Tickled Pink Again . . .

[cheapo_eddy]

. . . now that I've finally banished the last bit of barely audible 60 Hz ground loop hum from my ongoing little 12AU7 linestage DIY project, after seemingly endless (but fun) trial-and-error experimentation:

http://wp.cascadetubes.com/wp-content/uploads/2013/09/4S-schematic.jpg

Out of pure curiosity, one question remains: being quite satisfied with the "noise floor" of this thing, since it's plenty low enough for my crappy hearing, there remains a low-level white noise you can only hear in headphones while the amp is cranked way beyond any sane listening level, with no input signal present. Is this the fabled Johnson-Nyquist noise? Is it more likely coming from the cheap silicon diodes I used in place of the tube rectumfinder called for in the original design, or is it simply an endearing part of an audio preamp tube's thermionic personality?

Either way, I don't plan to "fix" it, being thoroughly happy at long last with this outcome. Just really curious.

 

[gadget73]

White noise is usually either the tube itself, or plate resistors. To a lesser extent it could be the cathode resistor. Basically any resistor current flows through can be a source of noise, and the larger the resistor value, the more noise it can make. Most of the time if you used some flavor of film resistor it will be fine but if this has carbon comp resistors for "vintage tone", I wouldn't be remotely surprised if thats what it is.

 

[trainbuftony]

Yes if it sounds like water rushing over a waterfall then you're probably hearing the current move through your plate resistors, or as you correctly stated Johnson Nyquist noise.

 

[Lavane]

. . . now that I've finally banished the last bit of barely audible 60 Hz ground loop hum from my ongoing little 12AU7 linestage DIY project, after seemingly endless (but fun) trial-and-error experimentation:

http://wp.cascadetubes.com/wp-content/uploads/2013/09/4S-schematic.jpg

Out of pure curiosity, one question remains: being quite satisfied with the "noise floor" of this thing, since it's plenty low enough for my crappy hearing, there remains a low-level white noise you can only hear in headphones while the amp is cranked way beyond any sane listening level, with no input signal present. Is this the fabled Johnson-Nyquist noise? Is it more likely coming from the cheap silicon diodes I used in place of the tube rectumfinder called for in the original design, or is it simply an endearing part of an audio preamp tube's thermionic personality?

Either way, I don't plan to "fix" it, being thoroughly happy at long last with this outcome. Just really curious.

I'm working on one of these myself.

 

[gadget73]

only thing I might suggest, depending on what you are running downstream, is to add a second tube as a cathode follower. Into a tube amp it won't matter but if you want it to be happy under 100k or so the buffer could be useful.

 

[trainbuftony]

Just for giggles, short the input jack.
That half meg resistor could easily pick up noise and the grid its attached to will amplify all of it.
The amplifier has no means by which to differentiate between music, and noise at that point. Even on a a feedback amplifier the grid usually isnt in the feedback current path loop and noise at that node will not be correctable with feedback.
Your tools are going to be things like layout and lead dress. If you find that the noise gets in via the grid. Obviously you'll need to use the quietest resistors you can budget for...

What if you replaced that resistor with a transformer secondary and had the primary terminate in 600 ohm balanced (XLR) connection?

That would be the final step in noise reduction before I simply chose a different circuit.

I'll stop redesigning your amp now.
Enjoy yourself. Each thing you build should be a building block towards bigger and better things.

 

[Lavane]

Grounding with this simple single stage preamp is the important bit.
If you go to the website and look at the underside picture it's all pretty basic parts other than using metal film resistors. I'm adding a switch to connect or disconnect the optional cathode bypass caps. Supposedly it sounds better at low volumes with the caps in circuit and better without them at higher volumes.
I haven't added the ground lift switch for the preamp circuit, but will later if needed. It can use a 12AU7,12AY7, 12AT7 and 12AX7. The 12AU7 should work fine to drive most amplifiers though.
https://www.cascadetubes.com/the-universal-preamp/

 

[gadget73]

I was going to make a comment about the volume control placement. I see its been discussed, but not my specific concern about it. With no way to reduce input, I can see the potential for overload with a really hot source. Not sure what the max signal input is on this though, so it may not actually be a problem.

 

[Lavane]

To the OP, how about a picture of your build? I've decided to add the ground lift for the grounds of the RCA jacks.

 

[Bob@FM]

I was going to make a comment about the volume control placement. I see its been discussed, but not my specific concern about it. With no way to reduce input, I can see the potential for overload with a really hot source. Not sure what the max signal input is on this though, so it may not actually be a problem.

That's a really good point - this is always running the tube at maximum volume, where the distortion characteristics are the worst. Many fixed sources put out 1-2 V rms, so you are dealing with peak to peak up to 5.6 V going in, with 20 dB gain (X10) could be 56 V pk-pk all the time.
If the answer is to vary the volume of the source, then you don't really need a volume control on the output. You can't control what the impedance of the next stage is, and the connection to it (cables) so now you have a variable low pass filter depending on volume position. The 250k pot will have ~ 62k ohm output impedance at 50% volume, plus the tube, so his numbers in the table for Ro (output impedance) are believable - but only for "that" volume setting. At different settings, it will be different.

