SE outputs and higher voltages?

[92Volts]

Hi, I made a post about a giant PP amp to take advantage of a 240/480v AC transformer I have (for 330v screens, 660v plates). I won't be going forward with that, at least not soon-- it's impractical, and very expensive (despite the part I already own).

But it's got me curious about higher voltages and the possibilities multi-voltage power supplies open up (they seem almost necessary for higher voltages, since screens rarely tolerate anything close to the uppermost plate ratings). In particular, neither datasheets nor any reports I can find mention a SE EL34 with much over 400v on the plates.

I can think of a few practical reasons for this:
1. SE amplifiers have a hard efficiency limit at 50% (which you still won't reach) so the possible benefit is limited.
2. Higher-power and higher-impedance SE iron is elusive and/or expensive.
3. High voltages suggest or require dual-rail power supplies as I mentioned... which leads to:
4. For all of these reasons if you want more power, push-pull is probably cheaper, easier, and more efficient.

Having mentioned the efficiency limits however, datasheet examples provide 11w, so there is room for improvement. PP also has theoretical limitations, but raising plate voltage is sometimes worthwhile, bringing those configurations from 50w to a "nearly ideal" 100w. So I don't see a reason a higher-efficiency (and power) SE output is impossible, though it would be impractical.

Any thoughts? I know push-pull is in many ways a better answer to the question of "more power", but I'm curious why I never see this tried. Am I missing something, which will make this harder on the tubes than high-voltage PP configurations, or some reason it will have effectively zero benefit with regard to performance?

[shooter]

Any thoughts?

HV comes with more problems than it's worth in amps;
G2 volts, arcing, better chance to die WHEN you screw up, drive signal required...........
look into the big ham rigs for ?solutions?

[John]

I know push-pull is in many ways a better answer to the question of "more power", but I'm curious why I never see this tried.

It's probably already been tried.


 I like SE amps myself, but (IMO) it's a great marriage because you get that overdriven/saggy kind of sound that you would, (I think) get in a 35W PP amp when it's dimed. But without the higher volume. The reason you "never" see it tried is because of all things you brought up. Extreme weight being the biggest deal to me, and probably most anyone else over 50. 


And Shooter brings up a great point about the higher voltage. Someone else put it best: when you're dealing with 400 volts, you find it. When you're dealing with 1000, it finds you. I've been hammered by a disharging (because I was careless) 450V cap, and I didn't like it. I'd hate to think what it would be like getting nailed with 800 or so.

[PRR]

> push-pull is probably cheaper, easier, and more efficient.

SE is for very small amplifiers. When you reach the limit of a single tube, you ALWAYS go to P-P.

> {no} ...SE EL34 with much over 400v

In SE your power potential is almost always limited by Dissipation, as you note.

Going higher voltage forces lower current to stay within dissipation limit.

High voltage and low current forces a higher load impedance. Which is typically a more expensive or less-bandwidth transformer.

Voltages over 400V eliminate the economy of electrolytic filter caps.

Nearly all the Audio Power Tubes were *designed* to do well at 250V to 400V in SE.

But see 845, 100W power triode. H-Fi guys build amps on these. Not always with happy results. The driver (145V-195V peak!) has to be as heroic as some 100W 6550 amps, to flog a lousy 30W from the 1200V 845 final.

[92Volts]

Thanks everybody. It sounds like there are "only" practical reasons not to do this, but those are legitimate reasons. "Standard" voltages are used for a reason so you will run into design and parts availability issues as you deviate from them.

Now, an EL34 (for example) should survive 650V SE plates with appropriately-low screen voltage, but the PSU and OPT requirements mean this isn't a cheap or easy experiment.

PRR, thanks for the info about the 845s. I feel like that case is extreme due to it being a triode, not only because it's a big SE amp but... I get the point. This is not an "elegant" (or efficient, or cheap) way to make lots of power.

