Okay, adding an anode resistor to the KT66 seems to fix the problem of overloading the tube AND reduce the distortion in Pentode mode. With the volume knob all the way cranked and the 2vrms input from the DAC, we see about 80ma and 290V on the anode, for about 23W, below the 25W recommended and 30W absolute max in the spreadsheet.
Even if I double the signal input to 4Vrms, because the signal clips at 300V you never get higher than about 300V&85ma or 25.5W. And that's an edge case.
Meanwhile, at a reasonable volume with the knob at 25% and the output producing about 2V, the anode is at 211V&53mA or 11W.
And the distortion is 0.18%, not bad! Gain is reduced to 9.6db which is less than the Crack but still pretty good and plenty for reasonable listening levels (or to use as a preamp).
Even if I double the signal input to 4Vrms, because the signal clips at 300V you never get higher than about 300V&85ma or 25.5W. And that's an edge case.
Meanwhile, at a reasonable volume with the knob at 25% and the output producing about 2V, the anode is at 211V&53mA or 11W.
And the distortion is 0.18%, not bad! Gain is reduced to 9.6db which is less than the Crack but still pretty good and plenty for reasonable listening levels (or to use as a preamp).
Last edited:
I've never been able to get rid of the decimal places either. I just live with it. I usually make my schematics bigger so they fit, or draw a little wire and put the .op reading on the end of that so it's horizontal and not vertical.
May I ask a question?
- Why aren't you using the usual cathode bias resistor for the KT66 cathode follower? I see R12 as a plate load, which I guess limits current enough to keep the KT66 from thermal runaway (at least in the simulation -- I don't know how that would work in real life), but doesn't that plate resistor rob the cathode follower of gm and so increase its output impedance? I think you want the Zout to be as low as possible to drive headphones, even 300 ohm ones.
It could be that you know something I don't, but I'm used to seeing cathode followers put together as below (note the addition of Rk and the removal of R12):
May I ask a question?
- Why aren't you using the usual cathode bias resistor for the KT66 cathode follower? I see R12 as a plate load, which I guess limits current enough to keep the KT66 from thermal runaway (at least in the simulation -- I don't know how that would work in real life), but doesn't that plate resistor rob the cathode follower of gm and so increase its output impedance? I think you want the Zout to be as low as possible to drive headphones, even 300 ohm ones.
It could be that you know something I don't, but I'm used to seeing cathode followers put together as below (note the addition of Rk and the removal of R12):
Last edited:
Because in this case the load resistor for the cathode follower can be so low in value that it also serves as the bias resistor.
Thanks. I see that now.
But does that make R12 necessary?
And doesn't R12 reduce the KT66 gm, thus raising its Zout -- at least partially defeating the purpose?
So it definitely does reduce the gm and it does seem to raise the output impedance from about 85ohm to 95 ohm.
Still that's perfectly fine for the high impedance dynamic headphones I'm using it with (still lower than the Crack), so it's a worthwhile tradeoff for not breaking the tube I think.
Agreed that 85R to 95R is a very small difference. Interesting...
Is the KT66's grid leak resistor value bootstrapped? I would think it's not, but I probably am not seeing the mechanism clearly.
Is the KT66's grid leak resistor value bootstrapped? I would think it's not, but I probably am not seeing the mechanism clearly.
Right click on your .op label (the voltage) and replace "$" with
"round($*100)/100" for two decimal places,
"round($*1000)/1000" for three and so on.
Neat little trick 😉
Last edited:
I don't think it is. Bootstrapping is another thing since the output cap value is relatively low, I'm going to need some kind of delay circuit if I'm not using a rube rectifier (haven't decided, need to think more about the power supply) or else the whole thing will blow up (from what I understand).
@ericrosenfield - Are you in a position where you are able to test out elements of this circuit safely? (high voltage test power supply, current limiting). I would start with the output section (with a higher value cathode resistor to reduce current) as this is the most crucial, ramp things up slowly and keep a close eye on current draw. you can feed a test signal into its plate (at close to unity gain) to start to get a good idea on how it's preforming.
