New 6L6GC Project. DC -> LTP

[sorenj07]

Originally Posted by ilimzn

5) You have an error on your 6N1P curves. If the supply is ~700V and the cathode is at 176V, thus the load line m8st be drawn for 700 - 176V = 524V or so, not 400.

Ok, this simplifies things a bit. I didn't know you don't have to subtract the voltage drop as a result of the plate resistor! This makes it a lot easier, and I'll probably re-draw the schematic keeping the original 460V supply. I don't want to risk arc-over on the noval tube socket just for superfluous headroom..

[sorenj07]

Originally Posted by Tony

yes they are basically the same....
there was a groupbuy sometime ago, lots of info here: IXYS IXTP01N100D for non-USA citizens Groupbuy

Great, thanks! Things are finally coming together.

[ilimzn]

Originally Posted by sorenj07

Ok, this simplifies things a bit. I didn't know you don't have to subtract the voltage drop as a result of the plate resistor! This makes it a lot easier, and I'll probably re-draw the schematic keeping the original 460V supply. I don't want to risk arc-over on the noval tube socket just for superfluous headroom..

You mean cathode resistor?
Keep in mind the tube curves show voltage respective to cathode. It does often say, eg. on the X axis, Plate Voltage - in fact it's Cathode to Plate voltage. Same thing with grid voltage, it's actually Cathode to Grid voltage.

The basic construction of a load line assumes a common cathode connection, i.e. cathode is grounded, which is why we use said curves that imply cathode = 0V = ground.FOr that scenario, the maximum current through the circuit would be if there was 0V Vkp, which is why we put B+ divided by plate resistor as one point the load line goes through, on the Y axis. The maximum voltage in the circuit is the actual B+ which is why we use this as a second point the load line goes through, on the X axis.

Here you might notice that in a realistic circuit, there is a cathode resistor used to develop bias voltage, but because this is fairly small with respect to B+ is, we tend to disregard that the cathode is not a 0V when drawing the load line, unless we are designing for something like a low voltage tube.

However, in order to get a LTP to work well, either we need to have a common cathode resistor with a rather high voltage drop, or a CCS which also uses a minimum of voltage to even work. With a CCS, there is an extra feature, that the cathodes will follow what you do on the grids in trying toi keep the current constant (which is what a CCS does).
Because you are connecting your LTP directly to the plate of the previous stage, the grids will be at the same voltage as the plate of the previous stage. This means the cathodes will be a few V higher, and in this case these voltages are not at all small with respect to B+ so you MUST take them into account. So, in this case, you can't use B+ on the X axis of the tube curves for the LTP, but the actual maximum Vkp, which is for all intents and purposes B+ less Vp of the first stage.

[sorenj07]

From the looks of things, especially on Pete Millett's page,

Current Source

the 10M45 isn't unhappy at 5mA, so I think to keep things simple (and the parts list shorter) I'll just use those, since I have 4 already. 600 ohms should provide the 5mA current, yes?

Originally Posted by ilimzn

Because you are connecting your LTP directly to the plate of the previous stage, the grids will be at the same voltage as the plate of the previous stage. This means the cathodes will be a few V higher, and in this case these voltages are not at all small with respect to B+ so you MUST take them into account. So, in this case, you can't use B+ on the X axis of the tube curves for the LTP, but the actual maximum Vkp, which is for all intents and purposes B+ less Vp of the first stage.

Really glad to clear this up. I've made a final final (I promise) design that I'll post as soon as I've written everything out nicely. There should still be plenty of swing, probably +117V -160V using 47K plate loads on the 6N1P, with the previous stage at 100V a-k.

[sorenj07]

newest iteration. neglected to draw 1K grid stoppers on the 6N1P inputs. also haven't sorted out the protection diode, everything else seems pretty solid.