Interestingly, I'll be working with a dual-rail supply soon enough. I found a deal on a broken Fender Champ 12, which uses a voltage doubler for 250v pre/screens and 500v for the plate of a 6L6. This amp has PCB construction with some unknown problem  as well as infamously-crappy reverb and overdrive, so I might put in a turret board and convert it to a Vibro-Champ. Should be interesting to play with that power supply and output (maybe 250v on the plates could be workable with a 6V6 or... something)

[kagliostro]

If you maintain ~250v on G2 and the ~500v on plates of the 6L6 you can use the stock OT

otherwise I think you must change OT

Franco

[PRR]

> extreme due to it being a triode

HK-257 is an indirect-heat Pentode of similar ability. Incidentally it will do OK at "just" 500V supply. Look at the Class A 1 Tube Operation data. The HK company folded but an equivalent tube was made under another number,  4E27. Some guys here got one, breadboarded, got it lit up, and decided it was a bit frightening and much too heavy to be practical as a guitar amp.

[shooter]

Incidentally it will do OK at "just" 500V supply

fwiw, when I 1st started dreaming big here, I envisioned a PSE KT88, I still have a pair of OT's, and an 11lb choke if you need parts

[DummyLoad]

> extreme due to it being a triode

HH-257 is an indirect-heat Pentode of similar ability. Incidentally it will do OK at "just" 500V supply. Look at the Class A 1 Tube Operation data. The HK company folded but an equivalent tube was made under another number,  4E27. Some guys here got one, breadboarded, got it lit up, and decided it was a bit frightening and much too heavy to be practical as a guitar amp.

i was that dolt. worked fine, sounded really good. impractical and would have weighed around 30lb for just the chassis. was driving it with a 6948 subminiature, (hopped up 6021) and running at 535V B+.  hum was prevalent but not bad when playing. IIRC, measured output power was disappointing from published specs but we were running auto-bias rather than in fixed bias.

wasn't scared of it, just was impractical since the anode connector NEEDS to be a heatsink as well, and that forced a custom build of a suitable anode cap to something that i wasn't prepared to fabricate at the time. we mulled over over an switched-locking enclosure, but lost interest at that point as it wasn't WOW better than properly designed SE KT88.

it was dubbed the "glo champ", see pics in link below for visual reference of the name.

pics of the experiment here

BTW, the HK257 is directly heated pentode and not indirectly heated. 

--pete

[92Volts]

That sounds like an interesting build but with the availability of KT-150s and similar scaled-up "audio" tubes it would be less necessary to work with a tube like that.

Incidentally it will do OK at "just" 500V supply

fwiw, when I 1st started dreaming big here, I envisioned a PSE KT88, I still have a pair of OT's, and an 11lb choke if you need parts

I have absolutely no idea what those are worth, but I might be interested in buying them... for some reason

[92Volts]

If you maintain ~250v on G2 and the ~500v on plates of the 6L6 you can use the stock OT

otherwise I think you must change OT

Franco

Just got the amp yesterday. Not super broken (plays clean and switches, with volume issues in the overdrive channel) and might just need a new pot or bypassing broken PCB traces.

If I did rework it, I'd keep voltages the same. Turns out the power supply is on its own PCB and out of the way, which is convenient. Tubes are also mounted in the chassis and wired to the PCB, not mounted to the PCB, making this a good platform for rewiring.

However the OPT is tiny, much smaller than the one from a blackface Champ (6V6). I don't know about impedance but in general this doesn't look like the "right part" for a 6L6GC! I guess OPT overdrive/saturation is part of the sound of these small amps, but this specific one was also built cheaply in a number of ways.

[jjasilli]

Here's some straightforward design criteria from the Valve Wizard:  http://www.valvewizard.co.uk/se.html


Some thoughts:


*  I wonder if it's possible to bias an SE tube in Class A at very high voltage?
*  Doubling the output of the amp in Watts yields only a 3dB increase in volume (SPL); so there's little bang for the buck
*  Output could be increased by paralleling SE tubes, w/o the issues of very high voltage


Still, those are practical considerations.  DIY'ers are happily not bound by practicality.

[kagliostro]

However the OPT is tiny ..........

