I have prototyped this on the bench and am hoping that someone with more experience than I can point out anything that I've gotten out of whack. It works fine and sounds amazing to my ear. I just don't want to build a seriously flawed design. The power supply shown in the schematic is using a toroid that I wound myself, using the original primary. I had to draw it as three transformers, but it's actually just three secondaries on the same transformer. I may or may not incorporate a filament rectifier. I'm very interested in regulated power supplies, but have only found information related to specific supplies and voltages. I don't know how to modify them to output different voltages. I'm open to some good reading material if someone can point me in the right direction.
The phase inverter bias is something that I'm particularly concerned about. I am having some trouble grasping how it works exactly. I really like the sound of the ECC81, despite it not being particularly linear. I also happen to have a few of them on hand. The preamp triodes are biased to 3.5mA and 112v plate voltage (310v B+) using a 56k plate load resistor. The 6CG7s are in push-pull and are biased to 13mA at 140v. They are driving an Edcor 12k:4R PP OPT and I am using 32 ohm Grado RS2e headphones. I am currently in the process of winding another toroid to ~450V so that I can run the ECC81 a little hotter without reducing the plate load resistor very much. I tried it with a 10k with the current power supply and it just didn't sound as good to me. The highs weren't as twinkly if that makes sense. I am considering some other preamp/phase inverter tubes as well, and would like to have the extra B+ to push them a little. Thoughts? I appreciate all the help that I've gotten here.
Edit: After posting, I realized that I had the 6cg7s running too hot. I replaced the 820R resistor in the PSU with a 1k6. That dropped the B+2 down to 120v and the 6cg7 plate current to 10.4 mA. 1.24 watts dissipation per triode now. Under the 5watt total (per tube) maximum.
The phase inverter bias is something that I'm particularly concerned about. I am having some trouble grasping how it works exactly. I really like the sound of the ECC81, despite it not being particularly linear. I also happen to have a few of them on hand. The preamp triodes are biased to 3.5mA and 112v plate voltage (310v B+) using a 56k plate load resistor. The 6CG7s are in push-pull and are biased to 13mA at 140v. They are driving an Edcor 12k:4R PP OPT and I am using 32 ohm Grado RS2e headphones. I am currently in the process of winding another toroid to ~450V so that I can run the ECC81 a little hotter without reducing the plate load resistor very much. I tried it with a 10k with the current power supply and it just didn't sound as good to me. The highs weren't as twinkly if that makes sense. I am considering some other preamp/phase inverter tubes as well, and would like to have the extra B+ to push them a little. Thoughts? I appreciate all the help that I've gotten here.
Edit: After posting, I realized that I had the 6cg7s running too hot. I replaced the 820R resistor in the PSU with a 1k6. That dropped the B+2 down to 120v and the 6cg7 plate current to 10.4 mA. 1.24 watts dissipation per triode now. Under the 5watt total (per tube) maximum.
Attachments
Last edited:
The 500 Ohm resistor in the B+1 line is not decoupled (meaning: there is no capacitor to ground after the 500 Ohm resistor). This creates feedback between the first and second stage (and if this is a stereo amplifier also between the left and right channel) and could lead to instability. I advise you to add an electrolytic capacitor after this resistor (which in the process gives you extra smoothing of B+1).
For more or less the same reasons it is better to also decouple the resistor in the B+2 line with an extra electrolytic capacitor.
The value of the cathode resistors of the two halves of the ECC81 is too low. You measure 3.5 mA so the cathode voltage is only 0,945 V (V = I x R = 0.0035 x 270). Gridcurrent starts already at about 1.3 V. So I advise you to higher the value to something like 680 Ohm.
I also advise you to higher the value of the cathode resistors of the 6CG7 and run them at some somewhat higher B+2. Now you are running them at about 10 mA so the cathode voltage is only 1V. With signal there's is hardly room for the cathode voltage to swing down.
The maximum cathode to heater voltage of the ECC81 is 90 V. With the cathode resistor of 270 Ohm, so with 3.5 mA of current, the voltage at the cathode is V = I x R = 0.0035 x 27270 = 95 V. So this is over the maximum value (with the risk of a cathode to heater short). If you higher the value of the cathode resistor of 270 Ohm like advised, the current will go down so this voltage will probably drop under the limit of 90 V.
You connect headphones of 32 Ohm to the amplifier. The OPT is 12K/4. So the primary impedance becomes (32/4) x 12K = 96K. That seems too high to me. The internal resistance of half a 6CG7 is about 7K so a primary impedance of about 20K would be more appropriate.
