This is a hypothetical question. I don’t see why a 12ax7 wouldn’t drive a DHT like a 45. The Rp worst case of the 12AX7 is say 80kohms. The miller capacitance of a 45 grid is roughly 28.5. This gives a -3dB 70khz, no issue here unless you are a bat?
Now as for as output impedance/input impedance voltage divider (loss) we have 80kohms/1meg ohm (max 45 grid resistance) lost so say 10% reduction of the signal (this is simply ohms law.)
I see people throwing a mofset in between a driver and a DHT output to give more “drive.”
I keep hearing a 12AX7 or similar low gm tube can’t “drive” a DHT output, not enough “current”. I honestly don’t understand these terms, what physics is involved here where are the “equations” that one uses to determine if a driver tube has enough current to drive a DHT output? Where is this coming from on the datasheets and how can I calculate?
Now as for as output impedance/input impedance voltage divider (loss) we have 80kohms/1meg ohm (max 45 grid resistance) lost so say 10% reduction of the signal (this is simply ohms law.)
I see people throwing a mofset in between a driver and a DHT output to give more “drive.”
I keep hearing a 12AX7 or similar low gm tube can’t “drive” a DHT output, not enough “current”. I honestly don’t understand these terms, what physics is involved here where are the “equations” that one uses to determine if a driver tube has enough current to drive a DHT output? Where is this coming from on the datasheets and how can I calculate?
You need current to drive the capacitance. The 12AX7 typically runs at about 1mA, so this may be barely enough to change the voltage quickly enough. Google slew rate limiting. Remember that you can only use a fraction of the current, as otherwise you get nonlinear distortion - assume maybe 10-20% of quiescent current is actually usable?
But I just showed the -3dB point from the miller capacitance is 77khz, why is this a worry?
I just don't see slew rate being a factor with class A SE tubes, they amplify voltage, I mean its a vacuum, the grid is a near null load as long as you stay away from positive. I am talking about deep class A SET's we are all found off. But I am here to learn.
I just don't want to drive a DHT with a 417 cause everyone is doing it, I need to see the datasheet parameters entered into a viable physics calculation, otherwise it just seems to be a bit audiophoolish😱
For a rule of thumb, we will have to simplify a little.
Assume the driver sees 40pF, including grid wiring, Cgf + Millerised Cga.
Take a class-A 45SE fixed bias at -50V
Lets say that the slewing of the driver is up to snuff if it can swing the grid from cutoff (-100V) to clip (0V) in the period of a 50kHz waveform: 20us. This allows a margin, for sure, but no point in adopting a wonderful 45SE unless you design with a margin.
Current i (A) needed to swing capacitance C through Voltage V in time t is :
i = CV/t
so
i = (40 e-12 x 100)/20 e-6
= 200uA
Af DF says, robbing this current from the driver is like a frequency dependent distortion, so you need to keep it as a small fraction of the anode current as possible. For 1% at 50kHz you'd need a 20mA driver.
Some constructors like triode-connected 6V6s for a driver, where you can run 45mA... you can see why.
This problem can be bypassed through using a stage architecture with current output, eg shunt-cascode, where the current diversion will not affect the driver at all.... may be one reason why 300B driver made this way is so much better sounding than the usual 6SN7, etc.
Assume the driver sees 40pF, including grid wiring, Cgf + Millerised Cga.
Take a class-A 45SE fixed bias at -50V
Lets say that the slewing of the driver is up to snuff if it can swing the grid from cutoff (-100V) to clip (0V) in the period of a 50kHz waveform: 20us. This allows a margin, for sure, but no point in adopting a wonderful 45SE unless you design with a margin.
Current i (A) needed to swing capacitance C through Voltage V in time t is :
i = CV/t
so
i = (40 e-12 x 100)/20 e-6
= 200uA
Af DF says, robbing this current from the driver is like a frequency dependent distortion, so you need to keep it as a small fraction of the anode current as possible. For 1% at 50kHz you'd need a 20mA driver.
Some constructors like triode-connected 6V6s for a driver, where you can run 45mA... you can see why.
This problem can be bypassed through using a stage architecture with current output, eg shunt-cascode, where the current diversion will not affect the driver at all.... may be one reason why 300B driver made this way is so much better sounding than the usual 6SN7, etc.
Thankyou this is what I am looking for.
