Those are some serious Spice skills, euro21! I had no idea that LTSpice could model things like EMI interference due to physical proximity. It's certainly valid to deduce that the filament PT could have been inducing hum in the anode choke, since the hum disappeared when the DC bench supplies were swapped in (and the AC PT removed from the breadboard). Hum also disappeared instantly when the 10Y filament power was cut, but not when either of the two high voltagerails or 2A3 filament power were cut. When the transformer was on the board, it was about four inches (10cm) from the right channel anode choke, and rotated 90 degrees. The hum was in both channels evenly. I can put it back on at a further distance and see if that changes anything.
I read Dmitry's page on filament-induced 120Hz hum when you first posted it. I learned a lot from it, and the phenomenon he investigates there seems exactly like what I'm experiencing -- at least the symptoms if not the underlying causes. The direction I've been following since then was to try the DC filtering and then AC hum-cancelling notions that I had, and then move on on to the approaches and solutions listed on Dmitry's page.
Having failed at my first attempts, I'm now trying to model in LTSpice what I did and what Dmitry did, but need to find a DHT symbol and model that has the two filament pins as in your schematic. I might have found one on the Duncan Amps wedbsite but haven't had a chance to use LTSpice today.
Thanks Thekak. I have not specifically modeled the filament circuit in LTSpice, and am teaching myself how to do that now. Last week I did find one wiring error in the DC filament supply, which I corrected. However, it had no effect on 120Hz hum. I've checked it over so many times, that I do believe it's all wired correctly.
Re: separate filament transformers for each channel: I started the 10Y stage of this project with a separate 10V transformer in each channel, with the diode bridge and DC filter, and had excessively high voltage and a pretty loud 120Hz hum. That was why I switched to a single 6.3V transformer (I had only one on hand) with separate diode bridges, with which the hum was actually less loud. When I tried the two AC methods last night, I was using separate transformers: the standalone 6.3V and a 6.3V secondary from one of the HV power transformers that wasn't being used. Believe it or not, the 10Y filaments ran at exactly 7.5VAC with the 6.3V transformers. The AC arrangements still induced the 120Hz hum -- even though only 60Hz AC was being applied.
It's certainly very tempting to just purchase the Coleman regs and be done with it. However, because I don't understand this yet, I don't know if those will actually help, espcially if there's some other error that I'm committing within the circuit. So I'd prefer to learn the physics of what's happening if I can, starting with modeling in LTSpice.
I really appreciate the time and effort that you folks are putting in!
I read Dmitry's page on filament-induced 120Hz hum when you first posted it. I learned a lot from it, and the phenomenon he investigates there seems exactly like what I'm experiencing -- at least the symptoms if not the underlying causes. The direction I've been following since then was to try the DC filtering and then AC hum-cancelling notions that I had, and then move on on to the approaches and solutions listed on Dmitry's page.
Having failed at my first attempts, I'm now trying to model in LTSpice what I did and what Dmitry did, but need to find a DHT symbol and model that has the two filament pins as in your schematic. I might have found one on the Duncan Amps wedbsite but haven't had a chance to use LTSpice today.
Thanks Thekak. I have not specifically modeled the filament circuit in LTSpice, and am teaching myself how to do that now. Last week I did find one wiring error in the DC filament supply, which I corrected. However, it had no effect on 120Hz hum. I've checked it over so many times, that I do believe it's all wired correctly.
Re: separate filament transformers for each channel: I started the 10Y stage of this project with a separate 10V transformer in each channel, with the diode bridge and DC filter, and had excessively high voltage and a pretty loud 120Hz hum. That was why I switched to a single 6.3V transformer (I had only one on hand) with separate diode bridges, with which the hum was actually less loud. When I tried the two AC methods last night, I was using separate transformers: the standalone 6.3V and a 6.3V secondary from one of the HV power transformers that wasn't being used. Believe it or not, the 10Y filaments ran at exactly 7.5VAC with the 6.3V transformers. The AC arrangements still induced the 120Hz hum -- even though only 60Hz AC was being applied.
It's certainly very tempting to just purchase the Coleman regs and be done with it. However, because I don't understand this yet, I don't know if those will actually help, espcially if there's some other error that I'm committing within the circuit. So I'd prefer to learn the physics of what's happening if I can, starting with modeling in LTSpice.
I really appreciate the time and effort that you folks are putting in!
I zipped the necessary files for simulating:
I hope that it's enough for testing.
Try it.
p.s.
Because it has AC filament PSU, only "Transient" and "AC Analysis" simulating command working.
It's works with 9 second delay, 1s run, because extra large LT PSU capacitors charging slowly.
If you want to test amplifier with tighter parameters (must to change LT PSU with 7.5V DC PSU), change the
" .tran 0 10 9 " line with
" .tran 0 {ncycles*period} {(ncycles/2)*period} {period/1e4} ".
