I read everything I could find on the LED biasing deal. Probably everything you've read. Seems a very biased opinion on its merits. As its real benefit is elimination of an electrolytic cap and getting max gain/frequency response, it's not used in instrument amplification as the cathode resistor/cap are a big part of tuning the overall sound/tone of the amp.
As for hifi, seems like a try it and see if you like it situation.
I've not read a lot about it lately, but I do remember some arguments from years ago about not using **any **silicon in a tube amp or preamp. Forget the fact that a CD player & DAC were the common sources for many audiophiles.
I always wondered about the powersupply rectification. Is there any real benefit to a tube rectifier? Or just another piece of glass to put on display for "purity"?
@DrewsBrews said:
I always wondered about the powersupply rectification. Is there any real benefit to a tube rectifier? Or just another piece of glass to put on display for "purity"?
In the old days replacing the rectifier tube with diodes was a common upgrade.
IMHO there are only three merits to using tube rectifiers:
1) Most common rectifiers tubes (not all of course) warm up slowly and don't apply the full B+ voltage to all the other tubes' plates within milliseconds, before those tubes own heaters get hot enough for them to start conducting. The negative effect of applying full B+ voltage to a cold tube is called cathode stripping. It is generally accepted that it reduces the life of the tubes.
2) Not HiFi but in guitar tube amps the B+ voltage will sag on long sustained notes. That will actually increase gain and thus improve the sustain, kind of like positive feedback.
3) If you're building a glorious tube amp, who doesn't want more glowing tubes (eye candy) up front for all to see?
I finally got REW and my external usb soundcard set up and calibrated to do some 6J1 frequency response and harmonic distortion sweeps. I ran a number of tests at different output voltage levels, driving into the 20k ohm input impedance of my soundcard, and the results are very interesting. I can also do RTA graphs showing noise, hum, and HD profiles at selected frequencies. Getting this loopback system set up and running properly was a real rabbit hole, but that will be a completely different thread. I'll stop back later today with some of my 6j1 screen shots.
I'm still using the stock circuit modded with 3.3uF coupling caps but without the feedback or cathode bypass cap mods. Here are a few REW screen grabs showing FR, HD, and RTA data of this version. For all graphs, the preamp is driving into a 20k ohm load through a total of about 330 picofarads of interconnect cable capacitance. I tested two preamp output voltages levels: 0.5 vrms and 1.0Vrms. As you can see, almost all of the distortion is 2nd order, which gives this preamp a very warm and pleasant sound quality. 3rd and higher order harmonics are almost nonexistent.
Frequency Response with 20k load and 330pF cable capacitance. Response is minus 0.1dB at 20Hz and minus 0.3dB at 20kHz:
Harmonic distortion at 0.5Vrms output:
Harmonic distortion at 1.0Vrms output:
RTA showing noise, hum, and HD. 1000Hz signal. 0.5Vrms output:
RTA showing noise, hum, and HD. 1000Hz signal. 1.0Vrms output:
That's looking great! I know after I made the feedback mod I saw THD was dominated almost entirely by the 2nd harmonic too. But that changed a bit when I swapped the tubes. That's something I'll have to investigate further.
I like what I'm seeing with your graphs in REW. I had never used it, but downloaded it and played around with it not long ago.
Thanks, Tom. When I do the feedback and cathode cap mods, I'll re-run the same tests to see how it changes the relationship of even and odd order harmonic distortion. I'm dirt cheap as well. So far, I've spent a total of $47 for the two kits, enclosure, both transformers, and extra power supply caps.
Tom, I installed the feedback mod (100K/.22uF) and cathode bypass mod (220uF) and have been running distortion and frequency response tests at various output levels and driving into various output loads. I've also run up a bunch of RTA/spectrum type plots at 1kHz. But I have been holding off reporting my results because my plots appear to be significantly at odds with the results shown in TheStuffMade youtube video above. So I have been double and triple checking my REW/Steinberg UR22 loopback setup, making sure I have it all hooked up properly. And double checking my PCB to make sure I installed all the resistors in the proper locations.
