The Janitor (on the drawing board)

[RicharD]

It's been a while since I designed an amp from scratch.  There's been some chatter as to why we don't see 12AU7's in guitar amps so this seemed like an appropriate challenge.  My goal is a "clean" guitar amp with as much headroom as possible.  Since I'm totally broke, I need to work around as many parts (ie transformers) on hand as possible.  Attached is the schematic of what I came up with.  If you want to see all the design docs, click HERE .  I'd love some feedback.

Thanks!

-Richard

[softwarejanitor]

If you want a clean amp with lots of headroom google up the schematics for the Motorola HS-544, HS-619 or HS-621.  They were originally late 1950s console record player amps, but I've converted several of them into guitar amps.  They use a 6AU6 and two 12AU7s for their pre-amp/driver stages and a pair of 6V6 tubes for the power tubes plus a 5Y3 tube rectifier.  Stock they are pretty much completely clean all the way up to full volume, of course I've modified mine to make them get distortion at lower volume by increasing the value of the feedback resistors and swapping the 12AU7s out for 12AT7s or 12AX7s.

You could probably pretty closely duplicate Motorola's design using transformers made by Hammond.

[Geezer]

google up the schematics for the Motorola HS-544, HS-619 or HS-621.

Hmmmm, apparantly radio folks are NOT as "giving" as the guitar guys   

I searched for a while & all I could find were sites that charge a fee for membership or sell the files for $6.95 & up. 

Can you provide any links that us poor 'ole guitar players can afford (meaning, free)? 

G

[RicharD]

>apparantly radio folks are NOT as "giving" as the guitar guys

Most of those guys remember the days when you got schematics by purchasing Sam's Books.  They kinda still run with that.  I don't like paying for PDF copies of other people's work.  If I'm gonna pay, I want originals or I want to pay the designer.  We here at Hoffman are much more of a Free Information Society than is the norm.

>If you want a clean amp with lots of headroom google up the schematics for the Motorola HS-544, HS-619 or HS-621.

Copying another circuit is not part of this exercise.  This is a "start from scratch" routine, kinda like a crossword puzzle.  I don't want to turn to the back page for the answers.

Here is how I designed this circuit:
I had a general idea of what my goal was so I needed to find a power transformer first.  The one I have that has been scream at me to use for ages is a 700VCT @300mA with dual filaments @4A, 5V @ 2A, and 30V at 50mA.  Smells a lot like a good candidate for 26L6GC's.  I have a donor amp that can cough up the OT so I don't have to buy any big ticket items.  Next I pulled the data sheet for 6L6GC and designed verbatim the 450V typical operating condition.  It calls for -37V.  Using a bridge on the the 30V tap will give me 41V at best so I opted for a voltage doubler.  I set the current at 10mA so that the math is easy, 100 ohms = 1V.  Next the data sheet says I need 70 volts of swing to make full power.  I weighed a 12AU7 long tail vs. a concertina.  I can make more output with a LTPI but I think I can make just enough with a concertina (100V).  This left me 1/2 a 12AU7 bottle.  I made a simple gain stage with an AV of about 11.  It has an input threshold of about 10V.  These 2 stages combined have a gain of about 10 vs the LTPI which had a gain of about 6 with roughly a 6V input threshold.  The tone stack is basically 3 decade boxes and will change if I do go beyond bread-board.  The idea here it to minimize my gain loss.  Pots are cheap, but switching caps is better IMO.  So now all I'm left with is a V1 stage.  I can't stand to waste 1/2 a bottle and I'm not looking for OD so cascading stages is out.  I decided to parallel V1.  It looks like when you do this, you lose a lot of Hi frequency response.  My V1 grid stopper is is 10X less than normal.  Time will tell if this will sound OK or not.  A single 12AX7 triode will make an AV of about 50.  This parallel circuit makes a gain of about 70.  I don't really need the extra gain for a clean amp where as V2 is only looking for 10V.  With 1V of input, V1 can kick out close to 70V.  I decided to keep on running.  Next I totaled up my currents and desired voltages and plugged them into PSU designer.  I crossed a few I's and dotted a few T's and here I am. 

