Greetings!
I have a couple of 805 tubes on order - I'm an electrical engineer with a fascination for tube amplifiers. I have a stereo tube amp with a Dynaco 70 topology - I bought it in the late 1990's and assembled it - I love it! Now I have become interested in single-ended amps. Here is my proposal, and I am going to take my time with this:
(1) Build an 805-based tube amp, and I can drive the 805 either with triodes or MOSFET's, or at least use MOSFET's in the bias circuitry.
(2) I also work with embedded computing in my job as an electrical engineer. Likewise, I would like to build the amp as a microprocessor controlled unit.
(3) Just to give you an idea as to what I'm thinking, I'm considering even being able to control the high voltage via microprocessor. Here is the scheme:
AC-DC power supply->
OP-AMP voltage adjustment (from digital to analog converter on CPU)->
inverter (oscillation either from 555 timer or CPU on-board timer) - capable of operating with varying voltage->
Transformer to high voltage->
DC rectifier
With this scheme, I could drive with a frequency higher than 60 Hz, but I would want to be careful there. I would likely have a band-pass filter so I get a near-sinewave from the timer output, or at least take out some of the harmonics.
The computer's role would be startup and monitoring. In the startup phase, the tubes would be warmed up and bias soft-started. During operation, currents and voltages could be monitored for faults, and the amplifier shut down if necessary.
I have seen different topologies for 805 based SE amps, and it seems doable to me. The high voltage is not a problem as I have been working with high voltage since I was a kid (my Dad taught me how to be safe - especially in working with lethal voltages in a TV set of the era).
Let me know any ideas you have. I know the 805 tube likes to have a transformer from the stage before it. Any recommendations? Also, I want to know any recommendations for the output transformer - I know Hammond makes SE type transformers.
Thanks!
Kevin
I have a couple of 805 tubes on order - I'm an electrical engineer with a fascination for tube amplifiers. I have a stereo tube amp with a Dynaco 70 topology - I bought it in the late 1990's and assembled it - I love it! Now I have become interested in single-ended amps. Here is my proposal, and I am going to take my time with this:
(1) Build an 805-based tube amp, and I can drive the 805 either with triodes or MOSFET's, or at least use MOSFET's in the bias circuitry.
(2) I also work with embedded computing in my job as an electrical engineer. Likewise, I would like to build the amp as a microprocessor controlled unit.
(3) Just to give you an idea as to what I'm thinking, I'm considering even being able to control the high voltage via microprocessor. Here is the scheme:
AC-DC power supply->
OP-AMP voltage adjustment (from digital to analog converter on CPU)->
inverter (oscillation either from 555 timer or CPU on-board timer) - capable of operating with varying voltage->
Transformer to high voltage->
DC rectifier
With this scheme, I could drive with a frequency higher than 60 Hz, but I would want to be careful there. I would likely have a band-pass filter so I get a near-sinewave from the timer output, or at least take out some of the harmonics.
The computer's role would be startup and monitoring. In the startup phase, the tubes would be warmed up and bias soft-started. During operation, currents and voltages could be monitored for faults, and the amplifier shut down if necessary.
I have seen different topologies for 805 based SE amps, and it seems doable to me. The high voltage is not a problem as I have been working with high voltage since I was a kid (my Dad taught me how to be safe - especially in working with lethal voltages in a TV set of the era).
Let me know any ideas you have. I know the 805 tube likes to have a transformer from the stage before it. Any recommendations? Also, I want to know any recommendations for the output transformer - I know Hammond makes SE type transformers.
Thanks!
Kevin
For a "first project", the 805 is one heck of a plunge off the deep end. The two concerns for a first project is the size/complexity/safety of
the B+ supply followed by the big design challenges for driving the 805's grid. Unlike "other" tubes, where you simply apply AC to the grid and you modulate plate current, the 805 will source current FROM the grid pin while you modulate plate current. This requires a very low impedance driver thats capable of swinging lots of volts _and_ dealing with sinking all the current that comes its way. Honestly, Class A1 is much easier to get your feet with then class A2.
