I'm building a class-D amp using the Purifi 1ET400A module and would like to add a little tube magic. I need to create a input buffer with a little bit of gain.
The requirements are differential input and output with about 10db of gain. I'd like to have an input impedance of 30K to 50K per leg, and the Purifi module has an input impedance of 2.2K per leg.
I've never designed with tubes before, and my circuit design skills in general are pretty rusty to say the least. It's been about 40 years since I did any serious analog design, so I could use a lot of help.
I'd like to be able to use the power supplies I already have to support the Purifi module and other circuits. These include +/- 65V unregulated, but heavily cap filtered from a linear supply (lots of current available), +/- 12V regulated with very low noise and ripple (~500mA available), +12V regulated to drive ancillary circuits (~300mA available), +15V regulated referenced to the -65V rail (~300mA available), and +5V regulated for driving digital logic (~50mA available). I could use a linear regulator on the buffer board tapping one of these to create another voltage if necessary.
Ideally, I'd like to avoid coupling caps, but I'm not sure this is practical (or possible). I was thinking of using a differential triode stage feeding an op-amp driver (like an OPA1632), but as I said I'm a total tube neophyte so not sure if this is the right approach, particularly if my goal is to add a little tube magic.
Any help or pointers would be much appreciated.
The requirements are differential input and output with about 10db of gain. I'd like to have an input impedance of 30K to 50K per leg, and the Purifi module has an input impedance of 2.2K per leg.
I've never designed with tubes before, and my circuit design skills in general are pretty rusty to say the least. It's been about 40 years since I did any serious analog design, so I could use a lot of help.
I'd like to be able to use the power supplies I already have to support the Purifi module and other circuits. These include +/- 65V unregulated, but heavily cap filtered from a linear supply (lots of current available), +/- 12V regulated with very low noise and ripple (~500mA available), +12V regulated to drive ancillary circuits (~300mA available), +15V regulated referenced to the -65V rail (~300mA available), and +5V regulated for driving digital logic (~50mA available). I could use a linear regulator on the buffer board tapping one of these to create another voltage if necessary.
Ideally, I'd like to avoid coupling caps, but I'm not sure this is practical (or possible). I was thinking of using a differential triode stage feeding an op-amp driver (like an OPA1632), but as I said I'm a total tube neophyte so not sure if this is the right approach, particularly if my goal is to add a little tube magic.
Any help or pointers would be much appreciated.
That's quite a low impedance to drive 2k2. Do you know how much voltage swing? So I guess you should look at some of the headphone type amps on diyaudio, if you want all valve. You could add some NFB to improve THD.
Last edited:
@batteryman - Thanks for the pointer. It looks like there is a lot of good information on that site.
@baudouin0 - It doesn't need to be all tube. I think it will be a lot easier to drive the 2.2K input impedance with solid state of some kind.
These are monoblocks, so I only need a single channel buffer, but it needs to be differential input and output. Of course, it can be single-ended in between the input and output stage if need be, so one option would be to use a differential receiver (e.g. OPA1611) followed by a tube gain stage, and then a SS discrete or op-amp driver output.
Most of the tube circuits I've seen use voltages in the 300V range. Is it necessary to use such high voltages? Would using the +/-65V supplies for the tube stage be adequate? Are there tubes that work better at lower voltages?
Thanks for the replies.
@baudouin0 - It doesn't need to be all tube. I think it will be a lot easier to drive the 2.2K input impedance with solid state of some kind.
These are monoblocks, so I only need a single channel buffer, but it needs to be differential input and output. Of course, it can be single-ended in between the input and output stage if need be, so one option would be to use a differential receiver (e.g. OPA1611) followed by a tube gain stage, and then a SS discrete or op-amp driver output.
Most of the tube circuits I've seen use voltages in the 300V range. Is it necessary to use such high voltages? Would using the +/-65V supplies for the tube stage be adequate? Are there tubes that work better at lower voltages?
Thanks for the replies.
@baudouin0 - To answer your question about voltage swing, the Purifi module will clip at about 6.5V pk-to-pk on each of the differential inputs (13v across), so that would be the maximum I'd want to drive. I'd like to have about 10db gain from the buffer, so figure about 2v pk-to-pk on each of the differential inputs.
You could consider ECC84/PCC84 or ECC88/PCC88. They are happy with lower anode voltages (the maximum anode voltage of the ECC88/PCC88 is only 130 V).
Nuvistors, such as 7586, are also good candidates for a +65V supply. They were often specified at anode voltages as low as 40V, but they can also work at higher values.
