Hi all,
Thanks for your fantastic inputs.
They gave me some thinking to do. I am new in this so please be kind to me.
Currently, I am building a headphone amplifier. It is a single stage grounded cathode 6H30 with a transformer on its anode. Cathode resistor is bypassed with Oscon, which is bypassed with a largish film cap.
The idea is to have minimal components in the signal path. A single active device in each channel, it doesn't get any less. But I am re examining that rationale to see if the amp can be further improved.
At the moment, the gain of the tube is roughly equal to the voltage step down of the OPT, meaning I get approximately unity gain overall. My listening position is at the 14th position of 24 step attenuator. I could use a bit more gain, but it is more of a luxury than a need.
Removing the cathode resistor bypass capacitor would likely leave me with less gain than I want. So if the cap goes, I'd need more amplification stage. But the upside is I can load the first stage with CCS, and along with cathode degeneration, distortion would be kept at minimal.
But I guess with the extra gain of 2 stages, I would always get more distortion compared to the lesser gain of a single stage. So it is not really a fair fight. On top of that, DF96 noted that distortion profile would shift to higher order. So it sounds like a bad idea in theory.
Just thinking out loud. Any suggestions would be welcome.
Thanks for your fantastic inputs.
They gave me some thinking to do. I am new in this so please be kind to me. Currently, I am building a headphone amplifier. It is a single stage grounded cathode 6H30 with a transformer on its anode. Cathode resistor is bypassed with Oscon, which is bypassed with a largish film cap.
The idea is to have minimal components in the signal path. A single active device in each channel, it doesn't get any less. But I am re examining that rationale to see if the amp can be further improved.
At the moment, the gain of the tube is roughly equal to the voltage step down of the OPT, meaning I get approximately unity gain overall. My listening position is at the 14th position of 24 step attenuator. I could use a bit more gain, but it is more of a luxury than a need.
Removing the cathode resistor bypass capacitor would likely leave me with less gain than I want. So if the cap goes, I'd need more amplification stage. But the upside is I can load the first stage with CCS, and along with cathode degeneration, distortion would be kept at minimal.
But I guess with the extra gain of 2 stages, I would always get more distortion compared to the lesser gain of a single stage. So it is not really a fair fight. On top of that, DF96 noted that distortion profile would shift to higher order. So it sounds like a bad idea in theory.
Just thinking out loud. Any suggestions would be welcome.
One more option is to use a triode with a higher gain for example the 12AX7 and then use a local feedback resister around just that one triode. Rig this to give exactly the gain you need. My simulations show this is the best over all the others. But do NOT use to much, you can run a 12ax7 with a closed loop gain of about 20 and get dramatically reduced distortion and lower out put impedance. The cost is only one resistor.
Just thinking out loud. Any suggestions would be welcome.
Your circuit is pretty hard to beat. Additional stages or poor sounding, low current tubes with feedback won't help. I would probably prefer soundwise an unbypassed arrangement of leds and diodes to get your desired bias. Oscons are great for digital, not so great for analogue. Bypassing them with films is more likely not to work.
you may find this interesting.....high tansconductance tubes need not operate with highish plate voltages unlike the 12xx7 types....Ultrapath = Ultra Simple...Except that it's not http://www.geocities.com/dmitrynizh/Ultrapath.htm
Last edited:
Regarding the choice of diode. I often seen "normal brightness" red LED being recommended. Is there a reason for this? Lowest noise?
Thanks. Seen the name being thrown around but never bothered to find out what it is, got some digesting to do.
First you choose the bias voltage you want. Then you notice that each color of LED has a differient voltage. So you pick a color based on that. Next there might be low-order secondary effects like the ability of an LED to take the required current or one having more noise than another. I think it works out that the smallest LED of the required color that still takes the required current is what you want.
But still "color" is the single most important feature as that determines the tube's operating point.
That said, a $1 voltage regulater can be configured as a constant current source. If you are going to allow solid state in your tube amp why stop with diodes?
Actually the use of local feedback (maybe in conduction with these bias methods) is nearly perfect if your over all amp can stand the loss of open loop gain.
One other bias method that should be nearly perfect is "fixed bias". I've never seen this done with pre-amp tubes but it is pretty common with power tubes. In a preamp it would let you simply ground the cathode. You get better gain and also it totally removes the cap from the signal path. Might be worth trying?
Any diode when forward biased has low impedance. This includes tube diodes as 6AL5 for example.
The fact that it's a headphone amp changes things a lot. You will by definition have a steep load line, and will have a lot of curent swing in the cathode. This is in my experience great as long as the distortion is mainly low order. You get a super clean and dynamic sounding amp. Unfortunately I think it also means that the unbypassed cathode resistor may not work well.
I'm working on a single tube power amp that has some similar qualities and found that even LEDs have a high enough nonlinear dynamic resistance, in the range of 3 to 10 ohms depending on current, to add significant high order distortion. Zener diodes are the worst.
http://www.diyaudio.com/forums/tube...multiplier-hybrid-parafeed-2.html#post3072318
YMMV, my experiments are with a 6S45pi so would show more effect than your 6H30. It may be informative to put a scope across the cathode bias and see how much the voltage reacts with bypass cap, LED, etc.