I also scratched my head when I read this "The single dominant high frequency pole here is due to the input capacitance of my test equipment. The tube input capacitance is at least an order of magnitude smaller and is swamped by the test equipment and the nest of wires on the table." He shows a typical 10:1 probe is used, which have pretty low C compared to RCA interconnect cables to an amp. And the typical scope has a 1M ohm input impedance, a really easy load to drive.

 

[cheapo_eddy]

To the OP, how about a picture of your build? I've decided to add the ground lift for the grounds of the RCA jacks.

In several previous versions, the ground lift switch proved useless, so I omitted it on this latest contraption, and star-grounded everything. Out of hum-phobia, I put the electronics and power supply in entirely separate enclosures, which is seriously ugly.

It ain't pretty. All of my Frankenjunk is cobbled from detritus found in and around dumpsters, post-yard-sale "FREE" piles, etc., with minimal exceptions made for shrink tubing, solder, etc. So the aesthetics tend to be deliberately whimsical, and way bigger & heavier than necessary because... who cares? Also, being huge makes it easy to modify. This particular random assortment of trash includes bits from a crusty old Gulbransen two-manual organ, a Zenith TV set, a Magnavox console stereo, a weird old Nikko receiver, a very ancient Burroughs punch-card reader, the bottom half of a GI surplus ammo can, a repurposed headphone amp enclosure with rear panel markings that are obviously no longer relevant, a "cathode-ray oscillograph" circa 1944 (even used some of the wiring from that, which was milspec and in surprisingly good shape) along with various items from the trash bins at a few aerospace manufacturing facilities where I was once employed, e.g. the "toggle" switch serving as an output selector, which is actually a rather exotic (for its day) cam-action leaf-type 6PDT with low-noise gold knife contacts, that came from the instrument panel of a Cold War-era helicopter or something like that, and would have otherwise ended up incorporated into a Twilight Zone prop had it not been thrown away. The stamped-aluminum gear serving as an escutcheon plate came from the innards of a Voice Of Music tape recorder just like one that was in my fourth-grade classroom during the Johnson administration.

Anyway, don't say I didn't warn you that this thing is ugly as homemade sin:

 

[Lavane]

It looks good. I recently repurposed an old chassis. It has so many holes in it now it should have plenty of ventilation. Luckily its hidden inside a wood cabinet I made. Those military cans can make for nice chassis. I've been wanting to build a small boom box in one of the larger ammo cans I have. I added the ground lift.
I do run into ground loop problems once in awhile. It happened with a tube preamp. It luckily had a ground lift switch. Reading through the thread on these over at DIY, the consensus seems to be that the 12AU7 and 12AY7 are the best choice for tubes. So you went with the SS rectifier it looks like. I had to order some 100k resistors. I have several boxes of resistors, but I guess I ran out of 100k's. I'm adding the 32uf cathode bypass caps on a switch. I'm not sure it's really needed at all though.

 

[cheapo_eddy]

That's a really good point - this is always running the tube at maximum volume, where the distortion characteristics are the worst. Many fixed sources put out 1-2 V rms, so you are dealing with peak to peak up to 5.6 V going in, with 20 dB gain (X10) could be 56 V pk-pk all the time.
If the answer is to vary the volume of the source, then you don't really need a volume control on the output. You can't control what the impedance of the next stage is, and the connection to it (cables) so now you have a variable low pass filter depending on volume position. The 250k pot will have ~ 62k ohm output impedance at 50% volume, plus the tube, so his numbers in the table for Ro (output impedance) are believable - but only for "that" volume setting. At different settings, it will be different.

I also scratched my head when I read this "The single dominant high frequency pole here is due to the input capacitance of my test equipment. The tube input capacitance is at least an order of magnitude smaller and is swamped by the test equipment and the nest of wires on the table." He shows a typical 10:1 probe is used, which have pretty low C compared to RCA interconnect cables to an amp. And the typical scope has a 1M ohm input impedance, a really easy load to drive.

Ha! I owe a debt of gratitude to you and gadget73 for having this conversation which led to a sinfully easy fix. I likewise noticed distortion with material that obviously overdrives the input, and not wanting to inadvertently diddle impedance, response curve, etc., I suddenly remembered that I'm only using the tube linestage to add a little even-harmonic warmth to my computer's onboard audio (surprisingly decent, with 192KHz sampling), so off to Windows 11 Control Panel I sallied thence, cranked the system volume down to a much more civilized setting, and now the beast seems tamed. All my CDs and vinyl are ripped to .flac at "full lossless" and reside on a non-boot 1TB SSD. So again, thanks for the feedback.

