GoodSoundClub - Romy the Cat's Audio Site - Are commercial PSETs made?

| Search | Login/Register

Home » Audio Discussions » How to get a LOT of SET power. (107 posts, 6 pages)

Print Thread |

1st Post |

Page 4 of 5 (107 items)

Select Pages:

« 1 2 3 4 5 »

	Target	Threads for related reading	Most recent post in related threads	Forum	Replies	Views	Started
»	New	More power from Melquiades? More powerful tube?..	A new Icon Audio's MB81 Mono Blocks...	Melquiades Amplifier Forum	19	249586	09-01-2007
»	New	NAT Audio Magna -160 Watts of Single-Ended Class A..	The KR Audio ways....	Audio Discussions Forum	5	73113	11-12-2007
»	New	Claude Perrier and his tube seremony..	Claude Perrier and his tube seremony...	Audio Discussions Forum	0	15399	01-05-2008
»	New	Getting more power from SET vs. properly distorting SS...	Sound Board...	Audio Discussions Forum	4	49380	05-09-2011
»	New	The ULF cannel for my new listening room...	The Organic Bass vs. ULF Drivers...	Audio Discussions Forum	43	130727	07-29-2018

05-12-2011

Post does not mapped to Knowledge Tree

Romy the Cat

Boston, MA
Posts 10,156
Joined on 05-28-2004

Post #:	76
Post ID:	16250
Reply to:	16249

Too many unknowns in both cases

N-set wrote:
That all will give 5 stages if I'm not overlooking some posibility: 3 filtering stages, with the first one bypassable, the driver being at the same time the last filtering stage

It is 3 stages total not 5. You do not need 3 dedicated filtering stages to write 24dB filter. The same Marchanhd crossover does it via 2 stages, I am sure it would possible to use only one extra stage to with two LC filters as a supplement to two-stages idea Of cause no one would use the low-pass OTP slope as at this slope a transformer has a LOT of distortions. You want to be at least 2-3 octaves from low-pass slope….

N-set wrote:
It may as well happen that the hybrid Romy is testing IS the way to go and Bass SET (BSET) is a waste of time and money and power.

If Romy were be testing any hybrid then it would be identically in complexity question: what would be proper harmonics structure at a given volume and how to control it. The ONLY excuse would be that it is very narrow band amp and it might be much more forgiving in harmonic inaccuracy…

"I wish I could score everything for horns." - Richard Wagner. "Our writing equipment takes part in the forming of our thoughts." - Friedrich Nietzsche

05-12-2011

Post does not mapped to Knowledge Tree

N-set
Gdansk, Poland
Posts 617
Joined on 01-07-2006

Post #:	77
Post ID:	16252
Reply to:	16250

The only way seems to try and compare

Sure, me idiot. I wanted for some reason multiply the harmonic imprints.
2 LC will not give a chance to try 3rd order, no? Or is this
option not really needed?
Also the driver should be kept for what it's meant--driving the output, so no filtering
duties on it as it'll be fighting the grid currents etc. Still looking for the output tube...

Cheers,
Jarek
STACORE

05-12-2011

Post does not mapped to Knowledge Tree

Paul S
San Diego, California, USA
Posts 2,657
Joined on 10-12-2006

Post #:	78
Post ID:	16253
Reply to:	16252

Overall Gain

Maybe the best way to think about the X/O is that one way or another, it's effectively part of the amplification chain.

Don't the "2 stages" in the Marchand include offset for X/O losses? Sure, if the X/O is interstage, then it will need to clean up its own mess before the output tube cranks it up. However, no free lunch no matter where you X/O, with respect to level, phase, distortion and noise. Smaller parts at line level is nice; but it does have to stay at that level.

The only advantages I see for transistor, op-amp or whatever are efficiency and/or signal/noise. Op-amps can be made very quiet -- with some serious feedback.

ULF might be servo, with the coil in the power loop. Talk about feedback...

Best regards,
Paul S

05-22-2011

Post does not mapped to Knowledge Tree

N-set
Gdansk, Poland
Posts 617
Joined on 01-07-2006

Post #:	79
Post ID:	16299
Reply to:	16252

Have the cake and eat it

Although the spiritus movens of the whole idea seems to have found the
bass solution, the idea continues to live.