Here is an article with some adjustable regulated power supplies: Valve (Tube) Regulated Power Supplies
For more or less the same reasons it is better to also decouple the resistor in the B+2 line with an extra electrolytic capacitor.
The value of the cathode resistors of the two halves of the ECC81 is too low. You measure 3.5 mA so the cathode voltage is only 0,945 V (V = I x R = 0.0035 x 270). Gridcurrent starts already at about 1.3 V. So I advise you to higher the value to something like 680 Ohm.
I also advise you to higher the value of the cathode resistors of the 6CG7 and run them at some somewhat higher B+2. Now you are running them at about 10 mA so the cathode voltage is only 1V. With signal there's is hardly room for the cathode voltage to swing down.
The maximum cathode to heater voltage of the ECC81 is 90 V. With the cathode resistor of 270 Ohm, so with 3.5 mA of current, the voltage at the cathode is V = I x R = 0.0035 x 27270 = 95 V. So this is over the maximum value (with the risk of a cathode to heater short). If you higher the value of the cathode resistor of 270 Ohm like advised, the current will go down so this voltage will probably drop under the limit of 90 V.
You connect headphones of 32 Ohm to the amplifier. The OPT is 12K/4. So the primary impedance becomes (32/4) x 12K = 96K. That seems too high to me. The internal resistance of half a 6CG7 is about 7K so a primary impedance of about 20K would be more appropriate.
Here is an article with some adjustable regulated power supplies: Valve (Tube) Regulated Power Supplies
47 nF is completely right. It is also 0.047 µF. Yes that is a µ.
They are not decoupling caps (as used to decouple rails) but coupling caps and like you I think they are on the small side. Just calculated it and 100 nF or 0.1 µF would be according normal HiFi standards. Normally one wants at least 20 Hz ... 20 kHz covered in good audio.
Since a few decades the world has seen a fast growing increase in HF/RF being in the aether literally everywhere. It is therefor advisable to build stuff to withstand that. So an input filter limiting to let's say 60 kHz is a right step to prevent issues.The amplifier building DIYers seem to lag behind but please consider building stuff 2021 proof. It might be the difference between satisfying and dissatisfying performance...
They are not decoupling caps (as used to decouple rails) but coupling caps and like you I think they are on the small side. Just calculated it and 100 nF or 0.1 µF would be according normal HiFi standards. Normally one wants at least 20 Hz ... 20 kHz covered in good audio.
Since a few decades the world has seen a fast growing increase in HF/RF being in the aether literally everywhere. It is therefor advisable to build stuff to withstand that. So an input filter limiting to let's say 60 kHz is a right step to prevent issues.The amplifier building DIYers seem to lag behind but please consider building stuff 2021 proof. It might be the difference between satisfying and dissatisfying performance...
Last edited:
DC blockers rather than smoothing 🙂 I've got into the very bad habit of calling everything decoupling recently..
You need a DC blocker on the input so the bias isn't at the mercy of the source.... Also RF suppression at the input
it a great thing to have.
Any rationale for 3k3 for the first grid-stopper? Smaller might improve noise floor a little.
it a great thing to have.
Any rationale for 3k3 for the first grid-stopper? Smaller might improve noise floor a little.
Last edited:
Thanks for all the replies. 🙂 I'll start reconfiguring for higher B+ and power supply topology. Those low-value cathode resistors were concerning to me, but I didn't know the actual consequences of using them. Thank you for clarifying that. The guy that recommended 12k OPTs was probably thinking that I'd be using high impedance headphones. Would it hurt to use the 32 ohm ones? I will plan on getting a different pair, but I do like the sound of those Grados. And thank you very much for that article on regulated power supplies. I will have to take the time to study it.
I also think in terms of µF rather than nF. I used the Falstad circuit simulator to draw that schematic though, and it automatically converts values like that. It has thrown me off more than once.
I'll change up the coupling cap sizes and add one to the input as well. Any recommendations on the input RF filter? Just a simple LC low-pass filter or is there something more function-specific that I should be looking into?
Thanks again for taking the time to help. I really do appreciate it.
I also think in terms of µF rather than nF. I used the Falstad circuit simulator to draw that schematic though, and it automatically converts values like that. It has thrown me off more than once.
I'll change up the coupling cap sizes and add one to the input as well. Any recommendations on the input RF filter? Just a simple LC low-pass filter or is there something more function-specific that I should be looking into?
Thanks again for taking the time to help. I really do appreciate it.
