I'm not sure you got what I said right.
The idea was to use a beefyer version of the output tube bias supply (before regulation) as a negative bias for the CCS negative rail. In truth, a few V negative would be enough for the first stage, and the second can easily work from ground. If you are using DC heating, getting a negative supply good for a few tens of mA should be easy.
Regarding the schematic, you seem to be doing something odd with the first stage. Given your values in the diagram, both the front end stages have a local supply of about 330V. The first stage has a plate DC operating point at 75V respective to ground. This stays the same regardless of what DC voltage you set on the grids of the 6SL7 as long as it is within the range of the CCS.
In using this configuration, you effectively need two coupling caps - one is the input cap, the other is the unmarked cap in the gate DC point filter. Hold this thought as I will get to it later on.
The plate DC point of the next stage is at 225V or thereabouts (I'm doing the math in my head so I may be off a few V). Since stage 1 and 2 are DC coupled, the grids are at 75V DC. For a tail current of ~~10mA, you have 5mA per half of 6SN7, shich puts the cathodes at about 70.5V (this is from the 6SN7 curves). In other words, your second stage CCS has a good 270V across it with no good reason - 70V is MORE than enough on the CCS, producing some 0.7V of dissipation. I am of course assuming the balance pots are of small value and do not drop signifficant voltage, as they shouldn't if you want the LTP to be a good differential amp.
Now, the big question is, what do you gain with this arrangement? If the first stage had a CCS referenced to ground or some small negative voltage, you could lose the input coupling cap and the input DC point network and it's associated coupling cap, which also passes audio current, so is really another coupling cap. Of course, you need to use coupling caps between stages then. The net plus is that these canbe smaller and easily be a foil type cap to you liking. In your current arrangement, you may need a large filter cap for the input DC network, even an electrolyitic, which is always much worse than even the most common foil cap.
You do, however, have two alternatives, based on the fact that the total gain of your two front stages is well over 500:
1) If you reference the first stage grids to zero, you still have 75V DC at the plates, the 6SL7 is still surprisingly quite linear under these conditions, and offers a gain of about 50 and 50Vpp output with reasonable distortion (well over what you could ever use as an input to the next stage, in other words). The next stage stays biassed exactly the same regardless. The positive side of this is, no input or DC bias network coupling caps at all. Given the very high total gain, you could also optimize the plate resistors of the first two stages for a better operating point for the 6SL7 (higher plate voltage) but slightly less gain. You could also add local feedback from plate to plate of the first and second stage (J. Broskie has an article at tubecad.com on this).
2) Given the high gain, and obvious desire for full DC coupling (which is to be commended), you may want to look into screen grid drive for the 807. Mr. Danielak has a screen grid drive SE amp using an ECL82 for the front end (triode part is broadly similar to the 6SL7, pentode part is triode strapped as a follower because screen grid drive needs screen grid current) driving a PL519 sweep tube. I have not been able to find an example of screen grid drive for the 807 or any curves for this mode of operation, but there you have it, a challenge 😉 . Doubling up something like his circuit and connecting the first stages as a LTP would roughly give you hwat you need. It may well be possible to use the same 6SL7 and 6SN7 (first as LTP, second as two followers), as long as the 6SN7 can provide the requisite 3-5mA of typical screen grid current. The reason why it is possible to make that sort of amp fully DC coupled is that screen grid drive voltages are positive and the DC point is easily set at around 100-150V, by varying the bias voltage on the (first) grid of the output tubes. This unfortunately also varies the gain from screen grid to plate of the output tubes, so a compromise is necessary.
P.S: given the various voltages on the cathodes of the input stages, you may also run into problems with heater to cathode maximum DC voltage...
The idea was to use a beefyer version of the output tube bias supply (before regulation) as a negative bias for the CCS negative rail. In truth, a few V negative would be enough for the first stage, and the second can easily work from ground. If you are using DC heating, getting a negative supply good for a few tens of mA should be easy.
Regarding the schematic, you seem to be doing something odd with the first stage. Given your values in the diagram, both the front end stages have a local supply of about 330V. The first stage has a plate DC operating point at 75V respective to ground. This stays the same regardless of what DC voltage you set on the grids of the 6SL7 as long as it is within the range of the CCS.