Thanks to the invaluable help of Stephie B, I've comeback to the GU81M like
a boomerang with some progress. There are 3 new things:

1) FFT vs Sound: as an ad hoc assumption let's accept that the limit of torelability is when
the 3rd hamonic starts to dominate the 2nd; weather it's right or not I guess only
the experiment will show

2) Selectable Ultralinear connection as one more degree of freedome to play with
power/distortion ratio

3) Incorporated low-pass filter around 12AX7 (I know nothing about x-overs, so correct me if there are mistakes).

Here are two versions: LC (Butterwoth???) filter and RC Sallen-Key type (Linkwitz-Riley):

Now the comments: there is a mistake, 6E5P has +214V on the plate of course.
Everything is DC coupled a la Chimera's Axiom,
which makes up for 73265876 power supplies
to supply 3 tubes (although the +/-200V 2mA is nothing). the -3dB point sims show at 5.5Hz.
The 6E5P driver is biased at -4V and not -3.5V to give it some
headroom. I'd love to have some provisions to adjust in some small limits
the bias of the driver and the ouput individually, as this all is very senitive.

Ad1) The sims show that the +/- optimum load is 2.9k giving 160W at the point when 3rd harm=2nd
(at I think around 10% THD).
(Stephie claims to have reached 187W with 6E5P pent. mode, but I could not reproduce it).
The tube can give much more, in excess of 200W but at considerable clipping. The search over
GU81M shows that amateur radio peple do use&abuse it quite hard, 450W free air is realistic,
and there are reports it can survive quite some more for years. Unfortunately I could not get
any feedback from audio users of GU81M.

Ad2) Thanks to Stephie I've resolved how to add an UL posibility -- since the grid needs it's own
supply the only chance is to add another winding on the OPT. This makes the OPT complicated, but
for a completely different reason than usually: the primary side now has 3 windings: the main plate load,
the selectable 0-20%-30%-40% screen load, and the compensating winding caryying filament current
to reach 130H@300mA with a reasonable core. I think the switchable UL shoud add some nice
degree of freedome, apart from changing the load. I could not simulate the UL as LTspice
gets stucked, so no idea how it "performs".

Ad3) I've assumed 12AX7 CF with 157k to be a good solution, following Romy's experience.
I know nothing about active filters, please educate me. The most "economical" seems to be
LC filters, but 99% of x-overs I've found googling (incl. the Marchand) are Sallen-Key RC
with positive feedback. Now, is S-K so popular because it sounds good or because
it uses cheap tubes and RC instead of costly chokes??? Any input highly appreciable.
I'm not even sure if my implementation of S-K is correct? Have to sim it.

The input sensitivity without the filters (still have to sim them) for 3=2 point is around 1.7Vrms.

Cheers,
N-set

Cheers,
Jarek
STACORE

05-22-2011

Post does not mapped to Knowledge Tree

Paul S
San Diego, California, USA
Posts 2,657
Joined on 10-12-2006

Post #:	80
Post ID:	16300
Reply to:	16299

Drifting Away

N-set, I am not the one to double check your mathematical solutions, that's for sure. I wonder if in practice the curve values here are effectively tied to variables? I like the Chimera idea, but it has no "locked curves" to contend with, and it uses diodes everywhere, even out, to help temper tank circuit effects and (it appears) to provide a sort of "feedback", or "anti-blowback".

I think the reasons for S-K RC with feedback include usual trafo noise and also to counter drift. This is what I meant by "corrected", above. Of course I like dial-a-bias, even if only to try to secure the active curve.

By the way, I somehow feel safer at "only" 500V. Now, caps might even fit in the chassis!

Best regards,
Paul S

05-22-2011

Post does not mapped to Knowledge Tree

Romy the Cat

Boston, MA
Posts 10,156
Joined on 05-28-2004

Post #:	81
Post ID:	16301
Reply to:	16299

This amp is OK only if you know the resulting Sound.

N-set, it is nice to see that you are keep punching the idea, somebody shall, I guess..

