Konnichiwa,
With screendrive you can be really neat. use the "Mu-Follower" output from an IXYS CCS loading (say) a 6SN7 to drive directly the screengrid of the lower KT88, then the upper valve also set up with mu-follower output (from the cathode) and with enough +B and anode current you should get an SE Amplifier with just one capacitor in the signal path (parallel feed output), very wide bandwidth and distortion and output impedance class leading for SE.... Except of course all stages would be push-pull, but without cancelling even harmonics... ;-)
Sayonara
tubelab.com said:
With screendrive you can be really neat. use the "Mu-Follower" output from an IXYS CCS loading (say) a 6SN7 to drive directly the screengrid of the lower KT88, then the upper valve also set up with mu-follower output (from the cathode) and with enough +B and anode current you should get an SE Amplifier with just one capacitor in the signal path (parallel feed output), very wide bandwidth and distortion and output impedance class leading for SE.... Except of course all stages would be push-pull, but without cancelling even harmonics... ;-)
Sayonara
Now I think that you are reading my mind, except that I am using a 6EM7. I may try a different tube later though.
Now if you build 2 of these amps, drive them out of phase, and hang the OPT between the two output stages, there will be no caps in the signal path, but it will be a push pull amp. I believe that DC offset correction will be needed.
Now if you build 2 of these amps, drive them out of phase, and hang the OPT between the two output stages, there will be no caps in the signal path, but it will be a push pull amp. I believe that DC offset correction will be needed.
GREAT WORK GEORGE!!!!
Well, during the weekend I made some tests too:
First of all, I'm moving in a different direction, I will use a pure SS CCS, without any tube in it:
This for few reasons:
1. I have a lot of big, nice to see, heatsink (so heat dissipation in the CCS is not an issue for me).
I'm planning to use one of them as a part of the chassis, making a really good-looking (and strange) arrangement. You will see.
2. I have only 4 KT88 tubes (I'm still a newbie), so I want to use them to build a parallel SE output stage, so I have no tubes left to use for the CCS.
3. I think a SS CCS is faster and more precise than a tube one: but this is only my opinion.
4. I can dissipate more power and have higher voltage across it than using a KT88 tube.
5. The minimum voltage across my CCS is about 3.5V, so I can use a lower power supply and have a bigger swing (for example, using a 550V PS I can swing the KT88 voltage from 150V to 540V...)
Back to my tests:
Of course you're right, BU508A worked fine but for a really short time, due to the secondary breakdown effect (I knew it before, but completely forgot about it cause the enthusiasm).
I substituted it with a high voltage mosfet (IRFP460N), it worked really fine without any modification.
Such a Mosfet can work up to 500V; within the current I choose (150mA) I'm into its safe area.
I will change it with a higher voltage mosfet, like a IRFPG30 or SK2610 device.
Perhaps I can also put my hands to some IXFH6N100, IXYS mosfet: 6A@1000V.
I think I wll go with the IRFPG30: they have a much lower input capacuitance (easier to be driven).
I do not have them now, I have to order them.
I built a 450V power supply, connected the CCS to it and, then, used a power resistor with a value of 1200Ohm as a load between its output and the ground.
The circuit worked perfectly for the whole afternoon, until I switched everything off.
Due to thermal drift of the sense resistors (R2, R6) the current slowly increased from 150mA (when cold) to 152 mA after half an hour (standard carbon resistor have a negative temperature coefficient): this is not an issue, I only have to substitute the resistors with temperature stable ones.
This is the last version of the circuit.
The PCB is exactly the same as before (the mosfet has the "same" pinout than the bipolar transistor).
The current was set to 150 mA, so the load was dissipating about 27W, while the transistor about 40W.
The Mosfet can dissipate up to 125W, I mounted it insulated on a BIG heatsink and had no problem at all (at the end the mosfet was really cool).
Tomorrow I will post some picture of the circuit (you know, when you need it, your digital camera batteries are always discharged...).
Now I'm building the SE circuit I posted before, with two KT88 paralleled and para-feed outout.
But,
What is "screen drive"? can you drive a tube by its screen? How? Benefits?
Ciao,
Giovanni
Well, during the weekend I made some tests too:
First of all, I'm moving in a different direction, I will use a pure SS CCS, without any tube in it:
This for few reasons:
1. I have a lot of big, nice to see, heatsink (so heat dissipation in the CCS is not an issue for me).
I'm planning to use one of them as a part of the chassis, making a really good-looking (and strange) arrangement. You will see.
2. I have only 4 KT88 tubes (I'm still a newbie), so I want to use them to build a parallel SE output stage, so I have no tubes left to use for the CCS.
