I'm planning to build a linestage and the threads on the 26 tube have caught my attention. My preference is to build it without any integrated circuit chips such as current sources or voltage regulators.
Jim De Kort's LC-loaded Phaeton design appeals to my sensibility, though I'm wondering what to do about a hum-free, DC filament supply for the 26. He seems to use a battery coupled with a LM1086 , and I'm wondering if that would eliminate the "openness" I like and which he claims to be affected negatively by the chip, though it runs hum-free with stability.
Has anyone successfully tried a non-regulated filament supply for the 26? Excuse my inexperience, but I wonder if something like a 3V CT filament transformer could be used as a starting point, with the center tap halving the voltage, and then maybe using something like a diode bridge to convert to DC.
Thanks for any suggestions you might have.
Jim De Kort's LC-loaded Phaeton design appeals to my sensibility, though I'm wondering what to do about a hum-free, DC filament supply for the 26. He seems to use a battery coupled with a LM1086 , and I'm wondering if that would eliminate the "openness" I like and which he claims to be affected negatively by the chip, though it runs hum-free with stability.
Has anyone successfully tried a non-regulated filament supply for the 26? Excuse my inexperience, but I wonder if something like a 3V CT filament transformer could be used as a starting point, with the center tap halving the voltage, and then maybe using something like a diode bridge to convert to DC.
Thanks for any suggestions you might have.
The 26 takes 1.05A at 1.5V. Remember that a rectified and capacitor filtered DC supply causes extra heating in the transformer versus an AC resistive load. Allow for about 2x the filament current to have a little margin, so use a secondary that is rated for at least 2A.
Transformer peak voltage is 1.414 times the RMS voltage.
You can use a 5VAC secondary with Center Tap. 2.5V * 1.414 = about 3.5Vpeak from each side to the Center Tap. That will give you a full wave rectifier with 120Hz ripple (or
100 Hz ripple if you have 59 Hz AC Power.
With a Schottky diode drop of about 0.5V that leaves 3V. 3V - 1.5V filament = 1.5V still left over.
Now, what to do with the "extra" 1.5V ... put it to good use, to get better hum filtering.
You can use CRC filtering.
1.5V/1.05A = 1.42 Ohm. You will find that what you will need will be a little less than
1.4 Ohm, but a little more than 1 Ohm (you will see why below). Lets start with 1.3 Ohms.
If you use a 10,000 uF first filter cap from the diodes, and a 1.4 Ohm series resistor, and then a 10,000 uF filter cap, the result will be the following:
Capacitive reactance, Xc is 1/(2 * Pi * F * C).
Xc of 10,000 uF at 120 Hz is 0.13 Ohms.
The first filter is:
1.05A (0.13 Ohms) = 136 mV peak-to-peak ripple.
The second filter is 1.3 Ohms/0.13 Ohms Xc; that is 1/10 of the original 136mV peak-to-peak ripple = 13.6 mV Peak to Peak.
Peak-to-Peak/2.828 ~= RMS. 13.6/2.828 = 4.8 mV rms.
That is too much ripple for a preamp with direct heated filament.
If you change both capacitors from 10,000 uF to 20,000 uf, the ripple will be 1.2mV rms.
That is still a little too much ripple.
Rather than increase the capacitance even more, you can use a 12.6V 2A Center tapped secondary. You will be able to use a resistor that is 5x larger, about 6.5 Ohms, and that will give 1/5 the ripple, 1.2mV/5 = 240uV rms ripple.
But keep in mind that the tradeoff will be that the resistor will dissipate lots of power (Isquared * R) = 1.05 squared * 6.5 Ohms = 7.2 Watts. A 25 watt resistor will work here without changing color, etc.
Keep in mind that you need to use 2 resistors in series, across the filament, and the center of those 2 is the path for self bias or ground. Doing that will reduce the effect of the very small residual ripple even further.
Transformer peak voltage is 1.414 times the RMS voltage.
