Just a couple of straightforward questions!
Say you have a typical valve amp, but the finals are driven via a MOSFET DC coupled to the output grids. The MOSFET drain is connected to a positive supply to allow positive peaks without blocking distortion (I'm not too bothered about AB2 operation in this scenario, there could be a fairly large grid stopper on the outputs)
The stage before is AC coupled and the bias voltage is fed to the gate via a 1Meg or so resistor. Nothing controversial here and this is known to work well.
My questions are as follows:
1. Has anyone ever run into Vgs drift with temperature having troublesome effects on bias stability?
2. Given the lack of concern for linearity into grid current, has anyone tried using a very light source load for the follower say 47K @ 1mA, as to avoid a heatsink on the MOSFET? The reason I ask is entirely for construction/practicality reasons. I realise this will potentially have a negative effect on slew rate - but as I intend to run pentodes it may be OK.
Say you have a typical valve amp, but the finals are driven via a MOSFET DC coupled to the output grids. The MOSFET drain is connected to a positive supply to allow positive peaks without blocking distortion (I'm not too bothered about AB2 operation in this scenario, there could be a fairly large grid stopper on the outputs)
The stage before is AC coupled and the bias voltage is fed to the gate via a 1Meg or so resistor. Nothing controversial here and this is known to work well.
My questions are as follows:
1. Has anyone ever run into Vgs drift with temperature having troublesome effects on bias stability?
2. Given the lack of concern for linearity into grid current, has anyone tried using a very light source load for the follower say 47K @ 1mA, as to avoid a heatsink on the MOSFET? The reason I ask is entirely for construction/practicality reasons. I realise this will potentially have a negative effect on slew rate - but as I intend to run pentodes it may be OK.
No issues Ive found with tempco/Vgs drift.
Instead of using a high impedance positive supply, you could drop down that voltage--say to a pos of 12-20V, but you'll still have your negative supply to consider for device dissipation.
You don't need a huge current through the MOSFET, 20-25mA is quite enough normally.
A small TO220 heatsink should be quite enough at this level....
Instead of using a high impedance positive supply, you could drop down that voltage--say to a pos of 12-20V, but you'll still have your negative supply to consider for device dissipation.
You don't need a huge current through the MOSFET, 20-25mA is quite enough normally.
A small TO220 heatsink should be quite enough at this level....
The main issue I have with a lowish positive supply - if using an amp with NFB with a driver stage with a lot of headroom - aren't you just "shifting" the problem up a few volts? - you're still going to block into the gate protection zeners around the MOSFET. If you give it a fairly hefty positive supply, it can swing positive as much as it wants, with the grid stopper limiting excessive current into the grids. If it's possible to get around that issue I'd happily use a lower positive rail - open to suggestions!
There's always that - but of course it adds more complexity - MOSFET followers seem more than linear enough into a resistive load. Vgs can still vary with temp with the CCS of course.
Last edited:
I use this type of driving (aka Powerdrive) in my amps.
In my 300B SE (LED biased D3a, CCS load -C-FET on TO220 heatsink-300B) the FET source load is an resistor. I tried CCS, the sound was better -especially in the lower registers-, but due to the limiting space I currently use resistor (Mills 12W).
The grid voltage is almost dependent of the FET temperature.
In my 300B SE (LED biased D3a, CCS load -C-FET on TO220 heatsink-300B) the FET source load is an resistor. I tried CCS, the sound was better -especially in the lower registers-, but due to the limiting space I currently use resistor (Mills 12W).
The grid voltage is almost dependent of the FET temperature.
Then again, I also use CCS tail to set the output section current, so if grids change voltage, the tail compensates and tube operation is still ok.
Of course the grids could change voltage in a dissimilar way, but that never happens according to my observations if the source follower FETs are same type same batch same kind of heat sink.
Most tubes aren't really that sensitive. +-10% ok. Balanced topologies of course need to make sure there's actual balance.
Of course the grids could change voltage in a dissimilar way, but that never happens according to my observations if the source follower FETs are same type same batch same kind of heat sink.
Most tubes aren't really that sensitive. +-10% ok. Balanced topologies of course need to make sure there's actual balance.
You can also use the driver tube in the source of the mosfet, in other words a mu-follower (sometimes called gyrator) load on the driver tube. Works great on my amps, and easily supplies up to 200mA of grid current to the output tube when called upon. See schematic linked below.
http://www.diyaudio.com/forums/tubes-valves/232484-midlife-crisis-my-833c-amp-build.html
http://www.diyaudio.com/forums/tubes-valves/232484-midlife-crisis-my-833c-amp-build.html
Last edited:
Gyrator driving the following grid is good, but for best results there should be nothing between the source and the grid, just a wire. That is hard to accomodate with regards to voltages if you use the preceding stage plate load to drive the grid directly.
200mA grid current? I want to see your output tubes! =D
200mA grid current? I want to see your output tubes! =D
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.