tube cascode help

[HTH]

I've been messing with valve/tube OD pedals for a while and have been messing with a cascode configuration (like the AX84 Firefly input stage) this weekend.

However, with just the cascode section plugged into my plexi, I'm getting only a very slight increase in gain when I was expecting quite a considerable gain boost.

I'm sure the wiring is okay, any ideas? (or am I expecting something that the cascode will never deliver?)

EDIT: oh, here's the link to the firefly schematic... http://ax84.com/media/ax84_m276.gif

[Jazz P Bass]

Heck of a pedal!
Do you measure 8.8V at the cathode output , as per the schematic?

[HTH]

Heck of a pedal!
Do you measure 8.8V at the cathode output , as per the schematic?

I'm only using the first bit of the schematic (V1), the cascode bit, not the full schematic.

[Steve Conner]

According to Valve Wizard,

"Total Gain: The gain of the cascode is equal to:
A = 1 / (1/gm1 * Ra2) + ((ra2 + Ra2)/Ra2) * (1/ mu1 (mu2 + 1))
When using two identical triodes this can be simplified to:
A = gm1 * Ra2
as was calculated earlier."

A 12AU7 has gm = 0.0022A/V so you'd expect a gain of 0.0022 * 100000 = 220. That's not a "slight increase" by any means, so your circuit probably isn't working right.

[HTH]

According to Valve Wizard,

"Total Gain: The gain of the cascode is equal to:
A = 1 / (1/gm1 * Ra2) + ((ra2 + Ra2)/Ra2) * (1/ mu1 (mu2 + 1))
When using two identical triodes this can be simplified to:
A = gm1 * Ra2
as was calculated earlier."

A 12AU7 has gm = 0.0022A/V so you'd expect a gain of 0.0022 * 100000 = 220. That's not a "slight increase" by any means, so your circuit probably isn't working right.

I've attached a schematic of whats in my OD pedal along with voltages. They look pretty close to the valve wizard website's voltages.

Any ideas why I'm not getting any distortion?

[J M Fahey]

Hi, I've never built a cascode, I don't like them, I think they belong in TV or oscilloscope circuits, not in Guitar Amps (by the way, *no* major guitar amplifier used them, ever, at least that I know of), but anyway, I do understand triodes. Considering this as two triodes in series, a common cathode one and a common grid one, I see very strange voltages in your schematic.
To begin with, the currents are consistent: 30V across the top 100K translate to approximately those .66V across the 1K8 cathode resistor. My calculator says .54V but that is irrelevant, we deal with real world measurements here.
*But*, -.66V is *nothing* to bias a 12AU7 into those 0.3mA flowing there.
In practical terms, I'd consider the lower triode unbiased and saturated.
Now onto the upper triode: with no cathode resistor, it shows -40V bias (69-29) WTF?
EDIT: now that I peek at the FireFly schematic: V1 is a 12AX7, not 12AU7, what did you use?
2)Considering you have +210V there, what's wrong with a single triode, 100K plate, 2k2//10uF cathode?

[Steve Conner]

it shows -40V bias (69-29) WTF?

I think the voltage discrepancies are caused by the meter loading down the upper grid. The grid voltage isn't 29V when it's not being measured.

I have no idea why it doesn't work. Unless maybe the 12AU7's gm is specified at a high current, and gets a lot lower at the current you're running.

The best tube for cascodes is the ECC88/6DJ8. If you're feeling heretical, you can make a hybrid cascode, replacing the bottom triode with a transistor or JFET.

[HTH]

thanks for the responses.

just to make things clear, I'm using an ECC83/12AX7 - I want LOTS of gain.

[HTH]

Considering you have +210V there, what's wrong with a single triode, 100K plate, 2k2//10uF cathode?

nothing at all - I tried that (except a 1uF cathode bypass) with the triodes in the ECC83 cascaded (alot like the preamp of a tweed champ). I couldn't get enough gain to get a clean/crunchy amp cooking.

[J M Fahey]

Hi Steve.
I agree, of course, there *is* a measurement error here, no doubt, but typical cheap digital multimeters now have around 10Meg. input resistance, so they will still load and change the measurement, but 40V seems too much, I would think perhaps 4 or 6 volts.
Probably measuring cathode-grid voltage would be somewhat more accurate.
Another discrepance is that the lower triode , with 1k8 bias resistor (admittedly small), drops around 70V , and the upper one, *no* bias, drops around 200V !!! .
If they were different triodes, well .... , but they both live in the same bottle .
EDIT: Hi HTH. A typical 12AX7 with cathode capacitor will provide around 50X real worls gain.
2 of them cascaded (not cascoded he he) then provide 2500X raw gain !!, enough for the most rabid heavy/gothic/thrasher/industrial out there.

[Steve Conner]

I agree! If you can't get enough gain to blow your brains out by putting two cascaded triodes in front of your amp, something is wrong.