All is not lost.... Keep those 805 you ordered. Spend another $80
and buy a pair of the Chinese 845s, or, the 211s. They fit in the same socket as the 805. They all need a 10V 4A filament supply.... The difference is Class A1 vs. Class A2. I'd recommend a 6600 ohm transformer.... The Hammond 1629SEA to be exact. Make sure you get the NEW edition (the SEA), and not the prior edition (the SE). 6.6K you can do all of them - 805,845,211. You could also get the Hammond 1638SEA (10K load). 10K is abit high for 845, but nice for 805s and 211. Do a simply Class A1 amp first, then shoot for A2.
As far as power sequencing, tungsten filament tubes heat up so fast you dont have to worry about power sequencing... Since you input and driver tubes are likely to be 6V, just apply power to the 6V filament transformer about 30 sec before enabling B+. Real easy. No digital needed. Just use a 6.3V heated 30-second Amperite glass thermal relay. 🙂
Regarding driving the 805, it's possible to deal with its insidiousness using various MOSFET followers, or, tube based cathode followers using big meaty power pentodes. Step-down interstage transformers are do-able, but require spending LOTS AND LOTS of money on a custom IITC. Cathode followers arents that bad.
If you want to be a trooper, how about a Power JFET driver since we all hate dealing with MOSFET gate fragility and feed through.
If you really want to be a trooper, how about you use your BSEE to
design and build for us an Ultrasonic filament supply for the 805/845/211. DC filaments are a problem with these tubes (suck the sound ouf of them). So, how about you figure out how to light
the 845 by giving it a steady 10.00 V at the required 3.25A at, oh, lets say 500 kHz. 🙂
-- Jim
the B+ supply followed by the big design challenges for driving the 805's grid. Unlike "other" tubes, where you simply apply AC to the grid and you modulate plate current, the 805 will source current FROM the grid pin while you modulate plate current. This requires a very low impedance driver thats capable of swinging lots of volts _and_ dealing with sinking all the current that comes its way. Honestly, Class A1 is much easier to get your feet with then class A2.
All is not lost.... Keep those 805 you ordered. Spend another $80
and buy a pair of the Chinese 845s, or, the 211s. They fit in the same socket as the 805. They all need a 10V 4A filament supply.... The difference is Class A1 vs. Class A2. I'd recommend a 6600 ohm transformer.... The Hammond 1629SEA to be exact. Make sure you get the NEW edition (the SEA), and not the prior edition (the SE). 6.6K you can do all of them - 805,845,211. You could also get the Hammond 1638SEA (10K load). 10K is abit high for 845, but nice for 805s and 211. Do a simply Class A1 amp first, then shoot for A2.
As far as power sequencing, tungsten filament tubes heat up so fast you dont have to worry about power sequencing... Since you input and driver tubes are likely to be 6V, just apply power to the 6V filament transformer about 30 sec before enabling B+. Real easy. No digital needed. Just use a 6.3V heated 30-second Amperite glass thermal relay. 🙂
Regarding driving the 805, it's possible to deal with its insidiousness using various MOSFET followers, or, tube based cathode followers using big meaty power pentodes. Step-down interstage transformers are do-able, but require spending LOTS AND LOTS of money on a custom IITC. Cathode followers arents that bad.
If you want to be a trooper, how about a Power JFET driver since we all hate dealing with MOSFET gate fragility and feed through.
If you really want to be a trooper, how about you use your BSEE to
design and build for us an Ultrasonic filament supply for the 805/845/211. DC filaments are a problem with these tubes (suck the sound ouf of them). So, how about you figure out how to light
the 845 by giving it a steady 10.00 V at the required 3.25A at, oh, lets say 500 kHz. 🙂
-- Jim
811A as a study case
Similar stuff going on here.... The 811A is a Class B transmitting tube. At high plate B+, the grid current is low. At moderate plate B+, the grid current is high. At low plate B+, its pointless to use this tube as its all grid current.