In using this configuration, you effectively need two coupling caps - one is the input cap, the other is the unmarked cap in the gate DC point filter. Hold this thought as I will get to it later on.
The plate DC point of the next stage is at 225V or thereabouts (I'm doing the math in my head so I may be off a few V). Since stage 1 and 2 are DC coupled, the grids are at 75V DC. For a tail current of ~~10mA, you have 5mA per half of 6SN7, shich puts the cathodes at about 70.5V (this is from the 6SN7 curves). In other words, your second stage CCS has a good 270V across it with no good reason - 70V is MORE than enough on the CCS, producing some 0.7V of dissipation. I am of course assuming the balance pots are of small value and do not drop signifficant voltage, as they shouldn't if you want the LTP to be a good differential amp.
Now, the big question is, what do you gain with this arrangement? If the first stage had a CCS referenced to ground or some small negative voltage, you could lose the input coupling cap and the input DC point network and it's associated coupling cap, which also passes audio current, so is really another coupling cap. Of course, you need to use coupling caps between stages then. The net plus is that these canbe smaller and easily be a foil type cap to you liking. In your current arrangement, you may need a large filter cap for the input DC network, even an electrolyitic, which is always much worse than even the most common foil cap.
You do, however, have two alternatives, based on the fact that the total gain of your two front stages is well over 500:
1) If you reference the first stage grids to zero, you still have 75V DC at the plates, the 6SL7 is still surprisingly quite linear under these conditions, and offers a gain of about 50 and 50Vpp output with reasonable distortion (well over what you could ever use as an input to the next stage, in other words). The next stage stays biassed exactly the same regardless. The positive side of this is, no input or DC bias network coupling caps at all. Given the very high total gain, you could also optimize the plate resistors of the first two stages for a better operating point for the 6SL7 (higher plate voltage) but slightly less gain. You could also add local feedback from plate to plate of the first and second stage (J. Broskie has an article at tubecad.com on this).
2) Given the high gain, and obvious desire for full DC coupling (which is to be commended), you may want to look into screen grid drive for the 807. Mr. Danielak has a screen grid drive SE amp using an ECL82 for the front end (triode part is broadly similar to the 6SL7, pentode part is triode strapped as a follower because screen grid drive needs screen grid current) driving a PL519 sweep tube. I have not been able to find an example of screen grid drive for the 807 or any curves for this mode of operation, but there you have it, a challenge 😉 . Doubling up something like his circuit and connecting the first stages as a LTP would roughly give you hwat you need. It may well be possible to use the same 6SL7 and 6SN7 (first as LTP, second as two followers), as long as the 6SN7 can provide the requisite 3-5mA of typical screen grid current. The reason why it is possible to make that sort of amp fully DC coupled is that screen grid drive voltages are positive and the DC point is easily set at around 100-150V, by varying the bias voltage on the (first) grid of the output tubes. This unfortunately also varies the gain from screen grid to plate of the output tubes, so a compromise is necessary.
P.S: given the various voltages on the cathodes of the input stages, you may also run into problems with heater to cathode maximum DC voltage...
To ilimzn:
I forgot to change the tail current for the first stage (dammit!). The new design value for that is: 2.0mA, for a plate current of 1.0mA for each section of the 6SL7. This is intended to put the plate-to-ground voltage at or near 0V(DC). The cathode voltage will then need to be: -147V(DC) for a plate -- cathode voltage of the same value (from characteristic). Since the bias point selected is -1.5V(DC), the grids would have to be at -147 - 1.5 (-148.5). This allows the DC coupling with the 6SN7's setting their own bias point.
You do, however, have two alternatives, based on the fact that the total gain of your two front stages is well over 500
This isn't right. Each section of the 6SL7 has a gain of 66. However, since they are wired in a differential topology, the stage gain is half that, or ~33. The second stage has a gain of ~8.5, as it's also a differential. This gives a front end gain of ~280, so that there is enough for inverse feedback with decent input sensitivity for whatever level of feedback proves best for optimizing distortion. Of course, that will require empirical determination once this has been built.
