Nachbaur Dogzilla Preamp Based Project

[alexc]

Hi

Could someone help me understand this line preamp of Fred Nachbaur's Dogzilla amp?

http://www.dogstar.dantimax.dk/tubestuf/dzart-8.htm

His designs are really interesting and I'd like some clarification on the V15 tube operation.
Seems to be a differential amp with one grid as input signal and other as a feedback signal with a dc voltage ref?

Been on  my mind for a while  ???

Addendum :  Fred gave a description of the workings of this circuit on http://www.dogstar.dantimax.dk/tubestuf/instpre.htm and http://www.dogstar.dantimax.dk/tubestuf/tonecont.htm

So that clears that up! NOT

So it is a quasi opamp - one side of the differential input is signal, the other is a dc coupled feedback network (?)

The input to that feedback network, being the plate of the AT7 - a large amplitude signal plus a high dc voltage (maybe 200V ?)

DC Coupled to 3 neon bulb string (3x65V ?) leaving a smaller dc voltage (maybe 20V)  to a 2/3 dc divider (14V ?) stiffened with some resevoir electrolytic caps and a dropping resistor(s).

Resulting in a smaller dc voltage offset (5v ?) on top of a smaller feedback signal.

So that dc voltage applied to one side of the differential pair would be some kind of bias?
And the ac signal would be the feedback applied to the non inverting input 

Or something.  :

 

[shabtek]

while I have no technical understanding to share--have a look at fred's 'hi-fi tube phono preamp' it uses a very similar layout with a 12ax7 as a differential input and a string a neon lamps to provide a d.c. reference.

edit: not sure of the merits of such a design as a line amp--may be fine for hi-fi hi-Z loads but seems paltry in 600 ohm +24dB land.
It does seem like an interesting way to achieve a nonlinear frequency response such as a phono cartridge requires...or other eq application.

 

[alexc]

Yep - he seems to use it as a preamp a lot - not as a line driver or anything, just a gain make up after eqs and after input stage amps.

I'm fairly sure Fred knew his stuff pretty good, so I would think it functions well for this purpose.

I just find intriguing the use of neons in a string to a pi filter kind of thing feeding the second grid in the pair, along with a feedback signal.

Kind of like a feedback amp using a differential pair with a kind of dc offset, reminding me of power supply rejection techniques which use a triode to blend in an antiphase version of the psu ripple.

Or not!  I may end up building one out of interest.

The rest of the Dogzilla is a monster indeed - lots of ideas which are 'off the beaten' track, including interesting psu, diode string limiters, single ended vari-gm compression, both Marshall and active baxandall style eqs  ...  the list  goes on.

Lots and lots of things going on there!  I'm a fan of his work for sure 

Rest in peace Fred!

 

[alexc]

Well I have been working on a small guitar preamp - amp - compressor  based on Fred Nachbaur's masterpiece "Dogzilla 8) Amp".  The scope of his vision in a single unit is panoramic!

Input  http://www.dogstar.dantimax.dk/tubestuf/dzart-9.htm
Drive + Sidechain  http://www.dogstar.dantimax.dk/tubestuf/dzart-7.htm


I can tell you it's a very interesting project. So much stuff going on in such a small place!
I've adapted some of Fred's preamp ideas to suit my twisted ends 


Basically it is a Champ amp driving either a speaker or dummy load. 

There is also a parallel clean/drive path.

Those two paths are  mixed and then sent to a balanced line amp main output.

The fun part is that it also uses single ended  vari-gm    compression.

I'm well into the building  - it's a whole lotta fun. Pushing my point-point skills to the max.
I've done maybe half a dozen pp builds so far. I'll put some pics up soon.

It's got 4 pieces of iron (including a utc a15) and 8 tubes.
Plus 11 knobs and 6 switches.
And HiZ In, Direct Out on the front and Balanced Out, Speaker Out, Insert
As well as a GR meter!

All of that in a roughly  1/2 width 1RU sized Hammond case.  8) 8)

Interesting working out all the gain staging, mixing and the vari-gm operations.

In use, it can be a Champ style amp (4W or so) with a high quality 'Boogie'd feed to a console
and a built in effects style compressor.

Or it can be a silent studio guitar pre with dual 'Boogie' and Champ dummy loaded paths and again, the compressor.

Now single ended compressors have 'thump' issues - so there are restrictions to the GR and time constants which can work. Should do well enough for guitar effects pedal style usage.

I am employing 2 stages of GR in my adaption. Hoping I can get more GR before artifacts get too bad.

2x 6BJ6 pentode tubes for this. Quite an interesting tube. Lots of emission for very small heater current.

Another one like this I am looking at is the 6SS7.
Metal can, 8 pin (not octal!) format with a little less 'puissance' than the 6SK7 or 6BJ6.


Only one way to find out if it will work. Surely one of the more challenging units I've attempted.
I'm nearly ready to attempt my Fairchild-y unit!

Thanks to all the tube guys whose educational discussions have taken me from tube noob to genuine vari-gm designer noob  ;D  You know who you are!

Cheers

 

[flaheu]

Nice project !!!
I like a lot that kind of "spirit" in tube builds.
I hope we will hear some samples 

 

[alexc]

I hope to hear something soon!

Tubes are EZ40 rectifier, 6BJ6 as the vari-gm input stage feeding a 12AY7 cathode follower driving a  fender style tone stack but with switchable 'Lift' on the 'Middle' tone control and a switchable 'Shift' of centre freq.

Tone stack make up is 12AY7 which is then sent to a pair of potentiometers which feed each of the :

6K6 pentode and a Champ style single ended transformer driving a switchable dummy load or 8ohm speaker in triode or pentode modes.

The dummy load is switchable from 2ohms to 16ohms so as to provide some variation in the impedance reflected back to the 6K6.

and

6BJ6 vari-gm switchable gain stage then 12AU7 cathode follower stage for buffering.

Those seperate paths are then sent to a dual deck mix pot and the mixed output goes to an insert point/side chain takeoff  and finally on to a 12AU7 with parallel stages as the line amp parafeeding a utc a15 10K:500 transformer output.

The remaining 12AU7 stage is is used as a grounded cathode amp feeding a 6AL5 rectifier tube sidechain and time constant network. That feeds each of the 6BJ6s giving 2 layers of single ended gain reduction.

All up it is using 2.5A of heater and around 90mA of HV. More or less.

Waiting on parts as usual but I think I'll have it ready to start passing some signal and check the dc operating points in the next few days.

--- bias  info changing without notice!

Of course the 6BJ6 will be the most interesting - I am biasing it to idle quite high : 1V grid and 9.5mA at 150V plate and screen is at 100V and 2.5mA or so.

It has a 15K plate resistor giving max gain of around x52 (min GR, plate 9.5mA) and min gain of around x12 (max GR, plate 2mA). It drives a cathode follower and sees a very high impedance.

That is for grid voltage from -1V to -4V  giving  gm 3500 to 800 umho, a GR of x4.4 or 12.8dB
CV extending to -15V for sub mA plate current.

---

And I have 2 stages of this available as neccessary.

The second one is biased somewhat cooler so as to provide for a higher input signal, less tube current and distortion  at the expense of amount of GR achievable.


Plenty of scope there for effects unit style GR I think.

In theory!

Additionally of interest is I am using a step-network to couple the first 6BJ6 to it's cathode follower.

Since I am using dc unregulated heaters (so far) and it is not easy to elevate them, I am using the step network to divide 150V down to around 60V at the CF so as not to exceed the max heater/cath voltage rating.

The second 6BJ6 GR stage's AU7 cathode follower I am running at the limit of its cathode dc voltage of 100V and hoping it will be OK. The reduced operating point voltage should be fine as the input is likely to be only 35Vpp or so.

It will be divided by 3 before driving the 'Mix' pot,  so as to be comparable to the Champ's output transformer of 8Vpp or so.
The mix of those two signals becomes the side chain takeoff and input to the final stage line amp.

If that doesn't pan out, I may try ac heaters and just elevate them.

DC heaters take 10000uF per A or so - a lot of filtering in this small space!

Not sure which will be better in this case - really, I should have some regulation in there. Both LV and HV.
If it were a proper differential GR stage I surely would.

That's it for now.

Thanks for reading 

 

[alexc]

...progress pic ...

Getting there - waiting on some parts. Design is pretty much done for now.

 

[flaheu]

really cool 

I'm eager to hear da beast in action !

 

[alexc]

Me too - I have to start passing current soon so I can check it before buttoning it all up.

It's getting kinda cosy in there  ;D

This one I will put up some mp3s when I get it going. Hopefully it will be really interesting.
Has enough gain pots at the key points in signal chain to handle pretty much any gain staging.

Still weeks away to finish this 'Ratzilla' -  rodent son of the mighty 'Dogzilla' (the thinking man's muscle amp!)

 

[rhiannon]

Hello everyone.  I hope I don't come across as a troll, but I am Fred Nachbaur's daughter, Rhiannon and I am here to discuss his amp.

Sadly, Dad passed away in 2004.  I miss him terribly.  He literally was a genius and invented things that I sadly don't understand as well as he could.  So I find myself in the situation of having something really awesome that I don't "get" but am sure one of your members will.

I inherited his epic Dogzilla amp.  The thing is just huge, loaded with hundreds, if not thousands, of dollars' worth of tubes.  And I know it will be of huge value to a collector or fellow geek.

I have posted it up for sale and hope you can help me assess the value and find it a suitable buyer.  I am located in BC, Canada (not far from where dad lives) and all proceeds from the sale will go to his grandchildren for their schooling.

Many thanks in advance for any help you can provide.  Dad sure was proud of this invention and I am sad to see it go, but it would be better off in a place where it is fully appreciated.

Best wishes, Rhiannon Schmitt (violin lady at www.fiddleheads.ca)

 

[alexc]

Hi Rhiannon

No problem with your request

So sorry too for the loss of your father - I'm sure I'm not the only fan of his really interesting work

There were lots of things I liked from Fred - his designs and articles were so very instructive as well as interesting. And the more elaborate of his designs still challenge my understanding of vintage tube circuits.

Truly he had designs from the very small to the very large, the Dogzilla being the most comprehensively featured and *brutally* powerful of them all

I've never seen it's like in any other design, combining flexibility and functionality with awesome strength. He made a point of overdesigning certaining aspects of the amp to provide safety and reliability.

As to ascribing a resale value, on the positive side of valuation it is an obvious one of a kind, fairly infamous on the interwebs and as I said, novel design, comprehensive and powerful.

As you said, parts alone would easily run to more than 2K I think. Time investment would be very, very high, making several K in 'hand made' valuation reasonable.

That would put it in the 2-4K region, I would think on a 'good day'.

Finding a buyer, particularly in today's world economic climate can be challenging and so reducing the price from it's value is probably required, depending on how quick you need to complete a sale.

So I would put the value somewhere in that region of 2-4K. 

People pay that sort of figure often in high end hifi and tube amp circles.

And the Dogzilla is truly a beast in a velvet suit  ;D

Wish you all the best of luck and I think of Fred each time I re-examine his designs.

AlexC

 

[alexc]

Well, back on this one, now I've cleared my diy decks somewhat, lately ....

I've done about 1/2 the building and have been testing that before I do the rest.

Happy to say it is all working well so far.

Uregulated DC heater with 19KuF of filtering for 2.5A - appears to be quite low ripple.
Unregulated tube rectifier + choke filter - again appears to be quite clean.

What's cool is that pretty much all the dc operating points I chose have panned out as expected.

2x 6BJ6 single ended pentode gain/GR stages
1x 12AY7 tone stack
2x AU7 signal/side chain  buffer/drivers

6K6 single ended speaker/dummy load driver
6AL6 side chain rectifier
EZ40 psu rectifier

Should be done and ready for 'tuning' in a week or so

The 'Ratzilla' is pretty over the top for a small unit   

Fred's ideas are quite addictive - I can already see the successor design. Even less common tubes this time.
Perhaps incorporating a GR stage from that wonderful Australian 'Trimax' limiting amp - a balanced variable impedance CF. I've only seen that in one other place and it is quite simple and apparently 'distortion free'

And much bigger  this time too 

It will be named somewhere between a rat and a hound, with a nod to my hometown where critters like Tassy Devils and even weirder  marsupials roam - like the spotted quoll. A terror to all small statured creatures of the night!

I will call it 'Quollzilla' - bad boy of the south  8)

 

[flaheu]

Like I said.......

Cool project !

Please, please, please.........we want samples  8) 8) 8) 8) 8) 8) 8) 8)

 

[alexc]

Thanks for the interest!

Will do - I'll post some samples as soon as I get to the first reasonable baseline.
Just have the sidechain amp and tone stack to complete now.

It's kind of like 'crotchet' in there! More like weaving than electrical assembly!

So far, it sounds .. crunchy!

Without a sidechain, those 6BJ6 are running pretty hard with lots of current and gain.

Progress pic ...

 

[alexc]

and another ..

controls are :

top panel  -  power amp  load (16R, 8R, 4R, 2R) rotary switch, pentode/triode mode 

front panel - power, illuminated Gain Reduction meter
                    compression knob, comp in switch
                    line level, amp level, mix knobs
                    tone stack defeat switch, tone stack shift switch, mid/lift switch, bright switch
                    hi, low, mid/lift  knobs
                    volume, gain knobs, normal/gain switch
                    input, direct out jacks

back panel - amp out 8R jack (switched to dummy loads when not connected)
                  balanced line out xlr (600R)
                  unbalanced direct out (hiZ)
                 
             

 

[alexc]

Hey Flaheu

How's things there in Belgium?  I spent years living in Antwerpen in my younger days.
Working on digital switches and such.

Remember it very, very happily.

Some of my favourite times were long weekends with friends staying at a farmhouse in the Ardenne.
And weekend trips to Amsterdam  8) As well as just about everywhere else.

I loved it there - the people were so well educated, they spoke english better than most native speakers.
As well as several other languages too.

Wish I were there again  :'(

Cheers

 

[alexc]

Some more testing - I've now measured all the operating points of the many stages and they are very close to my calcs. The whole thing is running as expected, sans tone stack and sidechain.

Couple of things I don't have quite right - some clipping at one of the signal routing points and some oscillation due to the massive gain without the tone stack losses present.

While I'm sorting that out, I've been looking at the gain reduction capabilities of the 6BJ6.

The bad news is that at my calculated and measured operating point, the zero-GR gain is lots *lower* than I expected. I'm seeing about 1/3 of what I expect.

That in turn limits the GR I can get for now. With the CV circuitry at the grid of the 6BJ6 in place but using an external DC voltage to simulate the sidechain, I get :

cathode bias voltage -1V, apply 0 .. -12Vdc at grid I get gains of x18 (no GR) and x5 (11dB GR).

So it definately works - I need to figure out why my basic no-GR pentode 6BJ6 stage gain is so low.
It is x18 and should be x50 or so. The measured dc voltages and currents match my calcs from data sheets pretty close. So wtf?

All I can think if so far is :

Lower than calculated gain in a small signal pentode can mean

- the stage is loaded too heavily and so the pentode is unable to maintain the expected voltage in the load
    -> I have several hundred K on the plate to ground so should not be an issue

- the screen voltage is too low and causing 'squashing' of the plate curves downward, reducing gain
    -> screen voltages/currents check out within 9V max of calculated/datasheet values
    -> plate voltages and currents spot on according to calcs/datasheets

- plate resistors not the correct values
    -> checked that I installed as calculated

- tube performance has degraded with age
    -> I tried several tubes with a change in gain of only 10% from best to worst. Not significant

-> screen bypass frequency dependent effects ?
    -> re-examining the screen supply circuit to see if anything awry there

-> datasheet innaccuracies ?
    -> I don't think this could account for such a big diff in measured and calculated gains

So I'm investigating the pentode operating point further to figure it out.

But it's getting there!

------------

I changed the screen bias to provide a higher screen voltage (from 100V to 115V) with no difference in gain.

Tried adding an electrolytic cap at the plate to ground (22uF) and that made a huge difference - I get x47 no-GR gain which is about right.

So I have some supply decoupling/screen bypassing issues - I'm back in the game now 

With my external grid bias at -20V, I get x47 (no GR) and x3.5 (max GR) -> 23dB or so 

So that confirms my basic GR operation. I have 2 of these stages and I am looking for a total of 12dB or so GR with minimised thump/CV passthru type of nasties, which are inherent to the single ended scheme.

----

By way of some more numbers :

pentode gain is approx gm*Ranode  and  I have 15K anode resistor here.

gm at -1V grid (no GR condition), datasheet says around 3300umhos at 100V screen, 250V plate
Expected gain is x51 (I'm running plate at 150V 9.5mA)

gm at -20V grid (max GR condition), datasheet says around 10umhos
Expected gain is x0.15

Measured is x47 and x3.5.

Datasheet says x3.5 should happen at -9V grid. I was actually not measuring directly at grid, but rather at the CV feed into the grid biasing network.  Also, using a couple of 9V batteries, so perhaps not enough joos to get the grid down and stay there.

So I'll do that again in more detail.  Just so you know  ;D

----

Onwards now to rework the screen supply circuit of the GR stages.

----

Well - addded some couple of uF electro's plate to ground and all is well. Have each stage now giving me the expected no-GR gains.

Turns out I had too little capacitance at the screen to ground and it was low Z path for dc to ground!
When I fixed, my rails stiffened somewhat and my heater also stiffened a little

(lucky for me an easily available 0.5R 25W heater resistor leaves the whole 2.5A circuit sitting at 6.27Vdc when warm )

On the subject of warm, this little guy certainly does do it! Warm and toasty for the cool climate here.



Looking at the sidechain now.

 

[alexc]

Considering the sidechain in Fred's Dogzilla,  it is a single stage grounded cathode amp 6SN7 stage with a relatively low plate voltage of 200V and Ra 22K, Rk 470R

So that would likely be a gain of x14 and a quiescent current of 4 to 5mA, bias of -3Vdc or so.

That's driving single ended-ly the two diodes of a 6AL5 as a 1/2 wave rectifier (I think!)
There's no dc bias applied to the 6AL5 and they are ac connection from the plate of the 6sn7 gca.

The time constant network is a 1uF cap and 1M pot in parallel. Time constant (release) would therefore be long, around 6s or so  (2*pi*RC). You can add some parallel fixed resistance to reduce it. I like 0.5s or so so i think 88K would do it.

The pot varies the amplitude of  sidechain signal applied to the GR tubes grid circuit- a rectified, smoothed varying version of the audio chain. The grid circuit is a voltage divider (1/2) with a cap to maintain dc conditions at the grid.

So - seeing as this is my first time doing this, I'll mess about abit first with regular diodes until I figure whats what.

Possible changes I'm thinking of, and noting I'm driving 2 grids instead of just 1 :

- maybe swapping the sidechain gca stage driving the diodes single ended 1/2 wave
  for a single stage phase splitter (concertina) and driving the diodes in full wave

That should give me a more symmetrical rectified signal and more current thru the 6AL5, meaning a faster attack.

It may be fine as is or the acceptable 'no thump' settings may dictate a longer attack (less C charging current).

I'll know more shortly when I get there! Tomorrow 

 

[flaheu]

Hi Alexc,

Hey Flaheu

How's things there in Belgium?  I spent years living in Antwerpen in my younger days.
Working on digital switches and such.

Things are going......you know.....pretty ok.
You know Belgium, at least for me, is an "artificial" country which is less than 200 years old.
That tiny territory is "squeezed" between France, Germany and Nederland. So we are a strange mix of those different cultures.
It's been said that it's what it mades us so particular. Well i don't know.
I'm living in the south french speaking side and I go in the north Flemish speaking side for work on a regular basis.
In these days of "crisis" the extremes reveals their ugly face, but Belgium is still strong !
And the most important, we have the best beers of the world !! more than 400 different kind !!! 8)  ok USA's got 2000 breweries but if you calculate the ratio brewery/amount of population we are still the first.....and the best  8)

Remember it very, very happily.

Some of my favourite times were long weekends with friends staying at a farmhouse in the Ardenne.
And weekend trips to Amsterdam  8) As well as just about everywhere else.

Amsterdam !!! I know what you mean !!! "pass da dutchie......lalalalallala"

I loved it there - the people were so well educated, they spoke english better than most native speakers.
As well as several other languages too.

Wish I were there again  :'(

Cheers

Thanks, but don't make a generality, we have also all sort of MF like everywhere else !

By the way, where do you live ?
If you come around one of these days, stop for a beer 

I'm following your project with attention and hope for some samples soon !! 8)

 

[alexc]

Yes, I suppose you are right.

But one forgets the MFs and remembers just the good parts after 22yrs have passed!

I live in the Tasmania as far south as you can go in Australia. Not likely to be in your part of the world any time soon, I'm afraid.

Cheers