I once did a similar mod where I added an EF86 stage to the front of one of the channels in an old Selmer. It didn't give "rabid" levels of gain, but it turned it into quite a nasty piece of work.

JM: The upper grid leak resistor is bootstrapped by the cathode, so the impedance at the grid is probably a lot more than 1M.

[loudthud]

What kind of cap are you using for the 1uF at the grid of the upper tube? I would expect more voltage at the cathode of the upper tube, perhaps a leaky cap is holding it down.

[HTH]

What kind of cap are you using for the 1uF at the grid of the upper tube? I would expect more voltage at the cathode of the upper tube, perhaps a leaky cap is holding it down.

Its an electrolytic cap. The voltage at that point eventually settled at 29v, but started off up close to the cathode voltage (69v).

I'll try replacing that cap tomorrow and see what it does. The circuit is very simple, so there must be something VERY simple screwing this up.

[Jazz P Bass]

Kind of curious that the second tube on tne schematic is labeled Gain.
That tells me that the first section was engineered as a buffer amp.
Why not build the whole circuit?

[HTH]

Kind of curious that the second tube on tne schematic is labeled Gain.
That tells me that the first section was engineered as a buffer amp.
Why not build the whole circuit?

I only want to use one ECC83/12AX7 in my OD pedal as thats what its wired up for right now. Plus, everything I've read on cascodes says that I'll get lots of OD, so I at least want to get this working.

As for building the whole circuit, in a sense my signal path into my amp (plexi clone) has a similar signal path if you follow it down the chain >> cascode OD pedal >> two gain stages then CF & PI in plexi.

[J M Fahey]

Its an electrolytic cap. The voltage at that point eventually settled at 29v, but started off up close to the cathode voltage (69v).

That *screams* leaky to me.
Specially because your meter, even if erring of the low side, will not drift its input resistance, so the only other factor there is the capacitor.

[Steve Conner]

No, it means the exact opposite, the capacitor must be good, or at least its resistance must be higher than the meter's.

When you put the meter on the grid, the initial voltage reading is the same as it was without the meter, but then it sinks down as the meter's resistance slowly discharges the capacitor.

DMMs are usually 10M ohms on the higher voltage ranges, and the time constant of 10M and 1uF is several seconds.

[guitician]

The circuit is very simple, so there must be something VERY simple screwing this up.

Very true, and the answer is probably right under your nose.

I have an old Lafayette PA amp that uses cascode and the circuit is just like the firefly's, except that cap on the upper grid is a .1 uf.

[J M Fahey]

Sounds reasonable.
Anyway I'd like to measure Vg-k straight on the socket, as I proposed before, which would give me a true Vbias measurement independent both from "distance" to ground, meter attenuation, etc,
Although very probably, as Loudtud suggests, the answer will be something very stupid, that lies right under our noses ..... because although I don't like cascodes in guitar preamps, cascodes *do* work.
It's not the first time similar things happen to me .... nor the last ...
Only worthwile advice here?: he he
EDIT: Just for kicks:
1) why don't you check that R 1 Meg is not open/unsoldered)
With everything off measure grid to cathode resistance, you should see 1 Meg.
2) use 2 meters. Clip one of them cathode-ground , you should have those almost 70V
Then, use the other to measure voltage grid to ground, check that it slowly lowers from 69 to 25 as before, and whether *Vk accompanies that downward movement at the same time*

[HTH]

ok, I decided to call it a day on this one for the minute as its taking up WAY too much time. I'd like to thank everyone for their help, its much appreciated.

instead, I've cascaded the two triodes into each other in a 'fairly' standard way...

* 1st stage is grid leak bias like a Fender 5C1
* 2nd stage is cathode bias like a fairly standard triode stage

I'll post some clips in the next few days - sounds pretty decent at the minute, though for all the gain I want, it would probably sound better with another 12AX7 added and build up the gain gradually. At the minute I'm playing with a single mosfet added before or after the valve stages for more saturation. Ain't decided which sounds best yet.

[imaradiostar]

I'd rip out the 1uf cap to the upper grid- you've biased the upper triode waaay into cutoff. Take a look at this:

http://www.freewebs.com/valvewizard/Cascodegraph.jpg

To quote Merlin's book, "generally this voltage will be in the region of 80v to 100v, or around 1/3 HT." This is referring to the voltage at the junction of the bottom plate and the top cathode. Of course the grid should follow these within a few volts. The picture I linked refers to that 80v as the "screen grid voltage."

I generally use a .1 uf 250 volt film cap in the position on my cascode builds. You'll find that 12au7 and 6DJ8 make much better cascodes than a 12ax7. I didn't get a ton of gain with a 12ax7 (around 90-100) but I get well over 200 with a 6dj8. They sound pretty neat when they're working right!

The other thing- when I've used larger caps in that position it takes FOREVER for the circuit to come up to voltage but has no positive effect sonically. Use a smaller cap!

jamie

[HTH]

thanks for the reply radiostar,

I did try a 0.1uF cap instead of the 1uF and it made no difference.

