The Rosenblit 6AS7 OTL

What people are working on at the moment
User avatar
Paul Barker
Social Sevices have been notified
Posts: 8863
Joined: Mon May 21, 2007 9:42 pm

#31

Post by Paul Barker »

Yes I enjoyed this one when I scratch built it many years ago.

Image
"Two things are infinite, the universe and human stupidity, and I am not yet completely sure about the universe." – Albert Einstein
Cressy Snr
Amstrad Tower of Power
Posts: 10552
Joined: Wed May 30, 2007 12:25 am
Location: South Yorks.

#32

Post by Cressy Snr »

Ray P wrote:Leaving aside all the stuff that accompanies Bruce Rozenblit, he does seem to keep coming up with the goods with his designs. My experience of his push-pull designs is very positive and I was staggered how good his original single-ended design (SE-OTL) was when I built one, significantly better than my Bottlehead Paraglow (2A3 SET). I'm currently scratch building one of his Minibeast SE-OTL designs (a few more watts available than the SE-OTL) and plan to eventually graduate to his latest SE-OTL design that uses four 300Bs per channel.

Ray
I'd agree with that Ray.
When it was working, my Son of Beast was a stunner.
I found out indirectly what was causing the problems with it.
The 100ohm resistors used to triode strap the PL519s were one by one, failing open circuit and shutting off the valves. They were cheap shite.

I discovered the problem when I tried to use two in series, in the cathodes of the EL84 push pull amp. I couldn't understand why the output stages wouldn't bias up on one channel. It turned out that half of the 16 I bought cheap off ebay (the ones that had been in the OTL) were open-circuit.

Trouble was by then I had dismantled the whole amp.
That I fully intend to rebuild it goes without saying.
User avatar
Ray P
No idea why I do this anymore
Posts: 6294
Joined: Thu Nov 22, 2007 5:18 pm
Location: Somerset

#33

Post by Ray P »

Son of Beast is a great amp Steve; heard one a while back at a friends house and it really sang (sadly, I can't recall offhand what it was partnered with). Good luck with the rebuild and no cheap stuff this time eh; nowadays I only buy parts from 'reputable' sources as it gives more piece of mind for the cost of a few beers.

Ray
User avatar
Paul Barker
Social Sevices have been notified
Posts: 8863
Joined: Mon May 21, 2007 9:42 pm

#34

Post by Paul Barker »

JamesD wrote:It's within the global feedback loop of the amplifier so I suspect he tried it without the resistor and found it unstable and so added the resistor to add a dominant pole due to Millar cap on the CF rather than add a physical capacitor... maybe it gave him more control of the stability margin that way? This might appeal to his sense of 'tweaking the nose of the conventional' :)

J.
Hi James I found my Rosenblit book. He calls it a cathode coupled push pull splitter. (herinafter and so on and so forth me lud shall become known as the CCPPS in the interests of brevity)

I couldn't find anything exactly like it in TCJ. It provides a third phase inversion needed for the feedback, but also even harmonics are cancelled, "resulting in a highly linear drive signal".

I think it is one of those cases of Rosenblit knowing best about his own circuits.

He dust suggest someone may like to try the 6au6 in place of the 12ax7, but with the proviso it wouldn't be as linear.

I still fancy using the original Futterman single 6au6 straight to concertina which provides the single phase conversion isntead of a necessary 3rd phase inversion. Feedback across a single inversion would be more stable. Though Rosenblit's intermediate gain stage is probably a secret gem in his design, the extra complexity and power supply difficulties encourages me to try the simpler Futterman.

Clearly futterman is the original inspiration.

OTOH this Rosenblit CCPPS might be a hidden nugget we should all be using somewhere in ever amp.

Something has conspired to make that circuit sound great. In his text it is clear he is very pleased with this element of the design.

