DIYHiFi.org

For the sake of audio
It is currently Sat Dec 16, 2017 11:15 am

All times are UTC




Post new topic Reply to topic  [ 125 posts ]  Go to page Previous  1 ... 4, 5, 6, 7, 8, 9  Next
Author Message
PostPosted: Thu Oct 05, 2017 9:06 pm 
Offline
Goat

Joined: Mon Nov 28, 2005 1:10 pm
Posts: 545
T, I am about to get these Philips PF86 tubes along with some other items from the same seller.

https://produto.mercadolivre.com.br/MLB ... das-ok-_JM

It is the same tube as the EF86 with a different heater. The seller has five used tubes at 5 dollars each, I think I´ll grab them... He says they are above 80% emission, but his transconductance tester isn´t ok.

"For both the Regensburg and the Notre Dame microphones Cathedral pipes uses a Old Stock PF86 vacuum tube. We have over 2000 of these in various classic brands... Or so we thought! As it turns out no matter what brand is stamped on the PF86 it is the code on the tube that tells the true story. Ours and we are guessing all NOS PF86 tubes are stamped with the two letters Dk then a digit. The letters tell us they all came from a well known factory in Hamburg Germany owned by Valvo/Phillips. The digit indicates which year within the 1950s the tubes were made. Ours so far are all between 1953 and 1958. So no matter what the box or label inidcates they are all in fact the same..."

Source: http://www.cathedralpipes.com/valvo_tubes.php


Last edited by alexandre on Thu Oct 05, 2017 9:30 pm, edited 1 time in total.

Top
 Profile  
Reply with quote  
PostPosted: Thu Oct 05, 2017 9:19 pm 
Offline
Goat

Joined: Mon Nov 28, 2005 1:10 pm
Posts: 545
A pity that he only has one TDA1540 ceramic. 3 dollars :D


Top
 Profile  
Reply with quote  
PostPosted: Tue Oct 10, 2017 9:04 am 
Offline
Muriel
User avatar

Joined: Thu Jul 26, 2007 5:45 pm
Posts: 1167
Hi,

hanski wrote:
Could we perhaps talk about a power amp to go with this?
I'll start the fishing with 5687 gain stage + lateral mosfet buffer suggestion?
(Hopefully the story goes to something sensible and clever as we've seen here.)


Well, lets work forward from the backend.

Powersupply, say we use multiple torroids "stacked" to get enough Volz and Ampz. We can easily get 50V AC this way and around 70V Rails under load. Add suitable caps etc. and "bob's your uncle" for the big power supply.

We can then use voltage doublers to make +/-140V unregulated (regulated to say +/-120V, ideally using shunt regulators). Add a small transformer for filaments and utility jobs, voila, done.

Now, how may we make an output stage?

We could use classic lateral Mosfets with added fast Bipolars as Fet/BJT Sziklai. The Mosfets can be run at enough quiescent current to just get into the negative tempco region (usually somewhere a trifle above 120mA or so) and the BJT's can run at very low Iq (see D. Self).

The resultant output stage can be used as classic source follower. If we use 3 pairs of the usual (say 2SA1302/2SC3281 et al) BJT's at (say) 26mA per transistor and with 0.47R current sharing resistors and a 5.1Ohm Drain resistor for our FET, we are looking at around 10A/V small signal transconductance in the Class A region, or 0.1 Ohm Z-Out and in class B at 30A current dropping to around 0.03Ohm. I think we can live with this, especially for a follower, we may want to play with the exact handover between the two Fets and the BJT Packs some to get the smoothest "birdwing" plot.

Now let us take 1/2 of the 5687 and make it a cathode follower. We can use bootstrapping for both anode voltage and a resistive cathode load, we now have almost constant current & voltage conditions for our follower. The grid should AC coupled to prior stages and is biased by our output servo. Coupling Cap can be as low as 100nF or even 30nF if we make the grid leak resistor large enough. This can directly couple to the Fet output stage (use suitable gate resistors, protection diodes etc.), we can easily run 15-20mA Anode current giving around 9mA/V transconductance and thus around 110 ohm drive impedance for our output stage.

Now we need a Gain-stage. We still have 1/2 5687, lets use this. Very simple, resistor & capacitor bias combo in the cathode from the -120V supply, resistor from the +120V supply and we can again bootstrap this resistor from the output, to increase the apparent load resistance (improve linearity). Lets pick 160V/10mA as operating point, this gives us around -8V bias and thus 820 Ohm Cathode resistor, bypass cap (say) 220uF/16V Elna Silmic II. Our Bootstrapped Anode combo would be 6.8k load, 1k decoupling resistor with a 22uF (or higher) decoupling cap (Film?) to the Amplifier output.