3. I think a SS CCS is faster and more precise than a tube one: but this is only my opinion.
4. I can dissipate more power and have higher voltage across it than using a KT88 tube.
5. The minimum voltage across my CCS is about 3.5V, so I can use a lower power supply and have a bigger swing (for example, using a 550V PS I can swing the KT88 voltage from 150V to 540V...)
Back to my tests:
Of course you're right, BU508A worked fine but for a really short time, due to the secondary breakdown effect (I knew it before, but completely forgot about it cause the enthusiasm).
I substituted it with a high voltage mosfet (IRFP460N), it worked really fine without any modification.
Such a Mosfet can work up to 500V; within the current I choose (150mA) I'm into its safe area.
I will change it with a higher voltage mosfet, like a IRFPG30 or SK2610 device.
Perhaps I can also put my hands to some IXFH6N100, IXYS mosfet: 6A@1000V.
I think I wll go with the IRFPG30: they have a much lower input capacuitance (easier to be driven).
I do not have them now, I have to order them.
I built a 450V power supply, connected the CCS to it and, then, used a power resistor with a value of 1200Ohm as a load between its output and the ground.
The circuit worked perfectly for the whole afternoon, until I switched everything off.
Due to thermal drift of the sense resistors (R2, R6) the current slowly increased from 150mA (when cold) to 152 mA after half an hour (standard carbon resistor have a negative temperature coefficient): this is not an issue, I only have to substitute the resistors with temperature stable ones.
This is the last version of the circuit.
The PCB is exactly the same as before (the mosfet has the "same" pinout than the bipolar transistor).
The current was set to 150 mA, so the load was dissipating about 27W, while the transistor about 40W.
The Mosfet can dissipate up to 125W, I mounted it insulated on a BIG heatsink and had no problem at all (at the end the mosfet was really cool).
Tomorrow I will post some picture of the circuit (you know, when you need it, your digital camera batteries are always discharged...).
Now I'm building the SE circuit I posted before, with two KT88 paralleled and para-feed outout.
But,
What is "screen drive"? can you drive a tube by its screen? How? Benefits?
Ciao,
Giovanni
Glad to see that your CCS works. I may try to copy it later and see how well it works. I am also going to try substituting the tube with a mosfet in the circuit that I am currently using. I see no reason why it shouldn't work. I like to use the Toshiba 2SK2700. It has an insulated case, and very low gate capacitance, but only 40 watts dissipation rating.
Screen drive is something that was derived by the audio community. Most tubes were never designed to be driven from the screen grid. Some tubes are linear in this respect, and some are not. The best candidates seem to be sweep tubes, since they have good screen grid sensitivity.
The advantage is the fact that the driving voltage is centered around a relatively high voltage potential, allowing direct coupling. This can eliminate the coupling capacitor. I will try using a mosfet follower with the gate tied to the plate of the driver, and the source tied to the screen. G1 is tied to the cathode, or a fixed bias source.
The disadvantage is that screen drive curves look like pentode curves, with the associated pentode characteristics, including a higher output impedance than a triode connection.
it might be possible to implement a "dual drive" circuit to get the best of both worlds.
Screen drive is something that was derived by the audio community. Most tubes were never designed to be driven from the screen grid. Some tubes are linear in this respect, and some are not. The best candidates seem to be sweep tubes, since they have good screen grid sensitivity.
The advantage is the fact that the driving voltage is centered around a relatively high voltage potential, allowing direct coupling. This can eliminate the coupling capacitor. I will try using a mosfet follower with the gate tied to the plate of the driver, and the source tied to the screen. G1 is tied to the cathode, or a fixed bias source.
The disadvantage is that screen drive curves look like pentode curves, with the associated pentode characteristics, including a higher output impedance than a triode connection.
it might be possible to implement a "dual drive" circuit to get the best of both worlds.
Sounds good...
OK, I will try to do something more classical, with a simple g1 driver...
I'm not so skilled to implement screen driver, I'm afraid.
About 2SK700: mosfets have a safe operating area too, like the bipolar transistors, and I'm afraid you cannot use such a mosfet with high currents (let's say not more than 100 mA).
If you want to use higher currents (like me) you should use something like IRFPG40, that has a bigger safe area, at the expense of a higher input capacitance.
But I think my circuit should have no problem driving it: I drive the mosfet with a maximum current of about 20 mA, more enough to drive it well above the audio band.
I can rise the current, however, up to 60-100 mA (changing the value of R1 and R5 rewsistors).
OK, I will try to do something more classical, with a simple g1 driver...
I'm not so skilled to implement screen driver, I'm afraid.