Lets say though that you aren't swinging that many volts, say a preamp or headamp. Say you are only swinging 10 volts into the grid of the 45, this would mean stealing less current from the driver correct, and one could use a #26 driver possibly? To me this is really damping factor that we are talking about, but the old rule of thumb 1:10 seems to be not good enough.
Damping factor relates to damping - reducing a resonance (e.g. in a loudspeaker). Here we are talking about current drive capability. This is a separate issue from frequency response. Rod has explained how the calculation works, to save you the trouble of googling about slew rate limiting. Slew rate is potentially a problem everywhere a capacitor needs to be driven, even a small capacitor. You are not stealing current from the driver, as providing this current is precisely what the driver is there to do.
Well, the peak current demand is proportional to the peak voltage swing, if you want to see how much current for lower swing. So half the swing needs half the current peak.
But be careful - music often has big peaks, and even modest volume levels can call for big swings. And if you have an MC cartridge, beware that riding over an LP scratch or defect can throw in huge overloads with up to 50kHz bandwidth, that may be exaggerated by a driver with inadequate slew capability.
You are right about the preamp, though, the demands are lower in this position. The 12AX7 may work well enough, depending on the signal level you specify. But I would not use low-current triodes in MC preamps, for the reason above.
The thing about it this philosophy that is counter intuitive is: the miller capacitance of everyone's favorite driver for DHT's, the WE417 is higher than a 45.
So essentially with a typicall 417a-45 SET we have an input tube that is harder to drive than the output tube! Now I understand that the input of the 5842 isn't swinging as many volts as the input to the 45, but it would seem the choice to use such an input tube would have repercusions to the preamp or source design criteria.
The other thing is a filterless NOS DAC would seem to be a horrible source for a DHT amp, I mean think of how much current is wasted driving ultrasonics into the output tube. Look at the input impedance of a 45 at 50khz assuming 50pf capacitance, hell of a load. Really makes one re-think their DAC analog stage.
So essentially with a typicall 417a-45 SET we have an input tube that is harder to drive than the output tube! Now I understand that the input of the 5842 isn't swinging as many volts as the input to the 45, but it would seem the choice to use such an input tube would have repercusions to the preamp or source design criteria.
The other thing is a filterless NOS DAC would seem to be a horrible source for a DHT amp, I mean think of how much current is wasted driving ultrasonics into the output tube. Look at the input impedance of a 45 at 50khz assuming 50pf capacitance, hell of a load. Really makes one re-think their DAC analog stage.
The question was more for discussion on the theory behind driver choices, I should have been more clear just looking at a common anode follower driver to DHT output to simplify the discussion.
I think Rod showed very well why the 12AX7 or similar tube isn't a great choice for a full power DHT output, where to draw the line is the issue I am getting at. Some applications you don't need full power, in other words I think a 12AX7(or #26)-45 preamp could be just fine if you only need say a little gain.
Regal, why don't you just build a prototype and test if the given theories are valid.
If I had a 45, I would definitely try this with 12AX7 as a driver.
If I had a 45, I would definitely try this with 12AX7 as a driver.
I still like the write-up done for the Bugle-45 amp. The pdf file has sidebar calculations for bypass, input capacitance, slew-rate, driver current capability, etc.
www.wavelengthaudio.com/bugle.pdf
My first clean-sheet 45 SET ended up using a 5814A DC-coupled cascade input amp / driver. The driver stage is biased at 2.8ma and easily drives the 45 grid at 120V pk-pk and flat within 1dB to 50KHz.
I've done some work with the 12AX7 as a driver for another (45 SET) design... I dumped it in favor of a 12AT7 with the 1st stage DC-coupled to the second as a cathode follower (biased around 2.5ma). Performance was much better.
Regards, KM
www.wavelengthaudio.com/bugle.pdf
My first clean-sheet 45 SET ended up using a 5814A DC-coupled cascade input amp / driver. The driver stage is biased at 2.8ma and easily drives the 45 grid at 120V pk-pk and flat within 1dB to 50KHz.
I've done some work with the 12AX7 as a driver for another (45 SET) design... I dumped it in favor of a 12AT7 with the 1st stage DC-coupled to the second as a cathode follower (biased around 2.5ma). Performance was much better.
Regards, KM
- Status
- Not open for further replies.