A Slightly Different 5-watt Amplifier.

[darryl]

I've been lurking on this forum for a few weeks now, so perhaps it's time for a construction project. So that I don't have to become bilingual, I will use terms with which I am familiar, so "tubes" will be called "valves" and weights & measures - if any - will be metric. If any of my terminology or bad jokes are unclear, then say so and I will make corrections to the terminology, but probably not the jokes. This thread will describe the process used to design and construct this particular amplifier. Other amp builders may find some useful ideas here, or suggest better ones of their own. All better ideas offered will be stolen and incorporated into my future builds.  

First a little background information - over the last two years I have posted several construction articles at the "Australian Guitar Gear Heads" ( "AGGH" ) forum, on the "In The Shed" board. http://www.guitargear.net.au/discussion/index.php?board=8.0 This current project has been spawned from a long and tortuous design at AGGH which began as a 3-valve 4-watt amplifier and ended as a 4-valve 10-watter on a completely different chassis.  http://www.guitargear.net.au/discussion/index.php?topic=20998.0

Very early in the design of that amplifier, 6BM8 valves were considered, then rejected as being too expensive ( in Australia ). That design then proceeded with 6T10 Compactron valves, but I thought I might return to further investigate the 6BM8's. They have already been used in very successful amplifier designs by a number of members here, but there is always room for one more amplifier!

It's probably time to state the objectives of this construction project. More history...  Early in 2009 I built a 2-watt push-pull amplifier, which was rather boldly named the "Impact". The construction method used was slightly unconventional, with an aluminium diecast box chassis. The limited dimensions of this chassis - 220mm x 144mm x 55mm made amplifier construction a challenge. This current project is an attempt to build a push-pull amplifier of over twice the power on the same size chassis. A mind-numbing five watts.



To build an amp with twice the power of an Impact, a number of design decisions had to be made, the first being the output valve(s). A push-pull output stage was selected, but the limited space meant that single function valves such as EL84's could not be used, as they would require a separate valve for the phase splitter. The 6DX8 valves used in the Impact were not suitable as their ratings would be exceeded in this design. A browse through some valve data books indicated that a pair of 6BM8's should easily deliver 5 watts with a B+ of about 200 volts.

Still more background information...  Australia is a very small market for valve amplifier components, so finding suitable parts at reasonable price can be challenging. This applies particularly to transformers, so a little ingenuity is required when sourcing these, especially for low-powered, low-cost amplifiers.The Impact amplifier uses two identical power transformers in a step-down, step-up configuration which is very inefficient, but quite suitable for a very low power amp. A more efficient method is to use low voltage transformers in a voltage multiplier. Both methods of obtaining valve amplifier voltages from relatively inexpensive off-the-shelf transformers have the disadvantage that several transformers are usually necessary - and in this design, chassis acreage was limited.

A suitable single transformer was available, however:



These are a 50VA toroidal transformer, with 240 volt primary and four separate 12 volt secondaries.

One secondary could power the heaters - the two 6BM8 valves would have their heaters wired in series, and the 12A*7 preamp valve socket would be wired for 12-volt operation. The other three secondary windings would be connected in series for a total output of 36 volts. A voltage quadrupler should then furnish about 200 volts DC.  

Another component requiring an upgrade in this design was the output transformer. The Impact transformer has a primary impedance of 20k ohms plate-to-plate and a quoted power rating of 5 watts. For this new design, a higher power rating was considered prudent, and the optimum plate-to-plate impedance was about 5000 ohms. A 15-watt PA line transformer which had been used successfully in a number of earlier projects was chosen. Using off-the-shelf line transformers as push-pull output transformers is another way to reduce the construction cost of small amplifiers. A detailed analysis of this approach is available here: http://home.alphalink.com.au/~cambie/6AN8amp/M1115.htm. Coincidentally this is the exact transformer type I am using.

To be continued...

[darryl]

On with the chassis bashing...

The first construction step was to prepare the chassis layout. Good practice is to keep preamp stages away from both the power supply and the output stage. To achieve this, the power transformer was placed at one end of the chassis, the output transformer at the opposite back corner and the preamp valve at the front, very close to the input socket.

The top of the chassis was marked out for the transformers, the valve sockets, holes for transformer leads and mounting points for tagstrips.