About 2SK700: mosfets have a safe operating area too, like the bipolar transistors, and I'm afraid you cannot use such a mosfet with high currents (let's say not more than 100 mA).
If you want to use higher currents (like me) you should use something like IRFPG40, that has a bigger safe area, at the expense of a higher input capacitance.
But I think my circuit should have no problem driving it: I drive the mosfet with a maximum current of about 20 mA, more enough to drive it well above the audio band.
I can rise the current, however, up to 60-100 mA (changing the value of R1 and R5 rewsistors).
I have been using these mosfets as drivers, in follower configuration, to drive the grid of large power tubes (845, 833A) directly. Here the average current is low, and the capacitance is an issue. I will look through my parts collection to see what bigger ones that I have. I tested a lot of different fets when developing the PowerDrive circuit before I figured out that bigger is not always better.
Has anyone tried IGBT's? I got some but haven't tried them yet.
Has anyone tried IGBT's? I got some but haven't tried them yet.
Konnichiwa,
Yeah, a good while ago, when they where still hot, cute and virginal. Not only did they have the bipolar 2ndary breakdown, they also had the "mosfet fog" input capacitance. But I'm sure in the last 20 years "they" have come up with something better, I just wish "they" would tell me....
Sayonara
tubelab.com said:
Yeah, a good while ago, when they where still hot, cute and virginal. Not only did they have the bipolar 2ndary breakdown, they also had the "mosfet fog" input capacitance. But I'm sure in the last 20 years "they" have come up with something better, I just wish "they" would tell me....
Sayonara
What do you mean, Kuei Yang Wang?
In my post I mean I do not know well tubes...
Well, bad news from my side:
I tried the circuit I drawn, this and, of course, the whole thing oscillates.
I connected a digital ammeter in series with the CCS, and a digital voltmeter between tubes anode and ground: both the meters got crazy, both show me out-of-range measurements (ammeter set to 2A, voltmeter to 600V).
I connected an old speaker to the output, I can ear a loud main hum.
So, time to make some serious tests.
First of all, can I substitute the CCS with a fixed resistor (let's say 2000 Ohm), connected between the B+ and anode, of course eliminating the load? In this way the amp will see a fixed load and, thus, I can see if the amp oscillate itself.
Ciao,
Giovanni
In my post I mean I do not know well tubes...
Well, bad news from my side:
I tried the circuit I drawn, this and, of course, the whole thing oscillates.
I connected a digital ammeter in series with the CCS, and a digital voltmeter between tubes anode and ground: both the meters got crazy, both show me out-of-range measurements (ammeter set to 2A, voltmeter to 600V).
I connected an old speaker to the output, I can ear a loud main hum.
So, time to make some serious tests.
First of all, can I substitute the CCS with a fixed resistor (let's say 2000 Ohm), connected between the B+ and anode, of course eliminating the load? In this way the amp will see a fixed load and, thus, I can see if the amp oscillate itself.
Ciao,
Giovanni
Konnichiwa,
An allusion to the fact that screendrive mode makes a pentode into an "enhancement" device, in other words, a lot like a transistor....
Try gridstoppers (1K) on the Valve Grids.
Also, you may wish to bypass the cathode resistors with capacitors of suitable value.
And yes, you can use a big old power resistor as anode load for testing...
Sayonara
croccodillo said:
An allusion to the fact that screendrive mode makes a pentode into an "enhancement" device, in other words, a lot like a transistor....
croccodillo said:
Try gridstoppers (1K) on the Valve Grids.
Also, you may wish to bypass the cathode resistors with capacitors of suitable value.
And yes, you can use a big old power resistor as anode load for testing...
Sayonara
Thanks for your help, Kuei Yang Wang
I will try tonight to place two grid stoppers.
A question: could the oscillation be caused by the separate bias of the two paralleled tubes?
I mean, having two searate autobias could it be that one of the tubes begin conduct more, than the other one and so on oscillating?
I know I was not clear, but it is too complicated for me to explain in English...
Could it be better to have a single negative bias voltage, and reference both the grids to it?
I mean, something like the solution used in
this amplifier (from Andrea Ciuffoli website).
Thanks again,
Giovanni
I will try tonight to place two grid stoppers.
A question: could the oscillation be caused by the separate bias of the two paralleled tubes?
I mean, having two searate autobias could it be that one of the tubes begin conduct more, than the other one and so on oscillating?
I know I was not clear, but it is too complicated for me to explain in English...
Could it be better to have a single negative bias voltage, and reference both the grids to it?
I mean, something like the solution used in
this amplifier (from Andrea Ciuffoli website).