You can use a 5VAC secondary with Center Tap. 2.5V * 1.414 = about 3.5Vpeak from each side to the Center Tap. That will give you a full wave rectifier with 120Hz ripple (or
100 Hz ripple if you have 59 Hz AC Power.
With a Schottky diode drop of about 0.5V that leaves 3V. 3V - 1.5V filament = 1.5V still left over.
Now, what to do with the "extra" 1.5V ... put it to good use, to get better hum filtering.
You can use CRC filtering.
1.5V/1.05A = 1.42 Ohm. You will find that what you will need will be a little less than
1.4 Ohm, but a little more than 1 Ohm (you will see why below). Lets start with 1.3 Ohms.
If you use a 10,000 uF first filter cap from the diodes, and a 1.4 Ohm series resistor, and then a 10,000 uF filter cap, the result will be the following:
Capacitive reactance, Xc is 1/(2 * Pi * F * C).
Xc of 10,000 uF at 120 Hz is 0.13 Ohms.
The first filter is:
1.05A (0.13 Ohms) = 136 mV peak-to-peak ripple.
The second filter is 1.3 Ohms/0.13 Ohms Xc; that is 1/10 of the original 136mV peak-to-peak ripple = 13.6 mV Peak to Peak.
Peak-to-Peak/2.828 ~= RMS. 13.6/2.828 = 4.8 mV rms.
That is too much ripple for a preamp with direct heated filament.
If you change both capacitors from 10,000 uF to 20,000 uf, the ripple will be 1.2mV rms.
That is still a little too much ripple.
Rather than increase the capacitance even more, you can use a 12.6V 2A Center tapped secondary. You will be able to use a resistor that is 5x larger, about 6.5 Ohms, and that will give 1/5 the ripple, 1.2mV/5 = 240uV rms ripple.
But keep in mind that the tradeoff will be that the resistor will dissipate lots of power (Isquared * R) = 1.05 squared * 6.5 Ohms = 7.2 Watts. A 25 watt resistor will work here without changing color, etc.
Keep in mind that you need to use 2 resistors in series, across the filament, and the center of those 2 is the path for self bias or ground. Doing that will reduce the effect of the very small residual ripple even further.
Thanks, 6A3sUMMER. So, if I take your meaning correctly, you're suggesting:
* 12.6 V CT filament transformer
* CRC filter comprising a 20,000uF cap, a 6.5 Ohm 25W resistor, and a 20,000uF cap
* Two more resistors, one on each leg of the filament, going to ground
If that's correct, then would the CRC filter connect to the two "F" points on the schematic from Jim de Kort's page? And would the two filament-to-ground resistors have a value of 22 Ohm each, as pictured here?
His page says, "When using rectified DC for the filament add a 470uF cathode bypass capacitor!" Since this application would be diode rectified, would I need the cathode bypass capacitor?
I've attached a hand drawn schematic to see if I understand your suggestion.
I'm still a newbie at this, so please forgive any gross errors.
* 12.6 V CT filament transformer
* CRC filter comprising a 20,000uF cap, a 6.5 Ohm 25W resistor, and a 20,000uF cap
* Two more resistors, one on each leg of the filament, going to ground
If that's correct, then would the CRC filter connect to the two "F" points on the schematic from Jim de Kort's page? And would the two filament-to-ground resistors have a value of 22 Ohm each, as pictured here?
An externally hosted image should be here but it was not working when we last tested it.
His page says, "When using rectified DC for the filament add a 470uF cathode bypass capacitor!" Since this application would be diode rectified, would I need the cathode bypass capacitor?
I've attached a hand drawn schematic to see if I understand your suggestion.
I'm still a newbie at this, so please forgive any gross errors.
Attachments
Last edited:
Yes, Bypass the 1k Ohm resistor with the electrolytic (+ to top of the 1k, - to the bottom of the 1k (ground)).
Yes, the DC goes to f and f.