At 430V, the grid has a transconductance of 1mA/V. Or, it has
source impedance of 1 Kohm. If you bias the 811 with +24V, the
plate resistance is going to be about 1200 ohms, so its a nice match
for a 5K:8 output transformer. However, your driver now has to deal with 24mA at zero signal, and further, it must sink upwards of 48mA at peaks and be able to hold the 811A close to 0V with a few mA moving around. Obviously, the driver is going to distort as you try to get it to take the cathode down to 0V.
The driver selection had two requirements: It has to have high enough of a Gm to drive 1K. Since the cathode has a Zsink of 1/Gm,
many tubes can drive the 1K as the typical cathode follower can do 200-300 ohm.... Second, it needs an operating point near 20V and around 24mA. Lots of stuff can do that: 6V6 family, EL84. Bigger stuff like 6L6 and EL34 is too much.
You could get an 805 off the ground in the same way.... I'm guessing around 650VDC you could use the same circuit.
BTW People liked the sound of this amp alot... It had power too. 9W of power! The second iteration I used a parafeed output transformer, which people really liked!
-- Jim
An externally hosted image should be here but it was not working when we last tested it.
Similar stuff going on here.... The 811A is a Class B transmitting tube. At high plate B+, the grid current is low. At moderate plate B+, the grid current is high. At low plate B+, its pointless to use this tube as its all grid current.
At 430V, the grid has a transconductance of 1mA/V. Or, it has
source impedance of 1 Kohm. If you bias the 811 with +24V, the
plate resistance is going to be about 1200 ohms, so its a nice match
for a 5K:8 output transformer. However, your driver now has to deal with 24mA at zero signal, and further, it must sink upwards of 48mA at peaks and be able to hold the 811A close to 0V with a few mA moving around. Obviously, the driver is going to distort as you try to get it to take the cathode down to 0V.
The driver selection had two requirements: It has to have high enough of a Gm to drive 1K. Since the cathode has a Zsink of 1/Gm,
many tubes can drive the 1K as the typical cathode follower can do 200-300 ohm.... Second, it needs an operating point near 20V and around 24mA. Lots of stuff can do that: 6V6 family, EL84. Bigger stuff like 6L6 and EL34 is too much.
You could get an 805 off the ground in the same way.... I'm guessing around 650VDC you could use the same circuit.
BTW People liked the sound of this amp alot... It had power too. 9W of power! The second iteration I used a parafeed output transformer, which people really liked!
-- Jim
Ultrasonic Power Supply?
The idea was mentioned to use an ultrasonic power supply to create the 10 volt 3.25 amps for filaments. That would be very possible. It would also be fairly easy for the higher voltages.
For starters, smaller transformers will suffice - but big iron transformers will not be hurt by these frequencies. Due to the intricacies of electromagnetics, bigger iron is needed for lower frequencies. Also, I would not want to step down due to the high DC bias going through them.
OK, now as for ultrasonic power supplies, here's how it would work:
Take the DC voltage off a power supply, say 6 volts, 12 volts - whatever. Use either a microprocessor timer or a 555 timer at the desired frequency, and feed into an inverter circuit (uses two transistors of some sort, whether BJT's, JFET's, or MOSFET's). Then you can use another transformer to "step up" the voltage accordingly. Also note that instead of transistors, I would probably use a power OP-AMP - with this device, I could feed in the timer signal and adjust the gain accordingly. There are OP-AMPS, typically 5 or 7 pin, in a TO-220 type case. You would first need to design the circuit for proper gain to give you the voltage you want (10 volts, for example). There you would get an output signal with no problems. As for the timer signal, it would need to be capacitively coupled to the OP-AMP so that you would get actual AC. Of course, you would need a bipolar power supply for the OP-AMP.