About the filters. For the low-pass and in this application I do not use inductors and much prefer RC filter. In LC filters the band-pass flows over coil. If we are taking about 20Hz filter then we have a coil of 50H-60H – it will be 5 kilometers of wire. If you use some kind of fancy cables of $1000 per meter and feel that this meter is important then in the coil you will have 5000 times of uncontrollably-sounding wire. In RC filter the signal the flow over the cap is the signal that shunting to ground and the band-pass signal flow only across the resistor, therefore it does not see any ugly reactive loads.
About the rest of the amp. I am not wild about it as it has a fixed design. Everything in this amp is moderated by one parameter: the current of the driver tube. What would happen if you do not nned 160W but would like to stray with 80W. Even if you redo your huge and expensive 100H plate choke then you will be force to lose power on the driver tube. You do not want to weaken the driver but only weaken the output tube.

I am not getting the idea about driving the filament from OPT transformer, can you to elaborate on it purpose and method?

The Cat

"I wish I could score everything for horns." - Richard Wagner. "Our writing equipment takes part in the forming of our thoughts." - Friedrich Nietzsche

05-23-2011

Post does not mapped to Knowledge Tree

N-set
Gdansk, Poland
Posts 617
Joined on 01-07-2006

Post #:	82
Post ID:	16302
Reply to:	16301

Demagogy

The 100W+ SE is pretty much uncharted teritory, not to mention Bass SE. Who can speak about the Sound?
Only that who tries.

Paul, the diodes in Chimera are for biasing purpose, but I agree, here there is no cathode feedback not ot loose power,
the whole thing can run away and explode. Will try to think of that.

Romy the Cat wrote:

About the rest of the amp. I am not wild about it as it has a fixed design. Everything in this amp is moderated by one parameter: the current of the driver tube. What would happen if you do not nned 160W but would like to stray with 80W. Even if you redo your huge and expensive 100H plate choke then you will be force to lose power on the driver tube. You do not want to weaken the driver but only weaken the output tube.

What did you expect from DC coupling? How much elasticity you'd expect from
2 stage minimalistic DC coupled amp? Either minimalism or elasticity.
But still there is some elasticity: the selectable UL. Also, as I said,
I'm thinking how to adjust the bias of driver and output individually in some limits.
BTW, very much to the oposite with the control-everything is controled by the bias of the input tube.

Romy the Cat wrote:

I am not getting the idea about driving the filament from OPT transformer, can you to elaborate on it purpose and method?

This is Dave Slagle's idea: the filament current offsets the DC bias of the core caused by the plate current.
This allows for a smaller gap and less turns on a given core to get the ind.

Filters--I feel I did not learn anything. If anybody wants to teach me, or better yet design the filter,
I'd apreciate :-)

Cheers,
N-set

Cheers,
Jarek
STACORE

05-23-2011

Post does not mapped to Knowledge Tree

Paul S
San Diego, California, USA
Posts 2,657
Joined on 10-12-2006

Post #:	83
Post ID:	16303
Reply to:	16302

Stability

N-set, have another look at the Chimera "bias", and the sophistocated way the diodes effectively buffer the gain stages, also the way the caps are staged over the tiered ground planes. Also check where the Ls are with respect to the signal. Perhaps a design like this only "works" with "perfect" parts; but here the diodes seem to effectively provide some breathing room and also a "calming effect" for the tank circuit, like a drop of oil on water. Of course, I have not heard this amp, and reports I have gotten about it were not ecstatic.

You obviously know about filters; certainly it is not for me to tell you. The same laws apply whether the curve is written with theoretical LC or RC; but in my experience the choke may become more problematic, at least in a situation like this, for reasons Romy cited, and also EM, ESR, and phase (drift...) issues that may be worth considering, as well.

Best regards,
Paul S

05-23-2011

Post does not mapped to Knowledge Tree

N-set
Gdansk, Poland
Posts 617
Joined on 01-07-2006

Post #:	84
Post ID:	16304
Reply to:	16303

Bias

Paul, again the diodes are there not because they are diodes and one needs a diode action there, but because they wanted
to substitute cathode bias resistor with something that would have a big static resistance (to set the DC point and stabilize)
and small dynamical resistance (to avoid cathode bias action). The price to pay is of course nonlinearity, esp. in the output stage
where the current variation is big enough to feel the curved characteristic of the diode.
But you are right, I forgot to split the bias into the grid part (the choke RDC) and the cathode part to give the amp some stability.
I also haven't heard the axiom. If someone knows a better the topology, please let me know!
Same with filters, I have no idea how they work.