Basically, I am getting THD figures only as low as 0.1%, but only at very low output levels of roughly 200mVrms, driving into a load of 120K ohms. If I boost the output level up to about 500mVrms, THD goes up to about 0.3%, again, driving into a 120k ohm load. At 500mVrms output, if I drop the load from 120k down to 20k, the THD goes up to about 0.5%. And if I boost the output to about 1Vrms driving into a 20k load, THD goes up to about 1%. This is much higher than shown in the video. So I must be doing something wrong.
Another thing that might account for some of the difference is that there is another mod that I have not done yet. It is the high frequency roll off mod, changing R12/R17 from 4.7k to 22k, and then adding a shunt capacitor of 220pF from grid to ground. This reduces the gain further and limits the high frequeny bandwidth to a reasonable level. I was wondering, have you installed this particular mod? Maybe I still have too much gain and bandwidth.
You may have hit upon a one thing that I have neglected - a typical real-world load impedance. I've been running straight into the inputs of my Motu interface. I was getting THD levels around 0.03% measured in REW this weekend, but I didn't note the input or output voltages. It was just a quick test to see if I had REW set up correctly and to make my loopback soundcard cal file. I'll post some screenshots later. I do have the input mod installed, so that may also be a contributing factor. I did remove the cathode bypass caps, since I don't need the extra gain.
My project is in pieces right now. I'm working on solving an oversight in my enclosure design. But once that issue is dealt with, I should have everything running again to take some new measurements.
OK, I'll install the input mods and run some more tests.
For testing, I have also been running straight into the inputs of my Steinberg UR22 interface. I'm looping back input/output channel 1 with the preamp included in the loop. Then I'm using a splitter to loop back the channel 1 output to the channel 2 input and then using channel 2 as the soundcard calibration loop. The input impedance of my UR22 is listed in the user manual as 20k ohms for the line level inputs, so if I run the preamp outputs straight into channel 1, the load impedance is 20k ohms plus the cable capacitance (about 300pF). If I add a 100k ohm resistor in series with the preamp outputs, then the input impedance is 120k ohms. But I have to boost the line input gain on the UR22 to compensate for the voltage drop across the 100k series resistor.
I don't know what the line level input impedance is for your Motu interface, but you should be able to do something similar to vary the load impedance of your test setup. To set and check the preamp voltage output levels, I am simply connecting my multimeter (DMM) across the 6J1 preamp outputs and then using REW's generator to play a 1kHz test tone.
I think the M2 unbalance TRS input impedance is really high - I think I read 1 Meg. That seems about right since most people will be plugging guitars & bases right in the front.
My UR22 has a "Hi-Z Input" with a 1 meg input impedance, but it is only available on channel 2. I could re-arrange my looback wires and try using this input, but I would have to be very careful not to overload it. The manual says that the maximum input for the "Hi-Z" is +8.5dBV, which is roughly 2.6Vrms. The line inputs that I have been using have a rated maximum input of +24dBu, which is roughly 12Vrms.
If you plan on using this tube preamp with a modern solid state amp, most of those have an input impedance of 20 kOhms or less. That's what I would use for an ouput load for testing bandwidth, gain, and distortion.
The problem with my chassis has been solved. I had screwed the panel holding the volume pot to the front panel of the shell. That left me no way to wire up the pot, since the bottom/back tray slides in from the back. Fixed that issue with 2 little L brackets attached to the tray. I have 5 more holes left to drill and then it’ll be on to wiring it all up.
Looking good, Tom. I see you ordered a 2nd kit as well.
Interesting REW graphs. Your distortion figures seem to be much more consistent with the youtube vid from The Stuff Made. I also see that your 2nd and 3rd HD plots are much more squiggly looking over the 20-20kHz bandwidth. My 2nd and 3rd HD plots are almost flat lines from 20-20kHz, which seems really odd. How can my 2nd & 3rd order HD be the same at all frequencies? I also notice that your frequency response plot climbs about 0.5dB below 40Hz. After I installed the feedback mod, my frequency response plot started to climb by about 0.5dB below 40Hz, just like yours. Whereas before I installed the feedback mod, my FR plot rolled off by about 0.1dB at 20Hz.
I figured these kits are so cheap, especially on Black Friday, I should start with a fresh board and have an extra set of the Chinese tubes. Like you, I'll only populate the audio section on this one, leaving the transistors and caps in the bag.
We'll see what happens to the THD once I have this running into a lower impedance load. I'm sure it will be higher, but curious to by how much. I may bump the output cap up to a 4.7uf to keep the low end from dropping.