Forgive me, but I'm thinking out loud. If it sounds complicated, it really isn't. The sequence is this:
1. DC supply
2. Timer stage
3. capacitor coupling (convert from 0-5 to -2.5 to 2.5)
4. OP-AMP set for proper gain
5. Step-up transformer
I'm thinking of using this scheme to get the high voltage so I can control it. I think solid state can really make tubes rock! (Or, as I should say "waltz" since I'm a classical musician. I love what tubes do to classical music.)
The idea was mentioned to use an ultrasonic power supply to create the 10 volt 3.25 amps for filaments. That would be very possible. It would also be fairly easy for the higher voltages.
For starters, smaller transformers will suffice - but big iron transformers will not be hurt by these frequencies. Due to the intricacies of electromagnetics, bigger iron is needed for lower frequencies. Also, I would not want to step down due to the high DC bias going through them.
OK, now as for ultrasonic power supplies, here's how it would work:
Take the DC voltage off a power supply, say 6 volts, 12 volts - whatever. Use either a microprocessor timer or a 555 timer at the desired frequency, and feed into an inverter circuit (uses two transistors of some sort, whether BJT's, JFET's, or MOSFET's). Then you can use another transformer to "step up" the voltage accordingly. Also note that instead of transistors, I would probably use a power OP-AMP - with this device, I could feed in the timer signal and adjust the gain accordingly. There are OP-AMPS, typically 5 or 7 pin, in a TO-220 type case. You would first need to design the circuit for proper gain to give you the voltage you want (10 volts, for example). There you would get an output signal with no problems. As for the timer signal, it would need to be capacitively coupled to the OP-AMP so that you would get actual AC. Of course, you would need a bipolar power supply for the OP-AMP.
Forgive me, but I'm thinking out loud. If it sounds complicated, it really isn't. The sequence is this:
1. DC supply
2. Timer stage
3. capacitor coupling (convert from 0-5 to -2.5 to 2.5)
4. OP-AMP set for proper gain
5. Step-up transformer
I'm thinking of using this scheme to get the high voltage so I can control it. I think solid state can really make tubes rock! (Or, as I should say "waltz" since I'm a classical musician. I love what tubes do to classical music.)
First check this out. Here a single ended 813 amp is controlled by a microprocessor.
http://www.pmillett.com/813_se_triode_amps.htm
I have experimented with microprocessor power sequencing before, and will use it again sometime. You need to be real careful not to let processor noise get into the audio chain. I used a PIC microcontroller. The power on reset starts the power up sequence for the tube amp section. After the amp is alive and all voltages and currents check out, the processor goes into deep sleep mode (clock stopped). This eliminates all processor noise. The processor has a wake up on change function that starts it up when any button is pushed.
I am (slowly) designing a microprocessor controlled guitar amp. The main duty of the processor here is saving setups and recalling them in a live environment.
The "ultrasonic" power supply is a bigger challenge. There have been threads about SMPS's (switch mode power supply) on this forum in the past. The tube audio world has not accepted these yet. That doesn't mean that it can't be done. A few forward thinkers have succeeded with these.
First, the frequency of operation needs to be quite a bit higher that the audio range. Second a considerable amount of power will be needed. I briefly experimented with ultrasonic power for filaments a few years ago. I believe that a DHT filament works better when fed with a sine wave. This means that you will need a power op amp (or other linear circuit) that can provide 30 to 50 watts of power at 50 KHz (or higher). This is not trivial, and my attempts resulted in melted mosfets.
The HV supply will want to operate with square waves for the sake of efficiency. A push pull smps converter can be used here. Conventional diodes will MELT in this application because they are too slow. There are now HV schottky diodes that did not exist when I tried this. The control loop is not trivial and will take some tinkering if a microprocessor is included due to the processing delay. People have used PIC processors for power supply controllers, and there are app notes on the Microchip web site.