One day I'll revisit this and try something like an ECC88 as these have been suggested as being better for cascodes than an ECC83.

[J M Fahey]

Hi HTH. Sorry to hear the news.
Hate to say so, but, why don't you just dump the cascode idea and build a straight, conventional two triode cascaded preamp?
Such as: Input Jack -> classic 12AX7 (100K-1k5//10uF or 2k7//.68)->.022->A1M pot.->more of the same. This would give you a *lot* of gain, guaranteed, no bias headaches, uses basically the same amount of hardware, you have two gain pots to tweak at will and after building the basic gain block, you can add grid stoppers, across-load capacitors, anti-blocking grid to ground diodes (á la JCM900), whatever you like.
Once you get what you want, you may replace one or both gain pots for fixed resistor networks, if you want to keep front panel clutter simpler.
Good luck.

[HTH]

Hi HTH. Sorry to hear the news.
Hate to say so, but, why don't you just dump the cascode idea and build a straight, conventional two triode cascaded preamp?
Such as: Input Jack -> classic 12AX7 (100K-1k5//10uF or 2k7//.68)->.022->A1M pot.->more of the same. This would give you a *lot* of gain, guaranteed, no bias headaches, uses basically the same amount of hardware, you have two gain pots to tweak at will and after building the basic gain block, you can add grid stoppers, across-load capacitors, anti-blocking grid to ground diodes (á la JCM900), whatever you like.
Once you get what you want, you may replace one or both gain pots for fixed resistor networks, if you want to keep front panel clutter simpler.
Good luck.

hi J.M.,

I've done just that - cascaded the two triodes. Its not quite the huge amount of gain you'd expect and its something I can't get my head around.

If I put an additional two gain stages inside my Marshall plexi, it would be in serious shred, saturated gain territory. However, with those two extra gain stages in a pedal and fed into the amp, its more hard-rock crunch than metal shred.

At present I've got:
* 1st stage: grid leak bias like a Fender 5C1
* 2nd stage: cathode bias like a fairly standard triode stage
* 3rd stage: mosfet gain stage

This is actually pretty much sorted now and more a case of tweaking than complete redesign.

[Guitarist]

Sometimes giving it break is the best thing. You come back with a new perspective and things get fixed

Does lots of gain may mean lots of crunchy clipping or just a a larger wave? You would think someone could make a squarish wave generator with one or two tubes It would be less noise than cascading stages with attenuations. Regarding 12ax7s, the lead designer at Bådcat told me about a particular amp, "...it was all about getting maximum gain from every stage." I know the view, I LOVE loads of distortion, but the statement didn't ring true to me somehow. Those amps have so much hand-wiring in there it is unreal. Only complaint was that, even with Engl's input with the design of the Badcat parallel loop, it sounded out of phase with the effects. Then again mine did too, (Ultimate) which is why I went series again

I hope you figure it out...

[FUCHSAUDIO]

I'm concerned about the 250-K pot on the output.

A cascode certainly wants a lower Mu tube than a 12AX7, and a 12AT7, 12AU7, or 6DJ8 would work fine, but the output impedance is very, very high and it won't be happy driving a tone stack or a comparatively low impedance like a 250-K pot...Perhaps the original idea was to limit the input gain, and count on overloading the output stage of the amp ? The output impedance being high would make driving long cables with this circuit a pain in the butt as well.

You could raise the output pot to 1-M, perhaps add a follower (like an FET or cathode follower) pre-volume control, but I'm not sure you'll get a ton of overdrive from just the single stage alone. You can use the gain increase from this circuit to drive something after this really hard, assuming the following load is high enough.

Just a few thoughts.

[Steve Conner]

The output impedance can never be higher than the plate load resistor, 100k in this case. So a 250k pot will only make a dB or two of a difference.

[cminor9]

HTH did you ever get this worked out? I have a Mesa boogie vtwin pedal that has never done anything for me. It would, however, make a killer shell for a real pedal and I have been thinking about a cascode design. Funny because I was thinking of grabbing the front end of an existing amp like you did. But if you could never get this to work maybe I'll look elsewhere though ax84 seems like a good place to start.

[imaradiostar]

If you're planning to make a high voltage pedal I have a few ideas:

1) something like an SIB varidrive- fabulous pedal. Fet into a single 12ax7. very usable distortion sound.

2) something like the H&K tube factor- once again, two gain stages, neat distortion tone that's really useful. It could easily be modded with a more useful tone stack.

3) 12ax7 cascode with a mosfet source follower- isolates the cascode from following loads. Put any of a number of interesting gain stages after that.

4) a triode into a pentode (6U8 or 6GH8a) to make the pentode sound more like a pentode....and possibly a marshall style gain stage and cathode follower afterward

5) 6dj8 cascode on its own, possibly with a low gain jfet in front of it to hit it a little harder. Not so much a distortion as a harmonic/tonal change box.

That's my short list of high voltage pedals I'd like to try to build!

jamie