I ran distortion profile plots in REW comparing the mod and stock channels. To make the comparison valid, I equalized the input and output RMS voltages applied for the test. This is necessary, because the mod channel has substantially less gain. For both channels, the input voltage was 0.43vac rms and the output voltage was equalized to 1.05vac rms. Both channels were driving into a 22k ohm load with approx 100pF of interconnect cable capacitance. As you can see, almost all distortion is 2nd order. The mod channel is slightly improved at about 0.11% compared to the stock channel at about 0.205%.
I ran RTA spectrum plots in REW comparing the mod and stock channels. Again, I equalized the gain levels at 0.43vac rms in and 1.05vac rms out @ 1kHz to make the comparison valid. Otherwise the noise and distortion levels will be overstated for the higher gain stock channel. In addition, I disconnected my input and output DMM's from the loopback and used very short, properly terminated RCA interconnects to complete the loopback. This is important, because you want to measure the noise and distortion of the DUT alone and it is very easy to mess up this test with unshielded DMM breakout jacks. As you can see, the noise level and distortion is slightly improved for the mod channel.
SOUND QUALITY: So how does it sound? Strangely, the stock version was the clear winner. I listened in mono to one channel at a time, equalizing the gain and swapping cables back and forth. I used the tubes that came with the kit for all listening. The stock version was simply more musical to my ears with a greater sense of depth and dynamic range. The mod channel seemed muffled, compressed and slightly dull by direct comparison. Confirmation bias? Not sure. But the difference in sound quality was very clear to me. Now, maybe my ears simply like the nice, warm sound of 2nd harmonic distortion, but I don't think that this was the cause. And I doubt that I can actually hear a difference between 0.11% 2nd HD and 0.205% 2nd HD. Was the difference due to the fact that the mod channel adds a coupling cap at the input? Don't know. All I know is that the stock channel is DC connected at the input and just sounds cleaner, crisper, and more dynamic. I tried to crank the volume up on the mod channel to get it to sound more like the stock channel, but I couldn't. It just did not seem to have the same level of dynamic "punch" that I could get out of the stock channel.
Interesting outcome. I would think any confirmation bias would favor the Mod version, especially since you knew THD was lower and bandwidth was flatter. Could this be an example of how we still can't measure everything our ears prefer?
@Tom_S said:
Interesting outcome. I would think any confirmation bias would favor the Mod version, especially since you knew THD was lower and bandwidth was flatter. Could this be an example of how we still can't measure everything our ears prefer?
I did my listening before I did the measurements. I was thinking that maybe I was biased in favor of the stock version because I have read, over the years, several reviews and articles about the negative sonic effects of negative feedback in amplfiers. After I took the measurements, however, I was kind of thinking that I should not have heard any difference at all, since the differences were so small. But I clearly did. So I would think, at this point, that if my mind was making this up, I would conclude that both versions sound the same. But they don't. So, now I have to give Craig's jumper wire idea a test to see where this goes.
I have heard (with my own two ears) negative feedback suck the life out of an amp. I get why you like the original design better. Too bad the gain is so high with the original circuit.
@PWRRYD said:
Forgot to ask... when doing your listening tests what was the input impedance of your power amp (ie how much preamp gain)?
For all listening tests, I was driving into a Lepai 2020 amp, which measures 10k ohms across the inputs. This would have made the preamp gain 14dB for the stock channel and 6.5dB for the mod channel. When I swapped out channels, I adjusted the volume control up and down to compensate and equalize the SPL from the speaker. Interconnect cable capacitance was about 200pF, so probably no HF roll off on either the mod or stock channels.
@PWRRYD said:
Try bypassing the input cap with a jumper.
I attached a short jumper wire and gave it a listen. As you can see in the photos, I soldered the jumper wire to the bottom of the board, but left it unconnected on the top side, roughly 1/16" away from the solder pad connection point. In this way, I was able to use a small wooden stick to quickly connect and disconnect the bypass jumper wire and switch the cap in and out of the circuit almost instantaneously as the music played. Result: No difference at all. The sound was the same with or without the cap.
So, the sound quality difference I am hearing between the mod and stock channels must be due exclusively to the feedback loop. Something about the feedback is sucking the life out of the music, making it sound dead compared to the stock channel.
