new jc circuits

All the parts have to be put together at some point. Have fun!
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

Oh... One additional point.... The "original" used the anode as reference point for the CF stage. That provides DC feedback, and stability for high gain circuits.

The "rev 2" biases the screen from the supply... No DC feedback. I think you will find in practice, that will be impossible to obtain a stable operating point with variation in parameters etc.

The alternative proposed last in this string used the "best of both worlds" the follower is returned to the anode, and the pickoff point is also the anode. That'll give you improved FB.

Stephie <3
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

Interesting circuit, tried simming it during my lunch-hour. Compared it with a "gyrator"-fed conventional D3a. Used a depletion-FET instead of the BJT as this seems to give the same result.

Without the 47k(last resistor to the left in pentofax schematic) loading, gain as expected was higher for the Gm with addition of factor Ia/Ig2. But when both circuits was loaded with 47k the difference was down to a few dB. Did I miss something in my sims? Can post the *.asc when I´m home from work. Could maybe also be interesting to see the simmed fourier-analysis.
Brgds
Lars
nanana
Posts: 143
Joined: Mon Jun 14, 2010 7:34 pm
Location: malmö sweden

Post by nanana »

of course, its preamps where frank's trick really shines... i have been using the sim to do different things and large signal performance is limited by the circuit. the noise thing is interesting but its not that big of an improvement over the straight up "Gm amp". 3 or 5 dB. but the gain... that is impressive. its easy to get 72 - 78 dB from one stage! combined in a Gm amp, where you can shape the bandwidth, its not that hard to get 65 dB of gain at 200kHz WITHOUT feedback! that is performance. if you do add some, it works really well, but you don't have to.

for amplifiers, the Gm amp can swing crazy volts at high frequency. you can make cascode Gm amps too. the last 211 amp i made, i use one starved fet/5755 cascode Gm amp for a gain of 7500 driving a 6AV5 cathode follower with feedback around the driver stage to adjust the gain. about 40dB for a 211. -3dB at 270kHz into 10k ohms (the grid resistor of the 211). it made the 211/output transformer stage the primary poles as the driver stage had at least 2 octaves on either side of it. so stable and really detailed. i don't usually work so "classically" as i prefer to really use the individual parts to fit into careful and particular analog combinations... still. these pentode circuits are thrilling in what they can do...

no, they are NOT the only way to do things... but they are a very good solution!
my skin is thin, my skull is thick.
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

These are the files. Doublechecked FET against BJT and no big differences when simmed.
Attachments
gml.asc
(5.96 KiB) Downloaded 125 times
gml.png
gml.png (95.25 KiB) Viewed 8428 times
Brgds
Lars
nanana
Posts: 143
Joined: Mon Jun 14, 2010 7:34 pm
Location: malmö sweden

Post by nanana »

hey lars, i'm not sure why you have such resistance to the resistor... but it does play an important role in the function of the circuit. have a look at the simplified block diagram. once you see this, you see why the gain goes up and down with the magnitude of that R, and why the circuit doesn't work so well without it.
Attachments
Gm amp037.jpg
Gm amp037.jpg (61 KiB) Viewed 8422 times
my skin is thin, my skull is thick.
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

The ".step param" was to compare gain with only 610k load against 47k. Off course there should be a load on a pentode, as this could/should be looked upon as current source.

And as I said my previous post, there is no big difference in gain, only 4dB, between the two when loaded with 47k. With 610k, 13dB.(In my opinion the D3a does normally work even better with harder load, sacrificing gain, at its sweet spot 150V/150V/20mA.)

Neither does the .four indicate any big differences. You find the .four results under View/Spice Error Log. Only works when running .tran .
Attachments
gml.asc
(5.3 KiB) Downloaded 140 times
Brgds
Lars
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

An "El Cheapo" example I made in Dave´s "Pentode phono"-thread a while ago.

This single stage RIAA might be bettered with the Gm-tecnique added.

A CCS might also substitute the "gyrator"(fixed voltage AC-CCS) to simplify.

