6418 tube experiments

[granite]

Long time reader, first actual post...

After I read the most excellent battery operated tube amp thread by iccaros, I got some of the mini tubes mentioned there.

(Here is that thread : http://www.diystompboxes.com/smfforum/index.php?topic=94778.0 )

So I started experimenting with the 6418 tube in triode mode (grid #2 is directly connected to power). Power is at 12V.

For people who are not familiar with this tube, it does not have a separate cathode, the heater is the cathode. Heater is supposed to get 1.25 V.

Datasheet for the tube : https://frank.pocnet.net/sheets/127/6/6418.pdf

Looking at the output vs the input on the scope, all 5 of my "NOS" tubes behaved a bit differently. One in fact seemed to not have any gain only a ton of distortion.

I almost gave up on that tube when I remembered to invert the output signal on the scope. This showed that this "broken" tube is clipping on the positive side a lot.

Having read through the triode theory from Merlin Blencowe's (ValveWizard) most excellent book (multiple times I may add), I figured that if I could somehow make the cathode more positive - the grid#1 more negative - that may help with this.

I ended up just using a diode to connect pin5 (one of the cathode+heater pins) to ground, raising the whole cathode up around 0.6V extra, at the same time giving the other side 1.8V instead of the normal 1.25. (Using a cheap ebay DC-DC converter.)

This had a very dramatic effect, the distortion is gone and the output swing is huge. Look at the following pic, using one of the tubes. Yellow (channel 1) is the input signal, blue (channel 2) is the output:




Here's another one, this is the tube I was going to give up on - see how the first picture the output swing is only 1.1V for the 0.5V signal. Also, the entire top of the wave is missing. Middle picture is with the diode added. Bottom is just cranking up the function generator to 1V peak-to-peak. At that point the output shows some distortion but the swing is 8.3V (on a 12V power supply)




So the lesson for me was that even in a tube like this, the cathode bias can be tweaked easily - just have to pay attention to keep the voltage across the heater pins in the specified range but the "bottom" of that range does not have to be GND.

I will try to put two of these heaters in series and see what an extra 1.25 V can do to the second.

[duck_arse]

welcome to the forum, granite. my personal favourite is basalt, but never mind that.

if you put two heaters in series, they will see half of the total applied voltage, not more.

[granite]

Thank you!

Sorry, I meant to say that I will crank up the Chinese DC-DC thing to 2.5V to do the series test - so both tubes can get their 1.25V and one will "start at" 1.25 instead of ground.

[amptramp]

If you intend to run filaments in series, there is a problem you may run into.  In some cases, the cathode current of a stage may be on the same order of magnitude as the filament current and since the filament is the cathode, there are some design techniques to ensure the filaments of each tube get the appropriate amount of current even when some of the current disappears from the cathode to the anode and screen.  If you look at the 3-way battery tube portable radio diagram below:



the filament string is brought out to the lower left in the schematic and it shows 470 ohm and 1000 ohm resistors going from ground to the filaments.  This is to provide extra current to the heaters to replace the cathode emission current going to the plate and screen of various stages.  For example, the 1000 ohm resistor has 4.5 volts across it so it adds 4.5 mA to the current to take care of the plate and screen current in the output tube.  Similarly, the 470 ohm resistor feeds 3.2 mA into the heater string.

If you are just doing small-signal stages where the plate current is a small fraction of the filament current, this may not be necessary but if you use it, the resistor current goes to both tubes on either side of the resistor-filament string junction so you only need one resistor for two tubes and you space them two filaments away from each other.

[granite]

Very interesting, thanks for the explanation.

I think with these particular tubes I probably won't encounter it - the datasheet shows 0.5 mA as the max cathode current vs 10 mA for the heater.

Just out of curiosity, how did you figure out the 4.5 V at that 1K Ohm resistor? Five 1.5 V batteries minus the heater voltage for the 3S4 tube?

Maybe it would be interesting for a guitar preamp to be able switch in/out that cathode diode so that a clear or dirty input could be obtained.

[PRR]

> in triode mode (grid #2 is directly connected to power).

That is Pentode mode.

Triode is G2 tied to Plate.

> invert the output signal on the scope

When you only have 12V of B+, and you can expect to be slamming the plate all-the-way, it is useful to put the 'scope in DC mode, with zero V at the bottom, +12V near the top. Then you can see all the action and what it is hitting.

> all 5 of my "NOS" tubes behaved a bit differently

All tube-work is +/-20%. Old filamentary types may be even worse.

