Valve mic preamp design incoherent rambling

[Winston OBoogie]

I think the use of pentodes goes hand in hand with superior winding techniques. It allows the use of high ratio xfmrs, cause low Miller cap.

Without a doubt. Some of those input transformers were 1:30 and still had excellent frequency response.
I enquired about purchasing a couple of 1:10's (CTV72/1) from a company that have the documentation and means to wind them still, and it 'ain't cheap! About 4 -5 times the cost of a very nice Lundahl. But it's a 5 chamber bobbin, not your typical jobbie, so that stuff costs money.
Too rich for my blood though.

I would like to see a comparison of a pentode with a cascode. Noise ought to be better with the latter... I don't see anything against cascode, except maybe the need for slightly higher B+.

Yep, your B+ would generally be higher by your artificial screen grid voltage so, probably in the 60v - 90v region.
Pick your operating point/anode load using triode curves, then shift everything to the right by the cascoding voltage for your pentode "knee".

Using inexpensive pentodes (6BR8) strapped as triodes, I was seeing about 2 - 3dB better noise figure
(over a single pentode) with 2 of those triodes on the bottom of the cascode. No partition noise in the equation now of course, so a little of the improvement no doubt came from there.

 

[abbey road d enfer]

Using inexpensive pentodes (6BR8) strapped as triodes, I was seeing about 2 - 3dB better noise figure
(over a single pentode) with 2 of those triodes on the bottom of the cascode. No partition noise in the equation now of course, so a little of the improvement no doubt came from there.

That's the beauty of cascode. You can use very high Gm tubes (or several tubes in parallels) and not worry too much about Miller.
I propose we refer to cascode as "the Miller Lite".
OK, I"m out...

 

[ruffrecords]

I think one of the UA mic pre designs used a cascode first stage.

Cheers

Ian

 

[abbey road d enfer]

The Brick also, but with a FET in the bottom. and a nuvistor on top. Never seen such a piece of c..p. I never understood why Terry Manning has so many of them (and uses them)...

 

[Winston OBoogie]

A couple of J-Fets (2SK-170) on the bottom of a 'Miller Lite', and you've wiped the floor noise-wise with tubes. Very easy to set up, just keep the drain/source V a couple of volts below 15.
Can't say a nuvistor would spring to mind when picking a tube for the top though? Strange choice.

Half a 6922 maybe? A triode strapped E180F looks pretty good - They're still cheap, high Gm, and they're small beefy little things.

The 'Budweiser Effect' not getting any mention in this?

Edit: The UA cascode was the 1008. But my humble opinion is that a 12AX7 is not a valve particularly suited to the job.
Better a 12AU7, or another 12AY7 like the output tube.

 

[ruffrecords]

Edit: The UA cascode was the 1008. But my humble opinion is that a 12AX7 is not a valve particularly suited to the job.
Better a 12AU7, or another 12AY7 like the output tube.

Definitely not a 12AU7. A cascode creates enough distortion by its very topology without compounding it by using a tube with the worst ever intrinsic distortion.

I have toyed with the idea of replacing the half a 12AX7 in my standard mic pre circuit with an ECC88 based cascode

Cheers

Ian

 

[Winston OBoogie]

Definitely not a 12AU7. A cascode creates enough distortion by its very topology without compounding it by using a tube with the worst ever intrinsic distortion.

That's fair enough. I was just suggesting that, within the scope of what UA were doing with the 1008, and looking at the usual suspects in tube types that folks were using at that time, a 12AU7 would make for a better cascode than a 12AX7.

Actually, I found that cascading allows some leeway in terms of playing with distortion type,

Unless I've lost the plot You could have the cascode voltage referenced to cathode so it tracks signal. and a lot of the high frequency distortion you can start to see appear with higher impedance sources such as from a 1:10 input transformer can be reduced.


In fixed gain amps, you could also have what's called a driven cascode where the amplifier output drives the "screen grid" with the same ratio of signal as your negative feedback.

Fine tuning the cascoding voltage is useful too as you can put the load line wherever you see fit around the knee.

Easy to do Ultra Linear by deriving the cascode V from the anode...



Lots to try and keep us tube folks busy anyway

Since the E88CC/6922's raison d'être was for cascoding, that's a good one to play with for sure.

 

[ruffrecords]

The main 'problem' with the cascode is that the top tube represents a much lower impedance load to the bottom tube than the usual plate resistor. The voltage gain of the bottom tube is therefore very low, which is why Mr. Miller is not a problem, but this lower than usual load is also what creates the distortion.

Cheers

Ian

 

[ricardo]

> I think the use of pentodes goes hand in hand with superior winding techniques. It allows the use of high ratio xfmrs, cause low Miller cap.

Without a doubt. Some of those input transformers were 1:30 and still had excellent frequency response.

It's not the ratio that causes problems but High Design Impedance. (This isn't the 150k recommended load but the Impedance Ratio, eg 150:15k on Jensen JT-115K-e1) Lundahl have a note that explains why they don't do trannies with >20k (IIRC) windings.

The design process for trannies goes something like this. You pick a LF cutoff (*) and this gives you the primary inductance you need for your primary L based on your source R. eg 200R 30Hz : L = 1.06H

You work out turns, devoting half your bobbin(s) to the primary and the other half to the secondary. For best results, you want the Cu losses about equal but this is often not possible cos wire sizes.
With loadsa turns & HiZ, the leakage capacitances become important and you have to use fancy sectioning etc to get good HF results. This is quite independent of what the rest of the circuit is doing ... triode, pentode, cascode, FET bla bla

There's RDH 4 and also Great Guru Baxandall's stuff in his last word on LN design. Get it all before it is gone for good.

(*) actually L in an 'iron' trannie varies with level so this is just a ballpark figure. What IS 'constant' is level / (THD * f ) So you pick your core size to give eg 3% THD @ 50Hz 1V

 

[Winston OBoogie]

The main 'problem' with the cascode is that the top tube represents a much lower impedance load to the bottom tube than the usual plate resistor. The voltage gain of the bottom tube is therefore very low, which is why Mr. Miller is not a problem, but this lower than usual load is also what creates the distortion.

Sure, it's not a free lunch all round, and none of the ideas I mentioned above will address that.
I don't know of a comparison with cascode vs pentode re distortion. If I did any comparisons myself, it's been so long ago I forget the takeaway bottom line. Anyway my instinct tells me it's similar and it's been OK for me when I used either.

Worth a shot playing with Ian, worst case is you learn that you prefer the Miller of a triode.


Edit: Pete Millet has a mic amp design with an E88CC cascode. One criticism I have of his design is the seemingly matched loading of the mic with his secondary pot termination. But I put that down to his misunderstanding of impedance with regard to mics (pro gear isn't his usual territory), his tube design is usually well engineered stuff.

http://www.pmillett.com/micpre.htm

 

[Winston OBoogie]

It's not the ratio that causes problems but High Design Impedance. (This isn't the 150k recommended load but the Impedance Ratio, eg 150:15k on Jensen JT-115K-e1) Lundahl have a note that explains why they don't do trannies with >20k (IIRC) windings.

Yep, probably should have been clearer in how I gave the example.
The 1:30 ratio transformers I was referring to were for 200 ohm source z so, 180K reflected. Specifically, the units within the V76 mic amplifier.
Loadsa primary inductance on those coils and, highly sectioned affairs to combat the C.

Personally, I'm OK with a good 1:8 (Lundahl Ll7903 looks very good to me) as the inherent resonance of the transformer can be a couple of octaves above audio. But of course, I get about a 2dB noise penalty over a 1:10 hence my recent messing about with a parallel pair of triodes on the bottom of a cascode to recoup that.

There's RDH 4 and also Great Guru Baxandall's stuff in his last word on LN design. Get it all before it is gone for good.

Both of those are great references. Peter Baxandall was in a league of his own.
Cheers Ricardo.

 

[solkatten]

I have fiddle a bit with a russian planar triode 6S17K-V. Gain: ca. 200. Very low microphonics and noise. It has some miller C and high GM. It can be loaded with anode resistor of 27K. Maybe that would make a good bottom tube in a all tube cascade. The only tricky part of this tube is the shared filament and cathode connection.

 

[Winston OBoogie]

... Russian planar triode 6S17K-V...

Scavenging parts from a MiG-23 guidance system Solkatten?

 

[solkatten]

Scavenging parts from a MiG-23 guidance system Solkatten?

Yep they use to sit in the radar navigation system of soviet missiles.

 

[solkatten]

6S17k-v can be grid biased at around 47K, with the cathode grounded. A very quite front end. If you want to combine it with higher step up trafo you could use filament bias with a 2,2 Ohm heather +cathode resistor, but you need a really quite filament supply and start to consider miller c.

 

[Winston OBoogie]

6S17k-v can be grid biased at around 47K, with the cathode grounded. A very quite front end. If you want to combine it with higher step up trafo you could use filament bias with a 2,2 Ohm heather +cathode resistor, but you need a really quite filament supply and start to consider miller c.

Interesting. 1:5 is probably the right ratio for that circuit then.
The only other example I see that uses the 6S17K-V is with it as the top device in a hybrid cascode with a J-Fet on the bottom.
This guy has it as the front end of a phono pre:

https://www.bartola.co.uk/valves/2015/08/16/lcr-phono-design-notes-part-iii/#more-4817
Are sockets available for it or did MiG and Ministry of Defense buy them all?

 

[solkatten]

Interesting. 1:5 is probably the right ratio for that circuit then.
The only other example I see that uses the 6S17K-V is with it as the top device in a hybrid cascode with a J-Fet on the bottom.
This guy has it as the front end of a phono pre:

https://www.bartola.co.uk/valves/2015/08/16/lcr-phono-design-notes-part-iii/#more-4817
Are sockets available for it or did MiG and Ministry of Defense buy them all?

No sockets but u can solder directly on the tube. Yepp I have seen Bartolas articles.. great design if you need 60dB of gain...

 

[Winston OBoogie]

No sockets but u can solder directly on the tube. Yepp I have seen Bartolas articles.. great design if you need 60dB of gain...

Got ya on the sockets.
Yeah, if you're up for going 'rogue', 'off the rails', and throwing some solid-state in with your tubes (I've no problem with it), then Bartola has some interesting stuff.

 

[solkatten]

I already have a plan for a grid bias 6s17k-v input stage with Lundahl 1538 1:5 trafo into CF and anode choke (I got 2 insanely big 200H 20ma at hand) parafeed line stage. Im planning on using 6n6p wich I like very much. PSU is around 200v so its hard to implement any cascade in that design... And I think without varible feedback to much gain becomes a problem. My plan is actually not to bypass the cathode resistor of the line stage to keep gain down a little. But I have nothing aginst solid state...

 

[solkatten]

Got ya on the sockets.
Yeah, if you're up for going 'rogue', 'off the rails', and throwing some solid-state in with your tubes (I've no problem with it), then Bartola has some interesting stuff.

That could be a cool idea for active Ribbon preamp 6s17k-v are tiny tubes and the ribbon neads a lot of gain