 

[cheapo_eddy]

That's a really good point - this is always running the tube at maximum volume, where the distortion characteristics are the worst. Many fixed sources put out 1-2 V rms, so you are dealing with peak to peak up to 5.6 V going in, with 20 dB gain (X10) could be 56 V pk-pk all the time.
If the answer is to vary the volume of the source, then you don't really need a volume control on the output. You can't control what the impedance of the next stage is, and the connection to it (cables) so now you have a variable low pass filter depending on volume position. The 250k pot will have ~ 62k ohm output impedance at 50% volume, plus the tube, so his numbers in the table for Ro (output impedance) are believable - but only for "that" volume setting. At different settings, it will be different.

I also scratched my head when I read this "The single dominant high frequency pole here is due to the input capacitance of my test equipment. The tube input capacitance is at least an order of magnitude smaller and is swamped by the test equipment and the nest of wires on the table." He shows a typical 10:1 probe is used, which have pretty low C compared to RCA interconnect cables to an amp. And the typical scope has a 1M ohm input impedance, a really easy load to drive.

Because my old but still functional Tek 2215A scope is mainly just for decorative purposes, and all the 250K pots in my junkbox - even the nice clean ones with low mileage - seem to badly degrade the output signal regardless of placement or shielding, and because I like to control volume at the amp anyway, I resorted to making a voltage divider with 200K & 40K film resistors, giving the equivalent of a pot set at about 80%. Seems to have been a workable guess in terms of audibly acceptable results.

 

[cheapo_eddy]

It looks good. I recently repurposed an old chassis. It has so many holes in it now it should have plenty of ventilation. Luckily its hidden inside a wood cabinet I made. Those military cans can make for nice chassis. I've been wanting to build a small boom box in one of the larger ammo cans I have. I added the ground lift.
I do run into ground loop problems once in awhile. It happened with a tube preamp. It luckily had a ground lift switch. Reading through the thread on these over at DIY, the consensus seems to be that the 12AU7 and 12AY7 are the best choice for tubes. So you went with the SS rectifier it looks like. I had to order some 100k resistors. I have several boxes of resistors, but I guess I ran out of 100k's. I'm adding the 32uf cathode bypass caps on a switch. I'm not sure it's really needed at all though.

I've taken to sticking a 3300/25V filter cap across the output of a 6.5V SMPS wall-wart to give nice smooth tightly regulated DC to tube filaments, while retaining the two 100Ω rocks specified for ripple cancellation in the original AC circuit - which I imagine is probably pointless, but at any rate, I'm now convinced that DC filaments are the only way to fly. And BTW, yes, the big old honkin' 5U4 or whatever it was supposed to be didn't even get considered, even though I have a few of them lying around. A couple of 1KPIV 3A diodes seem to work just fine.

 

[Lavane]

I have regulated DC heaters in in a couple SE amplifiers and a couple preamps.
Including my 300B. I started like DC heaters in my first 300B build. I may wind up doing it with this preamp also.

 

[Bob@FM]

Because my old but still functional Tek 2215A scope is mainly just for decorative purposes, and all the 250K pots in my junkbox - even the nice clean ones with low mileage - seem to badly degrade the output signal regardless of placement or shielding, and because I like to control volume at the amp anyway, I resorted to making a voltage divider with 200K & 40K film resistors, giving the equivalent of a pot set at about 80%. Seems to have been a workable guess in terms of audibly acceptable results.

Great solution, I would try getting the output from the top of the preamp output R to gnd, with no R in the way. See if you can hear a difference.
That 40K R just increases your output impedance, which usually most preamp designers engineer to be relatively low. Tube tuners and pre-amps can have output impedance of 5-10K ohms. Most vintage SS stuff is under 2K, and audiophile stuff under 100 ohms. With really short cables it may not matter, but for long ones, it does. Also depends a lot on the next stage impedance.

 

[cheapo_eddy]

Great solution, I would try getting the output from the top of the preamp output R to gnd, with no R in the way. See if you can hear a difference.
That 40K R just increases your output impedance, which usually most preamp designers engineer to be relatively low. Tube tuners and pre-amps can have output impedance of 5-10K ohms. Most vintage SS stuff is under 2K, and audiophile stuff under 100 ohms. With really short cables it may not matter, but for long ones, it does. Also depends a lot on the next stage impedance.

I might just do that, and leave the ground resistor at its present value, since the output of this thing is pretty "hot" anyway.

 

[Bob@FM]

The other thing you could try - put a series 500K R on the input, in front of the 500K to ground there. Doing so will reduce the signal by 50%.

 

[gadget73]

the bigger the series resistor, the more HF roll-off you'll get though. Thank you Mr Miller and your blasted "effect". Grid to plate capacitence * stage gain + 1 = the Miller load. Stick 500k in series with that cap value to ground and you have a low pass filter that you can calculate the roll-off points for.

basically the lower half of that voltage divider varies considerably with impedance, so its not a 50% split across the band.