Cheers, N-set

Cheers,
Jarek
STACORE

05-23-2011

Post does not mapped to Knowledge Tree

Paul S
San Diego, California, USA
Posts 2,657
Joined on 10-12-2006

Post #:	85
Post ID:	16307
Reply to:	16304

Active/Reactive

N-set, I am sure you realize by now that I do not properly understand "electronics". I see the "diodes" in the Chimera circuit as a sort of "active" bias that is also dynamic proof against blowback/ripple to and from the tank, at each of three "levels". I don't have the circuit where I can look at it now, and I don't really remember it all thet well, but I thought I remembered it was meant to be a "constant current" design, with the final diode (output) as "insurance"? I guess the corrected curve would "replace" the middle gain stage? I also think I remember that the designers considered the "speed"/rise/discharge time of the ground-plane caps to be very important. Likewise, I think the "diodes" are not just any old diodes. Naturally, I don't know how this circuit translates from ~7W to ~150W.

Best regards,
Paul S

08-23-2011

Post does not mapped to Knowledge Tree

N-set
Gdansk, Poland
Posts 617
Joined on 01-07-2006

Post #:	86
Post ID:	16851
Reply to:	16299

GU81M with a filter

I had some free time which I'd decided to waste with this "project".
For some unknown to me reason (I'm an idiot when it comes to filters)
the Sallen Key topology needed one more 12AX7 stage, unlike
I happily drew some posts ago. The resulting SPICE'd proposal is below:

I'd be happy to provide LT SPICE files should anybody want to play
with the model.
I've SPICE-settled for -86V @GU81M already few versions ago,
which I'd hapilly split between the coupling
choke RDC and some explicit -30V Romy bias supply to adjust power/operating point.
It all turns to be quite complicated when it comes to power supplies:

+/-200V for the input
negative bias for the input
300V driver
1530V plate
500V screen
auxilary negative bias GU81M
6! phew!

The power is a bit down according to SPICE to 150W@1R with 1.85Vrms input (@20Hz).
THis input is just about to touch 0V bias at GU81M.
Looking at the simulated spectra,
the two 12AX7 seem to add mainly higher order harmonics (4th, 5th, ...) to the
6E5P-DC-GU81M "engine" at full power. The spectrum looks a bit more even (the
monotonic decrease) with a healthy load of harmonics (I don't bother to evaluate THD,
I suspect some of the harmonic "richness" is a numerical artefact)

GU81Mv4_20Hz_1.85Vrms.JPG

The frequency response (the filters set at 50Hz), simulated at the load, is a mystery to me:

Why the corner freq. 50Hz is 7dB down wrt 20Hz???
It is at 12db (20Hz at 19dB, 100Hz at -9dB).
The roll-off towards 0Hz is due to the finite coupling inductances.
Apart from this issue, which perhaps somebody with knowledge will
explain/correct, it does seem to "work" in SPICE at least.

Cheers,
Jarek
STACORE

08-23-2011

Post does not mapped to Knowledge Tree

Paul S
San Diego, California, USA
Posts 2,657
Joined on 10-12-2006

Post #:	87
Post ID:	16852
Reply to:	16851

Model Citizen

The pisser is, you will eventually have to build one and heat it up, with a real speaker as a load, since even the best models are +/- static. As for corner frequencies, my own reality is model first, then remodel to suit. Sorry to be such a cynic, but I just can't seem to keep track of all the reasons why something does not work like it "should". If you can actually get it working (and keep it working...), 150W at 20 Hz/1R ought to do the trick, all right, at least in terms of power. Then, there's listening, and that's where it really gets dicey...

Again, as you must realize, 1530V, in and of itself, also creates "issues" that are not foretold with Spice.

Best regards,
Paul S

08-24-2011

Post does not mapped to Knowledge Tree

N-set
Gdansk, Poland
Posts 617
Joined on 01-07-2006

Post #:	88
Post ID:	16853
Reply to:	16852

Modeling

Paul, I would not touch this beast without modeling it first
at least to see if the electron flow is +/- what I'd expect.
The eventual sonics are of course hard to predict form the model FFT's.
The only hints are some posts at DIY audio and few words by haralanov.
But that of course pertains to a different topology, different OP,
full-bandwith amp, etc.
The filters--I don't know if they work like they should or not.
The goal was 50Hz corner and 24db/oct down, but it does not
look like that. It looks more 3rd order???