This (in the small signal region) makes the apparent anode load around 130k with a 2 Ohm load, giving a pretty good linearity. Theoretical peak/peak swing is around 260V (90V RMS) more than enough for the around 50V maximum we need. Gain would be around 17.5 or 25dB.

We have two problems remaining. First, our input signal is referenced to the negative -120V rail, not ground and really 25dB gain may not be enough, especially if we wish to add some global feedback.

Here we can take a VERY unusual approach. Say we use a P-Channel J-Fet as input (say J177), cascoded to withstand the the -120V that would otherwise from the drain voltage. Lets set the source resistor as 1kOhm (giving around 1.5mA drain current and 50mA/V transconductance) and allow a load resistor of equal 1kOhm, adding around 2V more to the 5687 gain stage bias, so we use 1k/220uF cathode combo.

If we add another 220uF Silmic cap on the J-Fet source, we can then use a feedback combo to apply some feedback, if we use 100R/3.9k as feedback divider (setting theoretical closed loop gain to 32dB) we get an open loop gain of around 150 (43dB) and an overall feedback factor of around 11dB, in first order approximation theory.

Yes, we have several LF time-constants in the system to be resolved, but that should not be too hard using modern sims.

The input J-Fet adds less than 20dB extra gain and the output stage is quite transparent. All stages are SE and with the exception of the tube gain stage (in effect the VAS in conventional amp's) are heavily degenerated - so the "tube sound" should dominate, though the tube should also be quite linear.

One instead may use another tube (say a 6072A) configured as SRPP with 2K cathode resistors running on around +100V Anode voltage and capacitor coupled to the 5687 Gainstage. This stage will produce around 26dB gain, allowing a little more feedback. A 75k Feedback resistor then gives around 32dB closed loop gain.

The 6072A SRPP -> 5687 Gainstage -> 5687 Cathode follower -> Output tube is the mainstay of Kondo San's Amplifier designs, including Ongaku, Kegon and Baransu, there is a certain "magic" to that combo. Combining the same with a neutral transistor output may very well be quite special. Imagine Ongaku Sound with 300W/8Ohm power on tap...

I'm sure Hanski will be soonest drawing this out, probably both variants, I lack the time...

Ciao T

_________________
"For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." Richard Feynman


Top
 Profile  
Reply with quote  
PostPosted: Tue Oct 10, 2017 9:21 am 
Offline
Muriel
User avatar

Joined: Thu Jul 26, 2007 5:45 pm
Posts: 1167
Hi,

alexandre wrote:
T, I am about to get these Philips PF86 tubes along with some other items from the same seller.


First, not sure what you want with EF86 (PF86) in the first place, second used to tubes with 80% emission are essentially worthless.

If you want to build with tubes, don't waste money of useless crap. Next time someone you know is in the USA, get them to buy a few NOS 5687 Tubes for you, they are still cheap.

Ciao T

_________________
"For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." Richard Feynman


Top
 Profile  
Reply with quote  
PostPosted: Tue Oct 10, 2017 1:00 pm 
Offline
Goat

Joined: Mon Nov 28, 2005 1:10 pm
Posts: 545
Thanks for the upshot. I have little experience with tubes.

I figured the PF86 could have been an useful low noise high gain post dac amplifier. It is used in microphones after all (even Neumann). The fact that it is a german vintage manufacture tube is appealing to me.

I could use AD1865 or 1862, both of which I have here, with Riv = 47 ohms or less, and a 5th or higher order analog LPF before the tube grid. The inductors are easy to wind because of the low impedance. Would need another tube (not PF86) for a cathode follower.


Top
 Profile  
Reply with quote  
PostPosted: Tue Oct 10, 2017 2:06 pm 
Offline
Cow

Joined: Fri May 30, 2014 6:24 am
Posts: 150
Kuei Yang Wang wrote:
The 6072A SRPP -> 5687 Gainstage -> 5687 Cathode follower -> Output tube is the mainstay of Kondo San's Amplifier designs, including Ongaku, Kegon and Baransu, there is a certain "magic" to that combo. Combining the same with a neutral transistor output may very well be quite special. Imagine Ongaku Sound with 300W/8Ohm power on tap...

I'm sure Hanski will be soonest drawing this out, probably both variants, I lack the time...