Pentode operation is very sensitive to bias. Cathode Bias would tend to even-out the tube to tube variations; as you say, you can't do that with naked filament cathodes. In resistor-loaded (opposed to transformer loaded) circuits, running Triode mode (G2 to Plate) will significantly tend to self-adjust bias (if it sucks too hard, G2 goes low, it sucks less). But probably also less voltage gain.

Some notes on what Plate Voltage you got would be clueful. As a general observation, we like plate "near half of B+", so it can swing both ways equally. Or since small tubes won't pull-down hard, we may aim nearer 2/3rd.

> how did you figure out the 4.5 V at that 1K Ohm resistor? Five 1.5 V batteries minus the heater voltage for the 3S4 tube?

That's one way. The other way is to see that this point feeds three 1.5V filaments in series. If everything is nominal, 3*1.5V= 4.5V. Which comes to the same place as 7.5V-3V, yes.

> the cathode current of a stage may be on the same order of magnitude as the filament current

I may have mis-laid a bottle. But I think the cathode I is significant compared to filament I "only" for the largest such tubes, and notably the 3S4-like type in your plan. While the 6418 is called a "power amplifier", it is a 10mA filament and more like 0.1mA of cathode current. (Max 0.5mA, but you need all of 17V at G2 to get there.)

[thermionix]

Paul, I just gave you your 1001th (1001st?) like...just to mess up your too-round number.

[granite]

Paul, thanks for the answer!

I found an old laptop power supply - 19.3V, so I started using it.

In triode mode (for real this time) I am getting 12.6V at the plate using a 100K Ra. This is very much in line with what you said about 2/3 B+.

With the diode still on the filament ground, it starts real distortion just below a 3V signal.

Thinking about putting another 6418 in front of it - either in pentode or triode mode - to amplify to that level from guitar pickup level signal.

The goal would be to build it into a preamp (or full amp) where the overdrive can be mostly controlled by the guitar volume.

[granite]

Just a quick update - after breadboarding 2 tubes in triode mode (g2 = anode) and hooking them up to an LM386, the sound was too clean for my taste.

Did a mu test (similar to the scope pictures I posted earlier). In triode mode I can get a voltage gain of 5-6 approximately. This seemed really low so I checked the datasheet - I am not that far off.

Using the "find mu" method from ValveWizard's book:



I get 7. I figured my old tubes doing 5-6 is pretty good. This does raise a problem though. As I said earlier, I need 3V swing of the input for the tube to start that nice overdrive tone I am after.

Tested with a Telecaster - normal strumming produces 200 mV-ish signal. With an ESP Les Paul I am getting around 4-500 mV on the first tube. So coming off the anode of the 1st tube I am measuring 1-1.2V for the Tele and 2 - 2.5V for the humbuckers.

That means that the 2nd tube amplifies things nicely and I am still left with a pretty clean tone. Not good  So what I did was add a 3rd 6418 tube and between tubes 2 and 3 I added a 1M pot to adjust the gain. Now tube 3 can be overdriven as much as one would tolerate as tube 2 signal will be around 5V for the Telecaster and 10+V for the ESP. That will make the output of tube 3 very square wave-ish.

Anyway, once the amp is out of experimenting, if someone is interested, I will draw up the schematics.

[PRR]

> I get 7. I figured my old tubes doing 5-6 is pretty good.

Mu would be Gain if the coupling and load impedances could be infinite.

You use a resistor to pass DC to the plate, and there's always some load.

12AX7 Mu of 100 tends to give gain of 45-55. 12AU7 Mu of 20 tends to give gain of 15-17. So you do OK.

Assuming Vp at 8 of 12V you have less than 4V of up-swing, and probably similar downswing.

Gain of 5, 4V/5 says you should need 0.8V peak to overdrive a bit. Yes, you probably need gain in front. As this is IAC clean gain, you can use an op-amp.

[granite]

I see.

Went back and re-read the load line parts of the book, I get it now. Mu would only be the voltage gain if the load line was horizontal - the triode basically running in constant current mode.

So I have 19V and a 100K Ohm resistor, my load line goes between (0 uA, 19V) and (19V / 100K Ohm = 190 uA, 0V)

Since it has a slope down, the actual voltage gain for every 1V in grid will be less. Green is the load line, the blue are the "measurement" lines for the 1V grid change:



5V or 4V for every 1V. That matches really well with what I measured.