Cheers,
Jarek
STACORE

08-24-2011

Post does not mapped to Knowledge Tree

Paul S
San Diego, California, USA
Posts 2,657
Joined on 10-12-2006

Post #:	89
Post ID:	16857
Reply to:	16853

Cap on Grid?

N-Set, at some point I wil try to actually pay attention. In the meantime, does effective RC change with signal/gain in this situation? If so, it might also appear to "change the slope".

Best,
Paul S

08-25-2011

Post does not mapped to Knowledge Tree

N-set
Gdansk, Poland
Posts 617
Joined on 01-07-2006

Post #:	90
Post ID:	16861
Reply to:	16857

OK

Paul S wrote:
N-Set, at some point I wil try to actually pay attention. In the meantime, does effective RC change with signal/gain in this situation? If so, it might also appear to "change the slope".

Paul, the FR does not seem to depend on the level, but I've sorted that out.
I played with the resistors a bit, lowering them to 15.5k,
which according to some formulas I've found (P. Millet) should give -3dB @approx. 70Hz...but
it gives -3dB@50Hz as wanted. Also the slope seems ok: 24db/oct. Me idiot.
What happens around around 370Hz I no idea. This is what the filter spitts out.
The green plot is the grid of the driver, the magenta is at the load:

GU81mv4.1_FR.GIF

So I guess the basic electron flow is +/- ok now. The next step would be breadboarding,
but I have no possibility to do it in the near future.
Any takers? I'd assist whith what I know.
Some HV ideas have been already put forward earlier in the thread, the main trick being
buying Nichicon 550V electrolytics cheaply directly in China (stacking 3x1500uF
for 500u/1650V) together with 1.2kV SiC's Smile

Cheers,
Jarek
STACORE

08-25-2011

Post does not mapped to Knowledge Tree

Paul S
San Diego, California, USA
Posts 2,657
Joined on 10-12-2006

Post #:	91
Post ID:	16864
Reply to:	16861

Z

N-set, have you tried it against any known impedance plots? Maybe you have or could get plots for a couple of real-world candidates?

Best regards,
Paul S

08-25-2011

Post does not mapped to Knowledge Tree

N-set
Gdansk, Poland
Posts 617
Joined on 01-07-2006

Post #:	92
Post ID:	16865
Reply to:	16864

Plots

Impedance not. Gain yes. I've basically copied Millet's implementation:

http://www.pmillett.com/tube_active_crossover.htm

and I looked at the plots there. This dip around 370Hz is a mystery,
the rest I think now looks ok....well, now when I look at the phase behaviour,
I'm not so sure anymore.

Cheers,
Jarek
STACORE

11-05-2012

Post does not mapped to Knowledge Tree

Romy the Cat

Boston, MA
Posts 10,156
Joined on 05-28-2004

Post #:	93
Post ID:	18712
Reply to:	2136

The high-power amp from Moscow.

There is Moscow-based Company “Next Sound”, the web site unfortunately only in Russian

http://next-sound.ru

… and they make some tube amps including one 400W.

I am sure it is not DSET configuration but it very much might be. The 400W as bass DSET is plenty of power to drive anything. The company looks like is very comfortable to do any customization and optimization for customers, including a proposal for the customer to “design” requirements for their own unique amplifiers. In my book it is a good direction….

Rgs,
Romy The Cat

"I wish I could score everything for horns." - Richard Wagner. "Our writing equipment takes part in the forming of our thoughts." - Friedrich Nietzsche

07-02-2013

Post does not mapped to Knowledge Tree

Romy the Cat

Boston, MA
Posts 10,156
Joined on 05-28-2004

Post #:	94
Post ID:	19600
Reply to:	2136

A good 150W project for ULF

Here are two guys in Belgium that made up 833A 150W amp. That might be a good candidate for driving ULF channels.

http://monolithmagnetics.com/index.php?option=com_content&view=article&id=77&Itemid=81

Why I do not have somebody like them as my neighbors?

The Cat

"I wish I could score everything for horns." - Richard Wagner. "Our writing equipment takes part in the forming of our thoughts." - Friedrich Nietzsche

07-02-2013

Post does not mapped to Knowledge Tree

N-set
Gdansk, Poland
Posts 617
Joined on 01-07-2006

Post #:	95
Post ID:	19602
Reply to:	19600

GU81M

...and nobody has taken up my GU81M idea ;(?
ah, low morals and small balls, guys...