The 12.6V transformers usually come with a center tap. If it does have a center tap, leave it unconnected (put shrink tubing over the end). Do not connect the center tap to ground.
If it does not have a center tap, that is OK too.
The 6.5 Ohm resistor may have to be adjusted. Check the 26 filament for 1.5V, and adjust the 6.5 Ohms as necessary (let the filament warm up, the voltage will rise when the filament comes to final temperature, then verify it is 1.5V). You can do this with the B+ unconnected (disconnect the high voltage secondary from the high voltage rectifier diodes, otherwise the unloaded voltage will rise up too high).
That is how I do an amplifier, I get the filaments up and running, and finalized first.
Then I activate the B+ circuit.
Yes, the DC goes to f and f.
The 12.6V transformers usually come with a center tap. If it does have a center tap, leave it unconnected (put shrink tubing over the end). Do not connect the center tap to ground.
If it does not have a center tap, that is OK too.
The 6.5 Ohm resistor may have to be adjusted. Check the 26 filament for 1.5V, and adjust the 6.5 Ohms as necessary (let the filament warm up, the voltage will rise when the filament comes to final temperature, then verify it is 1.5V). You can do this with the B+ unconnected (disconnect the high voltage secondary from the high voltage rectifier diodes, otherwise the unloaded voltage will rise up too high).
That is how I do an amplifier, I get the filaments up and running, and finalized first.
Then I activate the B+ circuit.
Thanks! This is incredibly helpful.
In the 25W range around 6.5 Ohm, Mouser has 6, 6.04, 6.2, 6.8, and 7 Ohm values. One of those should work, though I'll sit down to figure out the math tomorrow to see what kind of range I should purchase if I decide to go this route.
Any other suggestions out there? Would choke loading the filaments be an option at this low voltage level?
In the 25W range around 6.5 Ohm, Mouser has 6, 6.04, 6.2, 6.8, and 7 Ohm values. One of those should work, though I'll sit down to figure out the math tomorrow to see what kind of range I should purchase if I decide to go this route.
Any other suggestions out there? Would choke loading the filaments be an option at this low voltage level?
Not sure if this is convertible to 26, but certainly gets rid of IC from filament supply. Resistors will generate a lot of heat and iron will be very 'spendy'.
VinylSavor: UX201A
Cheers
Grant
VinylSavor: UX201A
Cheers
Grant
I've built many DHT line stages - 10Y, 26, 4P1L, 01A etc. With resistors, plate chokes, interstages and lately with the Bartola Gyrator.
If you want the best sound you will build the 01A Gen 2 line stage on Bartola Valves. With Ale's gyrator and Rod Coleman's V7 regs.
By all means do all the experimenting yourself. It's just that many of us have already trodden those roads and we're pretty much agreed on the Gen 2. And there is nothing out there approaching Rod's V7 for DC.
If you want the best sound you will build the 01A Gen 2 line stage on Bartola Valves. With Ale's gyrator and Rod Coleman's V7 regs.
By all means do all the experimenting yourself. It's just that many of us have already trodden those roads and we're pretty much agreed on the Gen 2. And there is nothing out there approaching Rod's V7 for DC.
I am always reading Ale's Blog and every now and then I write him with some questions - there is indeed a lot to be learned from going through his posts!
What I find very nice is to see the evolution towards the gyrator: that the best sound comes from a combination of one of the earliest valves together with latest SS technology (SMD and all). And indeed, I am using gyrators as well, latest one based on DN2540 + 2SK715 as plate load, gyrators as filter in PS, etc.
We live in a good time for valve amplifiers: best of the past with lots of new goodies!
For 01A, 26 or 4P1L I'd recommend filament bias so you get rid of the cathode bypass cap and its colourations. But this does make some demands on the cleanness of the filament supply which have to be met. Standard voltage regs are inadequate for filament bias, and I gave those up years ago - you need something a lot better. Thomas Mayer uses choke input, and that's a good solution. Most of us DHT builders use Rod's regs and they're very good. I think this matters so much that I use choke input plus Rod's regs. Yes, you can hear a difference in smoothness on a revealing system.