@PCL200 - Then 6DJ8, which I think is the same as the ECC88, is the tube I was thinking about using. But I've seen designs with >200v power supplies with this tube. The data sheet I found shows plate voltage transfer functions at 90v, 150v and 200v. Do you think this will still work well at lower voltages?
@pcan - the nuvistors look kind of interesting. It doesn't look like these can be socketed though, can they? The description I read mentioned really fragile leads.
@pcan - the nuvistors look kind of interesting. It doesn't look like these can be socketed though, can they? The description I read mentioned really fragile leads.
@ggetzoff - thanks for the pointer. I'll look into that some more. Looks like it's using a voltage multiplier on the board which is something I thought about. Interesting. It's cheap enough to pick one up just to play around with.
The OP is looking for a small amount of gain, say α approx. = 3 and "balanced". A differential gain block made from 12B4s will do that quite nicely.
Even with its remarkably low plate resistance (RP), the 12B4 will not do well into the 2.2 Kohm I/P impedance "chip". Solve that problem by DC coupling IRFBC20 source followers to the 12B4 plates. A protective 15 V. Zener diode between the gate and source electrodes of each FET is in order.
Even with its remarkably low plate resistance (RP), the 12B4 will not do well into the 2.2 Kohm I/P impedance "chip". Solve that problem by DC coupling IRFBC20 source followers to the 12B4 plates. A protective 15 V. Zener diode between the gate and source electrodes of each FET is in order.
I was just playing around with an 'anode follower' (aka 'plate follower' in the US) using a 6DJ8 (same as ECC88). That uses local plate-grid negative feedback to reduce the gain and distortion. You can set the gain using a voltage divider (2 resistors).
Follow that with a cathode follower to drive your 2.2k ohm load, which practically requires a cathode follower, a MOSFET source follower, or an opamp.
Something like the attached audio circuit should work very well for driving a Class D amp. The way I have it pictured yields 2x (6dB) voltage gain. You could adjust R2 and R3 to get to the gain you want. For instance, change R2 to something like 330k to get 3x (10dB) gain. The output impedance and distortion will stay low thanks to the cathode follower on the output.
The volume control in front is optional.
In something like this circuit where the cathode is 100V above ground, you'll want to lift the heater supply so that it's about 50V above ground, so that both cathodes are within the cathode-heater voltage limits.
Just an idea.
Follow that with a cathode follower to drive your 2.2k ohm load, which practically requires a cathode follower, a MOSFET source follower, or an opamp.
Something like the attached audio circuit should work very well for driving a Class D amp. The way I have it pictured yields 2x (6dB) voltage gain. You could adjust R2 and R3 to get to the gain you want. For instance, change R2 to something like 330k to get 3x (10dB) gain. The output impedance and distortion will stay low thanks to the cathode follower on the output.
The volume control in front is optional.
In something like this circuit where the cathode is 100V above ground, you'll want to lift the heater supply so that it's about 50V above ground, so that both cathodes are within the cathode-heater voltage limits.
Just an idea.
Attachments
Eli - that looks like an interesting tube and your proposed approach is pretty much what I was thinking.
I ran across a schematic for the Borbely Tube-Mosfet OpAmp which is along the same lines.
Thanks for your suggestion.
I ran across a schematic for the Borbely Tube-Mosfet OpAmp which is along the same lines.
Thanks for your suggestion.
Oooohhhh....
BALANCED!
Oops. Sorry, I missed that.
How about an anode follower with MOSFET 'source-o-dyne' output? Very low distortion this way...
--
Attachments
Last edited:
Differential input too?
Then Eli's idea of a long-tailed pair with some kind of solid-state output buffers likely is the way to go.
The problem will still be how to get only 3x gain.
I wonder if plate-grid feedback can be applied to an LTP...
--
Then Eli's idea of a long-tailed pair with some kind of solid-state output buffers likely is the way to go.
The problem will still be how to get only 3x gain.
I wonder if plate-grid feedback can be applied to an LTP...
--
How about this?
Balanced input 6DJ8 LTP DC coupled to IRF820 MOSFET source followers, with negative feedback from MOSFET source outputs to tube grid inputs. Gain calculates out to about 2.5x. This looks like it should work really well, even into 2.2k ohm loads.
(Sorry, I started to have some fun with the idea...)
Balanced input 6DJ8 LTP DC coupled to IRF820 MOSFET source followers, with negative feedback from MOSFET source outputs to tube grid inputs. Gain calculates out to about 2.5x. This looks like it should work really well, even into 2.2k ohm loads.
(Sorry, I started to have some fun with the idea...)