Once the chassis layout was marked out (and carefully checked!) a centre punch was used to accurately locate points to be drilled.



The centre-punch points were deepened with a 1.5mm drill so the larger drills to be used later in the process would centre correctly.



Front panel holes were required for:
                                                Power LED
                                                Input socket
                                                Gain switch
                                                Bright switch
                                                Volume, Tone and Master potentiometers

Rear panel holes required were:
                                                IEC power input socket with integrated fuseholder
                                                Power switch
                                                Line out socket
                                                Speaker socket

Once the layout was complete, serious drilling could begin. The old carpenter's maxim of "measure twice, cut once" also applies to metalwork. Drilling was done using a drill press but a hand drill could have been used. The holes were drilled in sequence, starting with the smallest diameter and moving up to larger sizes - far better to mistakenly drill a hole too small than too large. A chassis punch could have been used for the valve socket holes although the one available would probably have hurt itself attempting to punch 3.4mm diecast. The valve socket holes were finally cut using a step drill - one slight error being caused by the valve sockets being 19mm, while the step drill's steps were 18mm and 20mm.

The holes where output transformer leads would pass through the chassis were drilled large enough for protective rubber grommets to be fitted. Large irregular-shaped holes for the IEC connector and power switch were produced by drilling a string of closely spaced smaller holes and joining them up by GENTLY rocking the drill or chassis from side to side. (My metalwork teacher was spinning in his grave when I typed this...) An assortment of round, flat and triangular files were then used for smoothing and deburring hole edges.







Once the chassis bashing was complete, the chassis was cleaned and the marking out lines removed.

More to come...

[phsyconoodler]

I love how you call making the cutouts and holes 'bashing'. Aussie's are hilarious!

 Nice work by the way!

 Save yourself some drilling when making square holes:I use a large drill bit in each corner of a square hole and then use a jig saw with a metal blade to cut it out.Takes seconds on aluminum.Any touch up can be done with a flat file.

[andrew_k]

Howdy stranger 
LTP or cathodyne? I know Mal prefers the cathodyne, I vaguely remember you expressing something similar many moons ago?
Hope to see you again next Ampfest :clock:

[darryl]

I love how you call making the cutouts and holes 'bashing'. Aussie's are hilarious!

Often my metalworking could be described as an act of violence against an unsuspecting chassis. It also generally includes a degree of self-harm. A point for discussion - should blood spilled during the construction of an amplifier be left in situ, to add to the amp's mojo? :icon_scratch:

Thanks for the tip on the jig saw. :icon_thumright:

Howdy stranger 
LTP or cathodyne? I know Mal prefers the cathodyne, I vaguely remember you expressing something similar many moons ago?
Hope to see you again next Ampfest

Hi Andrew,
               Good to meet you here, so far from home...    I will be attending Melbourne Ampfest again this year, so you can finally pay up for the "use a 6V6 as a phase inverter" challenge of last year.

This amplifier will use a cathodyne PI, as do most of my amps, but you're pre-empting my narrative.

[andrew_k]

 

[phsyconoodler]

I lose lots of blood drilling and 'bashing' my amps.

[DummyLoad]

in that case, let the blood flow...

carrion...

--ISO

[darryl]

Back to the workshop.

At this point the circuit to be used in this amplifier should be outlined. My computer skills do not yet extend to drawing schematics - I'm currently attempting to conquer ExpressSCH, but it will probably take me much longer to draw a schematic than to build an amplifier.  

The amplifier flow chart is as follows:

Input ->   12AX7 -->   Tone & Volume   -->    12AX7   -->    Master   -->   6BM8 Triode   -->   2 x 6BM8
              Triode             Controls                 Triode           Volume          Phase Splitter          Pentode

The chosen chassis has very limited front panel space, so a simple control set was used to optimise the space available. There are three pots - volume, a one-knob tonestack ( is one knob a stack? ) and master volume. The tone circuit is the same as used in the Watkins Dominator and later in the 18-watt Marshall - it has fairly low insertion loss and little interaction with the volume control. A master volume was considered necessary, because even a 5-watt amplifier can be too loud in a domestic setting.