We're now about 10 years down the line from 26 preamps, Rod's early reg designs, Thomas Mayer's use of filament bias and Ale Moglia's experiments and gyrator designs. I've been building DHT line stages and amps and nothing else during this period, and I've watched it evolve. Rod is now on his 7th iteration regulator, Thomas has a thriving business making DHT based designs in Germany and a great blog, and Ale has second and third generation gyrators and an equally great blog. Put all these clever heads together and we get designs like the 01A Gen 2 line stage, which to my ears is the best expression of the fusion of all this joint evolution. Filament bias + gyrator + Rod's regs + DHT of your choice.
What's interesting is that this will surely continue to evolve, and Rod, Thomas and Ale are actively looking for better designs all the time. Plus now we have a whole community around DHT line stages and amps, and evolution will also come from all the other guys that are now using this design culture. Interestingly, it's quite a European, not to say global, thing as well - we seem to have adopted tubes and an audio culture that started in the USA in the 1920s but now spans the whole audio community. Quite a lot of fans in Scandinavia as well - and they have Lundahl of course. It's all very interesting.
In terms of the tubes, though, they're all in the USA with the exception of the 4P1L and some boutique suppliers, plus China for output tubes. So this does mean that those of us outside the USA need to scour ebay for stocks of 01A, 26, 10Y etc. That's an undertaking in itself - for shipping to be economical you need to buy in units of 20 or so, and you don't get that on ebay. And dealers are usually 150% the price or more. And the tubes need to be tested as well. Yes - it's all very interesting.
Last edited:
Thank you for the background on the community evolution. I've particularly enjoyed reading through your posts in a number of threads, andyjevans. They're very lucid and do a remarkably clear job of combining technical acumen with sonic description. I have only a power amp (Tubelab SE 300B) and a phono stage (VTA PH16) under my belt, so I'm still very new to this and need more hands-on experience before I'll be able to understand something like the 01A Gen 2.
I will admit that I'm engaged in a Proustian search for Lost Time and am chasing a particular impression. 15 years ago I was fortunate enough to live with a friend who had a George Wright system comprising a phono stage, line stage, and 300B mono blocks (prototypes of the WPA Mono 7) designed to sound more like a 2A3 -- none of which used any ICs -- and I found the liquid sweetness and three-dimensionality of that system utterly intoxicating. I guess it was partly the second-order harmonics that gave that effect, which probably means that I prefer a more colored, vintage sound than many people design for today.
Since Wright Sound equipment almost never goes on sale, I'm trying to learn enough to build components that provide that kind of sound. I feel like I need to reach that initial point of satisfaction (and basic knowledge) before I'll be ready to go off in other directions.
After a line stage, the next project will be a pair of 300B mono blocks based on the JC Morrison's Fi Primer 300B, one of the JE Labs versions, or a WE91A clone. I'll breadboard a few of them and then do a final chassis build with the one I like best.
After that, depending on the results, either a 2A3 or an IC-less phono stage (the PH16 uses several), maybe a build of the classic RCA circuit. I like the Tubelab and the PH16, both of which use FETs and other ICs, but they don't have that emotional, Romantic sound of the Wrights.
Sorry for the long post, but I hope this explains some of the motivation for doing the 26 linestage in the way I've proposed. Plus it's just nice to connect with people about our interest in audio since there's not much of a local community where I live.
I will admit that I'm engaged in a Proustian search for Lost Time and am chasing a particular impression. 15 years ago I was fortunate enough to live with a friend who had a George Wright system comprising a phono stage, line stage, and 300B mono blocks (prototypes of the WPA Mono 7) designed to sound more like a 2A3 -- none of which used any ICs -- and I found the liquid sweetness and three-dimensionality of that system utterly intoxicating. I guess it was partly the second-order harmonics that gave that effect, which probably means that I prefer a more colored, vintage sound than many people design for today.
