E80CF and a new phase-splitter

fredeb

AVForums Grandmaster
Joined
Mar 29, 2012
Messages
3,468
Reaction score
515
Location
Somerset West
I recently bought a batch of tubes from [member=24540]V4lve Lover[/member] , our friend in Holland . 13x 6080 , 10x E180f , 2x NOS Svetlana EL34 ( to go with the 2 lonely ones I had already ) and a single Phillips E80CF .

The E80CF was more thrown in with the rest , than something I wanted , but the more I research it , the more intrigued I get . A 9 pin minature triode/pentode

E80CF-04.jpg
   
E80CF-01.jpg



The E80CF is a special quality version of the ECF80 and has the same pinout , which is also the same pinout as an ECF82/6U8/6U8A . The ECF82 is becoming a popular substitute for the 7199 , in boards redesigned to drive the output tubes of Dynaco ST70 . Yes , ECF82 has a different pinout , and slightly different characteristics to 7199 , so it cannot be substituted directly .

This article found on r-type.org : http://www.r-type.org/articles/art-097.htm

New-Phase-Splitter.jpg


Arthur Bailey released this design in Wireless World, September, 1962 .

It seems this design was used in some Radford amplifiers . STA25 below

STA25.jpg


Here is the ST70 circuit for comparison

ST70.jpg


Seems like a good circuit to try . Comments appreciated .









 

fredeb

AVForums Grandmaster
Joined
Mar 29, 2012
Messages
3,468
Reaction score
515
Location
Somerset West
Moonraker said:
Probably not the  greatest circuit in terms of symmetry of drive impedances.

Which one ? The Dynaco or Arthur Bailey circuit ?

The Bailey circuit , with ECF82 , is meant to deal with this issue , with ECF80/E80CF anode and cathode resistor values will need to adjusted .



 

Moonraker

New member
Joined
Jun 18, 2021
Messages
26
Reaction score
7
Location
Free State
Radford use the Bailey circuit.  Their way of indicating capacitor values is really odd, but anyway the operation is clear from the Bailey schematic.

C6 is a decoupling cap and R8 is bias.  The circuit is therefore a paraphase type, but with a pentode amplifier stage which introduces an asymmetry.  The high overall NFB used in such a phase splitter (to ensure unity gain of the inverting stage) will tend to minimize any HF response advantages inherent in the pentode.

Maybe they got a stock of ECF80's cheaply.

The Dynaco circuit is simply a pentode amplifier followed by a 'concertina' type splitter.  Not nearly the same thing, but using a similar tube type.
 

handsome

AVForums Super Veteran
Joined
Jan 23, 2009
Messages
1,354
Reaction score
352
Not to split hairs but it?s not a paraphase but a Schmitt splitter. And if i remember correctly (which is highly doubtful) that particular configuration was discussed at length on this forum. The pentode eliminates a triode?s Miller capacitance and the resultant increase in bandwidth then allows for more feedback to be used - provided the output transformer has equally good HF performance of course. Even when Schmitt?s use similar tubes you still need to compensate for the differences in gain (anode resistors), so they never have been the most accurate splitters. Concertinas on the other hand can have excellent balance given matched resistors and that never drifts despite tube aging.
 

fredeb

AVForums Grandmaster
Joined
Mar 29, 2012
Messages
3,468
Reaction score
515
Location
Somerset West
handsome said:
Not to split hairs but it?s not a paraphase but a Schmitt splitter. And if i remember correctly (which is highly doubtful) that particular configuration was discussed at length on this forum. The pentode eliminates a triode?s Miller capacitance and the resultant increase in bandwidth then allows for more feedback to be used - provided the output transformer has equally good HF performance of course. Even when Schmitt?s use similar tubes you still need to compensate for the differences in gain (anode resistors), so they never have been the most accurate splitters. Concertinas on the other hand can have excellent balance given matched resistors and that never drifts despite tube aging.

Damn , you have a good memory ! Apologies - I had forgotten  :facepalm: , I had to go look for it now again , here : https://www.avforums.co.za/index.php/topic,92798.0.html

I'm still intrigued by this circuit , and reckon I'm going to have to try it .

 

Moonraker

New member
Joined
Jun 18, 2021
Messages
26
Reaction score
7
Location
Free State
Thanks - interesting info!

I suppose it makes sense to attempt to increase the HF cutoff of this stage, even if the amplifier is probably equipped with several triode stages anyway.
The phase splitter is not normally the stage where the dominant pole is created and extending its bandwidth makes sense.