Thanks again,
Giovanni
IT WORK!!!!
And it works wonderfully!
Yesterday night I modified my amplifier adding grid stopper resistors and cathode resistor by-pass capacitors: after that the amp worked fine.
Can you imagine my happyness when I eard Eurythmics's "Sweet Dreams" (one of my most favorite songs) from my speaker, clean and detailed in an awesome way?
I was so happy I screamed...
Remember this is MY FIRST TUBE AMP!!!!
I'm really, really happy.
Well, THIS is the updated schematic, and THIS is a gallery of my working amp.
Please note the big heatsink for the CCS: the first pictures show how the amp will look at the end, the tue in the front and, back, the heatsink becoming part of the chassis.
By the way, the mosfet is insulated from the heatsink, and remains cold after almost half an hour.
Tubes are now working at an anode voltage (referred to ground) of about 336V, the grid voltage is -28V, the CCS current is equally divided in the two tubes (75mA each).
Unfortunately nor my portable PC nor my CD player have enough output signal to drive the amp to full power, so I can't try it completely.
Another issue is my PSU: B+ voltage is now 400V, too low for such an amp. I need at least 550V, in manner to set the tube working point around 400V and have enough swing (remember my CCS can have a voltage across it as low as 4V) to reach the full available output power.
And, what about driver stage? Can a SRPP stage, made with two 12AX7 tube for each channel, make the job?
Or, what about a 12AX7 feed with a CCS, made perhaps with another 12AX7?
Again: I have no audio/spectrum analyzer, but I've seen there's a lot of softwares using PC audio cards; are you using one of those software, which one is the best?
Ciao,
Giovanni
And it works wonderfully!
Yesterday night I modified my amplifier adding grid stopper resistors and cathode resistor by-pass capacitors: after that the amp worked fine.
Can you imagine my happyness when I eard Eurythmics's "Sweet Dreams" (one of my most favorite songs) from my speaker, clean and detailed in an awesome way?
I was so happy I screamed...
Remember this is MY FIRST TUBE AMP!!!!
I'm really, really happy.
Well, THIS is the updated schematic, and THIS is a gallery of my working amp.
Please note the big heatsink for the CCS: the first pictures show how the amp will look at the end, the tue in the front and, back, the heatsink becoming part of the chassis.
By the way, the mosfet is insulated from the heatsink, and remains cold after almost half an hour.
Tubes are now working at an anode voltage (referred to ground) of about 336V, the grid voltage is -28V, the CCS current is equally divided in the two tubes (75mA each).
Unfortunately nor my portable PC nor my CD player have enough output signal to drive the amp to full power, so I can't try it completely.
Another issue is my PSU: B+ voltage is now 400V, too low for such an amp. I need at least 550V, in manner to set the tube working point around 400V and have enough swing (remember my CCS can have a voltage across it as low as 4V) to reach the full available output power.
And, what about driver stage? Can a SRPP stage, made with two 12AX7 tube for each channel, make the job?
Or, what about a 12AX7 feed with a CCS, made perhaps with another 12AX7?
Again: I have no audio/spectrum analyzer, but I've seen there's a lot of softwares using PC audio cards; are you using one of those software, which one is the best?
Ciao,
Giovanni
The 12AX7, even in SRPP will have a hard time driving the grids of 2 KT-88's. The input capacitance of the KT-88's is pretty high. A 12AU7, 12BH7, 6CG7, 6SN7 or similar tube may be a better choice for driving this capacitance.
Since you are not afraid of silicon, consider a CCS load on a triode, followed by a mosfet buffer to provide drive to the grid, and a high impedance load for the driver tube. In this situation a 12AX7 should work OK.
Look at my PowerDrive circuit for ideas:
http://www.tubelab.com/powerdrive.htm
The voltages and component values will be different for your particular design.
Since you are not afraid of silicon, consider a CCS load on a triode, followed by a mosfet buffer to provide drive to the grid, and a high impedance load for the driver tube. In this situation a 12AX7 should work OK.
Look at my PowerDrive circuit for ideas:
http://www.tubelab.com/powerdrive.htm
The voltages and component values will be different for your particular design.
Well, I studied your PowerDrive now...
Quite impressive, my compliments!!!!
I think I can do it, using for example a -100V bias voltage, and then using the mosfet to drive the grid of the 2 KT88s from 0 to -80V...
At the end I have a lot less grid voltage swing than a 845, I suppose.
So, imagine: the source pin of Mosfet connected to the -100V through a resistor (say 3K3 Ohm), the drain connected to my +550V, and the gate biased with a resistive network like yours (gate voltage adjustable from -20 to -60V).