May be they used a "poor" OT or they used high quality laminations

the use of high quality laminations permit to shrink the dimensions of the OT

Franco

[92Volts]

May be they used a "poor" OT or they used high quality laminations

the use of high quality laminations permit to shrink the dimensions of the OT

Franco

Maybe a bit of both. Part number matches the reverb transformer from a Vibro King, 5.4k ohms and driven by an EL84 in its original application. So it seems undersized (certainly not purpose-made for this application) but replacements are rated at 10w. It might perform better than a blackface-era OPT of the same size, and I haven't seen reports of them failing. Likely wouldn't make for a good bass or hifi amp at these power levels though!

[kagliostro]

Part number matches the reverb transformer from a Vibro King ..........

 

that isn't a good presentation card (for the use we are talking)

Franco

[92Volts]

that isn't a good presentation card (for the use we are talking)

Franco

To be fair, just replacement companies call it that. Fender (clearly) considered it multi-purpose, but this amp was a later and less common use for it than the Vibro King.

This is a super cheap amp though. The switching circuit is powered by DC from a second series cathode resistor, which sacrifices more screen voltage. But saves parts in the power supply! And of course the voltage doubler is already an unusual choice.

It's a fascinating design, in all the laziest/cheapest ways possible.

[PRR]

> I wonder if it's possible to bias an SE tube in Class A at very high voltage?

Why not?

Why??

High voltage and given Max Pdiss leads to low current. High voltage and low current leads to high impedance winding. Which costs more and has less bandwidth. Power output is essentially un-changed.

If the datasheet for the tube (or a similar tube) shows an SE condition, that is about as good as it gets. To suit an available PT, you can lever some higher/lower V or I, and thus Z, but the Power will not improve.

[jjasilli]

Ok, thanks for the explanations.  So, there's no point in designing for hi voltage Class A, whether SE or PP.  To get more power you at least one more power tube: either in parallel SE; or PP Class A. 

[shooter]

no point in designing for hi voltage Class A,

I've experimented enough now to figure out max plate can be had by many roads, but at a certain point you gain nothing.  I think a KT88 sounds better at 400v than 300, both running at or near max diss. by 500v I can't hear the difference, below 300 gets really un-usable, even at max diss.

I do believe DL is on the right path to get more good roar, but as you state, by going PSE

[jjasilli]

Good points.  Focusing on small bottle power tubes, the EL 84 states a max plate voltage of 300.  But Traynor runs them at 410; Trainwreck at about 325(???). In very old threads, Psychonoodler opined that it's the plate diss, not the plate voltage, that really counts.  Various voltages can be made to operate, but sound different.

[kagliostro]

Different Voltages equal different Load

different Load means different OT construction

May be the intrinsec caracteristic of different OT have a large impact on bandwidth and the final result ?

Franco

[92Volts]

There are challenges for making higher-impedance iron, this page talks about wider frequency response, but you can get the idea of some of the tradeoffs when you go for more "extreme" performance (particularly when it requires high inductance):
http://www.cascadetubes.com/2014/02/08/a-consequence-of-the-quest-for-20hz-roll-off/

Good points.  Focusing on small bottle power tubes, the EL 84 states a max plate voltage of 300.  But Traynor runs them at 410; Trainwreck at about 325(???). In very old threads, Psychonoodler opined that it's the plate diss, not the plate voltage, that really counts.  Various voltages can be made to operate, but sound different.

I've seen a lot of reports that screen voltage, and therefore dissipation, is the most common problem. I believe it-- the load impedance soaks up voltage (and therefore power) from the plate but not from the screen, meaning dissipation is many times higher than idle, often too high, but this is rarely taken into account.

EL84s and especially 6V6s happily run wayyyy over their ratings, but EL34s are considered reliable at way less than the ratings... partly because the screens aren't meant to be run as high as the plates. Or as others have pointed out in this thread, the expectation was 99% of the time the screens would be run at the same voltage as the plates, but that voltage would be a hell of a lot lower than 800.

[drgonzonm]

No one has discussed split-load amplifers, You can use PP o/ts and get the class A.  Ampbooks discusses using pentode tubes, in Push-Pull without a PI. 