All ultrasonic supplies should operate from the same clock. This is to avoid beat notes that may end up in the audio range. They should be derived from the microporcessor clock for the same reason.
Since there are a lot of unknown ground here, I would start with a much smaller amp, get it working with a conventional power supply. I use a lab supply from the 1960's that was designed for use with vacuum tubes. Then you can work on the unique stuff with a known working amplifier. I believe in minimizing the chances of failure, especially when forging new ground. You want only one unknown variable in each experiment.
http://www.pmillett.com/813_se_triode_amps.htm
I have experimented with microprocessor power sequencing before, and will use it again sometime. You need to be real careful not to let processor noise get into the audio chain. I used a PIC microcontroller. The power on reset starts the power up sequence for the tube amp section. After the amp is alive and all voltages and currents check out, the processor goes into deep sleep mode (clock stopped). This eliminates all processor noise. The processor has a wake up on change function that starts it up when any button is pushed.
I am (slowly) designing a microprocessor controlled guitar amp. The main duty of the processor here is saving setups and recalling them in a live environment.
The "ultrasonic" power supply is a bigger challenge. There have been threads about SMPS's (switch mode power supply) on this forum in the past. The tube audio world has not accepted these yet. That doesn't mean that it can't be done. A few forward thinkers have succeeded with these.
First, the frequency of operation needs to be quite a bit higher that the audio range. Second a considerable amount of power will be needed. I briefly experimented with ultrasonic power for filaments a few years ago. I believe that a DHT filament works better when fed with a sine wave. This means that you will need a power op amp (or other linear circuit) that can provide 30 to 50 watts of power at 50 KHz (or higher). This is not trivial, and my attempts resulted in melted mosfets.
The HV supply will want to operate with square waves for the sake of efficiency. A push pull smps converter can be used here. Conventional diodes will MELT in this application because they are too slow. There are now HV schottky diodes that did not exist when I tried this. The control loop is not trivial and will take some tinkering if a microprocessor is included due to the processing delay. People have used PIC processors for power supply controllers, and there are app notes on the Microchip web site.
All ultrasonic supplies should operate from the same clock. This is to avoid beat notes that may end up in the audio range. They should be derived from the microporcessor clock for the same reason.
Since there are a lot of unknown ground here, I would start with a much smaller amp, get it working with a conventional power supply. I use a lab supply from the 1960's that was designed for use with vacuum tubes. Then you can work on the unique stuff with a known working amplifier. I believe in minimizing the chances of failure, especially when forging new ground. You want only one unknown variable in each experiment.
Tubelab's Power Drive might be perfect to drive the 805's grid.
Questions for Tubelab. Do you think a high voltage bipolar transistor could do better than the 2SK2700 MOSFET? Do you think a constant current sink could do better than the resistor that is used to sink the 805 grid current when the grid voltage is negtive?
Questions for Tubelab. Do you think a high voltage bipolar transistor could do better than the 2SK2700 MOSFET? Do you think a constant current sink could do better than the resistor that is used to sink the 805 grid current when the grid voltage is negtive?
This has been suggested before in e-mail that I receive. It was also debated on a couple of threads here. I tried a few bipolars back when I was developing PowerDrive. They failed dur to secondary breakdown. When using a HV bipolar transistor, you need to use the SOA (safe operating area) curves.
I have not tried this, but it should work. I got an email from a reader who did make it work.
Real scary comment from jrdmedford . I have been reading reviews about commercial 805 amps and listen to pictures of several schema implementations with real components . Maybe its nonsense but I dont see so much reason to be afraid . One rule must be all components over standard . Of course 910v is deadly , 430 v either . Our cars too when badly droven .
Kerose think of the 45W SET powerful valve sound , i think its very tempting , an Audio Electric Chair .
Kerose think of the 45W SET powerful valve sound , i think its very tempting , an Audio Electric Chair .
- Status
- Not open for further replies.