Ciao T

I'll most certainly sim this!
Again, typical from T., an answer that I partly anticipated but still contains a lot of surprises, thanks!

Is bootstrapping the previous stage from amp output even legal in hifi circles? :cool:

It might take some time. At the moment I'm drawing the layout for the splendid RIAA & line stage. When I'm finished I'll get into this..


Top
 Profile  
Reply with quote  
PostPosted: Wed Oct 11, 2017 3:03 am 
Offline
Muriel
User avatar

Joined: Thu Jul 26, 2007 5:45 pm
Posts: 1167
Hi,

alexandre wrote:
Thanks for the upshot. I have little experience with tubes.

I figured the PF86 could have been an useful low noise high gain post dac amplifier. It is used in microphones after all (even Neumann). The fact that it is a german vintage manufacture tube is appealing to me.


They are not that quiet. There have been discussions of tubed (and J-Fet) post DAC Amplification with passive I/U conversion and the SNR's achievable before, here on this site.

Might be worth having a look.

The best tube for that job I know that remains fairly easily available NOS today is the E180F/6688.

Alternatively 4pcs BF861B etc. J-Fet paralleled with a 51 Ohm source resistor each to limit Id to around 4mA each, cascoded by 1/2 5687 with the second half acting as cathode follower would work rather decently, gain can be set quite flexible, filters can be integrated etc. It should give an Ein of around -135dBV and thus ~-115dB SNR for a signal of 100mV (RMS - 280mVP-P) after the passive I/U conversion.

If instead of this low noise hybrid you use an EF86 as triode, you get a gain of around 30 with an Ein of around -120dBV (so your usable SNR with 100mV after passive I/U conversion is ~-100dB). You could use another EF86 as cathode follower, but noise is just good enough for CD, not high res.

If you instead used the lowest noise tubes (e.g. D3a/7788 or the russian analogues or the russian 6S45PE) you could get a 10dB improvement in noise.

An option would be to operate the tube in grounded grid mode and perhaps several in parallel. You would still have -120dBV Ein, but the DAC's current can be send "current conveyor" style through the tube, allowing a more free tradeoff of the various circuit constraints.

Ciao T

_________________
"For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." Richard Feynman


Top
 Profile  
Reply with quote  
PostPosted: Wed Oct 11, 2017 3:05 am 
Offline
Muriel
User avatar

Joined: Thu Jul 26, 2007 5:45 pm
Posts: 1167
Hi,

hanski wrote:
Is bootstrapping the previous stage from amp output even legal in hifi circles? :cool:


I have not seen any law that says otherwise. Some of the reputedly best sounding medium power transistor amplifiers use this scheme. I suspect it relates to overload behaviour and the inevitable soft clipping this approach forces.

Ciao T

_________________
"For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." Richard Feynman


Top
 Profile  
Reply with quote  
PostPosted: Thu Oct 12, 2017 8:42 am 
Offline
Goat

Joined: Mon Nov 28, 2005 1:10 pm
Posts: 545
Kuei Yang Wang wrote:
An option would be to operate the tube in grounded grid mode and perhaps several in parallel. You would still have -120dBV Ein, but the DAC's current can be send "current conveyor" style through the tube, allowing a more free tradeoff of the various circuit constraints.

Thanks T for the very good explanations.

One particular of the current conveyor (grounded base/gate/grid) is that it makes a gain stage with constant current *and* single ended operation. Whereas the metaxas (for example) needs a differential (LTP) to achieve constant current operation.

(Jocko loves it. This guy too: http://tech.juaneda.com/en/projects/preamp4.html)


Top
 Profile  
Reply with quote  
PostPosted: Sat Oct 14, 2017 5:52 am 
Offline
Goat

Joined: Mon Nov 28, 2005 1:10 pm
Posts: 545
alexandre wrote:
One particular of the current conveyor (grounded base/gate/grid) is that it makes a gain stage with constant current *and* single ended operation.

Hmm, nope, not a constant current in the simple circuits such as Juaneda´s preamp.


Top
 Profile  
Reply with quote  
PostPosted: Fri Oct 20, 2017 5:59 pm 
Offline
Goat

Joined: Mon Nov 28, 2005 1:10 pm
Posts: 545
I´m listening to the TDA1545 with jfet I/V. Fully differential setup, the chip receives true and inverted 44.1KHz data with a 32fs bitclock.