The following was from another thread
an article found in Wireless Engineer by Amos dated February 1946, With the Title "Push-Pull Analysis, Subtitled "Cathode-Coupled Output Stage".
   c. Put together a library of resources.  This includes the article mentioned above, and copies of the two patents mentioned in the article. (#492,407 and #508,697)   I would recommend another ariticle: Compact Hi-FI power amplifier by melvin liebowitz/ Delaware Electronics Supply Co.  subtitled "Constructing a 6-watt amplifier in which the output tubes serve as their own phase inverter.  power tubes are EL84's.  This article is no where as detailed as the Amos article. 


 
I left a lot out,


 
 

[shooter]

the intrinsec caracteristic of different OT have a large impact on bandwidth

I've seen BW thrown around a bunch in this discussion, In guitar world what is a "good" BW? as a bonus what is the "typical" Center Freq you would design for?

[jjasilli]

@drgonzonm:  I mentioned (or referred to) Class A PP in Replies 17 & 19. 


@92Volts:  The cascade tube article is interesting.  But I'm not ready to agree that the small difference in load impedance is the cause for OT popularity, for a number of reasons.  The difference in load impedance is so small as to be withing the margin of error.  Also, popularity does nor correlate well with measurable sonic differences.  E.g., I love the Hammond Co., but would be reluctant to put a tranny with that name in a hi-fi build.  Even if that's objectively "wrong" I suspect it's a common prejudice. 


@shooter:  another good point.  For bass, we're talking about 40Hz bottom fundamental.  Objectively there might be measurable sub-harmonics in the lower 20Hz range, even if they're not "supposed" to be there.  Maybe you can hear them, but you can feel them.  Guitar can produce audible sub-harmonics below 80 Hz.  Subjectivity has crept in to an otherwise beautifully objective discussion.


Often there' more noise than signal below 40 Hz.  So arguably, that's why people prefer lower bandwidth OT's. 

[shooter]

In my old life, A to C was the “BW”, anything outside that and you lost brownie points, fast!

The pic is from the speaker driven by a PSE 6V6, I’m thinking A to B is the “effective” BW since A could be anywhere on the neck, also looks like anything less amplitude than B would get lost in the weeds.   When I spec’d the OT’s I bought, I wanted something good for guitar and not half bad for hifi, anything rolling off < 40, >14k was acceptable.

[jjasilli]

This thread is constantly reminding me how little I know.  How should guitar amp performance be measured?  This has led to lively discussions in the past!


We have commingled hi-fi power amp, with guitar power amp, with bass amp performance.  We have ignored speaker behavior which arguably might be more important in terms of both ultimate frequency response, impedance peaks, and SPL.

Hi-fi is easier to measure; the standard is 1000 Hz.  Shooters example is close enough.  It's the midpoint of human hearing.  And hi-fi speakers are designed so that their nominal impedance matches their actual impedance in the 1000 Hz zone.  Makes it all deceptively nice & tidy.

But guitars are nearly out of their fundamental range at 1000 Hz.  If you play in the key of A in the middle of the neck, you're mostly below 440 Hz.  So where to measure?  I don't know!!!

It would be a yeoman's effort to test & compare a number of venerable guitar amp - speaker combinations at various frequencies; spectrum analysis; THD measurements, etc. Sounds like a great grad school project.   

[PRR]

> guitars are nearly out of their fundamental range at 1000 Hz.

Naked steel strings on a hardwood axe have LOTS of harmonics.

Some guitarists use Fuzz etc, which makes lots of harmonics.

Non-harmonic "noise", like string-slide, also matter.

I think one reason Fender amps got popular is his 100K plate loads tend to be flatter to 15KHz, while 220K-470K plate loads used in other amps tend to fall at 5KHz. One stage may not matter, but a chain of stages is smothered by small shaves.

[jjasilli]

Fender amps sound "musical"  to me.  But guitar speakers typically drop-off at 4.5KHz.  I don't know how to reconcile this.

[shooter]

I don't know how to reconcile this.

divide everything by PI, then play guitar til you "get it"

[92Volts]

Fender amps sound "musical"  to me.  But guitar speakers typically drop-off at 4.5KHz.  I don't know how to reconcile this.