If the first stage had a CCS referenced to ground or some small negative voltage, you could lose the input coupling cap
Not gonna do that. I've seen way too many designs where the audio output is also used as a DC supply. The sound card for my computer does this: putting +12V(DC) on the audio outputs to operate some sort of speaker/amp combo. It may be convenient, but risks blowing the first stage grids (and in any case, will certainly cause horrendous distortion.
he plate DC point of the next stage is at 225V or thereabouts (I'm doing the math in my head so I may be off a few V). Since stage 1 and 2 are DC coupled, the grids are at 75V DC. For a tail current of ~~10mA, you have 5mA per half of 6SN7, shich puts the cathodes at about 70.5V (this is from the 6SN7 curves). In other words, your second stage CCS has a good 270V across it with no good reason - 70V is MORE than enough on the CCS, producing some 0.7V of dissipation.
Setting that up isn't that big of a deal anyway.
Given the high gain, and obvious desire for full DC coupling (which is to be commended), you may want to look into screen grid drive for the 807. Mr. Danielak has a screen grid drive SE amp using an ECL82 for the front end (triode part is broadly similar to the 6SL7, pentode part is triode strapped as a follower because screen grid drive needs screen grid current) driving a PL519 sweep tube. I have not been able to find an example of screen grid drive for the 807 or any curves for this mode of operation, but there you have it, a challenge.
I'll look into it.
I forgot to change the tail current for the first stage (dammit!). The new design value for that is: 2.0mA, for a plate current of 1.0mA for each section of the 6SL7. This is intended to put the plate-to-ground voltage at or near 0V(DC). The cathode voltage will then need to be: -147V(DC) for a plate -- cathode voltage of the same value (from characteristic). Since the bias point selected is -1.5V(DC), the grids would have to be at -147 - 1.5 (-148.5). This allows the DC coupling with the 6SN7's setting their own bias point.
You do, however, have two alternatives, based on the fact that the total gain of your two front stages is well over 500
This isn't right. Each section of the 6SL7 has a gain of 66. However, since they are wired in a differential topology, the stage gain is half that, or ~33. The second stage has a gain of ~8.5, as it's also a differential. This gives a front end gain of ~280, so that there is enough for inverse feedback with decent input sensitivity for whatever level of feedback proves best for optimizing distortion. Of course, that will require empirical determination once this has been built.
If the first stage had a CCS referenced to ground or some small negative voltage, you could lose the input coupling cap
Not gonna do that. I've seen way too many designs where the audio output is also used as a DC supply. The sound card for my computer does this: putting +12V(DC) on the audio outputs to operate some sort of speaker/amp combo. It may be convenient, but risks blowing the first stage grids (and in any case, will certainly cause horrendous distortion.
he plate DC point of the next stage is at 225V or thereabouts (I'm doing the math in my head so I may be off a few V). Since stage 1 and 2 are DC coupled, the grids are at 75V DC. For a tail current of ~~10mA, you have 5mA per half of 6SN7, shich puts the cathodes at about 70.5V (this is from the 6SN7 curves). In other words, your second stage CCS has a good 270V across it with no good reason - 70V is MORE than enough on the CCS, producing some 0.7V of dissipation.
Setting that up isn't that big of a deal anyway.
Given the high gain, and obvious desire for full DC coupling (which is to be commended), you may want to look into screen grid drive for the 807. Mr. Danielak has a screen grid drive SE amp using an ECL82 for the front end (triode part is broadly similar to the 6SL7, pentode part is triode strapped as a follower because screen grid drive needs screen grid current) driving a PL519 sweep tube. I have not been able to find an example of screen grid drive for the 807 or any curves for this mode of operation, but there you have it, a challenge.
I'll look into it.
807 is a lousy candidate for screen drive. Compared to sweep tubes, you've got to pump a LOT of voltage into that screen to get the tube to swing.
Miles Prower said:
My point was that there is really no need to force the plates of the first stage to 0V, since the second stage CCS just sees this as common mode voltage. As long as you have enough headroom on both stages to get the proper voltage swings, and can select a good low distortion operating point.