On with the construction:





The first step was to fit rubber grommets where leads would pass through the chassis, and then to rivet the valve sockets in place. The tagstrips were then riveted in place. A riveted tag can not be used as a reliable earth connector so the earth tags at the end of each row were bolted down using a machine screw and nut, with a star washer between the tag and chassis and another star washer between the tag and nut. 18-gauge tinned copper wire was then run beneath the tagstrips and soldered to each earth tag. These tags could then be considered reliable earth points. This "earth plane" construction is at variance with the conventional wisdom regarding star earthing but in this layout it does not introduce unwanted noise or instability.

Next was the power transformer - a fully encapsulated toroid, originally intended for mounting on a printed circuit board. Standard practice for toroidal transformers is to mount them with a large single bolt through the centre of the toroid. In this amplifier the single bolt mounting was not suitable, because even a small amount of free movement would have caused the transformer terminals to short circuit against the chassis. Fortunately the transformer designers must have had this possibility in mind, because there was an alternative mounting arrangement available, using four smaller diameter bolts.

The output transformer was also installed at this stage. Exposed tags on the primary side of the transformer were shrouded with heatshrink tubing before the transformer was mounted, and its leads passed through the chassis. Twisted pair heater wiring was run from one 12-volt secondary of the power transformer to the three valve sockets. The optimum heater voltage is 12.6 volts, but 12 volts was considered to be within an acceptable tolerance.



The IEC power connector was riveted into the chassis, and a lead run from its earth pin to a bolted down chassis lug. Next 240VAC power wiring was run between the IEC connector, the power switch and the power transformer. Heatshrink was used to cover exposed terminals.

Most of the power supply components were now installed. This power supply is a voltage quadrupler, so there are four diodes and four electrolytic capacitors in the B+ supply. Three additional power supply nodes were required for the output valve screens, the phase splitter and the preamp stages - these were installed as the other small components were being added to each stage. The large radial lead electrolytic capacitors were lightly glued in place with some contact adhesive, to reduce stress on their mounting leads.

Two 120 ohm 1 watt hum balancing resistors were fitted from each side of the 12-volt heater supply to earth. One potentiometer had to be fitted before the electrolytic capacitor immediately behind it was installed. This can not be considered good layout.    It will be corrected in the Mk II version of this amplifier...

Still more to come...

[Fresh_Start]

This "earth plane" construction is at variance with the conventional wisdom regarding star earthing but in this layout it does not introduce unwanted noise or instability.

Really interesting approach, and this proves that any grounding scheme is fine - provided it works

Did you do something with the PT terminals to shield them from the edges of the long, skinny access hole?

Thanks for the detailed blow-by-blow with pics.  

Any chance you can share the poweramp circuit with the paraphrase PI using 6BM8 tubes?  I understand that you're in business to make amps so please don't worry if you want to keep the schematic (or parts of it) to yourself.

Cheers,

Chip

[darryl]

Did you do something with the PT terminals to shield them from the edges of the long, skinny access hole?

The limited clearance was the prime motivation for using four mounting bolts on the power transformer. A single bolt through the centre of the toroid would have left the transformer free to rotate, causing the terminals to short against the chassis. I have used heatshrink to completely cover the 240 volt primary connections, but I haven't taken any extra precautions on the secondary side. I don't think the transformer is going to move, unless subjected to extreme violence. Now you have asked the question, I will give it more thought... 

Any chance you can share the poweramp circuit with the paraphrase PI using 6BM8 tubes?  I understand that you're in business to make amps so please don't worry if you want to keep the schematic (or parts of it) to yourself.

There are a couple of points here:

1.  Amplifier building is my retirement hobby, not a truly commercial business. I do sell a limited number of amps, in the hope that the hobby will at least be self-funding, although this is not yet the case.      All my designs use standard 'building blocks' - there is no secret mojo or proprietary circuits - although I will often make joking references to the contrary.

2.  I'm interested in the advice/experience of others here. I did post a circuit for one of my early projects at AGGH and the response was rather disturbing. I received a number of enquiries from constructors who obviously had absolutely no experience with electronics, let alone valve amplifiers. This leads to a concern about liability - if a forum member publishes a circuit, and an inexperienced constructor misreads it and fries in their own fat, can the forum or the member be held liable? This must also be a legal minefield for businesses supplying amplifier kits.

