• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Morgan Jones and cascodes

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
The EF86 is atypical for a pentode. It's very linear, with just a bypassed screen B+ dropping resistor. With regulated g2 B+, I think it will embarass quite a few triodes. See Eddie Vaughn's remarks here.

Ray, are you going to use a Dyna or a Mullard circuit? For the Dyna topology, a 5687 or ECC99 would be NICE as the phase splitter.
 
Hi lndm,

Originally posted by lndm Does the nonlinearity of a pentode limit its use to small swings in the same way as a cascode?

Not necessarily, it simply depends on the actual situation. Chosing the best fitted pentode for the task at hand and letting it work at a carefully chosen operation point can do little miracles regarding gain, swing and linearity.

Generally, small signal pentodes, like their big power brothers, want defined and constant loads for linearity. So, for example, driving a volume pot or baxandal tone stack is not what you want an EF86 to do, except you are out for distortion.

If you want linear driving capabilities into changing or severe loads from small signal pentodes, simply follow them up with a DC-coupled CF stage. Still easier to design and handle for the average DIYer than succesfully designing a cascode, I think.

Tom
 
ray_moth said:
A pentode has some practical advantages, including no second heater at elevated voltage and no sharing of the B+ across two tubes.


You don't need to share anything if you DC decouple common cathode stage and common grid stage.
 

Attachments

  • cascode.jpg
    cascode.jpg
    5.5 KB · Views: 457
Administrator
Joined 2004
Paid Member
I'm not a big fan of the Quad II amplifier although it does have a lot of adherents and sounds quite nice with the Quad ESL-57 And...

You guessed it... A pair of EF86's in the driver stage, in quite a novel arrangement, IIRC I think Peter Walker even got a patent on the topology - and the EF86 will do large signal swings with reasonable linearity, to the tune of at least 120Vpp to drive a KT88..

I haven't looked at a schematic recently so I don't remember all of the details.. I designed a similar circuit using 12AX7A and 6BQ5 years ago looking for something "clever" and inexpensive to mass produce as a oem'd cheap entry level 20W pp mono block.. Never made it to production, performance was promising though.

Some of the European telco pentodes ought to be interesting as an alternative to the cascode.. (Like C3M, D3A...)
 
Don't forget Telefunken EF804s, another awesone small signal pentode.


>>>If you want linear driving capabilities into changing or severe loads from small signal pentodes, simply follow them up with a DC-coupled CF stage. Still easier to design and handle for the average DIYer than succesfully designing a cascode


Agreed. A thread here discussed a pentode aikido to drive 300B. Just do a search.
 
Hey-Hey!!!,
The cascode is absolutely an acceptable pentode substitute. With a 6H30 triode pair on the bottom, a pair of IRF820B as the screen, you have a very nice high gm faux pentode. In a diff amp/LTP they do quite well.

If you want FB in a pentode amp, do the E-Linear thing. High gm/high plate z pentodes do the best, and the faux pentode an even better one for LTP building.
cheers,
Douglas
 
Some thoughts

No, Indm, an EC91 is a far sharper cut-off triode than a triode 6AQ5. E.g. at 250 Va the EC91 will draw 6,2 mA Ia for -2V grid bias, where the 6AQ5 will need about -30V bias for that Ia. My data shows the EC91 to have Gm of 8,5mmho at Ia=10mA and Mu=100 (Va=250V). I unfortunately do not have similar data for 6AQ5, but rough equivalency guess would be about half the Gm and a mu of only some 15 max. This apart from plate dissipation etc.

Generally, the 6AU6 was primarily intended as an RF amplifier, so microphonics are not inspiring. Our local supplier never had a shortage of no-name (oops - be careful) Russian EF86s, which I found to be OK on testing. I have not used them long enough to be able to comment on life.

The cascode is supposed to have lower 3rd harmonic distortion and noise for the same output as a pentode, but I have never confirmed that. But one needs a load resistor quite higher than rp to avoid distorton in the lower triode because of the low load it sees. A value of 2.mu.rp has been suggested, making Rl about 400K for a low mu triode. (mu about 450). I am aware of only the Connoisseur amplifier using a cascode input stage in the 50s, but there probably were others.

I would agree that the higher noise for a pentode is academic in practice - it is hardly noticeable when putting your ear right into a loudspeaker in my experience. But there is one problem. Data for the 7199 pentode shows quite a distortion variation with different Rl/Rg2 combinations, and also quite a variation between different tubes (classified as low, medium and high gm), which behaviour I believe to be general for pentodes. (I will put graphs up in a day or 2 when scanned; I have them on paper, perhaps they might be on the net.) I myself am fond of stabilising the operating point by dc feedback from a following triode cathode (cathode follower, concertina or ltp phase inverter) to g2, It helps a bit but a large Cg2 is required to prevent signal feedback.

Peter Walker did not have a patent for his circuit as far as I am aware, although he used distributed load (UL) output before the Hafler and Keroes claims.

Ray_Moth: Wishing you well in your venture! (Somewhat off-thread, remember the hint I posted on another thread about using a separate output transformer winding for feedback only. Being unloaded it avoids the influence of leakage reactance. The fact that the speaker winding is then non-feedbacked is of little consequence according to the source - your output transformer is still included in the feedback loop.)

Regards.
 
Johan Potgieter said:
But one needs a load resistor quite higher than rp to avoid distorton in the lower triode because of the low load it sees. A value of 2.mu.rp has been suggested, making Rl about 400K for a low mu triode. (mu about 450).

Yes, as far as I know the cascode (with two identical devices) has a stage mu of mu^2, and a total rp of mu*rp suggesting 2*mu*rp as a minimum.

I have watched people use Ra=3*rp (myself too), then claim the cascode has a limited swing. Could this be the issue?
 
Hey-Hey!!!,
One can treat a coacode just as a pentode. I do recommend staying away from plate voltage below Ec2, or twice the upper valve control voltage if the same type is used top and bottom.

One drives the load line out through the horizontal plate lines at g1=0. For a high gm lower valve, this can be a low value resistor compared to the plate Z( just like a penotde ).

If one takes a *VERY* low internal resistance device for the upper section, one can go closer to that section's control voltage. When a MOSFET is the upper device, one can go *VERY* close to the gate voltage...:)

You still treat it as a pentode for gain, gm*R_load

cheers,
Douglas
 
Doug,

A cascode with a FET on top is of interest to me. A sad fact is that the 12BY7 will become extinct, all too soon. That fact is of concern to H/K Cit. 2 owners, like us. If the 2 sections of a 12AT7 are paralleled, the net gm is ever so slightly higher than that of the 12BY7. 200-220 V. on the plate and 3 mA. Ib per section is a SWEET spot for the 'T7. I'm thinking of a ZVN0545A enhancement MOSFET on top of paralleled 'T7 sections. 6 mA. of drain current is well within the little FET's capability. DC couple the cascode to a cathode follower (either a 6FQ7 or 12BH7). Use the 2nd section of the med. mu twin triode as the pass device in a B+ regulator for the cascode. Voila, a replacement for the Cit. 2 voltage gain block. Our "old friend", a Schmitt circuit, will take care of phase splitting duties.

Does the concept appeal to you? Can you suggest an alternative FET?
 
Thanks Douglas,

The pentode rp/Ra relationship was lurking in the back of my mind there. The more I learn, the more I just want to call a cascode a DIY pentode :cool: (old news to some, I guess)

I have a Brimar EF86 sitting in a corner, hmmm...what can I do with it :D
 
Hey Eli,
Good to hear from you!

I have tried the 12GN7 and not liked the results all that much in what ever I have used it for. There is another nice 12BY7 alternative that also happens to work well in the Cit.II, the 12HL7. They are also only $2-3 each. The extra gm doesn't seem to bother it much, and it sounded good to me.

I just went to an antique radio swap meet up in Lansing, and scored about 20 12BY7 for $1 each. I think I have enough to keep any Deuce I run into up and running.

Your FET idea would certainly need a slightly different circuit arrangement for the upper gate, as the g2 in the Deuce counts on a bit of current draw to generate the proper voltages.

I am actually going to tear mine up sooner or later, and wind some output clones with a few pairs of g2 taps for use in my E-Linear 813 amp.
cheers,
Douglas
 
Eli,
Watch out with those ZVN0545A - They are stunning in the right place. I use them to switch a 300V gating pulse to a photomultiplier tube in <50ns. They have wonderfully low gate capacitance and a 450V rating, 90mA continuous or 600mA pulsed BUT ONLY 0.7W maximum dissipation. Its that power rating which may be a problem. They also have a P-Chan "mate", the ZVP0545A. In the switch circuit I mention above I use thm as a complementary push pull pair.
Cheers,
Ian
 
gingertube said:
Eli,
Watch out with those ZVN0545A - They are stunning in the right place. I use them to switch a 300V gating pulse to a photomultiplier tube in <50ns. They have wonderfully low gate capacitance and a 450V rating, 90mA continuous or 600mA pulsed BUT ONLY 0.7W maximum dissipation. Its that power rating which may be a problem. They also have a P-Chan "mate", the ZVP0545A. In the switch circuit I mention above I use thm as a complementary push pull pair.
Cheers,
Ian


Ian,

If Id = 6 mA., then Vdss = 50 V. is quite safe. Use a "top hat" heat sink on the TO92 case along with good air flow. Work out the drops across the load and cathode bias resistors. Add 260 V. for the 'T7 and the FET. "Raw" B+ in the Cit. 2 is approx. 475 V. Seems to me regulating it down is not a matter of great difficulty.
 
The ZVN/ZVP series of FETs from Zetex are in the peculiar flattenened "super E-line" package that looks like a TO-92 that's been squashed by a steam roller. Normal TO-92 type coolers won't fit on them. However, it would be possible to use thermal epoxy to glue the package to a heat sink and get a little more dissipation capability - not pretty, but beauty isn't everything. The Supertex series of small signal MOSFETs come in standard TO-92 packages, though they may not have a direct cross to the Zetex FETs. One can check at the Supertex web site to see. If there is an alternative from Supertex, it may be possible to order it through Mouser, who stocks some of the Supertex discrete line. Supertex used to offer their small signal devices in TO-220 packages as well as TO-92, but sadly, they stopped doing that quite some time ago.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.