Cheers,
Jarek
STACORE

08-19-2016

Post does not mapped to Knowledge Tree

xandcg

Rio de Janeiro, BR.
Posts 218
Joined on 09-07-2014

Post #:	96
Post ID:	22722
Reply to:	19602

212E PSE.

This topic is a bit old and I do want to pose as Audio Note (UK) promoter, but it is almost the only manufacturer of amplifiers I pay close attention so I usually talk about them.

That said, I know the project of the new amplifier from them, and as far I know still a prototype but a fully functional prototype, can be done/bought as 212E PSE. I think it would be about 150W of truly SE amplifier. Maybe there are still another more powerful possibilities on that project, but I do not know.

Cheers!
EDIT: Indeed, I think they said they can use that project to build the amplifier with any SE tube the client want, single or PSE.

Think for yourself, do not be sheep.

08-19-2016

Post does not mapped to Knowledge Tree

N-set
Gdansk, Poland
Posts 617
Joined on 01-07-2006

Post #:	97
Post ID:	22723
Reply to:	22722

212E crazyness

I propose looking at 1) the price of 212E 2) its nominal lifetime 3) calculate $$/hour of use 4) mutiply by 4 (2 PSE monoblocks) 5) think if you want to spend that to drive LF

Cheers,
N-set

Cheers,
Jarek
STACORE

08-19-2016

Post does not mapped to Knowledge Tree

xandcg

Rio de Janeiro, BR.
Posts 218
Joined on 09-07-2014

Post #:	98
Post ID:	22724
Reply to:	22723

I am not willing to do it.

@N-set

I am not walling to have it, I was just pointing out there is already a commercial option. Also, if I am not wrong, and remember correctly what they told me, the same amplifier can be done with any SE (single or PSE), so Eimac or other similar powerful but cheaper tubes theoretically could be used instead.

Not to say this is a un-launched experimental product, so all silver with all über exotic parts they have, with an astronomically price where the tube prices would be your minor issue.

Think for yourself, do not be sheep.

08-20-2016

Post does not mapped to Knowledge Tree

N-set
Gdansk, Poland
Posts 617
Joined on 01-07-2006

Post #:	99
Post ID:	22726
Reply to:	22724

Universal amp?

Well, I used "you" in an abstract rather than direct sense Wink

Its very starange what you say than any power tube can be used -
each power tube has its own drive requirements and making a
universal driver able to drive say 212E and eimacs is hardly possible.
But lets see what they come up with. BTW, tere is even more
rare bird - 308B (low mu version of 212E)

Cheers,
jk

Cheers,
Jarek
STACORE

08-20-2016

Post does not mapped to Knowledge Tree

xandcg

Rio de Janeiro, BR.
Posts 218
Joined on 09-07-2014

Post #:	100
Post ID:	22727
Reply to:	22726

I do not know details.

I do not know details, and I am not a technical person, but I think when they told me "any power tube" they was not talking about simple driver swaps but more likely they current amplifiers. Ongaku, Kageki, Kegon etc. seems to be all basically the "same" amplifier project/concept with just the minimum changes/adaptations they need to do to work with those different tubes.

But again, I am not a technical person.

Cheers!

Think for yourself, do not be sheep.

Page 4 of 5 (107 items)

Select Pages:

« 1 2 3 4 5 »

	Target	Threads for related reading	Most recent post in related threads	Forum	Replies	Views	Started
»	New	More power from Melquiades? More powerful tube?..	A new Icon Audio's MB81 Mono Blocks...	Melquiades Amplifier Forum	19	249586	09-01-2007
»	New	NAT Audio Magna -160 Watts of Single-Ended Class A..	The KR Audio ways....	Audio Discussions Forum	5	73113	11-12-2007
»	New	Claude Perrier and his tube seremony..	Claude Perrier and his tube seremony...	Audio Discussions Forum	0	15399	01-05-2008
»	New	Getting more power from SET vs. properly distorting SS...	Sound Board...	Audio Discussions Forum	4	49380	05-09-2011
»	New	The ULF cannel for my new listening room...	The Organic Bass vs. ULF Drivers...	Audio Discussions Forum	43	130727	07-29-2018