However, are the advantages that great?  Remember that the phase splitter gain is quite low due to local NFB and therefore the Miller capacitance is of relatively little concern.
 

handsome

AVForums Super Veteran
Joined
Jan 23, 2009
Messages
1,354
Reaction score
352
Schmitt?s don't have unity gain and the feedback loop encompasses the entire amplifier, so there is no explicit feedback just around the splitter. Thus the splitter does have Miller capacitance but, more importantly, it is fed by a pentode stage which means a high output impedance from that stage and thus less bandwidth due to Miller capacitance. So, the splitter used just triodes then it?s input probably would be the dominant pole since the output stage would be either pure pentode or ultra-linear (less Miller capacitance). So reducing the Miller effect of the splitter would give you more bandwidth which would then allow you more feedback to reduce any distortion. Using a pentode input stage gives higher distortion than a triode would so already you ?need? a fair amount of feedback. Imbalance as a result of a less-than-perfect splitter would create more 2nd order distortion, but a lot of that distortion would be cancelled by the push-pull action of the output stage. Odd order distortions are not cancelled by push-pull action. Its kind of a circular argument - the input pentode requires the lower Miller that the splitter pentode gives but then sort of demands more feedback to cancel the extra distortion the pentodes now generate. The Williamson amplifier used all triodes but also had 3 stages before the output and required a very high quality transformer to apply feedback due to all the poles and how close they were to one another - not much feedback can be applied around a Williamson. The subsequent Mullard and Bailey designs strove to achieve similar gain with only two stages so they could use more feedback to achieve similar or better distortion performance and probably a less stringently-specced output transformer. Its always fascinating to try and guess what drove the designers of these configurations.
 

Moonraker

New member
Joined
Jun 18, 2021
Messages
26
Reaction score
7
Location
Free State
No need to preach to the converted  :) :)

But seriously, as you say it is a tail-chase.  I doubt very much if the additional gain bandwidth , NFB, etc of this configuration will get one noticeably further after the tradeoffs have been considered.

The overriding deviations from the ideal, desired or not, of most (all) tube amplifiers are caused by the impedance matching transformer(s). 

Until someone does the logical thing and eliminates transformers by driving electrostatic loudspeakers directly.  Or by using a bank of say 6080 type tubes in parallel.  Or something.
 

fredeb

AVForums Grandmaster
Joined
Mar 29, 2012
Messages
3,468
Reaction score
515
Location
Somerset West
Moonraker said:
No need to preach to the converted  :) :)

But seriously, as you say it is a tail-chase.  I doubt very much if the additional gain bandwidth , NFB, etc of this configuration will get one noticeably further after the tradeoffs have been considered.

The overriding deviations from the ideal, desired or not, of most (all) tube amplifiers are caused by the impedance matching transformer(s). 

Until someone does the logical thing and eliminates transformers by driving electrostatic loudspeakers directly.  Or by using a bank of say 6080 type tubes in parallel.  Or something.

I happen to have stash of 6080 , and was wondering what a pair ( so parallel push pull ) would be like , without output transformer . A headphone amp , like a Crack clone parallel SE OTL should be simple enough . The only issue with 6080 is the high current filament supply . Purist probably reckon that a single triode SE OTL might sound better .

I quite like your idea of driving Electrostats directly , I never knew it was possible . I guess it could be dangerous if something went wrong . How does it work ? Does one disconnect the electrostat transformer ?

I know we're going waaaay OT now , but so what , seems like an interesting thing to dream about . 

I'd also like to add , that what first prickled my interest in the Bailey circuit , is that [member=2759]charles[/member] likes the Radfords so much , and that he does have much experience with many of the best classic amplifiers , probably owns them too . I respect Charles's opinion .

As far as decent output transformers go , they appear to be easily available again these days . Torroidy have some decent options that are affordable if you don't take the fancy covers . I was looking at some this morning , insane bandwidth and even cathode feedback windings if you so desired .

https://sklep.toroidy.pl/en_US/p/-TTG-CFB6600PP-Tube-output-CFB-transformer-6%2C6kOhm-Cathode-Feedback-Push-pull/660


 

Ampdog

AVForums Grandmaster
Joined
Aug 27, 2006
Messages
7,611
Reaction score
455
Location
Pretoria, South Africa
Grrrrr . . .

