Could check this pentode-mu stage thingy for me (disclaimer* it is for a gutar amp)

[aletheian]

OK. I have a thing for trying out Hi-fi concepts in guitar amps. What I'm trying to accomplish is a pentode input stage that is high gain and low noise, with a decent bandwidth, eg reasonably low lows (meh...30Hz or so) to about 15kHz-ish on the highs. For reference, the typical voltage from a guitar will range from 150mV to 1V at 15K to 40K ohms depending on the pickups. This pentode will be feeding a tone stack, which wants to see a fairly low impedance... not OTL headphone amp low, but under 4K to 5K-ish. Typical tone stack stage impedance to ground is about 1.2Meg.

So at first I thought, grounded plate triode cathode follower, but then I thought... why not wrap it up with a CCS and go with a mu stage instead. So I stacked a pentode on top of the pentode. I usually elevate my filaments anyway, so I didn't see too much of a problem.

I have about 400v to work with on the B+... unregulated but fairly stable and clean, but I will knock that down to about 250v-300v so that I don't fry the plate of the EF184 on top. I guess I'll be running the bottom tube around 3-5mA, but that is a guestimate. I'll have to look at the curves before I decide, but that sounds about right.

Anyway, if someone could take a peek at it for me, that'd be greeeeeat since I have never tried this before. Despite my desire for glowing glass, I am not opposed to using a single depletion mosfet up top, but I want to avoid cascoded ccs's, zeners and battery bias stuff if at all possible.

I want the thing to stay fairly clean. If it amuses anyone to know, this input stage will be feeding a LTP 6SL7 splitter with a B+ of 400v, pushing a gain of about 40 at around 25Kohms (before NFB is applied) into an ultralinear KT88 output stage, running at about 60W with 560v on the plates. There is a bit of local feedback from the kt88 plates to the LTP, and a maybe bit of global running from the OT to the input stage cathode...haven't decided on that one yet. It'll also have a second channel running the input stage into a 6AL5 clipper/clamper and then to a 6AU6 gain stage and out to the LTP.

[Sch3mat1c]

Looks okay, not sure what it's supposed to bias on though- I'd like to see, say, a voltage divider from +V to top grid.

Don't worry about melting plates, I doubt you can melt a nuvistor at 4mA.

Tim

[James D.]

Hi aletheian,

I've gone the high Gm pentode route as then the Zout is more or less set by the load resistor,

So 12GN7 loaded by 5K will give you a gain of 150 (Gm*Rl) and an output impedance of 5K. run at 30mA gives 150V olts accross Rl so the 1V input will just start to distort.

Run at 250V on the anode, 150 on sg, a little under 2volts on the grid.

Of course you could use this as an output valve instead of an EL84/6V6 too. Hey a Spud guitar amp! I've done this with an active pickup guitar and it is great for a practice amp :-)

James

[aletheian]

Originally posted by Sch3mat1c
Looks okay, not sure what it's supposed to bias on though- I'd like to see, say, a voltage divider from +V to top grid.

Don't worry about melting plates, I doubt you can melt a nuvistor at 4mA.

Tim

Hey Tim,
Thanx for the reply. The Top tube should be getting about 1 volt or so of bias from the junction of the 150r and 39K resistors, through that 1Meg resistor. So I should be good there.

Thanx

[aletheian]

Originally posted by James D.
Hi aletheian,

I've gone the high Gm pentode route as then the Zout is more or less set by the load resistor,

So 12GN7 loaded by 5K will give you a gain of 150 (Gm*Rl) and an output impedance of 5K. run at 30mA gives 150V olts accross Rl so the 1V input will just start to distort.

Run at 250V on the anode, 150 on sg, a little under 2volts on the grid.

Of course you could use this as an output valve instead of an EL84/6V6 too. Hey a Spud guitar amp! I've done this with an active pickup guitar and it is great for a practice amp :-)

James

That is somethng that I pondered. I don't have anything on hand that has GM as high as a 12GN7 though. I have an EF184 with a GM of 15.5 -ish, so a 10K Rp would buy me 150 gain and a 10K out, which is not too shabby, but I'll be losing the additional PSRR of the mu scheme. If I am not smoking crack here, I think that circuit that I sketched out can approach a gain in the high 200's, but that's on paper.

The original plan was to run a cascoded 6922 with a mosfet current source up top, but I heard some great pentode input amps lately, so I thought I'd go that route. My last endevor was a 6922/2sk369 cascode input stage with a large plate resistor, which was waaaaaay too much gain... around 1000, and noise, and it clipped like mad, but it sounded nice and thick. I used a voltage regulator on that one, and it dropped the noise down a bit, but it didn't sound as nice regulated. I actually kept the circuit, ran a 6922/12sk170 with a smaller Rp and ran it as a solo boost instead of an input stage.

[Sch3mat1c]

Originally posted by aletheian
Thanx for the reply. The Top tube should be getting about 1 volt or so of bias from the junction of the 150r and 39K resistors, through that 1Meg resistor. So I should be good there.

Problem is, how do you know if 1V produces exactly the right bias current you need? Especially with pentodes, a mere 0.01mA could make a difference of 50V!

Something you learn when dealing with transistors, which vary by a factor of oh 5 to 10, and that's across different operating points of the same transistor! Circuits have to be very stable.

Tim

[James D.]

Hi Aletheian,

PSRR can be an issue, I use two ways to minimise it, the first is to use a pentode Aikido stage (see www.tubecad.com) but is then more complex than the mu-stage - however it can give you a very low Zout, just work within the V/I limits of the output section. Or I use a VR tube on the screen grid returned to the cathode. This is sort of a halfway house as there is still the noise from the anode...

Same PSRR issue with cascodes... usually not too much of an issue. I often use both input stages in my guitar amps as they sound different. I haven't used hybrid cascode in my guitar amps - I'll give it a go and see how they sound Thanks!

One way I found around too much noise from the pentode input was to use a FET active guitar lead with the FET in the guitar jack end phantom powered from the amp end of the lead. This has the double benefit of boosting the signal and greatly dropping the impedance of the guitar lead - both reduce noise pick up and improve front end s/n ratio. Actually it has a triple benefit as the passive pickups and tone pots work into a nice well defined very high impedance - typically of the order of 3-7Megs

James

[aletheian]

Originally posted by Sch3mat1c


Problem is, how do you know if 1V produces exactly the right bias current you need? Especially with pentodes, a mere 0.01mA could make a difference of 50V!
...

Tim

Good call, That was a back of an envelope calc that I decided to Spice. it seemed likethat bias point would give me what I needed current wire, but I may have to varu the ration of the voltage divider to get what I need in real life.

[aletheian]

Originally posted by James D.
Hi Aletheian,
...One way I found around too much noise from the pentode input was to use a FET active guitar lead with the FET in the guitar jack end phantom powered from the amp end of the lead. This has the double benefit of boosting the signal and greatly dropping the impedance of the guitar lead - both reduce noise pick up and improve front end s/n ratio. Actually it has a triple benefit as the passive pickups and tone pots work into a nice well defined very high impedance - typically of the order of 3-7Megs

James

That's not a bad idea. I hate the thought of silicone in the signal path, but maybe a Jfet input buffer would be an interesting way to go. I may just scrap the CCS mu stage idea and just use a highly filtered power supply, and then shunt regulate it to just the input stage. I like input stages to have a wide frequency response and low noise to lay a cleam pallate for tonal shaping down the line.