Slow, boring work. I’m really questioning why I decided to use that double braided shielded cable for the inputs and outputs. I would love to finish the audio portion today, but I need to wrap a mountain of Christmas presents.
Probably right. I'll bet somewhere there's $$$ tool that does all the stripping in one quick step. Sadly, I didn't find that tool at the surplus store when I bought this cable.
After the audio portion is done and tested, then it's time for me relearn everything I've forgotten about basic Arduino coding!
Nice layout, with power supply wiring and xformers well away from the signal path wiring. Looks good. I agree, working with shielded cable connections is a major hassle. On mine, I used Mogami microphone cable to go from the PCB output pads to the rear panel RCA jacks. Since my cable run was only about 4 inches, I completely removed the black PVC jacket & outer shield and just used the two inner cables, twisting them slightly along the way. I don't hear any hum, so I think I lucked out. Your runs are a bit longer and closer to the xformers, so probably best to shield them completely.
Basic wiring is done. I was hearing a little hum and saw there was almost 2v of AC on the heater supply. I added a small board with an extra stage of RC. It’s now down to under 100mv. I’m hoping that took care of the audible portion. Time to run some more tests.
Some quick measurements in REW into a 19.6K load resistance. Flat as a pancake, as they say. THD is a little higher than before, but I'm running a different pair of tubes. Time to hook it up in the Audio Dungeon system.
I completed the input mod, changing R12/R17 from 4.7k to 22k. Since I had to remove the cathode bypass caps to get at these resistors, I decided to leave the cathode bypass caps out instead of soldering them back in place. This helped quite a bit. I now have much lower distortion and much less gain. Closed loop gain is now down to about 6dB on my unit, compared to about 10 to 15dB before. I ran loopback distortion and spectrum plots in REW at several voltage and loading levels and THD is now down to as low as 0.043% on some of my tests. 2nd order HD creeps up a little bit at the highest output levels, but remains well in check at what I would consider to be normal to moderate listening levels. Frequency response is ruler flat into all loads, except at the very highest and lowest frequencies (10Hz and 20kHz). I am really liking the way this turned out. Here are a few graphs from my tests:
Frequency Response into all output voltage levels and loads tested (varying from 115mVrms to 1.224Vrms output, driving into loads of 20k or 96k with 600pF of cable capacitance):
Harmonic distortion with 57mVrms input, 115mVrms output, driving into a 20k/600pF load:
HD with 103mVrms input, 206mVrms output, driving into a 20k/600pF load:
HD with 394mVrms input, 777mVrms output, driving into a 20k/600pF load:
HD with 627mVrms input, 1.22Vrms output, driving into a 20k/600pF load:
HD with 57mVrms input, 127mVrms output, driving into a 96k/600pF load:
HD with 103mVrms input, 227mVrms output, driving into a 96k/600pF load:
Preamp volume was turned all the way up, maximum gain, for all of the above distortion tests. I did not test higher voltage drive levels into 96k because the Steinberg UR22 did not have enough gain to complete higher drive voltage tests when the input was set to Hi Z. I need to build a unity gain buffer circuit and insert it between preamp and soundcard in order to test into 96k at higher levels.
I also ran another spectrum plot, which looks fairly good. The peak at 60Hz is a ground loop in my test setup that I need to address. I tried unplugging the laptop and running on batteries only, but that did not help. Probably picking up 60Hz hum from my DMM probes or something like that. Note added: The below spectrum was taken at the 394mVrms input level, 777mVrms output level, driving into a 20k/600pF load:
I found out what was causing the 60Hz hum spike on the spectrum plot that I attached to my previous post. I am using two digital multimeters to monitor the rms input and output voltages going to and from the preamp for all tests. I have the two meters connected to the RCA cables using Y splitters and alligator clips. The 60Hz spike was simply noise being picked up by improperly shielded cables. When I removed the two digitial multimeters and the Y splitters and then repeated the test with properly shielded RCA cables, the problem disappeared. Here is the spectrum plot with the proper shielding in place, using the same 394mVrms input level, 777mVrms output level, but now driving into a 20k/300pF load:
This is a fairly high playback level, and the power supply hum and noise are almost completely gone.