Yup, you are right of course, it was an obvious mistake, I forgot about the differential setup. Even so, this actually makes my point, that being, your first stage is not really required to swing very high voltages, so does not need a particulairly high Vkp. The second stage, on skimming over the 807 data, needs to provide 30ish something V, lets say 50. This brought about the reasoning that you may get away without a negative supply. You may be wondering what my dislike of a negative supply is, and I will get to that shortly.
I see what you mean, and actually I always provide a coupling cap for all my designs even if they could be DC coupled, but then, I do leave the option of puttinga jumper across it. A gate-cathode protection diode is also there...
To be brutally honest, if I came across a sound card with 12V on it's outputs, it would be flying into the trash bin in a nanosecond. That is just utterly irresponsible design, no two ways about this.
But then, if you have to use what you already have...
Agreed. It's just that sometimes, due to the small currents involved, people tend to forget about the SOA of the CCS BJTs, because it is very convenient to use small ones, to get good gain and with that, a high output impedance for the CCS. Given a choice, I'd rather see 70V instead of 270V across a BJT 😉
If you, however, need 0V at the grids of the 6SN7, then the CCS would obviously need a negative voltage.
[G2 drive]
SY said:
Yes, agreed - seems G2 drive is really not at all practical for this tube, now that I've finally managed to download a datasheet with some usable curves. Off hand, it seems that it would require hundreds of V...
Regarding my aversion towards negative voltages (and I acknowledge this is a knee-jerk reaction to an extent): Fotunately, in Miles' design, this problem is alleviated by the use of differential stages throughout. However, in a single ended stage, putting a cathode onto a negative supply makes the negative supply the reference point for that stage, and not ground. This requires a perfectly decoupled, filtered and smoothed power supply (any AC between it and GND will become the input signal for that stage), with a good low impedance AC path to ground. Even if we are talking a few mA of current from that supply, this tends to imply one more power supply cap passing AC signal, and that tends to be electrolytic. Of course, this situation is really unavoidable, if a signal is taken off a plate, it's reference really is the positive rail - it's just a case of choosing one evil instead of two.
Over here I'm experimenting with a hybrid amp that has the above exact problem, because I'm trying to keep DC coupling between the input tubes and output MOSFETs in order to avoid the standard AC coupling rectification problems on clipping, which could put the output stage into class B or even C if they last long enough. I've been on it for months trying to find the best solution to it... I appologise for it rubbing off here in this thread...
Second Revision
Made adjustment to the DC voltage distribution.
V(bias)= -33.75V(dc)
V(kk)= -20V(dc)
V(pp)= 365V(dc)
V(sgsg)= 270V(dc)
A(v1)= ~33
A(v2)= ~7.25
This gets rid of the need for putting negative voltages on the grids, and reduces the negative PS requirements.
Made adjustment to the DC voltage distribution.
V(bias)= -33.75V(dc)
V(kk)= -20V(dc)
V(pp)= 365V(dc)
V(sgsg)= 270V(dc)
A(v1)= ~33
A(v2)= ~7.25
This gets rid of the need for putting negative voltages on the grids, and reduces the negative PS requirements.
Any news on your design?
I did make one modification, by using the 6SN7 stage as a common plate driver, instead of a voltage amplifier. This simplifies the power supply considerably, even if it costs some open loop gain. This can be made up in a preamp (design pending: this is where the sonic controls will go.)
So far, now I'm in the process of winding the power xfmr and a second 8H ripple choke; waiting on the delivery of the Auricaps, VT hardware, and the OPTs. By the time these arrive, testing the LPT for correct operation and optimization can begin. Looking to a completion date by the end of August (sooner if the design comes together faster).
I did make one modification, by using the 6SN7 stage as a common plate driver, instead of a voltage amplifier. This simplifies the power supply considerably, even if it costs some open loop gain. This can be made up in a preamp (design pending: this is where the sonic controls will go.)
So far, now I'm in the process of winding the power xfmr and a second 8H ripple choke; waiting on the delivery of the Auricaps, VT hardware, and the OPTs. By the time these arrive, testing the LPT for correct operation and optimization can begin. Looking to a completion date by the end of August (sooner if the design comes together faster).
- Status
- Not open for further replies.