3.  Once I've mastered ExpressSCH I will post circuit details for discussion, but perhaps not a complete schematic, unless the liability issue is laid to rest.   

[Geezer]

a concern about liability - if a forum member publishes a circuit, and an inexperienced constructor misreads it and fries in their own fat, can the forum or the member be held liable?

In today's crazy world, I can see someone mearly suggesting on a forum "You should build an amp", then the guy he suggested it to gets fried, & then the court finds the original suggester liable. I know it goes against all common sense, but pretty much everything these days is against common sense. Oh, for those "good'ol days" of personal responsibility.

Once I've mastered ExpressSCH

The easiest thing to do is simply start with a schematic that is already posted & just move stuff around to make it match your circuit. The learning curve is really small once you get going.....

[PRR]

> an inexperienced constructor misreads it and fries in their own fat

Is there a case where this happened?

Sure, anything is possible. But researching (USA) accidental electrocutions, they are overwhelmingly construction-site or consumer products. Professional "electricans" get zapped a lot, but professional "technicians" have almost no electrocutions.

I've taken 600V through a finger, and I know better. I *am* surprised we don't see more dead amp-builders. But I can't find any.

Even musicians, who often use damaged gear on wet stages while hopped-up on strong drugs.... there was a cluster of deaths in the 1970s, but the death-toll since has been very slim. (And at least one notorious musician electrocution had nothing to do with music or amps: he touched a lamp while in the bathtub.)

[RicharD]

>Professional "electricans" get zapped a lot,
I got zapped on Tuesday.  I've gotten zapped 2 or 3 times in a single day.    I'm more likely to get killed falling off a ladder than by electrocution.  The likelihood of getting killed by an amp is tremendously lower than the possibility of simply getting hurt by an amp.  Who hasn't burnt themselves on a hot bottle?

[darryl]

The likelihood of getting killed by an amp is tremendously lower than the possibility of simply getting hurt by an amp.

Here in Australia the line voltage is a very dangerous 240 volts, so getting "zapped" is seldom a trivial matter. A cavalier attitude to electrical safety tends to severely limit an electrician's career prospects. 

[RicharD]

I have almost 20 years experience as an electrician.  I've done some rather cavalier tasks in my day.  I have a strict policy with my employees, "If you don't feel safe doing something, don't do it."  Of course this usually means I am the one who has to deal with it then.  There is seldom any situation where one can't turn off the circuit to service it.  Can't is a 4 letter word for, "I don't want to."  I am very safety conscience, especially with ladders.  Being cavalier isn't what gets you, plain ordinary carelessness is. 

Obviously risks increase with an increase in potential difference.  It's always amazed me that 1/2 the world uses 240V as standard wall power.  The obvious benefit is more watts for the same ampacity, but at an increased risk.  Here in the states, our larger office buildings operate at 480/277V wye.  The difference between a 120V and 277V shock is well..... shocking (sorry, couldn't resist).  Apply that to amplifiers and one could say working on a Champ is much less riskier than working on an old SVT with 6146 tubes and a B+ approaching 1kV.

You'll find a lot of encouragement here, but you'll find a lot of safety reminders too (bleeding caps, lamp limiter, probing single handed, etc.).  We should probably once in a while point out what many of us consider obvious such as proper foot-wear, eye protection, gfci receptacle on your work bench, don't hold your geetar while poking around in a live circuit.     

[darryl]

We should probably once in a while point out what many of us consider obvious such as proper foot-wear, eye protection, gfci receptacle on your work bench, don't hold your geetar while poking around in a live circuit.     

You're quite correct - although these basics are second-nature to us, we tend to forget that there are less experienced constructors who may not have any formal ( or even informal ) instruction in electrical safety.

Also, safe working procedures in one branch of electronics are not necessarily appropriate in another. I built guitar amplifiers commercially in the 1970's, but moved on to other things, eventually to repairing computers. I had some difficulty 'reprogramming' myself to always having one hand on the grounded computer case, to guard against static buildup, while probing a computer's gizzards with the other hand. Once I restarted building valve amps, I had to return to the one-hand-in-your-pocket rule, and it's taken some effort for it to become second-nature again.

A digression: I never worked on any of the SVT's which used 6146's but in the 1970's there was an Australian-made amplifier called the "Brick" which delivered in excess of 200 watts using four 6146's, with a B+ over 1000 volts! They had no bleeder resistors and could hold a dangerous charge for hours. Once you had one running on the bench, you didn't find the B+, it came looking for you.