One thing about these chips is that their calibration state machine is synchronized to the incoming I2S, obviously. It means that the calibration will be less frequent if you are feeding non-oversampled 44.1KHz. Shouldn´t be an issue. I can test 96KHz by removing the Tent Labs XO and putting in a generic 24.x MHz.

Sounds good, but I suspect the TDA1387 is an even better chip (an evolution of the 1545). I say this based both on listening and on the paper below. Although the paper is about the 1547 or 1549, it gives useful info about the continuous calibration technique that I believe applies to the other chips (similar or same CMOS processes). Such as section 5, current cell design. This explains an important difference about N-type versus P-type "memory cells", as Philips engineers call them. The N-type has poor rejection of substrate noise, the P-type is the better one.

Since the 1387 can only source current and has compliance down to 0V, its output must be made of P-type current cells such as the one shown on figure 8. The only disadvantage of the 1387 for me is that it is harder to work with (smd).


You do not have the required permissions to view the files attached to this post.


Top
 Profile  
Reply with quote  
PostPosted: Fri Oct 20, 2017 8:14 pm 
Offline
Goat

Joined: Mon Nov 28, 2005 1:10 pm
Posts: 545
I´m not so sure of this:

"The concept of absolute phase is rendered irrelevant for any instrument with strings (such as a guitar or piano), or for two or more instruments played together. Complex sounds such as these are known to have an undetectable phase relationship."
(https://en.wikipedia.org/wiki/Absolute_phase)

Now, the flip of polarity IS audible, obviously, because your brain is following the waveform and suddenly you flip the phase. That is very audible with the right signal.

The question is if one polarity sounds better than the other in the long term. To me the answer seems to be yes, true polarity sounds a tad more coherent, bodied, fuller.


Top
 Profile  
Reply with quote  
PostPosted: Mon Oct 23, 2017 5:59 am 
Offline
Muriel
User avatar

Joined: Thu Jul 26, 2007 5:45 pm
Posts: 1167
Hi,

alexandre wrote:
"The concept of absolute phase is rendered irrelevant for any instrument with strings (such as a guitar or piano), or for two or more instruments played together. Complex sounds such as these are known to have an undetectable phase relationship."


This such a debate, I find best is to listen.

If you hear differences on switching and long term, well then it matters to you, no matter what anyone writes on Wikipedia.

If you don't, well then it doesn't matter to you, no matter what anyone writes on Wikipedia.

Wikipedia is often edited by "activists", most audio topics tend to reflect the "debunker" position that "everything sound the same" (note, it is more complex and I know that), so as a source it is generally not reflective of the truth of any matter, but of who is more activist in editing, meaning generally the view on any topic which reflects that easiest to back by some references and which shouted loudest, not necessarily the truth. But heck, we live in the post truth era anyway.

Ciao T

_________________
"For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." Richard Feynman


Top
 Profile  
Reply with quote  
PostPosted: Mon Oct 23, 2017 2:48 pm 
Offline
Goat

Joined: Mon Nov 28, 2005 1:10 pm
Posts: 545
I agree.
Btw, that quote is of Douglas Self. But I do not wish to stir controversy, let´s leave it at that. His books are a good resource in many topics.


Top
 Profile  
Reply with quote  
PostPosted: Tue Oct 24, 2017 3:41 am 
Offline
Muriel
User avatar

Joined: Thu Jul 26, 2007 5:45 pm
Posts: 1167
Hi,

alexandre wrote:
Btw, that quote is of Douglas Self. His books are a good resource in many topics.


I consider his books an excellent source of showing why common circuitry should be avoided and of how bigoted prejudice prevents a smart and experience man from seeing obvious and patently plain problems and equally easy solutions. His continued championing of outdated concepts as the ultimate "blameless" design is counterproductive.

On balance I find for example Bob Cordell's Book on amplifier design by far more useful than everything ever put out by Mr. Self. And yes, I have all of them on my shelf, next to a complete set of Linear Audio (where Mr. Hegglun has some excellent work published, as does Mr. Stoccino which again puts Mr. Self's work in stark contrast) and others...

The latest edition of Horowitz & Hill "Art of Electronics" has some excellent audio related material not previously published in such comprehension as well. And naturally I would recommend Samuel Groener's commentary on Mr. Self's amplifier design book.

Ciao T

_________________
"For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." Richard Feynman


Top
 Profile  
Reply with quote  
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 125 posts ]  Go to page Previous  1 ... 4, 5, 6, 7, 8, 9  Next

All times are UTC


Who is online

Users browsing this forum: Bing [Bot] and 1 guest


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB® Forum Software © phpBB Group