Comments
I read everything I could find on the LED biasing deal. Probably everything you've read. Seems a very biased opinion on its merits. As its real benefit is elimination of an electrolytic cap and getting max gain/frequency response, it's not used in instrument amplification as the cathode resistor/cap are a big part of tuning the overall sound/tone of the amp.
As for hifi, seems like a try it and see if you like it situation.
I've not read a lot about it lately, but I do remember some arguments from years ago about not using **any **silicon in a tube amp or preamp. Forget the fact that a CD player & DAC were the common sources for many audiophiles.
I always wondered about the powersupply rectification. Is there any real benefit to a tube rectifier? Or just another piece of glass to put on display for "purity"?
In the old days replacing the rectifier tube with diodes was a common upgrade.
IMHO there are only three merits to using tube rectifiers:
1) Most common rectifiers tubes (not all of course) warm up slowly and don't apply the full B+ voltage to all the other tubes' plates within milliseconds, before those tubes own heaters get hot enough for them to start conducting. The negative effect of applying full B+ voltage to a cold tube is called cathode stripping. It is generally accepted that it reduces the life of the tubes.
2) Not HiFi but in guitar tube amps the B+ voltage will sag on long sustained notes. That will actually increase gain and thus improve the sustain, kind of like positive feedback.
3) If you're building a glorious tube amp, who doesn't want more glowing tubes (eye candy) up front for all to see?
I finally got REW and my external usb soundcard set up and calibrated to do some 6J1 frequency response and harmonic distortion sweeps.
I ran a number of tests at different output voltage levels, driving into the 20k ohm input impedance of my soundcard, and the results are very interesting. I can also do RTA graphs showing noise, hum, and HD profiles at selected frequencies. Getting this loopback system set up and running properly was a real rabbit hole, but that will be a completely different thread. I'll stop back later today with some of my 6j1 screen shots.
I think cheap and simple left this thread long ago....
I'm still using the stock circuit modded with 3.3uF coupling caps but without the feedback or cathode bypass cap mods. Here are a few REW screen grabs showing FR, HD, and RTA data of this version. For all graphs, the preamp is driving into a 20k ohm load through a total of about 330 picofarads of interconnect cable capacitance. I tested two preamp output voltages levels: 0.5 vrms and 1.0Vrms. As you can see, almost all of the distortion is 2nd order, which gives this preamp a very warm and pleasant sound quality. 3rd and higher order harmonics are almost nonexistent.
Frequency Response with 20k load and 330pF cable capacitance. Response is minus 0.1dB at 20Hz and minus 0.3dB at 20kHz:
Harmonic distortion at 0.5Vrms output:
Harmonic distortion at 1.0Vrms output:
RTA showing noise, hum, and HD. 1000Hz signal. 0.5Vrms output:
RTA showing noise, hum, and HD. 1000Hz signal. 1.0Vrms output:
That's looking great! I know after I made the feedback mod I saw THD was dominated almost entirely by the 2nd harmonic too. But that changed a bit when I swapped the tubes. That's something I'll have to investigate further.
I like what I'm seeing with your graphs in REW. I had never used it, but downloaded it and played around with it not long ago.
I think I might have spent $80 so far. That's dirt cheap for decent quality tube hifi.
Thanks, Tom. When I do the feedback and cathode cap mods, I'll re-run the same tests to see how it changes the relationship of even and odd order harmonic distortion. I'm dirt cheap as well. So far, I've spent a total of $47 for the two kits, enclosure, both transformers, and extra power supply caps.
Tom, I installed the feedback mod (100K/.22uF) and cathode bypass mod (220uF) and have been running distortion and frequency response tests at various output levels and driving into various output loads. I've also run up a bunch of RTA/spectrum type plots at 1kHz. But I have been holding off reporting my results because my plots appear to be significantly at odds with the results shown in TheStuffMade youtube video above. So I have been double and triple checking my REW/Steinberg UR22 loopback setup, making sure I have it all hooked up properly. And double checking my PCB to make sure I installed all the resistors in the proper locations.
Basically, I am getting THD figures only as low as 0.1%, but only at very low output levels of roughly 200mVrms, driving into a load of 120K ohms. If I boost the output level up to about 500mVrms, THD goes up to about 0.3%, again, driving into a 120k ohm load. At 500mVrms output, if I drop the load from 120k down to 20k, the THD goes up to about 0.5%. And if I boost the output to about 1Vrms driving into a 20k load, THD goes up to about 1%. This is much higher than shown in the video. So I must be doing something wrong.
