How do I calculate the input impedance on this preamp? JAN 6418 Tube based

[exabrial]

Hey guys,

I'm trying to redesign the Oatley Electronics k270 preamp. (see here: http://s13.postimg.org/patw5krxj/k270.png) The original parts are no longer available.

Here's my progress so far: https://123d.circuits.io/circuits/852190-jan-6418-vaccum-tube-valve-preamp/

And the image at this time of this writing:


First, do you see anything stupid about what I'm doing? I've had to change a few things, and I don't necessarily always understand the side effects.

Next, how do I calculate the input impedance of this preamp? I might be using it to boost a piezo, which notoriously require a preamp with super high input impedance. If the current design is too low, I may need to get it up into the 2-5mohm region

Thank you!

[PRR]

Welcome to the forum.

> how do I calculate the input impedance of this preamp?

The tube grid is "infinite" impedance (pencil 100 Megs for low audio range).

So the input also has to drive R1 in parallel with R2. 1Meg||47K is around 46.78K. (Counting 100Meg for the grid, 44.87K.)

C1 enters into this, but C1 is "large" (low impedance) compared to R2 down below the audio band.

> I may need to get it up into the 2-5mohm region

Then R2 *can't* be 47K.

And it does not "need" to be so tiny. These are small battery-heater tubes? The grid return may be several Megs. (1 Meg was good enough for V2.)

Change R1 to 22Meg. R2 can be 2Meg or 5Meg. (See if V1 plate voltage changes much when you use 5Meg and then short it; a 20% change is probably tolerable here.) C1 can be 0.005uFd (5nFd) and give full 17Hz bass into 2Meg.

If it comes right off a Piezo (not assorted mystery effects boxes), then R1 and C1 can be omitted. Then input Z is just R2.

Oh, there's some input capacitance which may matter at the top of the audio band, but if you are coming off a capacitive Piezo this does not affect frequency response.

You can try 50Meg or 100Meg at R2 but the tube bias (plate voltage) may shift a lot. You can try messing with R4 to get it back to a happy zone.

Since this is *apparently* a 35V circuit (most 3-pin regulators have 35V limit, and you would NOT want to drop 300V down to filament voltage), I have to wonder about high-gain pentodes and hot signals. Assuming V1 gives gain of 50, and can output 5Vrms clean, the maximum input level is 0.1V (100mV). Most hard-hit Piezos can deliver a lot more than that.

[amz-fx]

I would connect the tube heaters in series. Power them with a 5v regulator and a 250 ohm resistor (or 270) in series between reg and heaters. Even a 78L05 will work at this low current.

The 6418 is going to be happy with B+ up to 24v but the Oatley is powered by 9v. The lower voltage will limit the output levels and distortion will go up if you have a strong signal like a piezo. I'd use a 24v dc adapter if looking for clean tone.

I would do as PRR says and delete R1 and C1, then increase R2 to 10M. It won't matter then if it's a piezo or passive guitar pickups going into it.

One 6418 might do the job if you are just looking for a small boost, but having two allows you to crank up the Rpot as a drive control to add some grit. You'll need to keep it set low for clean tone. You could connect the 6418 in triode mode since you don't need a lot of gain.

I get a feeling that it is going to be gritty no matter what you do, but maybe in a good way!  Build it and find out!  Then tell us about it 

Best regards, Jack

[amptramp]

The R2 grid return seems to be sized to avoid contact potential which is a negative voltage at the grid caused by the impingement of electrons that are intercepted by the grid wires.  If you are running at 9 volts, the plate current would go to zero with less than a volt negative bias on the grid.  From about 1939 until the end of the tube era of radio in about 1970, the first audio amplifier usually had a 10 megohm resistor to ground and no other biasing.  This provided a grid bias of about -1 volt which was enough to let a radio signal through at normal volume (which was set by the volume control ahead of the grid) but this value would be too much for a 6418 with a 9 volt plate supply.

The 47K resistor has a much lower impedance than the volume pots on the guitar itself and a lower impedance than the piezo pickup, so there will be a loss of signal at a critical point where signal to noise ratio is already low.  You may want to use two resistors, one to ground in the R2 position and one to the +1.25 volts at the filament so the grid is pulled slightly positive to counteract the contact potential.  Then you can use the parallel resistance of these two resistors as the R2 value and set the bias on the grid independently of the input impedance.  The resistance in parallel with the filament would be too high to affect filament current to any great extent.

[duck_arse]

the original circuit shows C8 as 10nF, whereas your equivalent C3 is 10uF. any reason for that change?

the original shows the volume pot 'ground' returned to the heater mid-point. what effect is that going to have?

[PRR]

> connect the tube heaters in series

Note that they are not uni-potenital cathodes, they are naked filaments.



