Just popping ideas about a current drive amplifier driving a fullrange or midrange/tweeter section of a speaker.
What would be the simplest way? An output transformer with a pentode/tetrode output section? As a transformer winder I have the freedom to wind anything. Pentodes would favor high Ls low Cp/Cs OPTs.
But what about linearity? Pentodes are famous for potentially giving more odd-order distortion. Then I should introduce local NFB, but voltage NFB will decrease output impedance, which we don't want. Current FB perhaps?
What would be the simplest way? An output transformer with a pentode/tetrode output section? As a transformer winder I have the freedom to wind anything. Pentodes would favor high Ls low Cp/Cs OPTs.
But what about linearity? Pentodes are famous for potentially giving more odd-order distortion. Then I should introduce local NFB, but voltage NFB will decrease output impedance, which we don't want. Current FB perhaps?
I haven't tried any of this, but a circuit that keeps the screen grid out of the way "should" be less affected. Like perhaps the tdp style "enhanced triode" with screen grid drive, or one of those gu50 tubes where the screen grid needs to be lower than the anode. I intend to try this one day, but haven't yet. It doesn't seem such a bad idea with the current economic DSP crossover and EQ that is available.
1. Measure the impedance and phase angle of the mounted driver's impedance versus frequency (with the driver in the cabinet, horn, open baffle, etc that you will be using).
Find the lowest impedance and the highest impedance.
The electrical power will be the impedance times the current, times the Cosine of the phase angle.
0 degrees phase, max power.
90 degrees phase, no power.
For most frequencies, the cosine will be other than 0 or 90 degrees.
Calculate the power versus frequency for a given current.
This will tell you the power you are delivering to the speaker.
But the speaker has resonances, voice coil inductance, enclosure resonances, and other things that affect the efficiency of the sound level versus the electrical power that is delivered.
If you do not do these measurements, and do these calculations, you will not be able to predict the performance of the system (the amplifier and speaker and enclosure is a system).
Then there is the issue of damping the resonances, especially with a very high impedance current drive.
2. If the loudspeaker has impedances that have only a small variance over the frequency range, then you might consider using a fiairly high impedance amplifier; you do not need to use a true infinite impedance amplifier.
Suppose the driver impedance variations are from 4 to 8 Ohms.
In that case, if the amplifier output impedance is 80 Ohms, that is pretty much good enough.
But you, as a transformer winder already know that it is not all just about the high plate impedance of the pentode or beam power tube.
It is also about the transformer.
The distributed capacitance of the primary is reflected to the secondary (and the capacitive reactance is reduced by the impedance ratio).
The inductance of the secondary is proportionate to the primary to secondary impedance ratio. So if the primary is 40 Henry, and the impedance is 5k to 8 Ohms, then the secondary inductance is only 64 milli Henry.
The high impedance of the pentode or beam power plate is swamped out at very high frequencies and at very low frequencies; by the distributed capacitance of the primary, and the inductance of the primary.
There is no simple current drive amplifier I know of that uses tubes and an output transformer.
High impedance at mid frequencies, yes.
High impedance at low frequencies, no.
High impedance at high frequencies, no.
Just using feedback to create high impedance is problematic.
The un-dampened plate impedance, transformer, and loudspeaker impedances, phase angles, etc. makes applying very much feedback the cause of the amplifier becoming a power oscillator instead.
It is difficult enough to make the amplifier have a reasonably constant voltage output,
Much more difficult to have a constant current out.
Just my opinion.
Find the lowest impedance and the highest impedance.
The electrical power will be the impedance times the current, times the Cosine of the phase angle.
0 degrees phase, max power.
90 degrees phase, no power.
For most frequencies, the cosine will be other than 0 or 90 degrees.
Calculate the power versus frequency for a given current.
This will tell you the power you are delivering to the speaker.
But the speaker has resonances, voice coil inductance, enclosure resonances, and other things that affect the efficiency of the sound level versus the electrical power that is delivered.
If you do not do these measurements, and do these calculations, you will not be able to predict the performance of the system (the amplifier and speaker and enclosure is a system).
Then there is the issue of damping the resonances, especially with a very high impedance current drive.