This solution is great, also because you can change "on the fly" the operating point of the tubes simply changing the voltage at mosfet gate... isn't it?
My only doubt is (already reported yesterday): I have two KT88 tubes connected in parallel, biasing them in this way means they have to be matched (and mine are).
Now, as told before, the two tubes are sinking current exaclty in the same amount, making the output stage perfectly balanced (I hope it is clear what I mean): will I have problems, during time, with tube bias drift and operating curve modification due the age?
Using self bias tubes "self-compensate", in this way do not.
Of course, if tubes will age in the same way I will have no problem, I will have only to correct bias voltage sometime; but I'm afraid that if one of the two tubes "drift" with age and the other not, the behaviour is not anymore the same, and so I do not know how the amp will perform.
But anyway your solution is really good, my instinct says me it is a wonderful idea; and I trust my instint.
I think I will try it!!!
Sorry for my bad explanation, but my english is limited.
Ciao,
Giovanni
Quite impressive, my compliments!!!!
I think I can do it, using for example a -100V bias voltage, and then using the mosfet to drive the grid of the 2 KT88s from 0 to -80V...
At the end I have a lot less grid voltage swing than a 845, I suppose.
So, imagine: the source pin of Mosfet connected to the -100V through a resistor (say 3K3 Ohm), the drain connected to my +550V, and the gate biased with a resistive network like yours (gate voltage adjustable from -20 to -60V).
This solution is great, also because you can change "on the fly" the operating point of the tubes simply changing the voltage at mosfet gate... isn't it?
My only doubt is (already reported yesterday): I have two KT88 tubes connected in parallel, biasing them in this way means they have to be matched (and mine are).
Now, as told before, the two tubes are sinking current exaclty in the same amount, making the output stage perfectly balanced (I hope it is clear what I mean): will I have problems, during time, with tube bias drift and operating curve modification due the age?
Using self bias tubes "self-compensate", in this way do not.
Of course, if tubes will age in the same way I will have no problem, I will have only to correct bias voltage sometime; but I'm afraid that if one of the two tubes "drift" with age and the other not, the behaviour is not anymore the same, and so I do not know how the amp will perform.
But anyway your solution is really good, my instinct says me it is a wonderful idea; and I trust my instint.
I think I will try it!!!
Sorry for my bad explanation, but my english is limited.
Ciao,
Giovanni
Question:
Since I should never drive my KT88 grids over the ground voltage (curves are reported from 0 to -80V), I can connect the mosfet drain to a relatively low voltage, let's say 50V.
Is this correct or am I wrong?
In this way I still have enough swing at its drain to drive the output tubes (go down from 0V to -80V), and I can use a mosfet with a much lower breakdown voltage (200V).
I know several IRF parts that have exceptional linear characteristics (already used in the past for SS amplifier,like Nelson Pass's ones) with such a low voltage.
Ciao,
Giovanni
Since I should never drive my KT88 grids over the ground voltage (curves are reported from 0 to -80V), I can connect the mosfet drain to a relatively low voltage, let's say 50V.
Is this correct or am I wrong?
In this way I still have enough swing at its drain to drive the output tubes (go down from 0V to -80V), and I can use a mosfet with a much lower breakdown voltage (200V).
I know several IRF parts that have exceptional linear characteristics (already used in the past for SS amplifier,like Nelson Pass's ones) with such a low voltage.
Ciao,
Giovanni
Yes the drain of the mosfet can be connected to a lower voltage. I use +150 volts in my latest amps to lower the power dissipated in the fets. I need this much voltage since the grid of an 845 will swing to +50 volts on peaks, and higher on transients.
two output tubes? Just add a second fet follower (two coupling capacitors) with its own bias adjustment for the second tube. I tried this with 2 845's and it works good. I tried it with 3 845's and it works, but I decided to use an 833A for this much power. It is easier to drive.
The two tubes may be matched now but they won't be in a few years. This is not as much of a problem in a SE amp as it is in a push pull amp.
I have not tried using an EL-84 as a driver, but it might work.
I wouldn't wory about your english, since I have no problem understanding your explanations. Perhaps engineers think in the same language?
two output tubes? Just add a second fet follower (two coupling capacitors) with its own bias adjustment for the second tube. I tried this with 2 845's and it works good. I tried it with 3 845's and it works, but I decided to use an 833A for this much power. It is easier to drive.
The two tubes may be matched now but they won't be in a few years. This is not as much of a problem in a SE amp as it is in a push pull amp.
I have not tried using an EL-84 as a driver, but it might work.
I wouldn't wory about your english, since I have no problem understanding your explanations. Perhaps engineers think in the same language?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.