The "grid-cathode" voltage is measured to the "mid-point" of the naked filament. Not to an electrically isolated cathode sleeve.

If you stack (series) them, V2 grid return must be referred back to its own filament.

If we didn't, and say the filament voltage was 6V (un-likely!) each, then V2 would be biased at -7.5V, and probably cut-off (unless B+ is in the hundreds of Volts).

He does not say what the tube-type is?(*) Such tubes in US beach-radio service were usually scaled for 1.4V filament, and "very" low current (many types at 0.050A).

(*)Ahhh, if these are 6418 then it is 1.25V@0.020A. So running them in parallel is not a huge waste of heater power. And it keeps the biasing "K.I.S.S."

I do agree that if the raw supply is less saggy than a dry-cell battery on a hot beach, a simple series resistor is as good as anything. Switcher is a bit more efficient, but resistor is so "K.I.S.S."

10uFd on the main raw power rail makes more sense than 10nFd, unless you know there is a big electrolytic on the far end of the power wire. While we could probably use less than 10u, today low-volt Al electros are so cheap that we may as well brace the B+ good and solid.

[Puguglybonehead]

I built this, from the Oatley kit, a couple of years ago. Makes a nice, transparent clean boost actually. I agree with the suggestion of running the B+ voltage higher. This boost doesn't have as much 'boost' as it could. 18V at the plates, maybe? Might try that myself.

I would also suggest using 2 rotary switches if you're going to duplicate the high-pass and low-pass filters as they appear on the schematic. The original used DIP switches which are just not cool for a stomp box circuit. Also, western-made direct-heated tubes like the 6418s can be very microphonic. (Russian tubes are designed to reduce microphonics) Put some rubber grommets around those tubes to dampen vibration.

[exabrial]

Wow you guys are awesome, that's all I have to say. This is my first time designing any sort of audio circuit, so thank you for the kind encouragement and discussion.

Two design goals: One, get a little bit of boost, but two, I was hoping cranking the pot would give me a little bit of tube overdrive without farting out.

While a resistor in series with the power supply is the ultimate KISS method of providing a heater voltage, a 1.25v regulator and two filter caps was $1.50 in parts This will be powered by a supply, so runtime on a 9v battery is not a concern (which would be an acceptable 20-30 hours anyway). Great suggestion though, I need this to fit on a 40mm pcb, so I may have to cut parts.

I've changed up the input network a bit per your suggestions, take a look now over at 123d.circuits.io. I'm going to dive a bit deeper into the biasing, you guys have brought up a bunch of great points to think about.

[exabrial]

Ok, I'm still trying to understand biasing...

If the absolute maximum voltage of the filament is 1.25v, and you calculate the the cathode voltage by take the average at it's two connects (1.25,0)/2 = 0.6v. Given the coupling capacitor c3 which blocks DC, that means the pot is at 0vdc and my grid bias is -0.6v.

Is it even possible to have a larger bias than that somehow? Would I need a larger bias for any reason?

[PRR]

> Is it even possible to have a larger bias than that somehow? Would I need a larger bias for any reason?

Yes. No.

[exabrial]

Lol. Thanks. -0.6v it is. I'm guessing the bias isn't tied to gain or anything then

[exabrial]

Alright guys,

Another dumb question... I can't seem to find high quality .1uf bipolar audio caps on mouser. Since these are dc-blocking coupling caps, can I sub a larger value .47uf? My gut tells me that larger values would be ok.

Thanks!

[duck_arse]

use any plastic film caps you like/can fit/can afford/can find for values at or below 1uF.

[PRR]

> high quality .1uf bipolar audio caps

Save the extra-virgin olive-skin caps for fancy hi-fi.

Guitar effects use whatever is cheap.

At 0.1uFd I would avoid "ceramic" (which has several other names now), except as an experiment. Ceramic over 1,000pFd can distort, which may be good, or may be bad, depending what you want. For a total-wreck distortion effect, it may be the spice on the chop-meat. For a subtle enhancer, it may be a bit much.

As the arse says, Panasonic plastic/poly film is always a safe bet around 0.1uFd. (The several other brands are fine too; P-sonic may be selling-off their plastic-cap business and machines to focus on more exciting products.)

[exabrial]

Thanks again guys for the advice.

So... I breadboarded this thing. And it sounds like money. And the TUBES are fricken sweet. They glow like cat's eyes (which is what I think we're going to name this)




Plenty of clean boost, and if you keep going you get such a natural sounding distortion it's incredible!

I'm not stopping here, I think we want to try "starving" the second tube. I still can't believe this works!

[Cozybuilder]

Would you mind posting a schematic of your breadboarded project?

[exabrial]

Try my link @ 123d circuits