Another thing that might account for some of the difference is that there is another mod that I have not done yet. It is the high frequency roll off mod, changing R12/R17 from 4.7k to 22k, and then adding a shunt capacitor of 220pF from grid to ground. This reduces the gain further and limits the high frequeny bandwidth to a reasonable level. I was wondering, have you installed this particular mod? Maybe I still have too much gain and bandwidth.
You may have hit upon a one thing that I have neglected - a typical real-world load impedance. I've been running straight into the inputs of my Motu interface. I was getting THD levels around 0.03% measured in REW this weekend, but I didn't note the input or output voltages. It was just a quick test to see if I had REW set up correctly and to make my loopback soundcard cal file. I'll post some screenshots later. I do have the input mod installed, so that may also be a contributing factor. I did remove the cathode bypass caps, since I don't need the extra gain.
My project is in pieces right now. I'm working on solving an oversight in my enclosure design. But once that issue is dealt with, I should have everything running again to take some new measurements.
OK, I'll install the input mods and run some more tests.
For testing, I have also been running straight into the inputs of my Steinberg UR22 interface. I'm looping back input/output channel 1 with the preamp included in the loop. Then I'm using a splitter to loop back the channel 1 output to the channel 2 input and then using channel 2 as the soundcard calibration loop. The input impedance of my UR22 is listed in the user manual as 20k ohms for the line level inputs, so if I run the preamp outputs straight into channel 1, the load impedance is 20k ohms plus the cable capacitance (about 300pF). If I add a 100k ohm resistor in series with the preamp outputs, then the input impedance is 120k ohms. But I have to boost the line input gain on the UR22 to compensate for the voltage drop across the 100k series resistor.
I don't know what the line level input impedance is for your Motu interface, but you should be able to do something similar to vary the load impedance of your test setup. To set and check the preamp voltage output levels, I am simply connecting my multimeter (DMM) across the 6J1 preamp outputs and then using REW's generator to play a 1kHz test tone.
I think the M2 unbalance TRS input impedance is really high - I think I read 1 Meg. That seems about right since most people will be plugging guitars & bases right in the front.
My UR22 has a "Hi-Z Input" with a 1 meg input impedance, but it is only available on channel 2. I could re-arrange my looback wires and try using this input, but I would have to be very careful not to overload it. The manual says that the maximum input for the "Hi-Z" is +8.5dBV, which is roughly 2.6Vrms. The line inputs that I have been using have a rated maximum input of +24dBu, which is roughly 12Vrms.
If you plan on using this tube preamp with a modern solid state amp, most of those have an input impedance of 20 kOhms or less. That's what I would use for an ouput load for testing bandwidth, gain, and distortion.
Here are some screenshots from the last measurement I did. Not sure what the input or output voltages were. I just set the levels for ~-12dB in REW.
The problem with my chassis has been solved. I had screwed the panel holding the volume pot to the front panel of the shell. That left me no way to wire up the pot, since the bottom/back tray slides in from the back. Fixed that issue with 2 little L brackets attached to the tray. I have 5 more holes left to drill and then it’ll be on to wiring it all up.
Looking good, Tom. I see you ordered a 2nd kit as well.
Interesting REW graphs. Your distortion figures seem to be much more consistent with the youtube vid from The Stuff Made. I also see that your 2nd and 3rd HD plots are much more squiggly looking over the 20-20kHz bandwidth. My 2nd and 3rd HD plots are almost flat lines from 20-20kHz, which seems really odd. How can my 2nd & 3rd order HD be the same at all frequencies? I also notice that your frequency response plot climbs about 0.5dB below 40Hz. After I installed the feedback mod, my frequency response plot started to climb by about 0.5dB below 40Hz, just like yours. Whereas before I installed the feedback mod, my FR plot rolled off by about 0.1dB at 20Hz.
I figured these kits are so cheap, especially on Black Friday, I should start with a fresh board and have an extra set of the Chinese tubes. Like you, I'll only populate the audio section on this one, leaving the transistors and caps in the bag.
We'll see what happens to the THD once I have this running into a lower impedance load. I'm sure it will be higher, but curious to by how much. I may bump the output cap up to a 4.7uf to keep the low end from dropping.