Must to set input (peek) voltage of V6, because now 'Amplitude(V)' is zero (for testing hum).
The amplifier with these parameters at 5.9V Input (peek) reaches A1 limit.
p.s. 2.
The original dmitry_composites.lib extended with VT25_COMPOSITE model (borrowed by Ale Moglia measured parameters).
- 10Y DRD 2A3 SE_stacked PSU_jdrouin_2.asc
- dmitry_composites.lib
- triodedht.asy
I hope that it's enough for testing.
Try it.
p.s.
Because it has AC filament PSU, only "Transient" and "AC Analysis" simulating command working.
It's works with 9 second delay, 1s run, because extra large LT PSU capacitors charging slowly.
If you want to test amplifier with tighter parameters (must to change LT PSU with 7.5V DC PSU), change the
" .tran 0 10 9 " line with
" .tran 0 {ncycles*period} {(ncycles/2)*period} {period/1e4} ".
Must to set input (peek) voltage of V6, because now 'Amplitude(V)' is zero (for testing hum).
The amplifier with these parameters at 5.9V Input (peek) reaches A1 limit.
p.s. 2.
The original dmitry_composites.lib extended with VT25_COMPOSITE model (borrowed by Ale Moglia measured parameters).
Last edited:
At 10V AC filament transformer simulation I was "tricking" with CMC chokes DCR (use higher values, than presumable) to reach appropriate output voltage at given current (10Y filament current + 33R-33R current).
If you know (measured) DCR parameters of chokes, suitable series resistors (between filter stages) help to set 7.5V on 10Y filament pins.
6.3V AC is in the low side.
If the filament transformer gives 6.3V at -at least- 5, rather 6A (near 3A load!), theoretically possible to reach the 7.5V 1.25A output, if diode bridge and chokes eats quite a bit.
I sometimes use 7.5V AC (40-60VA) for raw DC filament but only with low DCR C-CMC-C PSU.
Success! I tried a symmetrical, passive filament supply on the 10Y with common mode chokes, caps, resistors, and a hum pot. The filter section is floating, with the hum pot wiper connected to the cathode resistor and bypass cap, which in turn go to ground. It's dead quiet and sounds very good.
Schematic is attached. At euro21's suggestion, I wired the negative leg of the CMCs in reverse, so that the opposite direction would cause ripple and magnetism to cancel out.
I wonder if the pot is really necessary because I can only hear the slightest amount of hum -- with my ear up to the speaker -- when rotated to the extremes. A virtual center tap might be fine.
So nice to be able to power up the breadboard and simply listen to music again.
Schematic is attached. At euro21's suggestion, I wired the negative leg of the CMCs in reverse, so that the opposite direction would cause ripple and magnetism to cancel out.
I wonder if the pot is really necessary because I can only hear the slightest amount of hum -- with my ear up to the speaker -- when rotated to the extremes. A virtual center tap might be fine.
So nice to be able to power up the breadboard and simply listen to music again.
Attachments
Looks like Brute Force has been used here, Anyone familiar with the common mode choke?
Check it out on Wiki.👍
The other concern is the 10Y's emission at low filament temp. Something to check out. 🙂
Check it out on Wiki.👍
The other concern is the 10Y's emission at low filament temp. Something to check out. 🙂
Jdrouin,
Have you ever looked at Russian DHTs as a driver? Ale Moglia has done a ton of curve tracing of various Russian types. They are affordable and sound very good in the drive position.
I apologize if his work has already been mentioned.
Have you ever looked at Russian DHTs as a driver? Ale Moglia has done a ton of curve tracing of various Russian types. They are affordable and sound very good in the drive position.
I apologize if his work has already been mentioned.
Thanks for the suggestion, ggetzoff. I'm aware of his work with Russian DHTs but haven't ventured to try any yet. Will look into them for potential rotation into the trials here.
Brief report on a recent change in the circuit. Again this is the choke-loaded 10Y direct-coupled to 2A3 with a stacked power supply and input transformers at a 1:5 step up. The 2A3 is always run at 250V/60mA, the standard operating point and also my personal preference after trying several other ones.
The 10Y had been running at 260Vak/14mA and sounded good.
I wanted to aim for the middle operating point on the regular type 10 datasheet, which is 350Vak/16mA. So I swapped in a 650Vct power transformer and a GZ34 rectifier, and now have it running at 335Vak/17mA. I'll adjust it further until it reaches that op point from the datasheet but already the improved performance is very audible. The 10Y definitely likes to be run at higher voltage with higher current, so that will guide the next sequence of adjustments and tests.
Brief report on a recent change in the circuit. Again this is the choke-loaded 10Y direct-coupled to 2A3 with a stacked power supply and input transformers at a 1:5 step up. The 2A3 is always run at 250V/60mA, the standard operating point and also my personal preference after trying several other ones.
