First of all, please have patience with my LazyCAD schematics. At the age of 44 I'm way too old to learn how to use new softwares...
Over the years I have developed a nasty habit of pushing some of my tubes into class A(B) territory by driving the grids positive. I currently have two pairs of monoblocks with triode wired 807s (SE and PP) that operate in peak A2 by using direct-coupled, high gm cathode followers between the gain stages and the output tubes. My most recent build, an 808 SET, works the other way around with the grids idling at +28V or so and only goes negative (A1) at close to max power. This amp used pentode wired 807s as CF drivers.
My next project will be a smaller SET using Taylor T20 and TZ20 as output tubes. These are smallish thoriated tungsten triodes capable of 20W plate dissipation and a mu of 20 and 66, respectively. I intend to operate them at ca. 400V 50mA with 10k plate loads and positive bias, around +12V for the T20 and +20V for its higher mu brother.
This calls for a driver stage with the following specifications:
* Adjustable to allow both types of output tubes to be used
* Capable of swinging +-50V or so into a non-linear load. Max grid current will probably be somewhere between 25 and 50mA at the peaks, hard to tell exactly from the old datasheet. The biggest problem appears to be the sudden change in load when the grid goes negative and ceases to draw current.
* Slow/delayed startup. My output tubes where hard to find and not particulary cheap so I want to be as nice to them as possible. Assuring that they are biased into cutoff until the filaments are hot and the B+ is up is one way to do this. Imagine where all the electrons will go if the filaments are hot and the grid voltage goes positive before the plate voltage...😱
* Good looks. Perhaps a bit childish but I've invested some money in a few pairs of these tubes and a nice set of shrouded transformers/chokes and it would be nice to keep the vintage feel intact by avoiding "modern"-looking tubes. Compromises could be made in this department though...
So, the easy way to do this would be to do it as I've done it several times before: A choke loaded cathode follower direct coupled to the output tubes grid. The small problem here is that the tubes that have the righ vintage look (octals or older) usually has too low transconductance to make good cathode followers.
"Modern" novals such as 6E5P, PL802 etc would be much better for the job but they simply don't look right next to the output tubes.
Mosfets would be even better (and would of course be hidden inside the chassis) but they lack the delayed startup that I would get from an indirectly heated tube. This could probably be solved by slow-ramping power supplies and/or a timer circuit that holds everything in place until all the voltages are where they should be.
Or, we could aim for the best of both worlds by going hybrid:
Drawing 1 shows a hybrid cascode follower where the drain voltage to the Fet is supplied by a trioded EL36. This would delay the drain voltage during startup, reduce it from over 200V to perhaps 30V and keep it much more constant during (which should reduce the impacts of the fets internal capacitances). Should do a decent job if it works at all.
Drawing 2 shows a sort of Super Triode/cascaded follower where a smaller triode CF drives the mosfet. Here the mosfet will have to handle more voltage/heat and large VDS variations but it should on the other hand be completely protected from voltage surges etc. on the input during startup. Bonus idea: Assuming a triode that can handle 400V at idle plus some swing, we could connect its plate to a tap on the output transformer and get a little bit of local feedback for free.
Any thoughts on this? None of these circuits has been tested (or even simulated) by me.
I've browsed through a lot of schematics where mosfet followers are used to drive control- and/or screen grids and there is rarely anything mentions of what happens during startup, perhaps I'm just creating problems in my head?
Over the years I have developed a nasty habit of pushing some of my tubes into class A(B) territory by driving the grids positive. I currently have two pairs of monoblocks with triode wired 807s (SE and PP) that operate in peak A2 by using direct-coupled, high gm cathode followers between the gain stages and the output tubes. My most recent build, an 808 SET, works the other way around with the grids idling at +28V or so and only goes negative (A1) at close to max power. This amp used pentode wired 807s as CF drivers.
My next project will be a smaller SET using Taylor T20 and TZ20 as output tubes. These are smallish thoriated tungsten triodes capable of 20W plate dissipation and a mu of 20 and 66, respectively. I intend to operate them at ca. 400V 50mA with 10k plate loads and positive bias, around +12V for the T20 and +20V for its higher mu brother.
This calls for a driver stage with the following specifications:
* Adjustable to allow both types of output tubes to be used
* Capable of swinging +-50V or so into a non-linear load. Max grid current will probably be somewhere between 25 and 50mA at the peaks, hard to tell exactly from the old datasheet. The biggest problem appears to be the sudden change in load when the grid goes negative and ceases to draw current.
* Slow/delayed startup. My output tubes where hard to find and not particulary cheap so I want to be as nice to them as possible. Assuring that they are biased into cutoff until the filaments are hot and the B+ is up is one way to do this. Imagine where all the electrons will go if the filaments are hot and the grid voltage goes positive before the plate voltage...😱
* Good looks. Perhaps a bit childish but I've invested some money in a few pairs of these tubes and a nice set of shrouded transformers/chokes and it would be nice to keep the vintage feel intact by avoiding "modern"-looking tubes. Compromises could be made in this department though...
So, the easy way to do this would be to do it as I've done it several times before: A choke loaded cathode follower direct coupled to the output tubes grid. The small problem here is that the tubes that have the righ vintage look (octals or older) usually has too low transconductance to make good cathode followers.
"Modern" novals such as 6E5P, PL802 etc would be much better for the job but they simply don't look right next to the output tubes.
Mosfets would be even better (and would of course be hidden inside the chassis) but they lack the delayed startup that I would get from an indirectly heated tube. This could probably be solved by slow-ramping power supplies and/or a timer circuit that holds everything in place until all the voltages are where they should be.
Or, we could aim for the best of both worlds by going hybrid:
Drawing 1 shows a hybrid cascode follower where the drain voltage to the Fet is supplied by a trioded EL36. This would delay the drain voltage during startup, reduce it from over 200V to perhaps 30V and keep it much more constant during (which should reduce the impacts of the fets internal capacitances). Should do a decent job if it works at all.
Drawing 2 shows a sort of Super Triode/cascaded follower where a smaller triode CF drives the mosfet. Here the mosfet will have to handle more voltage/heat and large VDS variations but it should on the other hand be completely protected from voltage surges etc. on the input during startup. Bonus idea: Assuming a triode that can handle 400V at idle plus some swing, we could connect its plate to a tap on the output transformer and get a little bit of local feedback for free.
Any thoughts on this? None of these circuits has been tested (or even simulated) by me.
I've browsed through a lot of schematics where mosfet followers are used to drive control- and/or screen grids and there is rarely anything mentions of what happens during startup, perhaps I'm just creating problems in my head?