[billcreller]

...."you didn't find the B+ , it came looking for you"    sounds like the IRS...

[darryl]

A little more progress...



Installation of small components around the output stage and phase splitter were next. The speaker and line out sockets were fitted, and a 10 ohm, 7 watt resistor was mounted between them. This resistor acts as a dummy load, allowing the amplifier to be used as a preamp without requiring an external speaker to be connected.

This amplifier has a conjunctive filter across the output transformer primary. To keep the very large signal voltages present at the OT primary away from more sensitive parts of the circuit, a small piece of tagstrip was bolted to the amplifier's back panel, and the resistor and capacitor which make up the conjunctive filter were attached to this strip.

One 6BM8 triode section was used as a cathodyne phase splitter, while the other 6BM8 triode section remained unused. All three terminals of the unused triode were tied together and earthed. Valve data books always quote earthing unused valve elements "good practice". It probably isn't that important at audio frequencies, but could be critical when designing RF circuits.

I'll try to have some circuit information in the next few days - but it might be hand drawn... 

[Fresh_Start]

The easiest thing to do is simply start with a schematic that is already posted & just move stuff around to make it match your circuit. The learning curve is really small once you get going.....

Given the circuit you've described, Geezer's "Little Wing" SCH file probably is a good starting point.  Here's the thread with links to schematic(s):  http://www.el34world.com/Forum/index.php?topic=562.0

On the safety front, a thread a week or two ago got me thinking that a "safety sticky" would be a good thing to add to this forum.  Most of us have good safety habits for working on amps, but good old repetition always helps.  I've already drafted part of it with multiple links to Aiken, Paul Ruby, and Drifter Amps safety-related info but would really appreciate input from the rest of you all.  If you have links and/or ideas you think should be included, please send me a PM.

Cheers,

Chip

[darryl]

There's no further construction detail to report, but there is now a schematic of sorts. I think the circuit is correct - feel free to point out any errors or incorrect use of ExpressSCH. Apologies to Geezer, who's "Mini Bassman" schematic was the one I plundered for useful symbols for my drawing.

I'll complete the amp construction details in the next few days, and attach some voltages to the schematic.

[darryl]

The schematic posted above has had a minor update. The plate resistors on V1a and V1b have now been corrected from 120k to 100k. The power LED has been included, and some no-signal DC voltages have been added.

On with the construction...



The three potentiometers were fitted, and were used to hold the engraved plastic faceplate. The three-position gain and bright switches, the input socket and the power LED were then installed. Shielded cable was run to and from the master volume pot. Work then continued on installing the small components around the preamp and tonestack. There were insufficient tag points on the tagstrips originally fitted, so a short 3 tag strip was carefully added near the power transformer secondaries. It was used to anchor components associated with V1b, including the zener diode clipper before the master volume. This is something else to be corrected in the Mk II model. 

Looking at the schematic, the tone circuit requires a 0.01uF capacitor, but the components installed are a parallel pair of dissimilar 0.0047uF caps. One of these is a polyester "mustard" cap, the other is a polystyrene "styroseal" cap. I could claim this is a source of tonal "mojo", but the practical reason is a shortage of 0.01uF, and a surplus of 0.0047uF capacitors. Perhaps some highly regarded amplifiers with similar strange capacitor combinations are so for similar prosaic reasons...

[darryl]

Above chassis views...





Once the under-chassis wiring was complete, the front panel knobs were fitted and identifying labels attached to the back panel and next to the valve sockets. The valves were plugged in for this photo opportunity, and then removed for initial testing of the amplifier.

First the line voltage was tested - it measured 244 volts AC. The line voltage in my workshop can range from 232 volts ( hot midsummer day, lots of airconditioners running ) to 250 volts ( mild weather, late evening ). A figure of 244VAC is quite close to the nominal 240, so test results have not been skewed by an incorrect line voltage.

Power was applied to the amplifier with no valves installed, while monitoring the B+ voltage, which rose quickly ( faster than the multimeter sampling rate ) to 240 volts DC. The heater voltage was 14.5VAC. There were no fireworks or clouds of acrid smoke, so power was removed, the capacitors allowed to bleed down, and the valves fitted. The speaker output was connected to an eight ohm dummy load, which was being monitored with an oscilloscope.

