• The move to the new server is done. There are some software and database maintenance updates in process. This has us passing the hat around to help out. We appreciate any donations. Seriously, even a dollar helps. The payment page may be found here - https://www.audiokarma.org/support.html

Edcor GXPP 8K high end?

hey Mike,
Try a simplified Schade I call E-Linear. Take the driver/phase splitter plate loads right to the corresponding final's U-L taps. Switch also to a pair of pentodes; Schade and E-Linear both want a stable and high output Z driver stage. With a tube like 6EJ7 the output Z will be *ABOUT* equal to the plate loads and it will not vary much across signal like a triode will( which will change the amount of delivered FB ). Try about 30k, and that will drive your 470k grid resistors in the EL84 stage too.
cheers,
Douglas
 
It looks like you currently have the high-pass filter inside the NFB loop, so you're working at cross purposes here. Cutoff frequency will be strongly dependent on feedback fraction. Any special reason for this arrangement?

Global NFB of 6dB or so is very unlikely to result in stability problems.

You're designing amps without a scope on the workbench? Why make life unnecessarily difficult?
 
Just for info, don't worry about the top end on any of the Edcor SE outputs, even those
inexpensive ones are out past 50k before they start dropping off any significant amount.
(I know your talking about the push pull ones here, just giving an Edcor reference)
 
PakProtector said:
It is also possible to choose source impedance so as to extend bandwidth...
Click to expand...

I see a lot of wisdom in that simple statement. I believe you could take a 2.5K:4 output transformer, put a 8 ohm load on it, and it would present a 5K load to the power tube. You could also take a 10K:16 transformer, put a 8 ohm load on it, and it would also present a 5K load to the tube. Assuming both transformers have similar bandwidth specifications, I'd expect the lightly loaded 2.5K:4 should offer better high end extension, while the heavily loaded 10K:16 might give superior low end extension.
 
ducati_EL34 said:
You said "your philosophy", but you also quotes Crowhurst and some other fellow, who stated, Feedback is needed in PP amps.
Click to expand...

Nope, no quotes -

"It has been a long time since I've thought about this and I do not claim to be Bill Whitlock or Norman Crowhurst; take it FWIW -"

Bill Whitlock, is the president of Jensen transformers.

FWIW is for what it's worth, i.e. not much.

;)
 
ducati_EL34 said:
Feedback is needed in PP amps.
Click to expand...

Huh? Feedback is desirable in pentode (and probably UL) designs to bring the output impedance (damping factor) down to a level that is normally accepted. What is it about push/pull that inherently "needs" feedback?

Do a quick Google search, and you will find plenty of push/pull designs with no global feedback. The first thing I notice is they are all triode (or triode strapped) amps. Here's a few to get you started:

http://www.hagtech.com/cymbal.html
http://www.caryaudio.com/products/classic/CAD211FE.html
http://www.audiodesignguide.com/my/pp3.html
http://www.kta-hifi.net/projects/amp_page/6ck4amp/6CK4.html
 
Last edited:
As long as you have a decent load an say like a triode tube, feedback isn't "needed".
Running triode strapped 6AV5 (basically 2A3) with 6k pl to pl load, no fbk required,. :music:
(triodes generally have low resistance, an no feedback can still give good dampening)

Running class A in push pull there is no crossover distortion, feedback is not needed
for that to happen, feedback generally lowers overall distortion an extends response.
(proper feedback in output stages lowers the amps output impedance, many choices)

---------

And global/loop feedback is not the only way to lower output impedance, UL is a form
of feedback, with it lowers output impedance over a straight pentode setup, Douglas's
E-Linear setup lowers output impedance, Schade plate to grid lowers output impedance.
(having outputs with cathode feedback windings can really do wonders there as well).. :)
 
Ty_Bower said:
I see a lot of wisdom in that simple statement. I believe you could take a 2.5K:4 output transformer, put a 8 ohm load on it, and it would present a 5K load to the power tube. You could also take a 10K:16 transformer, put a 8 ohm load on it, and it would also present a 5K load to the tube. Assuming both transformers have similar bandwidth specifications, I'd expect the lightly loaded 2.5K:4 should offer better high end extension, while the heavily loaded 10K:16 might give superior low end extension.
Click to expand...

Ty, I believe you've got source impedance mixed with load inpedance. A 2A3 has about 800 Ohms around a traditional Class A operating point. A KT88 has something in excess of 20k( depending on its OP and rigging, pentode, U-L, etc ). It is referring to what is driving the TX; speaking of the power finals.
cheers,
Douglas
 
Ty_Bower said:
Huh? Feedback is desirable in pentode (and probably UL) designs to bring the output impedance (damping factor) down to a level that is normally accepted. What is it about push/pull that inherently "needs" feedback?
Click to expand...

