022 27 minutes ago, captainbeefheart said: I think I'll wire up a microphone soon and go from tapping on a cymbal lightly to full crash like a drummer would do and record it to see the dynamic range in amplitude. This information could probably just googled. I'll check that first, google 'dynamic range recording of cymbals' or something like that. I'm curious about the dynamic range of cymbal crashes at the frequencies that reach the tweeter. In the case of Klipschorn AK 4 & 5, perhaps others, that would be Xover at 4,500 Hz rolling off below the Xover at 36 dB per octave. Quote Link to comment Share on other sites More sharing options...
jason str Posted January 30, 2022 Share Posted January 30, 2022 1 hour ago, Curious_George said: I assume the tweeter in the below diagram is a K77? The 561 bulb you found here confirms, this was for K 77. Quote Link to comment Share on other sites More sharing options...
tigerwoodKhorns Posted January 30, 2022 Share Posted January 30, 2022 15 minutes ago, jason str said: The 561 bulb you found here confirms, this was for K 77. I think that I have DJK's write up on the bulb in my server. I need to dig it up on Monday. 1 Quote Link to comment Share on other sites More sharing options...
garyrc Posted January 30, 2022 Share Posted January 30, 2022 3 hours ago, Curious_George said: Newer technology performs better such as poly switches, but they are not the final answer. @Curious George Since I have a Klipschorn AK4, I evidently have a polyswitch. In what ways are they better? Do you have any articles to recommend? Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.