Image
Brgds
Lars
nanana
Posts: 143
Joined: Mon Jun 14, 2010 7:34 pm
Location: malmö sweden

Post by nanana »

here is a pic of the "cheap" phono pre from above, for those who don't have spice. this is using a E80CF/6BL8, although any common dual pentode triode will work. 6AN8 or 6EA8, etc. the other tube is an 8416. this is super stable and easy to build... i like stacking power... i love it for amps best. there's something perverse about floating a 1300 vdc @ 250mA supply on top of a 220 vdc @ 12 mA supply made from a bacwards connected heater transformer... it just turns me on! this idea easily translates to amps. the old dinosaur dc 50 amp used a variation of this... i used a bendix dual diode for the floating ref. i only needed 2uF decoupling with that.
Attachments
dc phono.gif
dc phono.gif (231.94 KiB) Viewed 8389 times
my skin is thin, my skull is thick.
nanana
Posts: 143
Joined: Mon Jun 14, 2010 7:34 pm
Location: malmö sweden

Post by nanana »

ok. so this next circuit falls under the controversial heading of "high performance" tube circuits. this is not to say that audio isn't high performance... just that the performance necessary for audio is often not up to snuff for many other applications. sometimes we need more.

i am working on some tube based (there is some hybrid involved) voltage to frequency and voltage controlled oscillator designs that will end up in an analog synthesizer i am making. opamps are the most economical way to accomplish these functions, although there are many clever discrete resources to draw upon as well.... most digital techniques were created with tubes, remember. the first opamps were of course made with tubes, 12AX7s to be exact. i am making some higher performance circuits, but there are issues.

one is DC offset. the output of all amplifiers is level shifted from the input. getting the output to be the same DC level as the input is crucial for many applications. also, and more to the point, feedback around the opamp, which is so crucial to getting the various functions needed, is complicated by DC offsets even more than what the opamp is being coupled to...

so, level shifting circuits have been an obsession of mine for a few years now. you can see from the previous circuits that a lot of thinking has gone into simplifying DC coupling between circuit elements, even if there are transformers or caps at the input and output.

i need a tube gain block with a gain of 1 million and low DC offset between input and output. its coming, but not there yet.

on the way, i have come upon some ways to do DC feedback that don't require level shifting. this is oldschool analog computing stuff but i have incorporated a Gm amp into the middle of it. this is a current feedback gain block with an open loop gain of 90dB. 40 dB closed loop. check it out. this gives solid state discrete a run for the money, even though its big and hot and eats 300 volts for dinner.

i have made one of these with an E80CF and a 6922. this one is using better parts but does the same... in spice of course.
Attachments
here is the circuit
here is the circuit
tube opamp.gif (244.17 KiB) Viewed 8361 times
opamp sweep.gif
opamp sweep.gif (29.56 KiB) Viewed 8361 times
opamp fft.gif
opamp fft.gif (27.93 KiB) Viewed 8361 times
here is the compensation technique
here is the compensation technique
tube comp opamp.gif (232.97 KiB) Viewed 8361 times
op comp sweep.gif
op comp sweep.gif (29.32 KiB) Viewed 8361 times
my skin is thin, my skull is thick.
nanana
Posts: 143
Joined: Mon Jun 14, 2010 7:34 pm
Location: malmö sweden

Post by nanana »

ahh, the order is backwards... anyway, if you look at the uncompensated sweep, you can see that four tubes gets one flat to 20 MHz with a simple peak. just 2 pF between the plates pulls it down nicely. layout controls the bandwidth. the precision is interesting. it stays 40 1/2 dB even if you change tube types at the front and back. this is good for charge pumps. and many other things as well... like preamps. because it swings a hundred and something volts pp with .02 % thd out to several MHz, you can drive a 10 and then an 845/GM70 as well. not exactly audio... but it can be.
my skin is thin, my skull is thick.
nanana
Posts: 143
Joined: Mon Jun 14, 2010 7:34 pm
Location: malmö sweden

Post by nanana »

here is the opamp open loop... this is useful also if you need a MC or ribbon mic preamp. 100uV in gives about 8 volts pp out! noise is an issue. the 417A is about as good as tubes get in terms of quiet, but 100u is getting close to the johnson. you'll hear it no matter what you do. a transformer in front with a 1:3 will help... then use a step down at the output.
Attachments
opampopenloop.gif
opampopenloop.gif (126.49 KiB) Viewed 8342 times
my skin is thin, my skull is thick.
nanana
Posts: 143
Joined: Mon Jun 14, 2010 7:34 pm
Location: malmö sweden

Post by nanana »

this is a very stable circuit for such insane gain... which is of course why it works even better closed loop. only 2 out of phase gain stages. one of the areas i am working is using 2 of these amps in a feedforward (adding what is lacking) circuit with one more Gm amp. the gain is 130 dB but the stability isn't there. not quite there yet. the level shifting is the problem. i can fix it with a SS opamp servo, but that's cheating. i want to know that it was possible in 1960, to do what i am doing now... how's that for sick?

still, modding test equipment is sicker.
my skin is thin, my skull is thick.
nanana
Posts: 143
Joined: Mon Jun 14, 2010 7:34 pm
Location: malmö sweden

fear and loathing driver

Post by nanana »

okay, so i have had some interesting discussions on the joenet lately, and its brought up all sorts of deep sick thoughts. one thing that's come up is new/old awareness of general suspicion towards certain circuits, and of course circuit techniques... cathode followers are clearly the four letter words of the diy audio world. yes, that is clear. "un-necessary stages", is another interesting concept that i am feeling a lot of rough sentiment out there for... as if the no watt amp thing, the one stage amp thing, the one speaker thing, the DHT thing, and the no C thing, the no R thing (iron and tubes only, but of course!, thanx stephie!) and the "can't triode strap a beam tube" thing was all enough. so pretty much, the only thing you can build now with any cred at all is a 45 or a 46 amp, with iron i/o and a glo tube for filtering, and a mercury diode for AC/DC... okay, you can use a 71A too,,, well, that is fucking BORING!

i've decided, for my own sanity, to start a line of circuits i will call "fear and loathing" circuits. the criteria here is to make great circuits using all the most unpopular techniques and parts possible. call it my response to the "feral eye" (sorry herb, not making fun of you, really). i will eventually have a contest, maybe here in malmö, where a kit of parts will be made available using weird voltage TV and regulator parts for the contestants, and the whole point will be to DESTROY a perfectly constructed status quo triode amp with a "full range" speaker or horn setup as reference...

well, you gotta start somewhere, and here it is... this is watered down from work i'm doing for silbatone on a big pp AB2 GM70 amp. its a dirty job, but someone has to do it!

this is the fear and loathing driver. next, i will directly couple it to a CATHODE FOLLOWER. a down and dirty tube... a russian KT88. yes, surely there isn't a more hated tube. i think a sylvania 33JR6 has more street cred than that! it uses a 6V6 in triode and an EL-34 (i will build it with a 6AV5) for a gain of 70dB. it will swing 600 volts PP at 10% distortion open loop. i will use negative feedback around the driver to reduce the gain to 40dB (100), that is 1 volt AC in, 100 volts AC out. you can see the simmed results yrself.
Attachments
fear and loathing fft.gif
fear and loathing fft.gif (31 KiB) Viewed 8292 times
fear and loathing swing.gif
fear and loathing swing.gif (134.82 KiB) Viewed 8292 times
fear and loathing driver.gif
fear and loathing driver.gif (82.56 KiB) Viewed 8292 times
my skin is thin, my skull is thick.
nanana
Posts: 143
Joined: Mon Jun 14, 2010 7:34 pm
Location: malmö sweden

Post by nanana »

a couple of notes: the second stage is a Gm amp. the very important choke load allows it to swing the sick amounts it can with less distortion and above the rail. the 20H will have to be gapped for 5 mA. the swing doesn't go down much below 200 VDC as the screen kind of starts to put on the brakes. still, 215 to 920 is alot of potential swing. its distorted, but not as much as it would be if it was an EL34 in triode, suckas! the 6V6 has amazing low current performance in triode. it was used in analog computers for this very reason... direct coupled amplifiers no less. the buffer for the Gm amp screen is a 6C45P. i have too many of them.

yes, i'm a contrarian punk. unrepentant.
my skin is thin, my skull is thick.
nanana
Posts: 143
Joined: Mon Jun 14, 2010 7:34 pm
Location: malmö sweden

Post by nanana »

here's what the amp could be...
Attachments
fear and loathing amp.gif
fear and loathing amp.gif (228.93 KiB) Viewed 8290 times
my skin is thin, my skull is thick.
nanana
Posts: 143
Joined: Mon Jun 14, 2010 7:34 pm
Location: malmö sweden

Post by nanana »

here is a slightly more flushed out version... could be tweaked quite a bit more but... i tried to keep the B+ under 800 (the "death zone")

i still don't understand all the ins and outs of spice and transformers... some things don't make sense. but dave has a bigger clue than i. i know this works and it sims okay... but? LT spice gets weird when there are several power sources and several grounds...

just under 10 watts into 10 ohms... if the spectrum is anything like reality, i would imagine this amp to work fine.

the driver stage is interesting and i am going to make a version of it.

i am interested in huge linear swings with low distortion. hard to do it and tubes have an advantage over solid state... its not every area you can say that.

jc
Attachments
fear and loathing swfft2.gif
fear and loathing swfft2.gif (54.37 KiB) Viewed 8282 times
fear and loathing amp2.gif
fear and loathing amp2.gif (229.7 KiB) Viewed 8282 times
my skin is thin, my skull is thick.
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

About 10 years ago I played with outputs in the cathode to see what It was all about. I was using the Tango NY-15's (3K5) and with both a 300B and the 50 I preferred the sound of the transformer in the plate.

I always attributed the sound of the cathode connection to the use of the cathode follower but given the 20:1 stepdown, I suspect my driver was running out of swing.

dave
Get Your Fix
www.hifiheroin.com
nanana
Posts: 143
Joined: Mon Jun 14, 2010 7:34 pm
Location: malmö sweden

Post by nanana »

yes, that is the whole issue with followers... its all about the swing. but, it is very forgiving in terms of driving it. and if you have a driver that can swing like crazy but mustn't get loaded down...

one enormous advantage of the trans coupled/step down scheme is how it deals with back emf from a speaker... the current has Rp "behind" the transconductance of the amp (1/Gm) intead of in "front" (Rp*mu)... which is what i've been trying to tell you about the split load the whole time. the cathode is still more of a short circuit than the plate, even with unity gain feedback... even if it isn't much of a short circuit.

the grid is closer to the cathode than it is to the plate...! the grid really controls Rp, not the cathode. the plate can be modified to add or subtract from Rp very easily.

with the step down trans, you have better "damping" than is possible at the plate, and the buffering effect of the Gm means it doesn't effect the grid, which it does when pushing at the plate (changes Rp).

overload is very forgiving, especially with tubes (high B+).

anyway... its all interesting. i kind of want to try that step up trans/cathode follower thing again... it had amazing bass. it is absurd, of course! all of this is...

my kind of absurd anyway.

thorsten loesch is an asshole...
my skin is thin, my skull is thick.
Zelter
Posts: 1
Joined: Thu Sep 01, 2011 7:16 pm

Hi all,

Post by Zelter »

I´m new in this forum. Great stuff. Who made the Spice modells for all the tubes? Where could one get a copy of the models.

Dieter
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

They are collected from a bunch of sources. I made some, others made some.
They are in the "sticky" section in the "sim city" section

Stephie <3
anders80
Posts: 1
Joined: Wed Feb 27, 2013 5:03 pm

Post by anders80 »

What is the trans/cathode follower?
nanana
Posts: 143
Joined: Mon Jun 14, 2010 7:34 pm
Location: malmö sweden

Post by nanana »

it was short hand for transformer driven cathode follower....
my skin is thin, my skull is thick.
Post Reply