The 10Y had been running at 260Vak/14mA and sounded good.
I wanted to aim for the middle operating point on the regular type 10 datasheet, which is 350Vak/16mA. So I swapped in a 650Vct power transformer and a GZ34 rectifier, and now have it running at 335Vak/17mA. I'll adjust it further until it reaches that op point from the datasheet but already the improved performance is very audible. The 10Y definitely likes to be run at higher voltage with higher current, so that will guide the next sequence of adjustments and tests.
Yeah, the type-10 is a really neat driver and unfortunately a no-go for me personally.
The Russian DHTs are affordable and he’s provided lots of curves, it’s a great sight.
Another thing that would be fun would be to drive the 2A3 with a kink-less.
While you continue your quest / sage thing I’m having fun with some Tubelab builds
The Russian DHTs are affordable and he’s provided lots of curves, it’s a great sight.
Another thing that would be fun would be to drive the 2A3 with a kink-less.
While you continue your quest / sage thing I’m having fun with some Tubelab builds
Attachments
I ran a similar setup to the one described for a while - 10Y into 2a3 with a step-up in front, in my case a Hammond 1140-LN-C in 1:4. I wasn't entirely happy with it, although it was quite good. I think the step-up took something away from the sound.
I've ended up with PSE 4P1L with a 27 mesh plate driver. So just 2 stages. I don't know if it's objectively better but I'm more comfortable with it. Hugely cheaper, in fact very cheap indeed. I have about 40 of 4P1L which I've tested and labelled so I can match the output tubes forever. I think what I like about it is that it's much simpler to build and the 2-stage sound is honest and direct with a good amount of DHT subtlety. The 27 mesh should last well, but I have 8 of them anyway. This is probably my end game setup now.
I could also use a 26 as driver stage but I'm not rushed even though I have it ready. The 10Y didn't match well with the 4P1Ls - the sound was too lean and needed filling out with a richer driver stage. The 27 mesh seems to do that OK.
I've ended up with PSE 4P1L with a 27 mesh plate driver. So just 2 stages. I don't know if it's objectively better but I'm more comfortable with it. Hugely cheaper, in fact very cheap indeed. I have about 40 of 4P1L which I've tested and labelled so I can match the output tubes forever. I think what I like about it is that it's much simpler to build and the 2-stage sound is honest and direct with a good amount of DHT subtlety. The 27 mesh should last well, but I have 8 of them anyway. This is probably my end game setup now.
I could also use a 26 as driver stage but I'm not rushed even though I have it ready. The 10Y didn't match well with the 4P1Ls - the sound was too lean and needed filling out with a richer driver stage. The 27 mesh seems to do that OK.
Last edited:
ggetzoff: My very first amp was a Tubelab SE 300B, which I made about a decade ago (!!!). I liked it but the mosfets killed the sound, IMO, and it gave me the bug to start building from scratch -- so here we are! Though I would certainly not call myself a guru. I barely understand the physics or the mathematics beyond knowing what formula I need to use for this and that, and working knowledge of components.
Andy: Thanks again for the input. Many European and British designers seem to like the 4P1L, so I'll have to try it one of these days. For a time, on the breadboard I had a mesh plate 27 driving 6N6G, which sounded better than I thought it would. I'll probably try it with the 2A3 and again with 45 at some point. I had 26 driving 2A3, which was great, and 26 driving 45 -- a match that just sounded "right," as if it were meant to be. I'm going to do that 26 to 45 build in the future, after all this experimentation and after the final 2A3 amp is built. I even have a beautiful white oak chassis with medullary rays that was a cast-off mistake on a piece of furniture I made a couple years ago, but would be perfect for a tube amp. 😆
To be more specific about the sound improvements of running the 10Y hotter into the 2A3 -- it has more body and fullness, as well as spatial depth and width, more detail, attack/decay, air/sparkle and every audiophile adjective in between, all while sounding warm, musical, and involving. Like 26 direct-couple to 45, 10Y (with hotter op point) direct-coupled to 2A3 just sounds "right" to my ears.
I have output transformers to make a 26 line preamp, so at some point I'll implement that on the breadboard and remove the input trannies, to hear how it sounds. Someone very generously gave me a transformer volume control, so that can go in front of the preamp too.