In the following description I will round measurements to sensible levels of accuracy - 333 ohms will become 330, 199.5 volts will become 200, and so on...

Power was reapplied and the B+ voltage and output stage cathode voltage were monitored as the valves warmed up. Voltages were within an acceptable range, so a 400Hz signal was applied to the input and with the master volume full on, the volume control was turned up. A reasonably clean sine wave was produced, but it had obvious crossover distortion, so the output stage cathode bias resistor was too high in value. The original resistor was 500 ohms - two 1k 1 watt resistors in parallel. A third 1k resistor was added, making the total 330 ohms. Crossover distortion was reduced significantly, so this arrangement was retained for further testing.

Some voltage measurements were taken. The B+ voltage at first turn-on with cold valves was 233VDC, but once the valves had warmed up, it fell to 216VDC - close to the expected result. The output stage cathode voltage was 16.5VDC, so with a cathode resistor of 330 ohms, Mr Ohm declares the cathode current to be 49.5mA. The voltage drop from cathode to anode is 200 volts, so the total dissipation in the valves is about 10 watts, or five watts each. I'm not considering screen current in these calculations. The maximum plate dissipation quoted for 6BM8's is 7 watts, so we are around two-thirds of the maximum - a typical figure for a class AB output stage.

Some measurements were taken to check the power output. The amplifier was driven to the onset of clipping - as observed on the oscilloscope trace. The output voltage was then measured, with a true-RMS multimeter, at 6.7 volts, so the power output was 6.7 squared, divided by 8, or 5.6 watts. The waveform was showing signs of crossover distortion while running clean, and serious crossover distortion when overdriven.

One reason for this was the rise in cathode bias voltage as the cathode current increased. The bias voltage rose from 16.5 volts at no signal, to 23.4 volts at maximum clean output. A way of reducing this effect is to use a zener diode to clamp the bias at a value slightly above the no-signal level. In this amplifier, an 18-volt zener was appropriate, but as none were immediately available, a 12 volt and a 5.6 volt diode were connected in series. Retesting the amplifier produced an output of 7.1 volts, so a power output of 6.3 watts. The waveform also showed much less crossover distortion.

More to come, but for now... 

[supro66]

Why does the tube tag say 12A#7 and not 12AX7

[DummyLoad]

very nice work darryl. 

[darryl]

Why does the tube tag say 12A#7 and not 12AX7

Good spotting.   

That is to indicate that any of the 12A#7 family of valves ( 12AT7, 12AU7, 12AY7, 5751, etc. ) can be used in that position. Using different valve types will alter the gain and tone of the amplifier.

[phsyconoodler]

Very nice build!
  I would have used a tag board,but that's just me.
Point to point amps look cool.Your's looks like the best example of a Matchless or Badcat I've ever seen.
  Compact and functional.
Also tons of work!!

[andrew_k]

Lookin' good Darryl. Thanks for posting a schem, I enjoyed taking a closer look.

Also tons of work!!

You should see a gutshot of the monsters he built in the 70's!  :shocked:
I tried to find one, but no luck.

[PRR]

> around two-thirds of the maximum - a typical figure for a class AB output stage.

"AB" covers the whole range from pure-A to pure-B.

A happy self-bias amp must work nearer class A. Why? As you proved:

> serious crossover distortion when overdriven.  ...rise in cathode bias voltage as the cathode current increased. ...from 16.5 volts ...to 23.4 volts...

The idle current is whatever you want. The Full-Load current is whatever the load draws.

When these are not equal, the bias changes. Steady-state, the stage tends to crossover badly. Transient action is more complicated, but generally not nice.

You can adjust load impedance so that full-load is more similar to idle current. As a first-crack, go higher by 23.4/16.5 or 1.42 times, say 5K to 7K. However you won't get there with simple 4/8/16 speaker changes, and your OT selection is very limited. And this way reduces maximum power output, perhaps from 6.3W to 4.4W. 1.5dB is just about audible.