Slow, boring work. I’m really questioning why I decided to use that double braided shielded cable for the inputs and outputs. I would love to finish the audio portion today, but I need to wrap a mountain of Christmas presents.
Better than questioning why you didn’t, after the build is done… nice looking build
Probably right.
I'll bet somewhere there's $$$ tool that does all the stripping in one quick step. Sadly, I didn't find that tool at the surplus store when I bought this cable.
After the audio portion is done and tested, then it's time for me relearn everything I've forgotten about basic Arduino coding!
Nice layout, with power supply wiring and xformers well away from the signal path wiring. Looks good. I agree, working with shielded cable connections is a major hassle. On mine, I used Mogami microphone cable to go from the PCB output pads to the rear panel RCA jacks. Since my cable run was only about 4 inches, I completely removed the black PVC jacket & outer shield and just used the two inner cables, twisting them slightly along the way. I don't hear any hum, so I think I lucked out. Your runs are a bit longer and closer to the xformers, so probably best to shield them completely.
Basic wiring is done. I was hearing a little hum and saw there was almost 2v of AC on the heater supply. I added a small board with an extra stage of RC. It’s now down to under 100mv. I’m hoping that took care of the audible portion. Time to run some more tests.
Some quick measurements in REW into a 19.6K load resistance. Flat as a pancake, as they say. THD is a little higher than before, but I'm running a different pair of tubes. Time to hook it up in the Audio Dungeon system.
I completed the input mod, changing R12/R17 from 4.7k to 22k. Since I had to remove the cathode bypass caps to get at these resistors, I decided to leave the cathode bypass caps out instead of soldering them back in place. This helped quite a bit. I now have much lower distortion and much less gain. Closed loop gain is now down to about 6dB on my unit, compared to about 10 to 15dB before. I ran loopback distortion and spectrum plots in REW at several voltage and loading levels and THD is now down to as low as 0.043% on some of my tests. 2nd order HD creeps up a little bit at the highest output levels, but remains well in check at what I would consider to be normal to moderate listening levels. Frequency response is ruler flat into all loads, except at the very highest and lowest frequencies (10Hz and 20kHz). I am really liking the way this turned out. Here are a few graphs from my tests:
Frequency Response into all output voltage levels and loads tested (varying from 115mVrms to 1.224Vrms output, driving into loads of 20k or 96k with 600pF of cable capacitance):
Harmonic distortion with 57mVrms input, 115mVrms output, driving into a 20k/600pF load:
HD with 103mVrms input, 206mVrms output, driving into a 20k/600pF load:
HD with 394mVrms input, 777mVrms output, driving into a 20k/600pF load:
HD with 627mVrms input, 1.22Vrms output, driving into a 20k/600pF load:
HD with 57mVrms input, 127mVrms output, driving into a 96k/600pF load:
HD with 103mVrms input, 227mVrms output, driving into a 96k/600pF load:
Preamp volume was turned all the way up, maximum gain, for all of the above distortion tests. I did not test higher voltage drive levels into 96k because the Steinberg UR22 did not have enough gain to complete higher drive voltage tests when the input was set to Hi Z. I need to build a unity gain buffer circuit and insert it between preamp and soundcard in order to test into 96k at higher levels.
I also ran another spectrum plot, which looks fairly good. The peak at 60Hz is a ground loop in my test setup that I need to address. I tried unplugging the laptop and running on batteries only, but that did not help. Probably picking up 60Hz hum from my DMM probes or something like that. Note added: The below spectrum was taken at the 394mVrms input level, 777mVrms output level, driving into a 20k/600pF load:
I found out what was causing the 60Hz hum spike on the spectrum plot that I attached to my previous post. I am using two digital multimeters to monitor the rms input and output voltages going to and from the preamp for all tests. I have the two meters connected to the RCA cables using Y splitters and alligator clips. The 60Hz spike was simply noise being picked up by improperly shielded cables. When I removed the two digitial multimeters and the Y splitters and then repeated the test with properly shielded RCA cables, the problem disappeared. Here is the spectrum plot with the proper shielding in place, using the same 394mVrms input level, 777mVrms output level, but now driving into a 20k/300pF load:
This is a fairly high playback level, and the power supply hum and noise are almost completely gone.
Nice work Bill!