I would definitely try the TL431 in the cathode as one option. When I went to this in my amp the distortion settled down to mostly 2nd order. The TL431 has super low dynamic impedance and you can set the voltage with a trim pot. You only need the TL431 and the trim pot.
I didn't even consider a cathode capacitor in my amp due to the high transconductance and high current swing. Also my amp is CCS loaded and parafeed.
I'm working on a single tube power amp that has some similar qualities and found that even LEDs have a high enough nonlinear dynamic resistance, in the range of 3 to 10 ohms depending on current, to add significant high order distortion. Zener diodes are the worst.
http://www.diyaudio.com/forums/tube...multiplier-hybrid-parafeed-2.html#post3072318
YMMV, my experiments are with a 6S45pi so would show more effect than your 6H30. It may be informative to put a scope across the cathode bias and see how much the voltage reacts with bypass cap, LED, etc.
I would definitely try the TL431 in the cathode as one option. When I went to this in my amp the distortion settled down to mostly 2nd order. The TL431 has super low dynamic impedance and you can set the voltage with a trim pot. You only need the TL431 and the trim pot.
I didn't even consider a cathode capacitor in my amp due to the high transconductance and high current swing. Also my amp is CCS loaded and parafeed.
Last edited:
Thanks guys, again.
Regarding fixed biasing, I assume it is to ground the cathode and apply a negative voltage to the grid. Is this a good way of generating the required negative voltage?
http://www.aikenamps.com/BackBiasing.html
I suppose this will necessitate an input capacitor. I have concern if I can do it properly (sufficiently clean bias). Anyone has an example?
Regarding the shunt regulator on cathode, all I need is a trim pot and the chip? I have some idea on how it is done, but can I also have an example schematic to be sure?
Both sounds like worth trying. I began to think that I may have to build more than one of these.
Regarding fixed biasing, I assume it is to ground the cathode and apply a negative voltage to the grid. Is this a good way of generating the required negative voltage?
http://www.aikenamps.com/BackBiasing.html
I suppose this will necessitate an input capacitor. I have concern if I can do it properly (sufficiently clean bias). Anyone has an example?
Regarding the shunt regulator on cathode, all I need is a trim pot and the chip? I have some idea on how it is done, but can I also have an example schematic to be sure?
Both sounds like worth trying.
I began to think that I may have to build more than one of these. Thanks guys, again.
Regarding fixed biasing, I assume it is to ground the cathode and apply a negative voltage to the grid. Is this a good way of generating the required negative voltage?
Back Biasing
Yes. This is VERY common to do with power tubes. It was invented over 60 years ago and scehmatic above is a common way to do it. Although some power transformers have a 50V winding to be used for a negative bias supply.
This makes a lot of sense with power tubes becuase you get more power out of them for just the cost of a diode and a few caps.
For preamp tubes there is not much to be gained, more headroom I guess. and if you don't like the electro cap it takes that out of the circuit.
It is really all the same thing. All the normal cathode resistor does is move the cathode up to some small positive voltage so that a grounded grid "looks negative". But you can ground the cathode and use a smal negative supply on the grid. The tube will not know the difference.
This doesn't seem like a good idea to me now.
Looking at the the last schematic, signal path is pretty much the same thing as with using bypassed cathode resistor, except that the ground reference is moved from under the cathode resistor to above it.
And biasing the grid means I would need an extra input cap at the grid, so this looks like just adding more parts in the signal path without benefit. Unless there is a better way to do this.
Looking at the the last schematic, signal path is pretty much the same thing as with using bypassed cathode resistor, except that the ground reference is moved from under the cathode resistor to above it.
And biasing the grid means I would need an extra input cap at the grid, so this looks like just adding more parts in the signal path without benefit. Unless there is a better way to do this.
The is a small benefit but I will not argue it is worth the effort.
The cathod bias resister is NOT is perfect current sink and the voltage across that resistor depends on the current through the tube and hence on the audio signal. The bypass cap can help with this but grounding the cathode is a perfect voltage reference, the cathode says at EXACTLY zero volts at every possable current.
the other tiny benefit is that for the same B+ voltage you gain a few volts more headroom.
The two above effects are small in preamp tubes but much larger in power tubes. I think that is why the fixed bias (using a small negative power supply) is common with power tubes and not common at all in preamp and driver tubes. It is easy to see why. Preamp tubes are typically biased at only a 2 or 3 volts but a power tube might have 15V to 50V depending on the tube and of course the difference in current is even larger.
Again I'm not arguing to do this but I think it is worth trying in a Spice sim just to see the magnitude of the effect. It might turn out to be useful on guitar amps. I don't know.
Assuming that we are using the last schematic in the link (which is essentially the same as bypassed resistor). While the cathode voltage would remain fixed with current variation, it is now the grid voltage the would vary according to current variation. So effectively we still get the same effect as resistor biasing (because it is identical electrically, except the ground reference moved from the grid to the cathode), except that we now would need an extra input cap.
If we are using the second last schematic, it is essentially the same as LED biasing.
So unless there is another way of implementing the negative bias (a separate negative supply?) they are basically the same thing.
If we are using the second last schematic, it is essentially the same as LED biasing.
So unless there is another way of implementing the negative bias (a separate negative supply?) they are basically the same thing.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.