The nightmares of the various un-pentode executions of input pentodes!  First year basics fly out the back door as the screen grid is bypassed  - -  to common in the presence of relatively large nfb. ! This leaves a lamish kind of operation as in the equivalent circuit the nfb signal is applied in cancelling mode to the screen grid. (It vaguely resembles a kind of UL circuit, but in this application altogether non-sensical. [Not to open this heated opinion  generating debate yet again (although long time hence).  Various analyses show the compromises of this practice. (True pentode operation would of course have been when the screen bypass capacitor was returned to the cathode instead of common. (Even the worthy Harold Leak fell for this one . . .) The ST70 circuit has it right. . .

Mercifully the input stage adds little to total distortion unless the design is heroically poor. - still!  'It is not done that way! (snorrrrt)'

[Handsome - - careful always where you are concerned - but: pentodes give more distortion than triodes in an input stage? ?  My data has it that a pentode input can have some 10% - 15% of the distortion of an input triode at low signal!  This rather academic as input stages contribue little to the total distortion pattern when correcly designed.  (I prefer a pentode in the input because it imparts negligible Miller capacitance on an unknown pre-amplifier output impedance. And the screen bypass time constant can be used as a very handy pole when addressing stable l.f. response. Pentode can also operate at a wide  variety of anode voltages should direct coupling to the next stage is required]
 

Ampdog

AVForums Grandmaster
Joined
Aug 27, 2006
Messages
7,611
Reaction score
455
Location
Pretoria, South Africa
Apology! Not desiring to overwhelm - but I found the often encountered cathode leakage --to-heater in many ECF82s troublesome.  (Where I used them as input stage I had to resort to grid leak bias (20 meg  G1-common  resistance) and the cathode tied directly to common (or with a very low cathode resistor for NFB). (It is basically an R.F. tube with not much accent on heater-cathode insulation.)
 

fredeb

AVForums Grandmaster
Joined
Mar 29, 2012
Messages
3,468
Reaction score
515
Location
Somerset West
Ampdog said:
Apology! Not desiring to overwhelm - but I found the often encountered cathode leakage --to-heater in many ECF82s troublesome.  (Where I used them as input stage I had to resort to grid leak bias (20 meg  G1-common  resistance) and the cathode tied directly to common (or with a very low cathode resistor for NFB). (It is basically an R.F. tube with not much accent on heater-cathode insulation.)

Thanks for the warning [member=144]Ampdog[/member] . What is the way to detect such leakage ? My modded ST70 has a board that uses ECF82/6U8 instead of 7199 , I bought it like that . I suppose the change was made because of how expensive 7199 had become , if you can find them at all . What would the symptoms of above mentioned cathode to filament leakage be ?

Cheers
- Ari

 

Ampdog

AVForums Grandmaster
Joined
Aug 27, 2006
Messages
7,611
Reaction score
455
Location
Pretoria, South Africa
Usually hum. 

This could be quite soft, however, so also confirmed with an oscilloscope. (I have used a few other triode-pentodes without this symptom. Yes, the 7199 was never universally accepted; too many other similar but less expensive valves on the market.  (Although most have been developed for r.f. use, they work well in audio except at low signals eg. pre-amp input stages.)
 

fredeb

AVForums Grandmaster
Joined
Mar 29, 2012
Messages
3,468
Reaction score
515
Location
Somerset West
Ampdog said:
Usually hum. 

This could be quite soft, however, so also confirmed with an oscilloscope. (I have used a few other triode-pentodes without this symptom. Yes, the 7199 was never universally accepted; too many other similar but less expensive valves on the market.  (Although most have been developed for r.f. use, they work well in audio except at low signals eg. pre-amp input stages.)

Thanks [member=144]Ampdog[/member] , there was initially hum from the amp , with 2 different ECF82 that the amp came with , one a Zaerix and I can;t remember what the other was . I ordered some RCA 6U8A and 6AN8 , I haven't even tried the 6AN8 , but all seems good with the 6U8A .


 

fredeb

AVForums Grandmaster
Joined
Mar 29, 2012
Messages
3,468
Reaction score
515
Location
Somerset West
That said .... There are quite a couple of other ST70 driver boards out there , when I have time , I could try a different front-end . I wonder if it was the power and output transformers that limited the Radford to 25W , compared to the rated 35W the ST70 is said to put out ? Is the Bailey front-end incapable of getting 35W @ 8ohm from the ST70 output stage ? Or did the Americans and British have different methods for measuring output power ? Does the Radford offer better bandwidth at similar power levels ? Radford appear to have been quite proud of their output transformers . Are they really that good ?








 
Top