Since Wright Sound equipment almost never goes on sale, I'm trying to learn enough to build components that provide that kind of sound. I feel like I need to reach that initial point of satisfaction (and basic knowledge) before I'll be ready to go off in other directions.
After a line stage, the next project will be a pair of 300B mono blocks based on the JC Morrison's Fi Primer 300B, one of the JE Labs versions, or a WE91A clone. I'll breadboard a few of them and then do a final chassis build with the one I like best.
After that, depending on the results, either a 2A3 or an IC-less phono stage (the PH16 uses several), maybe a build of the classic RCA circuit. I like the Tubelab and the PH16, both of which use FETs and other ICs, but they don't have that emotional, Romantic sound of the Wrights.
Sorry for the long post, but I hope this explains some of the motivation for doing the 26 linestage in the way I've proposed. Plus it's just nice to connect with people about our interest in audio since there's not much of a local community where I live.
Last edited:
If you're really determined, you could use only chokes to build a quiet enough filament supply. But it's not an easy or cheap option, because the chokes must be very large, in order to:
- reduce the ripple-voltage to a low enough level;
- create somewhat high impedance looking out from the filament. (if you use a low-impedance supply, the differential music-signal across the filament is crushed by short-circuiting).
Raw rectified dc gives just about the worst sound possible, because the heating current is partly comprised of recharge and recovery pulses from the rectifier, and mains-noise is hardly suppressed at all. Raw dc has a low impedance, too, and the mains tolerance (often ±10%) shortens the life of the DHT.
It's true that IC regulators sound bad in this position: they use noisy bandgap references, have high leakage capacitance, and (e.g. LM317) have odd frequency peaks or notches. For all of these reasons, my regulator is designed with discrete transistors (no ICs!) to address all of the demands of the application.
- reduce the ripple-voltage to a low enough level;
- create somewhat high impedance looking out from the filament. (if you use a low-impedance supply, the differential music-signal across the filament is crushed by short-circuiting).
Raw rectified dc gives just about the worst sound possible, because the heating current is partly comprised of recharge and recovery pulses from the rectifier, and mains-noise is hardly suppressed at all. Raw dc has a low impedance, too, and the mains tolerance (often ±10%) shortens the life of the DHT.
It's true that IC regulators sound bad in this position: they use noisy bandgap references, have high leakage capacitance, and (e.g. LM317) have odd frequency peaks or notches. For all of these reasons, my regulator is designed with discrete transistors (no ICs!) to address all of the demands of the application.
Ah - so you're hoping to get to the last volume, "Le Temps Retrouve"....?
But consider this - 15 years ago you had a listening experience which transcended what you were used to. Isn't it possible that in 2017 you could equally have a listening experience that transcends Wright Sound?
My jaw literally dropped when I first tried out a 26 in my system, after years of EL84s and 6SN7s. I think I said out loud "This is the sound I've been looking for all my life!". Like you, I was transfixed. I still have my first 26 preamp build from around 2009. But it's not in the same class as the Gen 2 01A I use today. Time has moved on quite a lot since.
If you're just after nostalgia, then that's another story!
So you've been able to let go of your Albertine? ;-)
Don't get me wrong, I'm absolutely open to a new kind of transcendence, which a 26 linestage could potentially be. It's just that for now, and for admittedly irrational reasons, I'm compelled to recapture "the kind of sound" I heard back then. From reading the novel, I should know it's a fool's errand, but sometimes it's fun to be foolish.
At any rate, I've only been doing DIY audio for about 18 months, so I've got a lot of basics to learn, and this kind of project seems like a good pragmatic step from where my skills are now. It'll be a first time point-to-point build from schematic, using a simple circuit and parts that are not too expensive.
Speaking of basics, if I want to add a volume control, since the input resistor is rated at 10K, does that mean I'd use a 10K potentiometer or stepped attenuator?