Maybe the fool leaves it out. Hands up to that then!
"Two things are infinite, the universe and human stupidity, and I am not yet completely sure about the universe." – Albert Einstein
JamesD
Old Hand
Posts: 997
Joined: Tue Oct 09, 2007 4:26 pm
Location: North Yorkshire

#35

Post by JamesD »

Hi Paul,

Thanks for finding that! Interesting explanation as always from him :) I look at this with his words and it prompts some questions, such as:

Given the second grid is decoupled to earth at audio frequencies (1.5Hz turning frequency) and the output is single sided how is it push pull?

Given that the first valve is a cathode follower and for the second valve the signal is cathode coupled input where is the phase inversion?

And if you do look at it as push-pull then we have a long-tailed pair or differential amplifier with a single sided input and a single sided output - so where can the even harmonic get canceled without it also canceling the primary signal?

Not saying he's wrong - just that I don't understand how it does what he says it does so I think I need to sim or build this stage to see how it works as this is intriguing!

Paul many thanks for this its making me think!

James.
User avatar
Paul Barker
Social Sevices have been notified
Posts: 8863
Joined: Mon May 21, 2007 9:42 pm

#36

Post by Paul Barker »

I can't understand it either. But he says there are three inversions for the feedback. there is only this composite and a common cathode before it. It is odd.

I have heard it though, it does work.
"Two things are infinite, the universe and human stupidity, and I am not yet completely sure about the universe." – Albert Einstein
JamesD
Old Hand
Posts: 997
Joined: Tue Oct 09, 2007 4:26 pm
Location: North Yorkshire

#37

Post by JamesD »

I'm sure it does work and work well as his designs all seem too... its a puzzle though - he has a way of looking at things that is not conventional and I'm sure that is part of what makes him successful.

And on the other hand he seems to enjoy tweaking other designers world views :)

I'm having a look at it in LTSpice - just got to get the triode model working again... (Re-install of LTSpice as I'm now running Ubuntu Studio as my OS.)

James
User avatar
Paul Barker
Social Sevices have been notified
Posts: 8863
Joined: Mon May 21, 2007 9:42 pm

#38

Post by Paul Barker »

A little more information.

HE claims a voltage gain on the ECC83 of 92. That would be because of the 470k Anode resistor and the 1meg input resistance of next stage.

Then this couplet we can't fathom as a pair multiplies the 92 to 600. So he claims voltage gain of 6.52 for the couplet which is nothing for a 6sn7.

The total A of 600 or 55dB is reduced considerably by the high impedance of the output stage to 42dB (if the speakers are 16 ohm). then 28dB of this is lost in feedback. So he ends up with 14dB forward.

Obviously more output valves allows for reduced feedback requirements brings back some of the forward gain.

As the amp stands in the schematic it has 0.5 ohm output impedance.

The best solution is to double the number of output valves, but at what point does it turn into a monster that serves no useful purpose in the living room. (like most of what I have built.)

Back to the couplet. The two halves are out of phase at their grids so by vertue of the equal anode resistors and the shared cathode resistor they push and pull current from the power supply to a net zero. So it is an extremely interesting couple from that point of view.

I see where the 3rd phase inversion comes I think. Is it that the grid of second half receives antiphase of the 12ax7 through the cap to ground? So it inverts this on it's output which makes the output from this second half the same as were the only VA stage the 12ax7? ( remember the futterman only has one VA stage, the 6au6 as a pentode.

This might explain the forward voltage gain of only 6.52 X. Half one is not playing a part in the amplification, it has two purposes, to reduce the demand on the power supply and 2nd harmonic cancelation.

I will be very interested if anyone can sim it.

So recap, does the first half of the 6sn7 only balance PS demand and cancel even harmonics, while the second half provides the forward voltage gain with a resultant phase flip by taking it's signal from the 12ax7 from the flip side?

OR am I way off track?
"Two things are infinite, the universe and human stupidity, and I am not yet completely sure about the universe." – Albert Einstein
User avatar
Ray P
No idea why I do this anymore
Posts: 6294
Joined: Thu Nov 22, 2007 5:18 pm
Location: Somerset

#39

Post by Ray P »

If it's of interest, Bruce's recent book contains details for a 15watt pc 'low cost' push-pull OTL amp that reflects his current thinking. Valve complement consists of 12AT7s and EL509s.

