3Q5 Pentode Neurosis. Help.

[IndyNate]

Hi everyone! In my ever continuing efforts to build new 'Quickies', I stumbled upon the 3Q5. Research indicates it's basically an octal version of the 3S4, and the datasheets are fairly close to each other. So, here's where I am: I built a preamp copying my 3S4, but, octal sockets. I carefully followed the 3Q5 data, wiring the bottle in parallel for 1.4v on the heater, and using a 3.9k resistor for the plate load, and a 1k for the cathode bias. B+ is 33v (rechargeable batts). I checked and rechecked my work about a dozen times. Here's the problem!

Wired in triode mode, like the Quickie was designed, B+ is 33v, plate is 29v and bias is 1v. This reckons to be operating around 1ma, which is way off where it should be. Things get very interesting tho, as rewiring in pentode, holding the screen at 33v, the plate runs at 28v, and bias jumps to 1.5v meaning 1.5ma right? OK, hold the screen at 24v (tap off the batteries), and the plate runs at 30v with bias at .8v meaning 800 microamps? Fascinatingly enough, changing the plate or cathode resistors changes nothing. The pentode seems to pull the same current regardless. This is bizarre to me.

I thought about miswiring, but no matter how many times I check my work, I can't find any mistakes. I understand the plate resistance of pentodes is very high, running somewhat like a CCS, but that still doesn't explain this issue to me.

So, any thoughts out there? One last curiosity, replacing the cathode bias resistor with a blue LED (2.7v drop) makes the tube stop conducting. And I carefully checked orientation. The LED is working as other tests confirmed. I'm really stumped and apologize in advance for the confusing post! One final thing, I checked both tubes on my Hickok 6000, and they're NOS functioning.

[Paul Birkeland]

Were you able to find a 3Q5 datasheet with triode curves?

Is your bias 1V connected to the + or - end of the filament?

How do things look if you don't use rechargeable batteries?

[IndyNate]

I have tried alkaline batts in both B+ and filament supply just to rule out any variables. I've also tested voltages from just about every possible scenario. Things seem correct. The bias resistor is connected to the center tap of the filament, pin 8, just as the datasheet calls for. Unfortunately, I'm flying blind on this project, as I've yet to find a triode curve sheet like they published for the 3s4. Just spent another hour or so experimenting. If you strap the cathode to ground, holding G2 at B+ the plate draws 3.3mA. I think at this point I'm confident the 3q5 is NOT equivalent to the 3S4. Deeper studies comparing the datasheets show pretty big differences in transconductance and G2 voltages. It looks like the 1C5 is a lot closer to the 3S4, but again, I can't find triode plots, so, just not sure. i think at this point I'll just go back to the 3S4. Still, the strangest mystery to me is why I can't cathode bias the circuit with an LED. I don't understand why it ceases to conduct at all in that scenario. Makes me think I still have miswired it, or, I'm just stumped and need to abandon ship!

EDIT:

All is well. Turns out the operating point of 1.3mA is just great. Decided to finish up the wiring and give it a listen. These octal based batt tubes are just amazing. Massive thick gain and excellent imaging. Vanishing microphonics. From very rough measurements, this amp has about a mu of 4.5 or so, which is all the gain I needed.

So, for those keeping the Quickie battery pre ghost alive out there: 3Q5gt, ~32-38B+, G2 at B+, 3.9k plate resistor, 1k cathode resistor with your choice of capacitor bypass value (mine is 470uF), and off you go! Just follow the Quickie circuit minus of course the triode strapping and I think you'll love it. Flying blind and just giving up and trying it worked in this experiment.

Later all!

[Paul Birkeland]

Just spent another hour or so experimenting. If you strap the cathode to ground, holding G2 at B+ the plate draws 3.3mA.

I would recommend not attempting operational experiments that run a tube in a manner other than triode strapped with cathode bias.  If you zero bias the tube, G1 can draw grid current and that can destroy the tube.

I think at this point I'm confident the 3q5 is NOT equivalent to the 3S4.

Yes, you likely need to run more B+ voltage.

All is well. Turns out the operating point of 1.3mA is just great. Decided to finish up the wiring and give it a listen. These octal based batt tubes are just amazing. Massive thick gain and excellent imaging. Vanishing microphonics. From very rough measurements, this amp has about a mu of 4.5 or so, which is all the gain I needed.

1.3mA of current is not a particularly healthy target, especially if you are trying to drive something like a subwoofer plate amp or another solid state amp.

So, for those keeping the Quickie battery pre ghost alive out there: 3Q5gt, ~32-38B+, G2 at B+, 3.9k plate resistor, 1k cathode resistor with your choice of capacitor bypass value (mine is 470uF), and off you go! Just follow the Quickie circuit minus of course the triode strapping and I think you'll love it.

Again, now you've taken the lowish output impedance of the Quickie and multiplied it by a couple thousand.  It's totally fine to experiment, but I would strongly recommend getting back to making a proper circuit with the valve triode strapped.  With the tube pentode wired, you'll also have highly variable gain depending on the load (which gets really bad when your load impedance isn't perfectly resistive).

[IndyNate]

Good thoughts and ideas Paul. I can't disagree. The pre does work, and does sound good, but without good datasheets or curve tracers I won't be messing with it again. It is cathode biased to 1.5v, as yeah, I wouldn't want it fully conducting. That was just a quick experiment of now more than a few seconds for testing sake.

In my system it hits a 475k load into a tube amp, so impedance matching isn't too much an issue, and, my interconnect is 2ft. Makes sense tho about SS devices. This would probably NOT be a good idea. I would call this pre, um, boutique 

Thanks again for the advice and observations. I sincerely appreciate it!

[Paul Birkeland]

In my system it hits a 475k load into a tube amp, so impedance matching isn't too much an issue, and, my interconnect is 2ft.

If that 475K load is at the grid of something like a 12AX7, the input impedance is far lower than 475K.

[Bonzo]

In the past I've posted some pics of my quickie with 3q5 tubes (I used an adapter to fit the octal tube into the 7 pins socket of the quickie). I think the sound was great and the gain seemed higher than the 3s4. 3q5 tubes with metal body seemed even hotter.

Never get my head around the datasheets to calculate the operating point, so I'm reading with lot of attention Paul's comments on IndyNate experiment.

@Paul, do you think the standard circuit of the quickie must be redone in order to optimize the sound? How?

Thank you

[Paul Birkeland]

The place I would start would be with all of the pentode curves.  Where screen grid voltage=plate voltage, then you get points that are valid for the triode curves. 

For the rest, I would setup my bench supply and fix the grid bias voltage, then run the plate voltage up in 20V steps and record plate current, which would effectively generate a set of triode curves. 

Once you have the triode curves, you can figure out what the requirements are to design something like the Quickie with this tube. 

PJ likely has a more numerically focused approach to estimating what you would need to know, hopefully he'll weigh in as well.