Well, I think that pretty much wraps up my attempts to modify this version of the 6J1 preamp. I am going to remove the mod perfboard and change it back to the stock circuit. Gain will be about 14dB into a 10k ohm load, which is a little high by some standards, but I am OK with this. I intend to use this preamp with the Pass Labs Pearl I phono stage (59dB gain) in combination with a low output moving coil cartridge. A line amp with about 10-15dB of gain is just about right for that type of setup. If I had built either the Pass Labs Pearl II or III phono stage, then this line amp would probably have a little bit too much gain.
I tend to use low gain amps, like many of the passdiy's are around 15db and a hypex ucd250 with no input buffer that is 13db. Though also use a F4 diy sometimes which is 0db gain but have a FE2022 'preamp' that has some giddy up at 20db gain or so. But with weather getting warmer I use either the hypex or p101 lat mosfet amp (padded that down to 22db). All this preference of mine for lower gain amps likely due to my smallish 2 channel listening room. F5m has been my recent fav, like it much better than original F5, enough so that I sold off my F3. Stock F5m circuit gives nice 2nd harmonic (rather than needing distortion analyzer to set P3).
Have Pearl 3 kit that someday will build out, but my spinning vinyl not that often thus slow to get to it and suffice with mani2 when do, or the (gasp) built in phono stage of marantz receiver (modern) if in fam room.
I should really have a go at building tube power amp, 10w or less fine for me. Though my parts collection not for the high 200-400v used in tube circuits and the lurking subconscious fear of opening new can of worms.
@DrewsBrews said:
Any comparison to the lower voltage one? Maybe try some tube rolling and see if the differences are more or less pronounced?
Thinking about how to do this. I have two versions of the lower voltage preamps. From what I remember, they did not sound the same. One operates off the 12vac 500ma wall wart and uses a 58vdc voltage doubler circuit. The other has a custom "CRCRC" type 58vdc power supply.
We would probably want to start by hand selecting a group of four 6J1(6AK5, etc) tubes that sound the same in just one of the preamps. If we did not do this as a first step, then it would probably be difficult to separate the sound of tubes from the sound of preamp versions. Once the set of four tubes is selected, you would need to quickly switch from one preamp to the other before you forget what the other one sounded like. Would need to either DIY a custom switch box or swap out pairs of RCA interconnects at high speed. Not as easy task. Thoughts?
Comments
DISTORTION PROFILE:
I ran distortion profile plots in REW comparing the mod and stock channels. To make the comparison valid, I equalized the input and output RMS voltages applied for the test. This is necessary, because the mod channel has substantially less gain. For both channels, the input voltage was 0.43vac rms and the output voltage was equalized to 1.05vac rms. Both channels were driving into a 22k ohm load with approx 100pF of interconnect cable capacitance. As you can see, almost all distortion is 2nd order. The mod channel is slightly improved at about 0.11% compared to the stock channel at about 0.205%.
RTA SPECTRUM:
I ran RTA spectrum plots in REW comparing the mod and stock channels. Again, I equalized the gain levels at 0.43vac rms in and 1.05vac rms out @ 1kHz to make the comparison valid. Otherwise the noise and distortion levels will be overstated for the higher gain stock channel. In addition, I disconnected my input and output DMM's from the loopback and used very short, properly terminated RCA interconnects to complete the loopback. This is important, because you want to measure the noise and distortion of the DUT alone and it is very easy to mess up this test with unshielded DMM breakout jacks. As you can see, the noise level and distortion is slightly improved for the mod channel.
SOUND QUALITY: So how does it sound? Strangely, the stock version was the clear winner. I listened in mono to one channel at a time, equalizing the gain and swapping cables back and forth. I used the tubes that came with the kit for all listening. The stock version was simply more musical to my ears with a greater sense of depth and dynamic range. The mod channel seemed muffled, compressed and slightly dull by direct comparison. Confirmation bias? Not sure. But the difference in sound quality was very clear to me. Now, maybe my ears simply like the nice, warm sound of 2nd harmonic distortion, but I don't think that this was the cause. And I doubt that I can actually hear a difference between 0.11% 2nd HD and 0.205% 2nd HD. Was the difference due to the fact that the mod channel adds a coupling cap at the input? Don't know. All I know is that the stock channel is DC connected at the input and just sounds cleaner, crisper, and more dynamic. I tried to crank the volume up on the mod channel to get it to sound more like the stock channel, but I couldn't. It just did not seem to have the same level of dynamic "punch" that I could get out of the stock channel.
Try bypassing the input cap with a jumper.