The 12AU7 voltage gain stage is typical and will almost certainly be fine (may need to change resistor values to optimize), it is current limited by its plate resistor anyway.
As for distortion, this looks to simulate well and your looking at low enough numbers (in my opinion) that it should sound great!
The 12AU7 voltage gain stage is typical and will almost certainly be fine (may need to change resistor values to optimize), it is current limited by its plate resistor anyway.
As for distortion, this looks to simulate well and your looking at low enough numbers (in my opinion) that it should sound great!
That would be the safest and best option, as you know circuits in simulation are one thing, and another in real life.
Do you have a variac? you could build a conventional tube power supply and control it's output voltage?
Having a quick look at PSRR, it's better than I thought, I noticed connecting the screen grids capacitor to ground rather than the cathode helps here.
Having a well filtered power supply is obviously very important in a headphone amplifier.
Do you have a variac? you could build a conventional tube power supply and control it's output voltage?
Having a quick look at PSRR, it's better than I thought, I noticed connecting the screen grids capacitor to ground rather than the cathode helps here.
Having a well filtered power supply is obviously very important in a headphone amplifier.
@mr_zener I actually did just purchase a variac, because I'm making a SS power amp with a big transformer PS and I wanted to test it safely. So maybe I'll figure out the power supply for this thing next and just build that first and use that with the variac to test the circuits in the gain stages.
I'll play around with moving the screen grid to ground. The thing about that is when I'm experimenting with the actual prototype it seems like it'll be pretty easy to move the wire from one place to another to see which actually sounds better in practice. Someone elsewhere also suggested removing R1 and C2 and maybe C5, so those also seem like things to play with.
Thanks again for your help and insight, I really appreciate it!
I'll play around with moving the screen grid to ground. The thing about that is when I'm experimenting with the actual prototype it seems like it'll be pretty easy to move the wire from one place to another to see which actually sounds better in practice. Someone elsewhere also suggested removing R1 and C2 and maybe C5, so those also seem like things to play with.
Thanks again for your help and insight, I really appreciate it!
Yes, R1 is not needed as 12AU7 plate is already has a path to ground though the potentiometer, it will also interfere with the volume 'curve', maybe desirable, maybe not.
C2 is optional, it increases AC gain. Personally I would have a resistor in series with it so I could change the gain to whatever I wanted. You can always just put a wire link in place.
C5 - Removing this seems like a bad move as this makes PSRR much worse, I don't see any advantage.
On the subject of capacitors, I really like that the cathode is around 20V, this means you could use a 50V rated cap safely (maybe with zener diode protection from cathode to ground) unlocking much more choice and higher capacitance ratings 👍 this makes me think about making my own OTL amp.
Currently got my hands full with a high gain tube hybrid preamp, designed to drive unity gain output stages. Tested and works well, just got to select power transformers and build a custom case for it..
After that I got an old radio from 1949/50 I want to restore and convert to an internet radio, but keeping the same tube output section on a custom PCB. Looking forward to that as I can also brush up on my coding skills.
After that I got an old radio from 1949/50 I want to restore and convert to an internet radio, but keeping the same tube output section on a custom PCB. Looking forward to that as I can also brush up on my coding skills.
As an experiment I tried adding a CCS to one of the tubes to be like the Speedball upgrade to the Crack. This made the distortion worse and the anode voltage spike into redplate territory almost immediately:
It's entirely probable I'm doing this wrong. (I'm using an IXCP10M45S model that I found on this site here: https://www.diyaudio.com/community/threads/10m45s-ltspice-model.139948/
It's entirely probable I'm doing this wrong. (I'm using an IXCP10M45S model that I found on this site here: https://www.diyaudio.com/community/threads/10m45s-ltspice-model.139948/