[stingray_65]

I'm looking at this amp and thinking it might be a great amp for an acoustic guitar.

How well do you think it would work with a Fishman preamp and pickup?

After poking around the schematic for a few mins, I think I have maybe 90% of everything laying around here too, PT and choke from a Baldwin organ that pushed a pair of 6L6GB's, a Hammond 1750M OT or a Mag Comp 40-18010, a bucket of Raytheon and Amperex AU7's, etc. think I might even have a chassis that may not be too holey.

Is S1, S2 and S3 ganged? I wouldn't think so, but sometimes I miss-read things like that.

Then there's the question, what speaker to put behind all this  that is Uberclean with late breakup and not harsh?

LOL more questions than comments right now

Ray

[phsyconoodler]

Kind of off topic a bit,but why is it that most of us creative amp building guys are broke all the time?

Collectively some of the best minds in the amp world hang out on this forum but we all seem to be poor in the pocket and rich on amp ideas.

[Platefire]

I would answer that but I'm so broke I can't pay attention!    Back to Janitor, the future cleanest PP 6L6 on the planet. My only comment is usually when something is super clean it's not too touch responsive---so I hope your working that in the recipe. Platefire

[RicharD]

S1, S2, & S3 are not ganged.  S1 = lo cut, S2 = a different kin of lo cut, or unbypassed, S3 = bright caps or treble.  All them 5M resistors are just to cut down on the switch pops.  As for touch response, or lack there of, that may just come with the territory of a low gain amp.  Maybe by paralleling V1, I'll get a little better response.  I dunno.  Honestly I've never built an uber-clean guitar amp.... well not intentionally.  I'm gonna try to bread-board this circuit this weekend and see what it does. 

[octal]

Neat design. If you wanted to go even more 12AU7 crazy, you could cascade both sections of a 12AU7 for V1, perhaps with a small voltage divider between to keep the gain down to that of a single 'ax7 section.

Alternately, you could use both sections of a 12AU7 in a cascode configuration: http://www.freewebs.com/valvewizard/cascode.html
for almost the exact same gain as a single 12AX7 section. I'm not sure that a faux pentode pre is what you'd want for an ultra clean amp though....

[DummyLoad]

i'm with octal - make it all 12AU7 for pre-amp.

--DL

[kagliostro]

Hi Richard

very interesting idea, nice project

as very often happens you are trying something innovative

you know I can't give you any suggestion about schematic

only a practical consideration, you planed to use rotary switch and put some 5M resistors to avoid pops

but why you planned 5 position rotary switch instead to use more standard 6 position switch

building the amp it will be easier to find it, or did you planned to use something that is in your junk box ?

Ciao

Franco

[RicharD]

>why you planned 5 position rotary switch

I guess I was think 1 handful at a time.  6 position switches would be better.

>I'm not sure that a faux pentode pre is what you'd want for an ultra clean amp though....

I've never heard a parallel triode called that..... I like it.  Actually, the reason I went with a 12AX7 for V1 is because sensitivity.  An AU has so much input threshold that you can't really come close to saturating it's input with a geetar.  I do think the parallel tube might be a bad idea, too much gain, hi end killer, & potentially noisy.  I just can't stand to leave an unused triode.

So...  the whole enchalada is going through a complete redesign anyway due to parts on hand issue.  ISOload pointed out that the PT I have on hand is deceptive in it's labeling.  It's unloaded voltage is 390-0-390 and it's rated for 300mA continuous.  Yea....  That more along the lines of 520VDC.  DummyTone also didn't want me to hack out the OT I was planning on using so he set me up with a fine.... very fine Dynaco transformer.  The output section is now gonna be KT88's run in ultra linear mode which opens more that a few cans of worms.  The bias supply is gonna hafta be a voltage tripler and I need considerably more input swing.  To get there I'm gonna hafta rethink my PI.  Either a higher voltage tube such as a 6SN7 or go to a LTPI appear to be my choices.  I stayed up way too late last night melting my hair.  I'm gonna hit it fresh later on.

Oh yeah, I think I bunged up the treble control.  Those caps should range from something like 50pf to 470pF.  Needless to say, I won't be building this today.