Or you can set idle current high enough to cover full-load current. 23.5V/333r= 70.57mA times your 200V is 7.06 Watts per tube. The tube is supposed to stand 7W for thousands of hours; indeed that would be the typical use (as a TV's SE audio amp or as a video amp). The self-bias reduces any tube drift. The 7W rating includes some margin for mass production where you won't test/measure every unit; you are measuring your unit. And in guitar-amp work you spend much time at high power outputs, where dissipation actually drops. (I had one tube which creaked as it cooled when driven from zero to 13W out on dummy-load.)

And when tubes are cheap, a reduction of life from 5,000 hours to 3,000 hours may be very acceptable. I gather a lifetime supply of these bottles may cost less than one good pair of 6V6.

Of course at 1.43 times idle current the PT and chassis will run that much hotter as well. That may bring surface temperatures to uncomfortably hot; or caramelize spilt beer faster than you can grab a rag.

It is quite common to allow 20% rise of current. This improves power-output per Watt of costly dissipation. That suggests 70.6mA/1.2= 59mA idle, 6W per tube, plenty safe.

The fixed-bias situation is different. Bias is not affected by current. We may select idle bias current very different from full-power current with far less conflict. And we would typically be pushing other tube limits (particularly plate voltage and cathode current limits), so it would be wise to stay away from steady rated dissipation. Hence about 25% for battery jobs and about 70% for a wide range of wall-power amps.

Your Zener clamp transitions the stage from self-bias at idle to fixed-bias at high power. _I_ think this is a "trick", but it does indeed fix the bias-shift, somewhat abruptly. This gets into the question of "what impedance is the cathode looking into?" With unbypassed common cathode resistor, the transition from 100-200 ohms (the 333r plus the other tube's cathode) to the Zener's ~~3 ohm impedance may kink the curvature. Your 47uFd gives low-low impedance at 400Hz, so no kink; but 82Hz may be colored (which may not be a bad thing).

[darryl]

A little more testing. As PRR has noted, a single secondary winding doesn't offer many output impedance matching options. The optimum primary impedance was originally calculated to be 4500-5000 ohms plate-to-plate. The actual specification of the transformer used was 7000 ohms plate-to-plate, with an eight ohm secondary. Output voltage was measured with 4, 8 and 16 ohm loads, and the output power calculated.

        4 ohms     8 ohms     16 ohms
        4.8VAC     7.1VAC     8.6VAC
         5.8W        6.3W       4.6W

This amplifier can be used to drive either a 4 or 8 ohm load and it will deliver its quoted output of 5 watts. It can be used with a 16 ohm load, but the power output is less than 5 watts. Plate and screen dissipation was within the valves' design limits at all load impedances.

There are several aspects of the phase splitter and output stage which need some explanation. These are design choices made to produce a "smooth" overdrive, without the typical harsh and "fizzy" sound of a low-powered amplifier being driven hard:

1. The conjunctive filter across the output transformer primary.

2. Output stage screen supply. This takes the form of a 1k ohm, unbypassed resistor, common to both screens, with individual 47 ohm "stopper" resistors on each screen. The 1k resistor produces a substantial drop in screen voltage under load, protecting the screens from excessive dissipation, but also producing a noticeable compression effect when the output stage is driven hard. Having no bypass capacitor on this resistor allows some negative feedback via the screens, which slightly reduces the gain of the output stage, but also reduces crossover distortion - always a good thing! I note with interest recent discussion of this very subject: http://www.el34world.com/Forum/index.php?topic=9025.0

3. Large value grid stopper resistors on the phase splitter and output valves. This is a very useful circuit device championed by Merlin on his website and in his excellent book.

4. The zener diode clipper across the master volume. This serves two purposes depending on the master volume level. If the master is full on, the voltage necessary to clip the output stage can be exceeded by 50% before the zener clipper operates. This allows maximum power output and smooth overdrive, while limiting the harsh distortion possible if the phase splitter and output stage are grossly overdriven. As the master volume is turned down, the clean output of the amplifier can be reduced, and an overdriven output produced at lower volume. This is preamp and clipper distortion rather than output stage distortion - not as smooth, but still surprisingly acceptable.

After all that talk, a couple of pictures. The amplifier can be used in the "naked chassis" form shown above, but if it is to be moved about, the chassis can be fitted to a basic protective sleeve:

[tubenit]

Very nice!  Impressive work.  I like it!

With respect, Tubenit