The source would be the phono stage (which has a buffer), CD player, or a phone/iPad streaming music from the Internet, for which I plan to add a 3-position selector switch.
Don't get me wrong, I'm absolutely open to a new kind of transcendence, which a 26 linestage could potentially be. It's just that for now, and for admittedly irrational reasons, I'm compelled to recapture "the kind of sound" I heard back then. From reading the novel, I should know it's a fool's errand, but sometimes it's fun to be foolish.
At any rate, I've only been doing DIY audio for about 18 months, so I've got a lot of basics to learn, and this kind of project seems like a good pragmatic step from where my skills are now. It'll be a first time point-to-point build from schematic, using a simple circuit and parts that are not too expensive.
Speaking of basics, if I want to add a volume control, since the input resistor is rated at 10K, does that mean I'd use a 10K potentiometer or stepped attenuator?
The source would be the phono stage (which has a buffer), CD player, or a phone/iPad streaming music from the Internet, for which I plan to add a 3-position selector switch.
If you now desing the line stage, the 10k input impedance is uncomfortable.
The tube stage's grid resistor/potentiometer is the previous stage load, so 10k is too low for average tube equipment (this load is good for SS device).
Why not use 50-100k?
Filament.
I continuously develop -and use- equipments with rare -mostly thoriated tungsten filament- direct heated triodes (01/201, 10/VT25, 26, 801/801a/VT62, 841/VT51 etc.).
The most hum sensitive device is #26 (almost equally to 841, only sightly better than 4P1L
), especially the globe ones (sadly they "sing" better than ST ones). I'm not sure, that simple L-C-L-C filtration is satisfactory (even 1A heater current via the filtration chokes!).
My #26 preamp is humless, but I'm select #26 tubes for lowest radiated hum (even 20dB difference between same -aged and manufactured- tubes at 50Hz !!! radiated hum), use apropriate (C-CMC-C-L-C-r-C) raw DC supply.
But..... the raw DC supply is "cost no object" solution. I made chokes on amorphous -large!- cores, use high ripple current Panasonic capacitors, non inductive wirewound resistors etc.
The result: 50/100Hz hum is less than 1mV on the raw DC output!
Rod's regulator (which has minimum 60dB PSRR!!) finished the work.
If you want to use similar parameters with -good- chokes and capacitors, it will be cost a lot.
Yes, the CRC filter does not filter the recharge/recovery pulses too well - and whatever remains of these current-pulses is mixed directly with the (small) anode/plate current in the filament itself.
The final C presents a low impedance to the differential music signal on the filament, which is not desirable, as described above.
It's easy to build the raw dc and IC regulator dc filament supplies - and maybe it's useful to try them as a benchmark - not as a good one though.
Yes, you could use a choke to aid the filtering. But the choke will have to handle 1.5A without saturating, have not too much DCR, etc. And you will need the room.
Watch out for the magnetic spray from the choke, it can destroy a preamp or an input stage (hum).
I never had a choke that was able to handle the 1.25A 5V 300B filament that I powered off a 6.3V filament winding (that thankfully actually about 6.9V rms).
Instead, I used a Schottky bridge, 20,000uF, ~ 2 to ~ 3 ohms resistor, and another 20,000uF. That asI remember was about 1 to 2 mV ripple (Ok for an output stage, but not for a preamp.
Brute force can work sometimes.
Watch out for the magnetic spray from the choke, it can destroy a preamp or an input stage (hum).
I never had a choke that was able to handle the 1.25A 5V 300B filament that I powered off a 6.3V filament winding (that thankfully actually about 6.9V rms).
Instead, I used a Schottky bridge, 20,000uF, ~ 2 to ~ 3 ohms resistor, and another 20,000uF. That asI remember was about 1 to 2 mV ripple (Ok for an output stage, but not for a preamp.
Brute force can work sometimes.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.