2. If the loudspeaker has impedances that have only a small variance over the frequency range, then you might consider using a fiairly high impedance amplifier; you do not need to use a true infinite impedance amplifier.
Suppose the driver impedance variations are from 4 to 8 Ohms.
In that case, if the amplifier output impedance is 80 Ohms, that is pretty much good enough.
But you, as a transformer winder already know that it is not all just about the high plate impedance of the pentode or beam power tube.
It is also about the transformer.
The distributed capacitance of the primary is reflected to the secondary (and the capacitive reactance is reduced by the impedance ratio).
The inductance of the secondary is proportionate to the primary to secondary impedance ratio. So if the primary is 40 Henry, and the impedance is 5k to 8 Ohms, then the secondary inductance is only 64 milli Henry.
The high impedance of the pentode or beam power plate is swamped out at very high frequencies and at very low frequencies; by the distributed capacitance of the primary, and the inductance of the primary.
There is no simple current drive amplifier I know of that uses tubes and an output transformer.
High impedance at mid frequencies, yes.
High impedance at low frequencies, no.
High impedance at high frequencies, no.
Just using feedback to create high impedance is problematic.
The un-dampened plate impedance, transformer, and loudspeaker impedances, phase angles, etc. makes applying very much feedback the cause of the amplifier becoming a power oscillator instead.
It is difficult enough to make the amplifier have a reasonably constant voltage output,
Much more difficult to have a constant current out.
Just my opinion.
Last edited:
The simplest path
You want to make it simple. Then it will be single-ended, no doubt. I am talking about the output stage only.
Single-ended Class A amplifiers are nice in the special way that they follow the "First Watt" -principle.
A pentode without any feedback would be the simplest possible option. Do not triodise the pentode, because you wish a high output impedance which is characteristic to a current-drive. If you bias the pentode with a cathode resistor, then you need also a bypass capacitor. This is probably the simplest way to make a pentode stage without feedback.
If you wish to have some feedback to linearize the pentode, connect a feedback resistor in series with the bypass capacitor. That is the simplest feedback I can think of.
There is also the possibility to add a current sense resistor in series with the load, which means the secondary side of the output transformer. There is a drawing available: Current-drive by a valve amplifier
I think that those are the simple options. Tweaking and optimizing further the circuits become something else than simple.
You want to make it simple. Then it will be single-ended, no doubt. I am talking about the output stage only.
Single-ended Class A amplifiers are nice in the special way that they follow the "First Watt" -principle.
A pentode without any feedback would be the simplest possible option. Do not triodise the pentode, because you wish a high output impedance which is characteristic to a current-drive. If you bias the pentode with a cathode resistor, then you need also a bypass capacitor. This is probably the simplest way to make a pentode stage without feedback.
If you wish to have some feedback to linearize the pentode, connect a feedback resistor in series with the bypass capacitor. That is the simplest feedback I can think of.
There is also the possibility to add a current sense resistor in series with the load, which means the secondary side of the output transformer. There is a drawing available: Current-drive by a valve amplifier
I think that those are the simple options. Tweaking and optimizing further the circuits become something else than simple.
If you drive a speaker with a current source/sink output, the speaker will see a very high source impedance, and will not get any electrical damping from the amplifier. That could have consequences for sound quality, since speakers are designed with voltage source drive in mind. If you want to experiment with that, however, a Howland current source would fill the bill nicely.
I believe PP is the way to go because it cancels the dominant order distortion and has huge power reserve compared to SE. Good PP beats SE in sound quality, but requirement of accurate phase splitting and fully differential operation all the way through are difficult and expensive to implement.
Aperiodic loudspeaker, regardless of implementation
That is true. That is a reason to choose the loudspeaker accordingly, to to utilize the current-drive amplifier properly. Practically speaking, some kind of an aperiodic loudspeaker is required.
That is true. That is a reason to choose the loudspeaker accordingly, to to utilize the current-drive amplifier properly. Practically speaking, some kind of an aperiodic loudspeaker is required.
Push Pull Pentode / Beam Power amplifiers can be very simple, especially when we are not using negative feedback. With no negative feedback, we get the highest output impedance possible with a simple circuit.
Yes, the distortion may be a little high, especially at large signal levels.
But . . . if the need/desire to have a current source amplifier is Paramount, you get it with this topology.
And the push pull amp is almost as simple as single ended (see below on how simple it can be), but the distortion is much lower than
a simple single ended current source amp.
The current source will not be perfect at the frequency extremes, but can be much much higher impedance than a fed-back amplifier.
Pentodes / Beam Power tubes have lots of gain, so the splitter/driver does not need to be complex, and it does not need more than 1 driver/splitter stage (a pair of triodes). And because there is no feedback, the gain is higher.
A 6SN7 or 12AU7 is all that is needed to drive the pentodes / beam power output tubes.
Use a simple current source (current sink), like an LM334, or an LM317, (LM334 for lower current, LM317 for higher current).
Be sure to check the minimum voltage specification, and minimum current specification that makes the current source work (in this case the current 'sink' for the cathode coupled splitter/driver). Make the splitter plate loads be the same (1% or better). Use RC coupling, and make the output tube g1 resistors the same (1%).
All this is simple for those of you who have built anything close to this simple amplifier.
No negative feedback, only 3 tubes, etc.
This push pull amplifier will symmetrically drive the speaker (less "Woofer Walk" than single ended output).
That is especially important, because the woofer is not dampened by the amplifier, the amp is high impedance.
Now, if you have either a solid state amplifier or a tube amplifier that has an output that is Not a current source (probably what you are listening to today), connect a 50, or 100 Ohm power resistor in series from the hot tap (example 8 Ohm tap) to the loudspeaker hot connection in.
Connect the amp common (0) lead to the common (0) lead of the speaker.
Be very careful! Do not start with the volume turned up.
The sound level is reduced a lot, but the last thing you want to do is turn the volume control all the way up.
The amp will go into clipping, and in some cases may destroy itself (no appreciable load).
But this simple test will allow you to get an idea of how your loudspeakers will sound when they are driven by a true current source amplifier.
I hope that gives some of you an idea of what to build, if you decide to go the current source output amp journey, into another sound adventure.
Yes, the distortion may be a little high, especially at large signal levels.
But . . . if the need/desire to have a current source amplifier is Paramount, you get it with this topology.
And the push pull amp is almost as simple as single ended (see below on how simple it can be), but the distortion is much lower than
a simple single ended current source amp.
The current source will not be perfect at the frequency extremes, but can be much much higher impedance than a fed-back amplifier.
Pentodes / Beam Power tubes have lots of gain, so the splitter/driver does not need to be complex, and it does not need more than 1 driver/splitter stage (a pair of triodes). And because there is no feedback, the gain is higher.
A 6SN7 or 12AU7 is all that is needed to drive the pentodes / beam power output tubes.
Use a simple current source (current sink), like an LM334, or an LM317, (LM334 for lower current, LM317 for higher current).
Be sure to check the minimum voltage specification, and minimum current specification that makes the current source work (in this case the current 'sink' for the cathode coupled splitter/driver). Make the splitter plate loads be the same (1% or better). Use RC coupling, and make the output tube g1 resistors the same (1%).
All this is simple for those of you who have built anything close to this simple amplifier.
No negative feedback, only 3 tubes, etc.
This push pull amplifier will symmetrically drive the speaker (less "Woofer Walk" than single ended output).
That is especially important, because the woofer is not dampened by the amplifier, the amp is high impedance.
Now, if you have either a solid state amplifier or a tube amplifier that has an output that is Not a current source (probably what you are listening to today), connect a 50, or 100 Ohm power resistor in series from the hot tap (example 8 Ohm tap) to the loudspeaker hot connection in.
Connect the amp common (0) lead to the common (0) lead of the speaker.
Be very careful! Do not start with the volume turned up.
The sound level is reduced a lot, but the last thing you want to do is turn the volume control all the way up.
The amp will go into clipping, and in some cases may destroy itself (no appreciable load).
But this simple test will allow you to get an idea of how your loudspeakers will sound when they are driven by a true current source amplifier.
I hope that gives some of you an idea of what to build, if you decide to go the current source output amp journey, into another sound adventure.
Last edited:
One more thing about the push pull pentode / beam power tube current source amp:
The speaker may have relatively high impedance at some frequencies, bass resonance(s), crossover from woofer to tweeter, etc.
At that time, the plates alternately head towards 50V or so.
That means there is a big rise in screen current.
And, that is also the region of the classical "Kink" in the plate curves.
There will be additional distortion there.
Fix that. The remedy is also simple.
A series resistor from B+ to the screen, and an 0D3 gas regulator tube from the screen to ground is all that is needed. The screen will be regulated at 150VDC (much better than having the screen at the same B+ voltage that the push pull transformer center tap is.
I should never have considered writing about the current source amplifier.
Now I have to build one (kind of wish I never thought about it).
But thanks to all of you who have contributed ideas about this kind of system (especially to the starters of 2 threads on these subjects) . . .
. . . Nikolas Ojala, and 50AE.
And, one thing that made me curious . . . I have on several occasions taken a standard solid state amp, and a standard tube amp, and performed the series resistor experiment on both kinds of those amplifiers, and used 30, 50, and 100 Ohm series resistors to see what that did to the sound of different systems that I had at the time.
Just as I suggested you to try in my Post # 8.
I remember one 2 way speaker I tried the series resistor experiment.
It had a port, but the port was covered with a metal plate that had small holes in it (dampened ported enclosure).
But the single pole crossover L and C on woofer and tweeter respectively caused a high impedance at the crossover frequency.
Curiosity may have killed the cat, but curiosity makes me work harder.
Happy experimenting everybody!
The speaker may have relatively high impedance at some frequencies, bass resonance(s), crossover from woofer to tweeter, etc.
At that time, the plates alternately head towards 50V or so.
That means there is a big rise in screen current.
And, that is also the region of the classical "Kink" in the plate curves.
There will be additional distortion there.
Fix that. The remedy is also simple.
A series resistor from B+ to the screen, and an 0D3 gas regulator tube from the screen to ground is all that is needed. The screen will be regulated at 150VDC (much better than having the screen at the same B+ voltage that the push pull transformer center tap is.
I should never have considered writing about the current source amplifier.
Now I have to build one (kind of wish I never thought about it).
But thanks to all of you who have contributed ideas about this kind of system (especially to the starters of 2 threads on these subjects) . . .
. . . Nikolas Ojala, and 50AE.
And, one thing that made me curious . . . I have on several occasions taken a standard solid state amp, and a standard tube amp, and performed the series resistor experiment on both kinds of those amplifiers, and used 30, 50, and 100 Ohm series resistors to see what that did to the sound of different systems that I had at the time.
Just as I suggested you to try in my Post # 8.
I remember one 2 way speaker I tried the series resistor experiment.
It had a port, but the port was covered with a metal plate that had small holes in it (dampened ported enclosure).
But the single pole crossover L and C on woofer and tweeter respectively caused a high impedance at the crossover frequency.
Curiosity may have killed the cat, but curiosity makes me work harder.
Happy experimenting everybody!
Last edited:
Yes, LTP with current sink should do it, ideally driven from a balanced source. Even 12AX7 or 6SL7 is adequate as output pentode driver, because there is no problem of Miller capacitance. Driving triode output is PITA by comparison, need way higher voltage swing at way lower output impedance.
One problem with LTP driver though is capacitor coupling. For an audio purist, transformer-coupled differential driver would be preferable. That takes us back to low Rp triode driver and considerable expense of high quality interstage transformer.
One problem with LTP driver though is capacitor coupling. For an audio purist, transformer-coupled differential driver would be preferable. That takes us back to low Rp triode driver and considerable expense of high quality interstage transformer.
For those who pick the right R and the right C, RC coupling is wonderful.
But you have to pay attention to many other details too.
First, pick an R that is less than the maximum grid resistor rating for the output tube.
The maximum resistance is in detailed data sheets. It has 2 values, one for fixed adjustable bias, and another (higher) maximum resistance value for self bias.
Then, you must keep the signal amplitude so the grid never goes closer to the cathode than -1V Grid-Cathode. The grid starts to draw some grid current, even before 0V Grid-Cathode).
Now, there is not any blocking distortion that some complain about. Turn the volume down.
Talk about clipping, wait until the speaker impedance is high at a frequency of a note that is being played, and one plate will head very close to 0V, while the other plate is at 2x B+. Clip!
In that case, RC coupling is not the issue.
Need more power out, pick a more powerful pair of output tubes, so that you do not have to draw any grid current.
Do not draw grid current for RC coupled stages.
Interstage coupling is also wonderful.
But there are many other factors you must consider, not just cost.
Magnetic hum pickup
Bandwidth, low and high frequencies
Square Wave response
Perhaps other tradeoffs too.
As to the output stage, since there is no negative feedback, the gain of that stage is merely the tube transconductance times the load impedance.
Simple.
Input/phase splitter/driver need only be one stage.
"You should make things as simple as possible, but no simpler" - Albert Einstein
But you have to pay attention to many other details too.
First, pick an R that is less than the maximum grid resistor rating for the output tube.
The maximum resistance is in detailed data sheets. It has 2 values, one for fixed adjustable bias, and another (higher) maximum resistance value for self bias.
Then, you must keep the signal amplitude so the grid never goes closer to the cathode than -1V Grid-Cathode. The grid starts to draw some grid current, even before 0V Grid-Cathode).
Now, there is not any blocking distortion that some complain about. Turn the volume down.
Talk about clipping, wait until the speaker impedance is high at a frequency of a note that is being played, and one plate will head very close to 0V, while the other plate is at 2x B+. Clip!
In that case, RC coupling is not the issue.
Need more power out, pick a more powerful pair of output tubes, so that you do not have to draw any grid current.
Do not draw grid current for RC coupled stages.
Interstage coupling is also wonderful.
But there are many other factors you must consider, not just cost.
Magnetic hum pickup
Bandwidth, low and high frequencies
Square Wave response
Perhaps other tradeoffs too.
As to the output stage, since there is no negative feedback, the gain of that stage is merely the tube transconductance times the load impedance.
Simple.
Input/phase splitter/driver need only be one stage.
"You should make things as simple as possible, but no simpler" - Albert Einstein
Last edited:
Aperiodic mid-high speaker is not a problem, the problem is main resonance. Aperiodic subwoofer is possible - a large array of drivers whose main resonance is above the working frequency range.
Alternatively, a bass driver can be acoustically dampened. Scan Speak used to sell acoustic resistance panels, or one can DIY them.
Alternatively, a bass driver can be acoustically dampened. Scan Speak used to sell acoustic resistance panels, or one can DIY them.
Be sure to use all battery power for your amplifiers.
No capacitors needed.
I know Gary Pimm, and I heard his all battery power amplifier, great sound!
I also know Lynn Olson, I have heard some of his amplifiers and his speakers, great sound!
I even played one of my amplifiers on his speakers in his home.
For a while, we worked in the same division of a major test and measurement company.
Oh, Gary Pimm and I worked at that T&M company too.
Even Ultra Path amplifiers have capacitors in them, I have heard them, great sound too!
"All Generalizations Have Exceptions" - Me
Start by peeling the outer layers off of the onion, before you peel the inner layers.
No capacitors needed.
I know Gary Pimm, and I heard his all battery power amplifier, great sound!
I also know Lynn Olson, I have heard some of his amplifiers and his speakers, great sound!
I even played one of my amplifiers on his speakers in his home.
For a while, we worked in the same division of a major test and measurement company.
Oh, Gary Pimm and I worked at that T&M company too.
Even Ultra Path amplifiers have capacitors in them, I have heard them, great sound too!
"All Generalizations Have Exceptions" - Me
Start by peeling the outer layers off of the onion, before you peel the inner layers.
Last edited:
sser2,
I apologize.
I did use too high of a number, 120dB.
The youth full ear may have 120dB of dynamic range, but Not at the same time.
At 0 dB, the ear's bones are in the lowest attenuation position.
At 120dB, the ears bones will adjust as quickly as possible to the highest attenuation position.
I used 120dB for effect.
Often, taking both the worst and the best extremes is helpful to illustrate the range of the problem.
There are so many tube types, pentodes, beam power, tetrodes, triodes, etc.
I look at the triode and see:
Grid, high impedance
Plate, medium impedance
Cathode, low impedance.
I prefer individual self bias, with all the tradeoffs and advantages that it has.
But you are correct, a capacitor can affect the sound of an amplifier.
Using self bias, and RC coupling:
Capacitor current is least for the grid.
Capacitor current is equal for the cathode and the plate.
The plate impedance, rp, and the plate load Rp are typically higher impedance than the construction and materials of the B+ capacitor-caused distortion. So the B+ filter cap gets a "medium layer" assignment to the onion.
But the cathode bypass cap is the most subject to capacitor sound.
That is because the amplified current has to be bypassed by the cap.
I will use the term "grunge" to describe some kinds of distortion.
And any "grunge" that is developed across the bypass cap, is also amplified again by the tube, because the "grunge" voltage that develops across that cap affects the cathode to grid voltage, and that is amplified in the plate.
This is one of the "outer layers" of the onion.
The grid RC coupling cap, is one of the most "inner layers" of the onion.
Consider the capacitance of that cap; versus the grid resistor, rg, and the grid capacitance.
Just my opinion.
I apologize.
I did use too high of a number, 120dB.
The youth full ear may have 120dB of dynamic range, but Not at the same time.
At 0 dB, the ear's bones are in the lowest attenuation position.
At 120dB, the ears bones will adjust as quickly as possible to the highest attenuation position.
I used 120dB for effect.
Often, taking both the worst and the best extremes is helpful to illustrate the range of the problem.
There are so many tube types, pentodes, beam power, tetrodes, triodes, etc.
I look at the triode and see:
Grid, high impedance
Plate, medium impedance
Cathode, low impedance.
I prefer individual self bias, with all the tradeoffs and advantages that it has.
But you are correct, a capacitor can affect the sound of an amplifier.
Using self bias, and RC coupling:
Capacitor current is least for the grid.
Capacitor current is equal for the cathode and the plate.
The plate impedance, rp, and the plate load Rp are typically higher impedance than the construction and materials of the B+ capacitor-caused distortion. So the B+ filter cap gets a "medium layer" assignment to the onion.
But the cathode bypass cap is the most subject to capacitor sound.
That is because the amplified current has to be bypassed by the cap.
I will use the term "grunge" to describe some kinds of distortion.
And any "grunge" that is developed across the bypass cap, is also amplified again by the tube, because the "grunge" voltage that develops across that cap affects the cathode to grid voltage, and that is amplified in the plate.
This is one of the "outer layers" of the onion.
The grid RC coupling cap, is one of the most "inner layers" of the onion.
Consider the capacitance of that cap; versus the grid resistor, rg, and the grid capacitance.
Just my opinion.
Last edited:
Hey folks, nice discussion!
So far it seems a good and easy compromise is not a full current source amplifier, but a high impedance amplifier. Let's say we're aiming at a Zout of 30-50R. It's possible the same can be done with an OTL, but we need tubes with high current output capability. That's why I was aiming for an OPT.
The OPT itself must have as low as overall capacitance as possible, while leakage inductance won't be a problem. Here we're talking about an almost to none interleaving. Let's suppose the amplifier will be used for midrange, so it can be limited on primary inductance.
Pentode push pull sounds like a good idea, especially LTP with a CCS at the bottom.
When it comes to the speaker, I will be building my own, so I will be fully aware of the impedance plot and will do correction networks if needed
So far it seems a good and easy compromise is not a full current source amplifier, but a high impedance amplifier. Let's say we're aiming at a Zout of 30-50R. It's possible the same can be done with an OTL, but we need tubes with high current output capability. That's why I was aiming for an OPT.
The OPT itself must have as low as overall capacitance as possible, while leakage inductance won't be a problem. Here we're talking about an almost to none interleaving. Let's suppose the amplifier will be used for midrange, so it can be limited on primary inductance.
Pentode push pull sounds like a good idea, especially LTP with a CCS at the bottom.
When it comes to the speaker, I will be building my own, so I will be fully aware of the impedance plot and will do correction networks if needed
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.