I did not say that. I was only quoting a portion of an earlier post from Soundmotor... :thmbsp:
 
ducati_EL34 said:
I did not say that. I was only quoting a portion of an earlier post from Soundmotor... :thmbsp:
Click to expand...

Which was in context to the OP -

BTW the application is PP EL84

I was not thinking about the class-A aspect w/ SE or PP triodes or triode-connected pentodes when I wrote my reply. None of that had entered the discussion yet. My conclusion though is still the same, spending on better output transformers is money well spent.
 
soundmotor said:
Which was in context to the OP -

BTW the application is PP EL84

I was not thinking about the class-A aspect w/ SE or PP triodes or triode-connected pentodes when I wrote my reply. None of that had entered the discussion yet. My conclusion though is still the same, spending on better output transformers is money well spent.
Click to expand...


No argument there. especially for the small price difference. :thmbsp:
 
Last edited:
soundmotor said:
Which was in context to the OP -

BTW the application is PP EL84

I was not thinking about the class-A aspect w/ SE or PP triodes or triode-connected pentodes when I wrote my reply. None of that had entered the discussion yet. My conclusion though is still the same, spending on better output transformers is money well spent.
Click to expand...

I don't think you'll get much argument on getting a best output you can depending on
frequency response you want/need in a given application, it was the fact of saying an
push pull amp "always" needs feedback to perform/test well, making it sound like single
ended doesn't always need it but push pull does, where in reality that right load on the
right tube neither topology inherently "need" to have the feedback to perform well... :)

-----------

Reading into Douglas's point, take a single ended 2A3 with no feedback and proper load
with good circuitry, then take the 2A3 and do the same push pull and it will test better.
 
BinaryMike said:
It looks like you currently have the high-pass filter inside the NFB loop, so you're working at cross purposes here. Cutoff frequency will be strongly dependent on feedback fraction. Any special reason for this arrangement?

Global NFB of 6dB or so is very unlikely to result in stability problems.

You're designing amps without a scope on the workbench? Why make life unnecessarily difficult?
Click to expand...

By high pass are you referring to the .00147uf coupling caps? The reason is that the output is intended for 80Hz and above.

mike
 
Thought it might be helpful to show a more up to date look at what I am proposing. This is the current iteration of the design. Note that the treble control is between the VAS and PI (R20 and R21 are the treble pot and R17, R19 and C2 make up the treble pre-emphasis that allows for + and - in the control.

The SS part is a 2nd order LP eq to compensate for the intentionally "too small" subwoofer enclosure. It is sort of a simplified Linkwitz in a way. The separate plate amp will provide the actual crossover and the gain control on the plate amp provides the bass control.

I have made the caps coupling the PI to the drivers small enough to get the desired HP for the main channel now so that they are outside the Schade loop.

In the end I may well end up at different bias points for the early stages of the amp after taking a closer look at the curves for the tubes in question. The bias on the 6BQ5s is straight out of the Mullard data sheet so I suspect that it ought to be pretty close to ideal. :)

attachment.php


attachment.php


And a question for Kegger. I notice that in similar situations you use about 50K on the 12AT7 plate in a Schade situation (at least in SE). My simulation uses 68K. Do you think that there would be a significant difference in performance between these two? Also is normal load line analysis valid for the driver in a Schade output?

Thanks all.

mike

P.S. Just realized I forgot the coupling cap after the VAS. Without it there is DC current through the treble control pot so would add a .68u cap there. Also note that treble control has 100k pot in series with 22k shelving resistor.
 
Last edited:
Basically I go by the 3 times plate resistance "rule" meaning it is just enough, I prefer
a 5x plate resistance plate load ratio when the supply offers enough voltage headroom.

That is how I tend to setup plate loads on triodes, be it in a schade setup or not even.
 
mashaffer said:
I have made the caps coupling the PI to the drivers small enough to get the desired HP for the main channel now so that they are outside the Schade loop.
Click to expand...

I don't see a good place to connect global NFB; assuming that you intend to experiment with it. Also, global NFB will shift your HPF cutoff frequency, making it difficult to evaluate other aspects of performance.
 
Good point Mike. I could change input triode to resistive bias and connect gNFB there but as you say that mucks with the HP. Seems like it would require entirely different approach or adding more stages (not an attractive option).

mike
 
Just last night I noticed another (this time fundamental) error. In the tone control circuitry I put the treble cut pot before the pre-emphasis voltage divider. Of course this allows a tremendous variation in load on the previous amplification stage such that the overall gain of the circuit changes dramatically with the setting of the pot. By putting the constant voltage divider first this interaction will be dramatically reduced.

One other minor issue is that I modeled a 150k (the 97k on the bottom of the pot includes the 22k shelving resistor) pot which of course does not exist. :)

mike
 
You must log in or register to reply here.
Back
Top Bottom