[softwarejanitor]

google up the schematics for the Motorola HS-544, HS-619 or HS-621.

Hmmmm, apparantly radio folks are NOT as "giving" as the guitar guys   

I searched for a while & all I could find were sites that charge a fee for membership or sell the files for $6.95 & up. 

Can you provide any links that us poor 'ole guitar players can afford (meaning, free)? 

G

This would be the link where I found mine.  Unfortunately it is up and down and appears down at the moment.

http://techpreservation.dyndns.org/schematics/Motorola.htm

[octal]

>I'm not sure that a faux pentode pre is what you'd want for an ultra clean amp though....

I've never heard a parallel triode called that..... I like it. 

Actually, that's what I was calling the cascode circuit I linked to, not parallel triodes. It's two 12AU7 sections "vertically stacked" in such a way that they perform almost exactly like a pentode, with the gain of a single 12AX7 stage.


 The output section is now gonna be KT88's run in ultra linear mode which opens more that a few cans of worms.  The bias supply is gonna hafta be a voltage tripler and I need considerably more input swing.  To get there I'm gonna hafta rethink my PI.  Either a higher voltage tube such as a 6SN7 or go to a LTPI appear to be my choices.  I stayed up way too late last night melting my hair.  I'm gonna hit it fresh later on.

Why not follow the cathodyne with two 12AU7 gain stages (one for each phase) like a Marshall Major? I believe that uses KT88s, and would fulfill your need to use 12AU7s as well! It's a "Williamson" style circuit.

Good luck,
Nathan

[Platefire]

Wouldn't hurt to have MV after C15 so you could dial in a little sensativity and if you didn't need any leave it full open 100%. 

[DummyLoad]

drums fingers... well, we're waiting...


    

[RicharD]

drums fingers... well, we're waiting...
    

Sheesh!!!!!  Gimme a break already. 

so....  Rev 2 is completely redesigned.  Working backwards, ISOload loaned me a 4K UL OT designed for KT88's.  That right there changed everything.  I need more input swing & higher neg bias voltage to drive KT88's.  I totally had a major dyslexia fit last night while trying to draw the negative voltage tripler.  Somebody please check it because I literally had to walk away for fear my brain would get stuck in reverse.  99% of the time I can deal with it but sometimes it gets real scary.  I feel fine this morning and my spelling seems OK.  I switched to a LTPI because I couldn't quite get satisfactory output from a Concertina.  The LTPI got me there no problem.  I also switched to a 6SN7.  As similar as the 2 tubes are, the SN7 is considerably more robust and just calculated out better.  The tone stack needed to be revisited and probably still needs tweaking by ear.  V1 is a cascode circuit as Octal suggested.  I stuck with the 6SN7 here just to keep the amp all octal.  So Janitor 2.0 ready for review.

-Richard

[DummyLoad]

that looks kool...

i'd like to see some NFB - either back to LTPI or global. problem is with global, where do you insert into a cascode?

--DL

[octal]

In your schematic, the output point for the cascode is incorrect. The output should come of the plate of the top tube.

Nathan

[RicharD]

It's not incorrect, it's completely missing. DOH!!!!!!

I Fixted it.

[octal]

Oh, that's funny. I didn't look carefully to see that the cathode was attached to that switched cathode cap network. I don't mean to nit-pick!

One more thought (and I might add that I have never built a cascode circuit in real life) I wonder if it would be preferable to use smaller value resistors for the voltage divider feeding the top grid of the cascode- if you keep the ratio of the resistors the same the upper grid will be held at the same voltage but the noise level should be lower. I think that's right.

Nathan

[RicharD]

You are correct.  That gif is hard to read but R7=2M7 and R8=1M2.  Nit picking it totally encouraged.

[DummyLoad]

yer missing 2 dots! :p

[RicharD]

Power supplies are up -n- running.  B+ unloaded is about 525V.  Bias loaded is about -130.  Right now I have (from ground to hot) 4k7, 1k, 2k5 pot, and a 4k7 (everything is 2W).  I can dial from -57 to -82V so I'm right at 10mA.  The cut sheet for the KT88 calls for 2 separate adjustable bias sources, so I need 2 - 5k pots in parallel to replace the single 2k5 pot.  The used the word "essential" so I guess it is....... well essential.  Of course I have no 5k pots.

[PRR]

> smaller value resistors for the voltage divider feeding the top grid of the cascode- ...noise level should be lower.

Thermal (hiss) noise from this source will be insignificant.

Power supply ripple gets right in (1/3 attenuation, not much). Even that may not matter because the upper plate has nearly zero power supply rejection. Effective plate resistance is about Mu*Mu/Gm, over 100K, so the 33K lets all (3/4) the B+ ripple to the output.

My main fear would be stray noise and howl-round from the very high impedance there.

Normally there would be a bypass cap to ground. You want much less than 1Meg audio impedance. You want much-much-much less than the ~~1pFd coupling from later stages. 0.01uFd-0.1uFd seems fine.

Overall I suspect this will be clean as my new PEX pipes.

[RicharD]

I've only experimented with a Cascode amplifier once.  IIRC, thangs didn't go so well.  Now I've gotten simply splendid results from the SRPP totem pole circuit.  I may still swap to a 6SL7 and go there with it.  The new guy suggested the cascode circuit and it caught my attention.  I certainly wouldn't want him to feel left out.

I've got the circuit built on the bread board and the PSU's have been tested.  I'm short a few components.  I need to make a jaunt over to Fry's and ISO-Load's house.

2 - 33k 2W resistors
1 - 39k 2W resistor
1 - 10k 2W resistor
2 - 5k linear pots
1 - KT88 

I need to put the puffy farad decade box in a box.  Perhaps it's time to put all 3 decades in 1 bigger box, or perhaps zilla's front panel?   

[DummyLoad]

i have all except the 2 pots - i may have 10K pots.

--DL

[RicharD]

I kinda wanna stick with 5k pots.  They're for the bias supply and are in parallel in the middle of the voltage divider to give me separate bias adjustments.  They replace a single 2k5 pot.  The bias current is 10mA so that 100 ohms = 1V for a single supply.  I wanna K.I.S.S.

[DummyLoad]

I kinda wanna stick with 5k pots.

looks like a fry's kind'a day for you bud! ;)

they have the PEC 2W in 5K, and the El-cheap-O variety as well... 

[DummyLoad]

there is a care package for you in the flower pots on my front porch.   

[darryl]

...in the flower pots...

Are your flower pots log or linear? 

[DummyLoad]

...in the flower pots...

Are your flower pots log or linear? 

no, ceramic... sheesh! 

[octal]

I've only experimented with a Cascode amplifier once.  IIRC, thangs didn't go so well.  Now I've gotten simply splendid results from the SRPP totem pole circuit.  I may still swap to a 6SL7 and go there with it.  The new guy suggested the cascode circuit and it caught my attention.  I certainly wouldn't want him to feel left out.

You're too kind! I'm planning on trying a 12au7 cascode in my next build, so I'll be interested in hearing your results. Sometimes I tend to be seduced by novel circuits when refining a more conventional circuit might be a better route.

[octal]

> smaller value resistors for the voltage divider feeding the top grid of the cascode- ...noise level should be lower.

Thermal (hiss) noise from this source will be insignificant.

Now, in fairness, I said I thought it might help with the noise level. :-)  Is the screen resistor also an insignificant contributor to noise on preamp pentodes? (Obviously, I am talking pentodes without the screen AC bypassed to ground, as an AC short would shunt any resistor thermal noise to ground so resistor noise would be a moot point.)

[RicharD]

>You're too kind!

I play this game for fun.  It's a give -n- get situation.  You'll find this forum to frighteningly polite and informative.  I have a spare tirode so why not try a Cascode?  I too get seduced by different circuits as well as different tubes etc.  It's all brain candy for me and hopefully I'll learn something along the way.  Thanks to the Dummy mang, I'll probably get to light this thang up tonight.

[RicharD]

Hey Dummy Load,
You need to come over tomorrow after work and let's light this thang up.  I'd kinda like a 2nd set of eyes beforehand since the OPT you loaned me is not exactly cheap.  If it smokes, you should be here.

[stingray_65]

Up and running?

I'm curious, was there smoke or music that came out of this amp 

Ray

[PRR]

> pentodes without the screen AC bypassed to ground,

Interesting question which "never" arises, because the gain goes way down so why use a pentode at all?

Of course musicians may find a reason to pay for a pentode not working at full gain.

[DummyLoad]

Hey Dummy Load,
You need to come over tomorrow after work and let's light this thang up.  I'd kinda like a 2nd set of eyes beforehand since the OPT you loaned me is not exactly cheap.  If it smokes, you should be here.

oki...see ya mañana dude. 

--DL

[RicharD]

Man I am slow.  Just now after reading Octal & PRR's discussion, It suddenly dawned upon me why the cascode is called a pseudo pentode.  I get it now.   

[octal]

Butterylicious-  Thinking about what PRR said about the gain of a pentode with an unbypassed screen-- I notice that you don't have a bypass cap on the "screen" (the upper grid) of your cascode. Did you run a sim on this circuit?

Nathan

[RicharD]

Nathan,

That circuit is copied directly from TubeCad.  Broskie doesn't show it on the "text book" cascode but he does on the "self bias" cascode circuit.  I've got a .1uF on standby.  Anxiously awaiting Dummy Load's arrival.  All the decade boxes are removed for start up.  I think I put a .01uF interstage for the moment, no bright cap, no cathode bypass.  Gotta get it all biased in and tuned up.  I'll grab some telemetry and then we start playing.

-Richard

[RicharD]

Wah!  The circuit sorta works but the PI has a miserable 60Hz hum.  Flopped to DC filaments so that's not it.  Disconnect the PI, hum goes away.  Pi input grounded, still hums.  Something like 6Vp-p DC supply looks clean with a scope.  Grounding scheme seems right but now I'm too tired to think about it.  Not a bad 60Hz ring modulator.  Just compared DC measurement to TubeCad and we're in the ball park standing on the pitcher's mound.  Even though it's 60Hz and not 120, I'm smelling a grounding error.  60Hz present at the cathode too.  I'll recheck it tomorrow and try again. 

[RicharD]

I fixted the hum.  2 words........ bias supply.  Of course now it's waaaay to late to light up a geetar.

Drawing is wrong.  400uF cap is on the other side of the pot... it's late.

[DummyLoad]

I fixted the hum.  2 words........ bias supply.  Of course now it's waaaay to late to light up a geetar.

Drawing is wrong.  400uF cap is on the other side of the pot... it's late.

did you revert to AC filaments? should be nearly quiet. never thought to look there.    

[RicharD]

I'm still on Zilla DC filaments.  Playin with other stuff for the moment.  I'm fixing to update the schematic then go on a tangent.  

So I set the VOM to AC and started looking for AC where it shouldn't be.  Clean in front of the PI.  Like 6 volts at the plate.  I accidentally went to check bias while still in AC mode and saw like 30VAC.  Epiphany.....

So the amp is clean, too clean for geetar.  It's super schweet with a bass guitar.  I dumping all the tone controls.  Dialed in the fullest values and I'm happy.

Updated schematic to follow shortly.
  

[RicharD]

Version 2.2

Dual input:  6SN7 cascode & 12SJ7 pentode.  The 12SJ7 sound better.  Yes I tried cascading them.  It does what you'd expect.  I'd have to add some lo cut if I wanna go OD.  Not my goal this time.  The Cascode 6SN7 just doesn't quite have enough push.  The 12SJ7 seems just right although this tube is inherently microphonic which doesn't really bother me.

[DummyLoad]

ah got it! we'll work up a number 6 on 'em...  

how about we give 'em the 'ol CnC treatment? ....cascaded cascodes.

that's a " whole lotta pseudo goin' on ".

[RicharD]

Something I noticed about the 6SN7 cascode circuit:

As you roll back the volume on the guitar, the hi frequency attenuation is very pronounced.  This is not at all an uncommon situation, but this circuit is very pronounced.

[PRR]

What is the middle 1K||33u doing? Seems pointless. Unless you intended to take the 1Meg to the other end of it??