Ray
User avatar
Paul Barker
Social Sevices have been notified
Posts: 8863
Joined: Mon May 21, 2007 9:42 pm

#40

Post by Paul Barker »

Thank you Ray I have read detail from a friend's book. He really is out there worlds apart from the rest.!
"Two things are infinite, the universe and human stupidity, and I am not yet completely sure about the universe." – Albert Einstein
JamesD
Old Hand
Posts: 997
Joined: Tue Oct 09, 2007 4:26 pm
Location: North Yorkshire

#41

Post by JamesD »

Well I'm puzzled by the written words....

I don't get the same gain figures as Rosenbilt - neither does LTspice... I don't get any 2nd order cancellation - neither does spice...

I get a normal common cathode first stage, single input into a long-tailed pair with single side output without phase inversion...

The cap on the second input of the LTP connects it to ground with an F3 of about 1.5Hz so no phase inverted input there. (And doesn't a cap give 90degrees of phase shift only?)

What I also get is that the Millar cap on the LTP input is the dominant pole inside the feedback loop and that without feedback the EEC83 would slewrate limit into the millar cap at CD signal levels... as does the sim...

James
User avatar
Paul Barker
Social Sevices have been notified
Posts: 8863
Joined: Mon May 21, 2007 9:42 pm

#42

Post by Paul Barker »

Odd. But it works.
"Two things are infinite, the universe and human stupidity, and I am not yet completely sure about the universe." – Albert Einstein
User avatar
Nick
Site Admin
Posts: 15706
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

#43

Post by Nick »

I might be missing something, but it looks to me to be a cathode coupled phase splitter, but only taking one phase as a output.

http://www.bonavolta.ch/hobby/en/audio/ ... led%20Pair

I guess it will provide a constant group delay so wrapping feedback around it will get complex at increasing frequency.

Also, as the load on the two anodes are not equal, that may have some odd effect, but maybe not. Maybe the sim might show that.

But as a cathode follower driving a common grid amplifier I dont see more than one 90 deg shift.

I suppose it avoids the miller of the second stage from causing roll of because of the first stage output impedance.
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
JamesD
Old Hand
Posts: 997
Joined: Tue Oct 09, 2007 4:26 pm
Location: North Yorkshire

#44

Post by JamesD »

Hi Nick,

I see two stabilising mechanisims both at the input to the LTP one for open loop and one for closed loop.

Initially, i.e. before the feedback loop has time to respond the slew rate limit into the Millar capacitance will prevent the transient growing large i.e. limited open loop large signal response and once the feedback loop is established and the input signal to the LTP truncated then the Millar capacitance sets the upper frequency response limit for the closed loop. I know its the Millar cap in both cases and both relate to the source resistance and cap value but open loop the slew is likely to bite before the F3 limit and closed loop its the F3 limit that hits first...at least in my mind :)

Unequal but same order loads on a true push-pull implementation has surprisingly little effect on its operation, as does different amplifying devices on each side - I forget where but there are a series of articles on using a 45 as one side and a 2A3 as the other side in push-pull and it works really well...
Andrew
Eternally single
Posts: 4206
Joined: Thu May 24, 2007 2:18 pm

#45

Post by Andrew »

JamesD wrote:
Unequal but same order loads on a true push-pull implementation has surprisingly little effect on its operation, as does different amplifying devices on each side - I forget where but there are a series of articles on using a 45 as one side and a 2A3 as the other side in push-pull and it works really well...
On a very similar note, I too read somewhere of a chap who had a LTP and used a very long tail and dissimilar valves, like an ECC88 and an ECC83, the AC gain behaviour at both anodes was surprisingly similar despite the different valves.
Analogue, the lost world that lies between 0 and 1.
Post Reply