There are tons of things other than plate resistors which affect frequency response, including response within the speaker's range. You could be hearing any number of other things.

But the rated frequency response of a speaker is usually when output decreases by -3dB (and does not come back up)... which makes more sense for hifi than guitar applications. Guitar speakers have easily 10dB of variation within their range, if we tolerate that much decrease the "highest" frequency is far above 5k or whatever is rated. Not quite 15k (looking at curves for some random Eminence speaker it looks closer to 7k) but you can definitely hear some sounds above the numbers on the spec sheet.

[jjasilli]

Good point.  Depending on the Emi speaker, you could be down say 15dB @ 7K or 8K Hz.  Overtones would still be audible.

[drgonzonm]

Just for giggles, look at the Landahl site, they make some very pricey and very good transformers. Look at the weight of SE O/T's. and compare them to the PP O/T's.  This is one source for high wattage SE O/T's. 

Regarding Hz, many of the Fender schematics show voltages with a 1kHz input.  PA were designed for 400Hz

As long as you are designing tube amps, Low frequency requires iron.  And once you limit your self to SE designs, even more iron is required. 

[PRR]

Guitar speakers typically rise (on-axis) 1KHz to 4KHz before the sharp drop.

Many classic tone-stacks, when set to middle settings, are flat 50Hz-1KHz, rise to 2KHz, then flat (+/- treb knob).

This must be an important zone, even though above the basic frequency of the notes.

An amplifier with 3 stages each down 2dB at 5Khz will be down 6dB at 5KHz, relatively "muffled" compared to a flatter amp, and not bringing-out the top-boost designed into many speakers.

[jjasilli]

Guitar speakers typically rise (on-axis) 1KHz to 4KHz before the sharp drop.

Yes, there's typically a big hump there.  See, e.g., http://www.eminence.com/speakers/speaker-detail/?model=Legend_1275

The hump is accompanied by a significant rise in impedance, about 20 Ohms under the hump, from a nominal 8.  Actually the nominal 8 Ohms is only true as a dip around 300 Hz.  This seems to indicate that sound "quality" from the speaker, at and above the "hump" frequencies, derives a lot from speaker resonance rather than from pure signal output from the amp.  This in turn would corroborate that speaker-cab selection is significant to tone.

[shooter]

Just for giggles, look at the Landahl site

I've been there 1/2doz times, I 1st drool, then cry, haven't giggled yet

[PRR]

> See, e.g.,  .... Legend_1275

That's an extreme example.

And will only be like that ON-axis. A Twelve is wide-spread below 600Hz, very beamy by 2KHz. This design combines paper property, cone angle, and beaming to give that big bump. Most designs aim for a lesser bump. And that bump is not evident off-axis or at a distance in a room.

> This in turn would corroborate that speaker-cab selection is significant to tone.

Really?? Yes, of course, the speaker is the solid guitar's real sound-board. Speaker selection is probably more important than anything you do in the amp; sadly more costly to experiment with.

[DummyLoad]

http://www.electra-print.com/singleended.php


owwwwccch!


--pete

[92Volts]

> See, e.g.,  .... Legend_1275

That's an extreme example.

And will only be like that ON-axis. A Twelve is wide-spread below 600Hz, very beamy by 2KHz. This design combines paper property, cone angle, and beaming to give that big bump. Most designs aim for a lesser bump. And that bump is not evident off-axis or at a distance in a room.

I actually have that speaker, and when I was looking at a datasheet it was that one too. I wouldn't consider it exceptionally bright or "peaky" in use. The peak is more extreme according to the datasheet, but the peak is still near 10dB for others too. Even the "American voiced" equivalent 1258 has a wider, but equally tall spike: https://www.eminence.com/pdf/Legend_1258.pdf

I would agree it has to do with the direction and/or measurement technique though.

[PRR]

HK-257  ...... equivalent tube was made under another number,  4E27.

Historical footnote. At the end of WWII, H-K ran a series of ads endorsing JAN standardization. Mar 1945 notes the new JAN number for the H-K product.

EDIT- Found announcement.