And in case anyone is interested, in LTSpice I've designed a 3-stage all-dht all-direct-coupled amp, with a triple stacked power supply, which is 26 --> 10Y --> 300B. That's definitely going on the breadboard after the 2A3 circuit is finally settled. 😈
Andy: Thanks again for the input. Many European and British designers seem to like the 4P1L, so I'll have to try it one of these days. For a time, on the breadboard I had a mesh plate 27 driving 6N6G, which sounded better than I thought it would. I'll probably try it with the 2A3 and again with 45 at some point. I had 26 driving 2A3, which was great, and 26 driving 45 -- a match that just sounded "right," as if it were meant to be. I'm going to do that 26 to 45 build in the future, after all this experimentation and after the final 2A3 amp is built. I even have a beautiful white oak chassis with medullary rays that was a cast-off mistake on a piece of furniture I made a couple years ago, but would be perfect for a tube amp. 😆
To be more specific about the sound improvements of running the 10Y hotter into the 2A3 -- it has more body and fullness, as well as spatial depth and width, more detail, attack/decay, air/sparkle and every audiophile adjective in between, all while sounding warm, musical, and involving. Like 26 direct-couple to 45, 10Y (with hotter op point) direct-coupled to 2A3 just sounds "right" to my ears.
I have output transformers to make a 26 line preamp, so at some point I'll implement that on the breadboard and remove the input trannies, to hear how it sounds. Someone very generously gave me a transformer volume control, so that can go in front of the preamp too.
And in case anyone is interested, in LTSpice I've designed a 3-stage all-dht all-direct-coupled amp, with a triple stacked power supply, which is 26 --> 10Y --> 300B. That's definitely going on the breadboard after the 2A3 circuit is finally settled. 😈
@jdrouin
"Someone very generously gave me a transformer volume control, so that can go in front of the preamp too."
TVC always need small/er/ source impedance for "good sounding".
If you have -classical- tube output sources (phono, DAC), this doesn't always happen.
Because of this I use -in my #26 preamp- TVC as output of preamp .... but the design (gyrator loaded tube) guaranteed low source impedance for TVC.
"Someone very generously gave me a transformer volume control, so that can go in front of the preamp too."
TVC always need small/er/ source impedance for "good sounding".
If you have -classical- tube output sources (phono, DAC), this doesn't always happen.
Because of this I use -in my #26 preamp- TVC as output of preamp .... but the design (gyrator loaded tube) guaranteed low source impedance for TVC.
Generic VT25/10Y has enough large Ri.
If you use this tube with usual load (resistor, choke, transformer), the tube's output impedance (and the load) strongly influenced the "tone".
If Ri smaller (hotter operating point), the "tone" usually would be more lovable.
If you use these tubes with more complex load (for example CCS, gyrator etc.), the effect is less noticeable.
If you use this tube with usual load (resistor, choke, transformer), the tube's output impedance (and the load) strongly influenced the "tone".
If Ri smaller (hotter operating point), the "tone" usually would be more lovable.
If you use these tubes with more complex load (for example CCS, gyrator etc.), the effect is less noticeable.
Thanks euro21, I'll keep that in mind. I've got a DAC with 600 ohm output impedance, a CD player with 50 ohm output imedance, and a MacBook that automatically adjusts its output impedance to suit the device it plugs into.
Just a note to say that I've determined the final build for the 2A3 amp will be the direct-coupled design I've developed most recently, with a stacked power supply and 10Y driving 2A3. It just sounds superb with the 10Y running at 350Vak/18mA choke-loaded with cathode bias, and the 2A3 at 250Vak/60mA in grid bias.
I still have to assemble the Rod Coleman regulator, and will try it in normal filament supply mode with cathode bias as well as in filament bias mode, and then do a final build with whichever sounds best. More details to come once it's finally settled.
Also, I decided to try this same topology with 300B output tubes, and -- WOW! It's just a wonderful sonic presentation, with a huge sense of space and notable authority in complex passages. At first it sounded underwhelming with the budget Psvane Hi-Fi Series 300Bs running at 350Vak/60mA, and marginally better at 400Vak/60mA. But when I swapped in the pair of Svetlana winged "C" 300Bs, it was an almost cosmic transformation. Tonight I boosted the current, running them at 400Vak/70mA, and it's even better.
The signal chain is the same as the 2A3, using a stacked power supply but with an 800Vct PT and different OPTs:
The 300Bs really like being driven by 10Y at that operating point.
I still have to assemble the Rod Coleman regulator, and will try it in normal filament supply mode with cathode bias as well as in filament bias mode, and then do a final build with whichever sounds best. More details to come once it's finally settled.
Also, I decided to try this same topology with 300B output tubes, and -- WOW! It's just a wonderful sonic presentation, with a huge sense of space and notable authority in complex passages. At first it sounded underwhelming with the budget Psvane Hi-Fi Series 300Bs running at 350Vak/60mA, and marginally better at 400Vak/60mA. But when I swapped in the pair of Svetlana winged "C" 300Bs, it was an almost cosmic transformation. Tonight I boosted the current, running them at 400Vak/70mA, and it's even better.
The signal chain is the same as the 2A3, using a stacked power supply but with an 800Vct PT and different OPTs:
The 300Bs really like being driven by 10Y at that operating point.
Last edited: