Aside from more complicated considerations....
The load for SE is often a compromise with 2nd harmonic distortion. Too low makes an ugly number.
In push-pull the 2nd is cancelled. While the residual appears as 3rd, it may be a far smaller number. It is then rational to decrease the load impedance to suck-out more power per precious tube dollar.
Also a SE stage has to idle hot. In P-P the idle can be lower, which allows much leeway in B+ or load impedance.
So no, there is no simple 2X rule which always gives a best answer.
The load for SE is often a compromise with 2nd harmonic distortion. Too low makes an ugly number.
In push-pull the 2nd is cancelled. While the residual appears as 3rd, it may be a far smaller number. It is then rational to decrease the load impedance to suck-out more power per precious tube dollar.
Also a SE stage has to idle hot. In P-P the idle can be lower, which allows much leeway in B+ or load impedance.
So no, there is no simple 2X rule which always gives a best answer.
For a PP pentode stage I like to set the load impedance lower than that which generates the largest output, up into the knee of the plate family. The load is usually a loudspeaker, for much of the program spectrum the speaker impedance is higher than nominal. And the 3rd is reduced at the same time.
In some ways the power output spec of pentodes is misleading. it happens only under optimum conditions. Triodes are much more tolerant of load, less affected by moderate mismatching. But for them better the load be on the high side. The UL connexion is a partial fix.
My opinions only, others may differ.
Should be fine as long as dissipation is kept reasonable.
Voltage limits on a lot of tubes, especially miniature tubes, has to do with the ability of the socket to resist breakdown. It's why non-top-cap sweep tubes usually have a few empty pins on either side of the anode. The internal parts are unlikely to arc as long as the vacuum is intact, it's the air's ability to ionize that is the main problem. A 7-pin miniature tube has ~2.5mm between pins which gives you until ~450V standing DC before it's almost guaranteed to arc, assuming it's clean. Impulse voltage can go way above that, depending on how fast it is, before it happens. On the 6AQ5 your highest risk is pin 5 to 4(plate to heater).
Internal arcs do happen in good tubes when extremes are reached. For example, a frame grid tube like a 6DJ8 with REALLY close grid to cathode spacing seeing a couple hundred volts difference between the two. DC coupled cathode followers can sometimes arc upon startup with a solid state supply and no grid to cathode zener protection.
For the 6AQ5, do you guys think it's safe to use it with an anode voltage of 310V and a reduced screen voltage of 200V? Assuming operation within the maximum dissipation limits.
When plate & screen are operated from the same supply but at large voltage differences, some kind of stabilization of the screen voltage is required. Otherwise the screen voltage will drop a lot, depending on the screen instantaneous current & the screen supply impedance.
Easy way out was to use VR tubes. Now zeners, but there are problems with temp coefficient of voltage. Better to use two 100V diodes rather than a single 200V part. That gets a better (lower) internal resistance as well.
The loadline needs to be adjusted, higher than the published numbers.
When plate & screen are operated from the same supply but at large voltage differences, some kind of stabilization of the screen voltage is required. Otherwise the screen voltage will drop a lot, depending on the screen instantaneous current & the screen supply impedance.
Easy way out was to use VR tubes. Now zeners, but there are problems with temp coefficient of voltage. Better to use two 100V diodes rather than a single 200V part. That gets a better (lower) internal resistance as well.
The loadline needs to be adjusted, higher than the published numbers.
If the objective is maximum power output the reason for the low screen voltage is maximizing plate efficiency while avoiding melting the screen grid.
In any amplifier tube or solid state a certain amount of DC power is fed to the active devices in the amp. Some of this gets turned into audio and the rest is turned into heat. In most amplifier designs the great majority of the wasted power occurs in the output stages, so improving the efficiency in the output stage offers the most benefit, and that's where we look for improvements. We will also neglect the poaer turned into heat in the tube heater / cathode structure.
Lets look at the data sheet value for the 6AQ5 in class AB1. We supply 250 volts to the plate and screen and crank up the drive until we get 10 watts from our class AB1 pair. The plate supply is feeding the OPT 79 mA, or 19.75 watts, burning 9.75 watts in the output tubes, or 4.875 watts per tube. This is 10.6 % plate efficiency. Obviously we could do something to generate more tube current, since the plates are running at less than half of their rated dissipation.......
This same amplifier draws 13 mA from the screen supply or 3.25 watts, or 1.65 watts per tube. This is 81.25% of the maximum screen dissipation. If we just drove the tube harder and cleaned up the distortion with feedback we would soon over dissipate the screen grid leading to internal element warpage, a hot spot, and eventual meltdown.
I have long stated that the 6AQ5 was limited to 10 to 12 watts of power output, and melted a couple dozen tubes to find a way past this limit. I have had several hundred 6AQ5's in a box for over 10 years because I believed that, but........
I found the benefits of driving the control grid positive (A2 or AB2 operation) several years ago when playing with 45's and 300B's. This is a documented method of improving the efficiency of the output stage by reducing the saturation voltage of a tube. It however does not help the screen dissipation issue, and makes it worse, because as the plate voltage is pulled lower more electrons are diverted to the screen grid, especially when the screen grid is more positive than the plate. Dissipation in the plate or any grid is determined by the voltage on the element multiplied by the current through it. So the method for reducing screen dissipation is simple, turn down the voltage and it will draw less current, a double hitter in the direction we want to go. Simply reducing the screen voltage will reduce efficiency and power output, so we must use another means to retrieve the losses. There are three, and a balance between them must be struck. It will be different for each application.
We can turn down the screen voltage and drive the control grid somewhat positive to keep the losses down. We can also increase the plate voltage, which also reduces the screen current since the higher plate voltage will attract more electrons. We can also reduce the load impedance if the cathode emission is high enough to support the peak current demands.
Most tubes can be operated well beyond their published plate voltage ratings if a CLEAN modern ceramic socket is used. The limiting factor is usually idle dissipation. As the plate voltage is increased the plate dissipation also increases. You can reduce the idle current to keep the dissipation down, but there will be a point where you would need a bunch of feedback to keep crossover distortion down. That voltage on most 25 to 30 watt TV sweep tubes is 600 to 650 volts, while a 6AQ5 can run a 400 volts and still make 0.2% THD at 100 milliwatts out with zero GNFB and a few dB of local feedback.
The plate dissipation can be briefly exceeded if the average is kept to a reasonable level.
Bob Carver exploited the fact that music has at least a 10 dB peak to average ratio (often called crest factor), and usually 20+ dB. This means that your 10 watt amp, if pushed to clipping on peaks, is producing 1 watt or less on average, and the heat sinks (or plate dissipation ratings) can be sized accordingly. Despite warnings to the contrary, I beat my Carver M-400 mercilessly with my guitar for over 10 years and it never blew up. It was still working when I gave it away a few years ago. We can do this with tubes too.
If you are building an amp for musical instrument amplification, PA use, or cranking uber compressed dance music well into distortion, or plan to run sine wave tests for hours, you shouldn't venture too far beyond the published ratings. For typical HiFi music listening the plate dissipation spec can be violated by a factor of two or more at full power IF the idle dissipation is well below the spec. The average dissipation will still be less than spec.
It is still important to avoid violating the screen grid ratings since it is would with tiny wire and has virtually zero thermal lag. One cranked guitar note can take a screen grid from cold to glowing before it dies out. Some means of limiting the drive or screen current is needed if the amp will be pushed into clipping since the screen current skyrockets when the plate is pulled to near zero. I find that regulated screen supply is helpful and current limiting is easy to implement, but can cause rather ugly distortion depending on how it's implemented. I find that a simple series resistor in series with the screen fed from a simple stabilized (zener and mosfet) supply works without undue distortion, although I wouldn't run this int heavy clipping for long periods of time. For guitar amp use I tie both screen grids together, and use a series resistor from a stabilized supply about 20 to 30 volts higher than needed. I then hang the entire preamp chain on the screen side of this resistor. This reduces the peak drive at the onset of clipping and makes for some good touch sensitivity and creamy smooth distortion.
So haw can we do this with a 6AQ5?
I set out to find one possible recipe with the following criteria. Maximum power output, per tube plate dissipation limited to 15 watts, slightly over spec, but the amp must run at full power without tube glow in a dark room for an hour (tested with two different tube sets). Idle dissipation should be 5 to 8 watts per tube. Screen grid dissipation below the 2 watt spec for all operating conditions, load and B+ voltages could be anything, but commonly available OPT impedances should be chosen.....
I got my desired results at 400 volts of B+ with a 6600 ohm OPT and 200 volts on the screen. The control grid is driven positive for power outputs beyond 8 to 10 watts. Idle current was set at 17.5mA or just over 7 watts per tube. THD is 0.2% at 100 mW, 0.4% at 1 watt, and rises to 2% around 30 watts. On one set of tubes the screen dissipation hits 1.9 watts at 35 watts out, while the other set is at 1.6 watts at 35 watts out. Plate dissipation is right at 15 watts per tube for a plate efficiency of 53 to 54 %.
Both sets are sucking 3 to 4 watts per tube when pushed to 40 watts out, and the THD is over 10%, so most people would want to turn it down. Extended sine wave operation at 40 watts provoked mild screen grid glow in one tube and a pale redness on the plates in a dark room on three tubes. Plate dissipation was 17 watts per tube.
My guess is that 310 volts on the plates with 200 volts on the screen is OK, but monitoring the screen current during initial testing is needed to avoid over dissipation in the screen grid. The higher plate voltage I used actually reduces the screen current. This is all load dependent, and you didn't specify a load impedance. Higher impedances tend to reduce average G2 dissipation, but may raise it during clipping.
In any amplifier tube or solid state a certain amount of DC power is fed to the active devices in the amp. Some of this gets turned into audio and the rest is turned into heat. In most amplifier designs the great majority of the wasted power occurs in the output stages, so improving the efficiency in the output stage offers the most benefit, and that's where we look for improvements. We will also neglect the poaer turned into heat in the tube heater / cathode structure.
Lets look at the data sheet value for the 6AQ5 in class AB1. We supply 250 volts to the plate and screen and crank up the drive until we get 10 watts from our class AB1 pair. The plate supply is feeding the OPT 79 mA, or 19.75 watts, burning 9.75 watts in the output tubes, or 4.875 watts per tube. This is 10.6 % plate efficiency. Obviously we could do something to generate more tube current, since the plates are running at less than half of their rated dissipation.......
This same amplifier draws 13 mA from the screen supply or 3.25 watts, or 1.65 watts per tube. This is 81.25% of the maximum screen dissipation. If we just drove the tube harder and cleaned up the distortion with feedback we would soon over dissipate the screen grid leading to internal element warpage, a hot spot, and eventual meltdown.
I have long stated that the 6AQ5 was limited to 10 to 12 watts of power output, and melted a couple dozen tubes to find a way past this limit. I have had several hundred 6AQ5's in a box for over 10 years because I believed that, but........
I found the benefits of driving the control grid positive (A2 or AB2 operation) several years ago when playing with 45's and 300B's. This is a documented method of improving the efficiency of the output stage by reducing the saturation voltage of a tube. It however does not help the screen dissipation issue, and makes it worse, because as the plate voltage is pulled lower more electrons are diverted to the screen grid, especially when the screen grid is more positive than the plate. Dissipation in the plate or any grid is determined by the voltage on the element multiplied by the current through it. So the method for reducing screen dissipation is simple, turn down the voltage and it will draw less current, a double hitter in the direction we want to go. Simply reducing the screen voltage will reduce efficiency and power output, so we must use another means to retrieve the losses. There are three, and a balance between them must be struck. It will be different for each application.
We can turn down the screen voltage and drive the control grid somewhat positive to keep the losses down. We can also increase the plate voltage, which also reduces the screen current since the higher plate voltage will attract more electrons. We can also reduce the load impedance if the cathode emission is high enough to support the peak current demands.
Most tubes can be operated well beyond their published plate voltage ratings if a CLEAN modern ceramic socket is used. The limiting factor is usually idle dissipation. As the plate voltage is increased the plate dissipation also increases. You can reduce the idle current to keep the dissipation down, but there will be a point where you would need a bunch of feedback to keep crossover distortion down. That voltage on most 25 to 30 watt TV sweep tubes is 600 to 650 volts, while a 6AQ5 can run a 400 volts and still make 0.2% THD at 100 milliwatts out with zero GNFB and a few dB of local feedback.
The plate dissipation can be briefly exceeded if the average is kept to a reasonable level.
Bob Carver exploited the fact that music has at least a 10 dB peak to average ratio (often called crest factor), and usually 20+ dB. This means that your 10 watt amp, if pushed to clipping on peaks, is producing 1 watt or less on average, and the heat sinks (or plate dissipation ratings) can be sized accordingly. Despite warnings to the contrary, I beat my Carver M-400 mercilessly with my guitar for over 10 years and it never blew up. It was still working when I gave it away a few years ago. We can do this with tubes too.
If you are building an amp for musical instrument amplification, PA use, or cranking uber compressed dance music well into distortion, or plan to run sine wave tests for hours, you shouldn't venture too far beyond the published ratings. For typical HiFi music listening the plate dissipation spec can be violated by a factor of two or more at full power IF the idle dissipation is well below the spec. The average dissipation will still be less than spec.
It is still important to avoid violating the screen grid ratings since it is would with tiny wire and has virtually zero thermal lag. One cranked guitar note can take a screen grid from cold to glowing before it dies out. Some means of limiting the drive or screen current is needed if the amp will be pushed into clipping since the screen current skyrockets when the plate is pulled to near zero. I find that regulated screen supply is helpful and current limiting is easy to implement, but can cause rather ugly distortion depending on how it's implemented. I find that a simple series resistor in series with the screen fed from a simple stabilized (zener and mosfet) supply works without undue distortion, although I wouldn't run this int heavy clipping for long periods of time. For guitar amp use I tie both screen grids together, and use a series resistor from a stabilized supply about 20 to 30 volts higher than needed. I then hang the entire preamp chain on the screen side of this resistor. This reduces the peak drive at the onset of clipping and makes for some good touch sensitivity and creamy smooth distortion.
So haw can we do this with a 6AQ5?
I set out to find one possible recipe with the following criteria. Maximum power output, per tube plate dissipation limited to 15 watts, slightly over spec, but the amp must run at full power without tube glow in a dark room for an hour (tested with two different tube sets). Idle dissipation should be 5 to 8 watts per tube. Screen grid dissipation below the 2 watt spec for all operating conditions, load and B+ voltages could be anything, but commonly available OPT impedances should be chosen.....
I got my desired results at 400 volts of B+ with a 6600 ohm OPT and 200 volts on the screen. The control grid is driven positive for power outputs beyond 8 to 10 watts. Idle current was set at 17.5mA or just over 7 watts per tube. THD is 0.2% at 100 mW, 0.4% at 1 watt, and rises to 2% around 30 watts. On one set of tubes the screen dissipation hits 1.9 watts at 35 watts out, while the other set is at 1.6 watts at 35 watts out. Plate dissipation is right at 15 watts per tube for a plate efficiency of 53 to 54 %.
Both sets are sucking 3 to 4 watts per tube when pushed to 40 watts out, and the THD is over 10%, so most people would want to turn it down. Extended sine wave operation at 40 watts provoked mild screen grid glow in one tube and a pale redness on the plates in a dark room on three tubes. Plate dissipation was 17 watts per tube.
My guess is that 310 volts on the plates with 200 volts on the screen is OK, but monitoring the screen current during initial testing is needed to avoid over dissipation in the screen grid. The higher plate voltage I used actually reduces the screen current. This is all load dependent, and you didn't specify a load impedance. Higher impedances tend to reduce average G2 dissipation, but may raise it during clipping.
Attachments
Thank you all for your replies.
That should be no problem, I have a PCB with a regulated supply.
Because I already have a 230V(AC) power transformer which gives me 320V(DC).
I have an output transformer with 8000 Ohms plate-to-plate impedance.
That should be no problem, I have a PCB with a regulated supply.
It wud be a lot simpler to run plate & screen at something like 250V if lower HV supply is the objective.
Because I already have a 230V(AC) power transformer which gives me 320V(DC).
I have an output transformer with 8000 Ohms plate-to-plate impedance.
There will be ~320v at no load. A practical supply with cap input to filter will be more like 280 - 290V. Then the regulator will need some volts, depends on ripple at the output of the filter. So might be 260V available in a real cct.
With the screen so much lower the optimum load will be more like 10 to 12K plate to plate.
There will be ~320v at no load. A practical supply with cap input to filter will be more like 280 - 290V. Then the regulator will need some volts, depends on ripple at the output of the filter. So might be 260V available in a real cct.
With the screen so much lower the optimum load will be more like 10 to 12K plate to plate.
260V? Do you think so? Then I can just as well use the 6AQ5 datasheet values, no?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.