[ilimzn]

Looks good... protection diode between grid and cathode of top 6N1P, anode to grid, cathode to cathode. 1N4148 will do but you might want to look up something a bit more sophisticated, maybe BAS21 or something like this, as you want low capacitance.
I would also recomend a bleeder resistor between 463V supply and ground, just in case. Also, grid stoppers (1k for now) on 6L6GC. These may need to be made a bit higher, depending on how the amp will clipping (blocking distortion due to attempting to drive the grids of the 6L6 positive).
Finally, check your swing considering that you have 47k plate resistors for the 6N1P and 100k grid leaks for the 6L6, the swing will be determined by a dynamic load line through your operating point on the 47k lstatic load line, but for 47k in parallel with 100k, roughly 31k or so, which is signifficantly smaller than the 47k plate resistor.

[sorenj07]

Originally Posted by ilimzn

protection diode between grid and cathode of top 6N1P, anode to grid, cathode to cathode. 1N4148 will do but you might want to look up something a bit more sophisticated, maybe BAS21 or something like this, as you want low capacitance.

Looks like the 1N4148 has only 4pF judging by this datasheet..?
http://www.nxp.com/documents/data_sh...148_1N4448.pdf

When you say "cathode to cathode", what does that mean? I get that the grid is always supposed to stay negative with respect to the cathode, and anode should always be positive with respect to the grid and cathode, but the last part stumped me..

Originally Posted by ilimzn

I would also recomend a bleeder resistor between 463V supply and ground, just in case.

Maybe 220K 5W or something?

Originally Posted by ilimzn

for 47k in parallel with 100k, roughly 31k or so, which is signifficantly smaller than the 47k plate resistor.

Ugh.. can I just make the plate resistors 68K and the grid leaks 130K to average out 44.6K? Or is that cheating?


Thanks so much for all the help so far. Can't wait to get this project up and running, and post some in-progress build pictures

[ilimzn]

Originally Posted by sorenj07

Looks like the 1N4148 has only 4pF judging by this datasheet..?
http://www.nxp.com/documents/data_sh...148_1N4448.pdf

When you say "cathode to cathode", what does that mean? I get that the grid is always supposed to stay negative with respect to the cathode, and anode should always be positive with respect to the grid and cathode, but the last part stumped me..

Ah, sorry for that.
Diode anode to tube grid
diode cathode to tube cathode
This will make the diode cinduct whenever the grid of the triode goes about 0.5V positive WRT cathode of the tube. In all normal conditions the diode will be reverse-biassed and will not conduct.
The problem with the capacitance is that it varies with the reverse voltage on the diode junction. Check that it is small at around -2V or so, if yes - 1N4148 will do. Don't have the diode data at hand so I'm leaving this to you

Maybe 220K 5W or something?

Yes... although I think up to 470k will do just fine too.

Ugh.. can I just make the plate resistors 68K and the grid leaks 130K to average out 44.6K? Or is that cheating?

Probably... in either case the position of the actual dynamic load line will not be the same but you might wiggle around the values and get what you need. Incidentally, IIRC 100k is given as a minimum grid leak for a 6L6 with fixed bias... so that should be more or less fixed.

[sorenj07]

Originally Posted by ilimzn

Ah, sorry for that.
Diode anode to tube grid
diode cathode to tube cathode
This will make the diode cinduct whenever the grid of the triode goes about 0.5V positive WRT cathode of the tube. In all normal conditions the diode will be reverse-biassed and will not conduct.
The problem with the capacitance is that it varies with the reverse voltage on the diode junction. Check that it is small at around -2V or so, if yes - 1N4148 will do.

1N4148 it is. Capacitance stays under 1pF at all times.

Originally Posted by ilimzn

100k is given as a minimum grid leak for a 6L6 with fixed bias... so that should be more or less fixed.

Sounds good! I'm polishing some transformer bells now.

[sorenj07]

got 'er built. all is well except those 72K plate resistors on the 6N1P are dropping way too much voltage, I'm reading like 180V on the anode. looks like it's back to 47K as planned. that said, it doesn't distort audibly and there isn't any audible hum, so i'm happy so far

overall wiring


bias supply


outside


inevitable glow shot