I'll add a coupling cap when I get the new power transformer ready to test. I just more or less picked a random value for that grid stopper. I'll try some lower values. I appreciate the advice.
Make that an RC filter. This tool is great but beware to use the correct way of writing unit values!
RC Low-pass Filter Design Tool
Why do we use nF? Because it prevents writing mistakes with either a 0 too much or too little. One can't go wrong with 220 pF. Please write that one in µF 🙂 That is (not coincidentally) a good value for the filter. Use polypropylene or styroflex or the like.
Last edited:
If you post an updated schematic, can you label its measured voltages too?
It would help to be able to see cathode voltages (pins 3 and 8 of the 12AT7 and 6CG7) and plate voltages (pins 1 and 6 of the 12AT7 and 6CG7).
Also, the actual measured B+ and heater voltages would be helpful.
🙂
It would help to be able to see cathode voltages (pins 3 and 8 of the 12AT7 and 6CG7) and plate voltages (pins 1 and 6 of the 12AT7 and 6CG7).
Also, the actual measured B+ and heater voltages would be helpful.
🙂
with 47nF coupling caps in conjunction with 270K and 100K grid resistors you have high pass filters with -3dB points of 78Hz and 213Hz. You won't have much bass. I would recommend 1 Meg and 470K respectively. that will set the -3dB points at 21Hz and 45Hz.
I don't have it set up on the bench at the moment, so unfortunately I can't get you measured voltages. The B+ voltages on the original schematic are actual measured voltages +/- 2V under load though. I'll get some better measurements when I get it set back up. Loaded filament voltage was 6.5V, if I remember correctly. I've got everything cleaned off of the bench at the moment so I can wind this new toroid. I've been working on it for a few hours and I'm only up to 91V unloaded on the first winding so... It might take me a few days. Maybe next weekend since I don't have much time during the week. Either way, I'll certainly draw up a new schematic with better measurements when I get it set back up. Thanks, guys.
You missed out the "2pi" in the formula. -3 dB points are 12.5Hz and 33.9Hz, still too high but not so bad.
What is to grasp? It biases exactly the same as the stage before it. The plate resistor has been sawn in half and moved around through the power supply to "under" the cathode. But the tube has no way to know that.
A common dropper on just *TWO* stages is normally quite stable. Moreso when one of them is a cathodyne ("unity" gain). Myself, I fear it needs more filtering against buzz, more than filtering against leak-back through the supply. But maybe it is fine as is. And the point about L<->R sneakage deserves consideration.
Just doubt, really. The positioning of the grid leak resistor had me concerned that I was doing something wrong. It's all good now though.
I finished winding the transformer. The unloaded secondaries are 520vAC, 240vAC and 8vAC(will use a filament rectifier). (1920 total windings.. phew...) I'm working on a rudimentary power supply now. I priced high voltage electrolytics and nearly soiled myself when I saw how much they cost. (~$150 EACH) Luckily, large value (>100μF) HV film caps are "only" ~$10-$15 each, so I'll likely be using them instead. All I currently have on hand are 7.5μF AC motor run caps, so yeah. I am looking into using a 555 timer, mosfet and relay for a delayed B+ startup circuit so that the ECC81s' plates won't be slammed with 730vDC before their filaments heat up. Now that I think of it, a DPST relay for both B+ would be nice. Are there any other special considerations that I should look in to because of the new high B+? Aside from not electrocuting myself of course.
Last edited:
220mmf. That's how we used to write it. And say it. "MMMMMf"
Where do you find a 730V DC relay? (The DC is critical... the arc will never stop.)
Photonicinduction showed a massive DC circuit breaker the size of a mini fridge, so I'm sure one can be had. He said that one probably came out of a submarine. Interesting arc-breaking mechanism. I was planning on switching the secondary before it hits the rectifier though.
Edit: I went back and looked at the video. Only 50 volts, but 1600 amps continuous. Very impressive thing nonetheless.
Edit: I went back and looked at the video. Only 50 volts, but 1600 amps continuous. Very impressive thing nonetheless.
Last edited:
Yes I know the silly habit and may only hope it was "µµF" which of course is way simpler than "pF" or picoFarad. Strange enough "mF" or milliFarad so 1000 µF is then also used which is relatively unknown elsewhere. These are exceptions to the international standards just like the habit to use "mho" for Siemens (reciproke value of Ohm and then the unit name is written backwards 😀) and the general persisting habit to write unit names wrong. Like "mmf" and "Kilo volt" which are examples of 2 errors in one unit name.
Last edited: