Solid State Guitar Amp Forum | DIY Guitar Amplifiers

Solid State Amplifiers => Tubes and Hybrids => Topic started by: ilyaa on September 09, 2014, 12:39:34 AM

Title: kay model 720 => AB mod
Post by: ilyaa on September 09, 2014, 12:39:34 AM
got this amp the other day because i was intrigued by the cabinet design - its a 1x15 but with a grill on each side and the controls and stuff on the traditional 'rear' to encourage you to face the speaker butt out - and thought itd sound good to use together with my sound city concord -

recapped everything but the main can and its sounding good but its a little quieter than i expected - took another look - didnt realize it was a class A design!

i was thinking itd be an interesting and fun learning experience/experiment to modify it to run AB instead. ive looked into tapping off a negative bias supply (http://music-electronics-forum.com/t2545/ - thats a useful thread!) and am deciding between the ampeg style capacitive divider and this simpler one: http://music-electronics-forum.com/attachments/547d1175892694-kscan_0001.gif. what kind of diode is appropriate for this - just any ole high voltage silicon style 1N400x?

but before i get into that i want to make sure im not missing something -

1) i cant find any info on the specific output transformer - if its meant to run two 6L6's Class A, should it be able to handle AB operation?
2) and a follow up to that: because of the greater range of input voltage of to AB designs, will the amp actually be making more power - larger voltage swing?
3) is this a stupid idea?

Title: Re: kay model 720 => AB mod
Post by: Enzo on September 09, 2014, 02:37:56 AM
How did you determine it was a class A circuit?   You are aware that cathode bias is not the same as Class A?  What is the B+ on the plates and what is the cathode voltage?  It may well be Class A, but just having cathode bias does not tell us that.

A bias supply has negligible current needs, and the voltage is not high, but if you drop off the high voltage, use a 1N4007.


Is it stupid?  Not for me to say.  But be aware that of you take a 25 watt amp up to 50 watts it will only be 3 decibels louder.   That's all you get.  Loudness is not watts
Title: Re: kay model 720 => AB mod
Post by: ilyaa on September 09, 2014, 03:33:25 AM
B+ ~ 400V
Cathode ~ 30V

Title: Re: kay model 720 => AB mod
Post by: Enzo on September 09, 2014, 01:00:55 PM
That is about 22 watts a tube, not very warn for class A.

But nothing prevents you from making this conversion either way.
Title: Re: kay model 720 => AB mod
Post by: Roly on September 09, 2014, 01:22:56 PM
Are the output bottles wired in parallel as Single-Ended, or is it Push-Pull with a PI?

If it's the former it's Class-A, but if it's the latter it's Sinny to a brick that it's cathode biased Class AB1, and I'd guess rated for about a 40 watt (RMS) output.


(http://www.triodeel.com/6l6gc_p4.gif)

Note; "Class AB1" and "fixed bias" and "cathode bias" conditions.

A Single-Ended amp, pre or main, must be Class-A.

A Push-Pull amp may possibly be Class-A (a Hi-Fi), but in guitar service will almost always be in Class AB1, and sometimes AB2 and B1 and B2.

Very generally speaking the divide between cathode and fixed bias for most of the commonly used bottles is the 40 to 60 watt step ('tho I've see cathode bias used up to the 100 watt level).

Weak output?

Q1. How old are the OP bottles?  (in terms of work hours)

What's the peak HT current under drive?


{Got enuf drive volts out of the PI?}
Title: Re: kay model 720 => AB mod
Post by: ilyaa on September 09, 2014, 01:55:57 PM
QuoteAre the output bottles wired in parallel as Single-Ended, or is it Push-Pull with a PI?

push-pull - but cant it be class A push-pull? i thought the main distinction was just where the control grid was biased (not how the tubes are set up) - around cutoff for AB and in between cutoff and saturation for A. i figured if the grid is only 30 volts below the cathode, as opposed to like -40 to -50 volts, then the power tubes are conducting fully even when idle and its class A...

on that datasheet, where it says "max signal power output," is that for one tube or two? and is that in RMS watts or Peak? it says 24.5 watts under cathode bias and 50 watts under fixed bias. it says "values are for 2 tubes" up top - does that mean two tubes in push-pull cathode bias are only making 25 watts together? (would that be because of cathode feedback/degeneration)? there is no cathode bypass cap - see schematic attached.

the op bottles are new - i havent run any real tests or anything - i was just making grand plans.
Title: Re: kay model 720 => AB mod
Post by: g1 on September 09, 2014, 01:59:11 PM
 Ilyaa:  You probably read about it, but I'll mention it for others who may read this thread, the bias circuit you posted with the resistor that connects to the HV AC,
(http://music-electronics-forum.com/attachments/547d1175892694-kscan_0001.gif) that circuit will only work if the PT is center-tapped.
Title: Re: kay model 720 => AB mod
Post by: Roly on September 10, 2014, 02:27:54 AM
Quote from: ilyaapush-pull - but cant it be class A push-pull?

It can, but outside Hyper-Fi you are very unlikely to find an amp running Class-A push-pull {and you can't understand a Class-AB1 push-pull loadline until you understand single-ended  Class-A loadline anyway}.


Let's recap; Class-A means that cathode current flows during the entire duty cycle, the full 360º of each cycle, at no time in cutoff.  In Class-A, as with Class-AB1, the bottles are biased to pass enough standing current to push their anodes to full dissipation (25W ea for 6L6GC's). 

In Class-A the anode current will go just to cutoff (Vbias ~38V, Vcutoff ~ 45V, i.e. about a 7V input swing).

Now we can put in 38Vpk before the grids start to go +ve, but for swings over about 7Vpk (about 6W out from a pair of 6L6's) one or other of the output valves must be effectively cut off, and we are no longer in Class-A but Class-AB1 which is defined as conduction for less than 360º but more than 180º of each cycle.

Quote from: ilyaai figured if the grid is only 30 volts below the cathode, as opposed to like -40 to -50 volts, then the power tubes are conducting fully even when idle and its class A...

That's true; in effect all Class-AB1 amps operate in Class-A for the first 10% or so of their range.  If you want to stay in Class-A then you can't drive the grids into cutoff and you give up 90% of the available power and 50% of the potential efficiency of the stage.  I think most guitarists would consider only 6W from a pair of 6L6GC capable of 55 watts to be a waste of good valves.

While we are here: the Vox AC-30 is widely known as a "Class-A" amplifier.  The only problem is that it isn't.  While it has a very hot bias which means a wider Class-A operation around zero, it is actually a Class-AB1 stage.  If you look at this (annotated) Vox "works drawing"...

(http://www.ozvalveamps.org/vox/ac301960.gif)
(dwn: R.D = Richard "Dick" Denney, Tom Jennings' designer/tech)

...you will see that an idle cathode voltage is given, and a higher voltage for full output "12.5V at 30 watts, quiescent 10V" - and that's it right there, the average supply current in a true Class-A stage does not change between zero and full drive, but it changes just like this in a Class-AB1 stage. 

Rk = 50 ohms
I = E/R
10/50 = 0.2 200mA (50mA/valve)
12.5/50 = 0.3 or 300mA (75mA/valve)

So there's a 50% increase in output stage cathode current between idle and full output leaving no doubt at all that this is a Class-AB1 stage.


Quote from: ilyaaon that datasheet, where it says "max signal power output," is that for one tube or two?

"Unless otherwise specified values are for two tubes"

WRMS.

The Watts "PMPO" madness started after this was printed, and we try to ignore that nonsense as far as possible.  Engineers and techs deal in "watts RMS" because it is not only the most conservative rating, it is also the thermal rating.  It is perfectly valid (and surprisingly accurate) to measure power by putting your load resistor in a thermos of cooking oil and measuring the temperature rise against time, then crunching a few numbers - in fact this is how you might calibrate a dummy load/wattmeter.  {n.b. when you are dealing with irregular waveforms such as music then (failing a kilobuck HP wattmeter) thermally is the only practical way to measure the average power.}




{muse aside: how do we know that electrical/electronic theory is real/correct (at least as far as it currently goes)?  There are two powerful intersections between electrical and physical forces, and between electrical and chemical reactions.  We can use scales to measure the physical forces of electrostatic attraction and magnetic attraction due to current.  These are based on our understanding of things like the charge on an electron, and evolve onto Ohms Law, the Power equation, etc.

In electro-chemistry (batteries in particular) the voltage of a cell is given by the electrochemical potentials of its component parts.  When you, say, charge a lead-acid battery we can make predictions about how much H2 and O2 gas will be evolved, and how much energy will be stored, how much heat will be lost, based on our understanding of molecules, and electron orbital energy levels (reaction potentials).  Similarly with electro-plating and weight gain by the plated object (which is directly related to the Coulombs of charge that have flowed, indeed some very early DC electricity "meters" were a silver plating cell with the electrodes weighed to "read").

Then we have discharge lighting which gives an almost direct observation of electron activation then light emission on a characteristic set of wavelengths (related to the orbital energy levels of the gas, mercury, sodium, neon, argon, hydrogen, etc).

All of these mesh neatly together having been arrived at from different scientific directions, and because the mesh so neatly we can be pretty damn sure that they are all correct, only different aspects of the same thing looking from different directions.

Muse = off}





Quote from: ilyaa(would that be because of cathode feedback/degeneration)?

Essentially yes, 'tho it is determined by the load line chosen for the particular application.

(http://www.valvewizard.co.uk/SELoadline1.jpg)
(A topical set of load lines thanks to Merlin)

N.B. for simplicity this is a Single Ended Class-A amp, not push pull, but the basic theory remains the same.


This is the anode characteristic of an EL34 (6CA7).

The red line Wa = 25W is the absolute anode power limit; all operation must be below this curve.

The anode supply voltage is across the X axis an represents the HT supply voltage.

The cathode current in on the Y-axis and represents the maximum current allowed by the load at the anode voltage (in this case Za = 4k ohms).

The magenta, grey, and blue lines represent how the anode will track for a 4k load and differing supply voltages.

As the grid voltage changes the operating point signal slides along the load line from high voltage cut off to high current saturated, e.g. Vg = -12V -> Va = 50V, Ik = 125mA; Vg = -32V -> Va = 500V, Ik - = 25mA.

For the magenta load line the resting, idle, quiescent or Q-point (your bias setting/cathode resistor value) is the magenta dot at Va = 300V, Ik = 75mA, Vg = -24V.

The gradient of the load line gives us a clue to the load impedance, flatter is higher, less current per volt, and steeper is a lower load, more mA/V.

{for push pull you do much the same, only offset the load line down so the bottle is close to cutoff, then flip the graph and merge from below to see the full push-pull load line, just a repeat upside down really.}


Finally, just because it's Class-A doesn't mean it's clean.  Even down low in Class-A a push pull stage still has the matter of transfer distortion, similar in concept to crossover distortion.  Unless both valves have identical characteristics each will amplify an identical signal differently, and that's what we call "distortion".  To get the full Hi-Fi advantage of Class-A operation the OP bottles need to be matched at least for the Q-point and maximum and minimum currents (and good luck with that).



{then you go and stick a fuzzbox in front of it...  ::) }

HTH
Title: Re: kay model 720 => AB mod
Post by: ilyaa on September 10, 2014, 11:36:00 AM
phew thanks, Roly!

ive been reading richard kuehnel's power amp book and i was interpreting his explanation of operating class a little incorrectly, i guess. now i understand the distinction between A and AB and its relationship to bias much better - so then let me take a step back and ask about the kay amp:

looks like its set up to make a bit less power than id like

1) because of the unbypassed cathode bias
2) and the lowish supply voltage
3) which together are giving me a smaller cutoff to saturation swing
4) and some cathode current feedback which is even further reducing the grid voltage during high current operation

in otherwords, if i wanted more power out of the amp id have to

1) mod it to fixed bias
2) raise the supply voltage
3) find a different DC operating point
4) get a different amp

?

but along these lines, how much P-P voltage should the PI be giving them? normal AB style, double grid bias, so like 60V P-P? (i also see the lower 6L6 is fed off the screen grid of the 7199  - not familiar with this kind of PI, any notable differences?)
Title: Re: kay model 720 => AB mod
Post by: Enzo on September 10, 2014, 01:14:51 PM
The lower 6L6 gets its grid signal from the cathode of the PI, not the pentode screen.  The voltage at the cathode of the split load PI is used to bias the screen grid of the pentode.  Unusual, but they used it elsewhere.


Do you want more power or do you just want it louder?   If you double the power out, you only get 3db louder.


The power of any amp comes from its power supply.  All the amp circuits do is control the power.  SO to get actual increases in power, you will usually have to provide a larger source of power.  6L6 tubes don't make power, they only act as its valve.  A lot of time the increases you add along the way serve mainly to move the volume knob.  In other words however loud it used to be on 5, it now is on 4.
Title: Re: kay model 720 => AB mod
Post by: g1 on September 10, 2014, 02:00:06 PM
  Just to add a little to what Enzo said, to increase the power supply usually means installing a larger power transfomer, and some amount of re-design.
  So your point about needing a different amp to get (a significant amount of) more power is probably correct.
Title: Re: kay model 720 => AB mod
Post by: Roly on September 11, 2014, 12:31:02 AM
When I'm designing an amp often the starting point is the desired power output, and after having decided I want 50 watts (or whatever) then the process moves on to which bottles, how much HT voltage, cutting and trying each potential combination on paper/simulation until something appears that is workable.

Very often you can simply pick one of the sets of conditions already worked out for you in the valve data sheets.  One of the reasons that guitar amps tend to be highly generic is that most manufacturers do the same thing, pick it off the data sheet.  It's simple, easy, and reliable for a builder who wants to sell amplifiers, not take on an open-ended research project to find something novel.

You might be able to get more voltage out of the power supply, but you also have to get more current, and this will generally mean a power tranny with a higher VA rating (and possibly OPT as well where the amp has one).  When you get to that point you really are looking at a situation where you would be better off starting fresh from scratch rather than trying to turn a shoe into a gumboot.



{BTW it's a really good thing you are doing here because this close examination of somebody elses design exposes the design compromises that were made; homebrew or commercial, people cut corners for economic and other reasons and you need to know that these designs all come from people with all their complications, prejudices, and panic, not handed down to Moses.

I laugh when I hear some Hyper-Fi-ist flogging some gear as "no compromises".  That's bulldust because every design compromises something somewhere.  When I see claims that "everything has been optimised" I'm reminded of the old mathematical truth - "you can't optimise for more than one variable at a time".  There is no magic spot where noise AND distortion AND output power AND ... are all at their best values - it just doesn't happen with real devices.

Mic preamps are a classic case where low noise matters more than gain or even a tiny bit of distortion.  It's like a net, if you pull on noise then it will also tug on distortion and gain, if you tug on gain...  It's all "deeply intertwingled".

To the first order resistors are pretty simple, Ohms Law, but even at the first order all active devices are far from simple, gm, hfe, etc., even a transistor doesn't start to conduct until it has almost a volt on its base, and this is an "imperfection" that requires a bias network, and thus impacts on input impedance.

Design is the process of trying to find a workable combination against the constraints of the final object (e.g. servo amp, guitar amp, PA, Hi-Fi) and the constraints of the, generally active, devices available; a long circular process of tradeoffs until all the constraints have been met, or proving the job is impossible unless a constraint is changed (see: Linear Programming (http://en.wikipedia.org/wiki/Linear_programming)).  A great deal of time is spent ploughing through catalogues just to see what combinations of parts are reasonably available.}



I thought Enzo was going to go there, but as he didn't; you want louder?  The most straightforward and effective way to get more SPL is to fit more efficient speakers.  I know this isn't as exciting as redesigning the OP stage, but lifting driver efficiency from 3% to 5% will give you more bang for your buck.  The poorer the original speakers the greater the opportunity for considerable improvement.


A little point about load lines.  Where the load is resistive, say in a preamp stage, then the conditions are sufficiently similar for DC and AC conditions and a single load line will suffice.

In a push-pull output stage the load is the OPT which is an impedance (R+jX), it's effective resistance as a load depends on frequency.  If the cathode is bypassed then it is also frequency dependent.

The conditions that the OP valves sees at DC are quite different to the conditions it sees at signal frequencies.  This can mainly be explained by the fact that for DC only the OPT primary winding resistance of a few tens of ohms has any effect, while for AC the transformer couple the secondary and the impedance (resistance plus inductive reactance, the referred load on the secondary) of the winding rises to a few K ohms.

So this means we have two quite different loadlines, a DC one where the OPT could be replaced by, say, a 47 ohm resistor, and the cathode looks like 300 ohms; and the AC loadline where the load impedance is perhaps 3k6 ohms plate to plate and the cathode effectively shorted to ground via the bypass cap.

If you actually do both the AC and DC load lines on a valve anode characteristic curve you will find that the AC loadline is fairly flat, perhaps running for 300V supply to 300mA anode current (for ZL = 1k), but when you plot the DC loadline you will find that it rises almost vertically from 300V because the saturated current with a 47 ohm (winding resistance) load will be;

I = E/R
300/47 = 6 AMPS! - clearly neither possible nor healthy (this is what tries to happen when you accidentally short the bias supply and both anodes light up bright red).   :o


Quote from: ilyaai was interpreting his explanation of operating class a little incorrectly

It is normally understood rather than specifically stated, but the class of an amplifier stage is defined by its operating mode at full rated output.  The Vox AC-30 is Class-A up to a few watts, but its rated output power is 30 watts and at that level the valves are deep in Class-AB1 (and if you look at that it is implied by the "AB", it's not pure Class-A or Class-B but a transition class).

{as it happens I just stumbled across an old Wireless World article for a 6 watt push-pull Class-A Hi-Fi amp - using a pair of KT66's, more-or-less 6L6's.  You could get maybe 50 watts out of them, but it's "Class-A" because its maximum rated power is only 6 watts.}


The 7199 PI is somewhat unconventional but is similar to the 6BL8 PI used in the Playmaster 116/7 and many Goldentone amps (and no doubt others).  The triode section is a basic spilt-load or "concertina" PI, and the pentode section is a Voltage Amplifier Stage (VAS).  It looks a bit odd because it is directly-coupled, and the pentode derives its screen voltage from the cathode end of the PI.

I haven't examined this one in detail but it would seem that the DC feedback from the PI cathode to the VAS screen acts as a local NFB to keep the PI at the right bias point, i.e. one-third of the HT across the anode load resistor, one-third across the valve, and one-third across the cathode load resistor.  If the PI cathode voltage should happen to drift down then the pentode screen voltage will (eventually, CR = 0.02uF+180k) follow causing a reduction in current through the pentode and its anode to rise in voltage, applied to the PI grid this will correct the downward drift.  And of course v.v.  The CR network prevents this feedback happening at signal frequencies so it is effectively DC NFB around the two directly-coupled sections.


HTH