Interesting outcome. I would think any confirmation bias would favor the Mod version, especially since you knew THD was lower and bandwidth was flatter. Could this be an example of how we still can't measure everything our ears prefer?
Exactly what I was thinking.
Good idea! Will give this a try and report back.
I did my listening before I did the measurements. I was thinking that maybe I was biased in favor of the stock version because I have read, over the years, several reviews and articles about the negative sonic effects of negative feedback in amplfiers. After I took the measurements, however, I was kind of thinking that I should not have heard any difference at all, since the differences were so small. But I clearly did. So I would think, at this point, that if my mind was making this up, I would conclude that both versions sound the same. But they don't. So, now I have to give Craig's jumper wire idea a test to see where this goes.
Ha Ha, waiting for yo to catch the rabbit down the hole...
this is all very interesting, though i am absorbing just maybe 10%....
I have heard (with my own two ears) negative feedback suck the life out of an amp. I get why you like the original design better. Too bad the gain is so high with the original circuit.
Forgot to ask... when doing your listening tests what was the input impedance of your power amp (ie how much preamp gain)?
For all listening tests, I was driving into a Lepai 2020 amp, which measures 10k ohms across the inputs. This would have made the preamp gain 14dB for the stock channel and 6.5dB for the mod channel. When I swapped out channels, I adjusted the volume control up and down to compensate and equalize the SPL from the speaker. Interconnect cable capacitance was about 200pF, so probably no HF roll off on either the mod or stock channels.
I attached a short jumper wire and gave it a listen. As you can see in the photos, I soldered the jumper wire to the bottom of the board, but left it unconnected on the top side, roughly 1/16" away from the solder pad connection point. In this way, I was able to use a small wooden stick to quickly connect and disconnect the bypass jumper wire and switch the cap in and out of the circuit almost instantaneously as the music played. Result: No difference at all. The sound was the same with or without the cap.
So, the sound quality difference I am hearing between the mod and stock channels must be due exclusively to the feedback loop. Something about the feedback is sucking the life out of the music, making it sound dead compared to the stock channel.
Well, I think that pretty much wraps up my attempts to modify this version of the 6J1 preamp. I am going to remove the mod perfboard and change it back to the stock circuit. Gain will be about 14dB into a 10k ohm load, which is a little high by some standards, but I am OK with this. I intend to use this preamp with the Pass Labs Pearl I phono stage (59dB gain) in combination with a low output moving coil cartridge. A line amp with about 10-15dB of gain is just about right for that type of setup. If I had built either the Pass Labs Pearl II or III phono stage, then this line amp would probably have a little bit too much gain.
I tend to use low gain amps, like many of the passdiy's are around 15db and a hypex ucd250 with no input buffer that is 13db. Though also use a F4 diy sometimes which is 0db gain but have a FE2022 'preamp' that has some giddy up at 20db gain or so. But with weather getting warmer I use either the hypex or p101 lat mosfet amp (padded that down to 22db). All this preference of mine for lower gain amps likely due to my smallish 2 channel listening room. F5m has been my recent fav, like it much better than original F5, enough so that I sold off my F3. Stock F5m circuit gives nice 2nd harmonic (rather than needing distortion analyzer to set P3).
Have Pearl 3 kit that someday will build out, but my spinning vinyl not that often thus slow to get to it and suffice with mani2 when do, or the (gasp) built in phono stage of marantz receiver (modern) if in fam room.
I should really have a go at building tube power amp, 10w or less fine for me. Though my parts collection not for the high 200-400v used in tube circuits and the lurking subconscious fear of opening new can of worms.
Any comparison to the lower voltage one? Maybe try some tube rolling and see if the differences are more or less pronounced?
Thinking about how to do this. I have two versions of the lower voltage preamps. From what I remember, they did not sound the same. One operates off the 12vac 500ma wall wart and uses a 58vdc voltage doubler circuit. The other has a custom "CRCRC" type 58vdc power supply.
We would probably want to start by hand selecting a group of four 6J1(6AK5, etc) tubes that sound the same in just one of the preamps. If we did not do this as a first step, then it would probably be difficult to separate the sound of tubes from the sound of preamp versions. Once the set of four tubes is selected, you would need to quickly switch from one preamp to the other before you forget what the other one sounded like. Would need to either DIY a custom switch box or swap out pairs of RCA interconnects at high speed. Not as easy task. Thoughts?