baronbeehive said:

 

sonictrance said:

  Haha! No Eurovision for Sabaton I think

Click to expand...

 
Why not? I seem to remember Norway had a very heavy metal band one year. Seriously I still love Abba, to me on a par with the Beatles.

Click to expand...

Yes, Abba is by far Swedens biggest music export of all time. They were (and still is) huge worldwide!

It turns out that the first mod tried in the other thread is much more important than ever contemplated. ..

I am talking about the power resistors (330ohm in MK6) for the plate/anode.

Even so much as a slight variance in resistance will affect sound, depending what headphone you ate using!

Why?

Because this resistor directly affects the output impedance operational points. ..
And the resistance change is an exponential one, not a linear progression at all..
A few ohm difference can make a tenfold difference in the circuit/tube operation for headphone impedance setting..

So I was correct all along to insist on using tight tolerance, non-inductive wirewound MILLS resistors. ..

I have to credit to :
http://www.soniccraft.com/index.php

As they went above and beyond the already tight tolerance of these resistors,
And offered an amazing "precision matching",
Down to level "L3":
L3 = 1/4 factory tolerance 1/4% matching.


So all my resistors for the next mod (headphone impedance matching mod),
are down to tolerance levels beyond what test equipment use!

Also, my next set of WCF caps came in, for testing..

Of course they will be doubled to achieve the .22uf ,
But the the interesting point is in using two caps instead of one..
As the actually measure .115uf each:


Using two smaller caps should make for a more responsive performance.

Maxx, I cannot grasp how the plate resistor in a servo design can affect the output impedance. The only way there will be 0 volts DC at the output is if both tubes have matching impedances, the servo adjusts the top triode impedance so there is 0 volts on the output at all times therefore maintaining a constant output impedance correct? The only thing I see the plate resistor doing is setting the voltage/current to fit the linear range on the load lines.
Have you personally tested that the plate resistor has any affect on a servoed WCF?

coinmaster said:

  Maxx, I cannot grasp how the plate resistor in a servo design can affect the output impedance. The only way there will be 0 volts DC at the output is if both tubes have matching impedances, the servo adjusts the top triode impedance so there is 0 volts on the output at all times therefore maintaining a constant output impedance correct? The only thing I see the plate resistor doing is setting the voltage/current to fit the linear range on the load lines.
Have you personally tested that the plate resistor has any affect on a servoed WCF?

Click to expand...

 
I don't know what is a "servo design", but the Zo of a WCF is calculated like that :
 

and the plate resistor ("Ra" here) is part of it.
 
I had run some simulations from this formula some months ago and the variation of Ra had a significant impact on the overall Zo.
 
The funny thing with this formula is to play with the Rk, when considering the Rk as an high-pass RC filter, and see how the Zo drops with the increase of frequency.
From that, you have a better idea of what's the "best" value for the Cathode bypass caps.
 
 
[EDIT]
and obviously, the second reason for choosing the best possible Plate resistors is that the "signal" current will run through them.
Any "resistor coloration" will be heard straightaway, which is not the case for the Cathode resistors for instance.

SMLD MODS:
 
Stage 3 - Identify changes on PCB.
 
Wiring: The 2 brown, 2 blue and 2 yellow output wires in the centre of the board tied together. Replace these with 20AWG OCC copper which should not need to be doubled or twisted. Also the 4 runs of wiring around the chassis from board to output jacks to be replaced with 22AWG silver plated copper wires, easy to see.
 

 
Resistors:
1. The 8 x 330 ohm power stage green resistors. 4 on either side of the board, (2 of which in pic above, reverse of board). Replace with 8 x Mills MRA5 330R.
2. 4 x 220k ohm driver stage blue resistors, 2 together in the top right of pic next to driver tube. Replace with 4 x TKD 2w 220K metal film.
 

 
3. 1 x 4k1 resistor on the right of the 3 resistors on the bottom edge next to the small brown vertical cap. Replace with 2 x Mills MRA5 8K2.
 

 
Capacitors:
1. Coupling caps, the 2 large black caps in centre pic in the middle. Replace these with 4 x Audyn True Copper 0.22uF 630VDC.
2. WCF caps, the 2 horizontal red oblong caps in the centre of the centre pic above the power tubes. Replace these with 4 x Mundorf Silver in Oil 0.33 uF 1200V.
3. Decoupling cap. Add 4 x Mundorf Mcap 47uf, 250V for power stage, and 2 x MKP 40 µF Wima 800V for driver stage. Also 6 x 0.33uF 275Vdc Evox Rifa bypass caps. Pics to follow.
4. Cathode Bypass caps, The 2 large blue 220uF caps in bottom centre of the pic in the centre in between the 2 power tubes. Replace with 4 x Mundorf E-cap 220uf, 63V. Also 4 x 0.1uF 275Vdc Evox Rifa bypass caps.
5. PSU caps, the 2 black vertical caps on left of pic above. Replace with 2 x Mundorf M-Lytics AG 1500µF 160V.
 
Next section should be the mods themselves with locations etc. . I've found out that it's possible to put the tubes in to test the mods with the board as it is half in and half out providing you are careful. I haven't tried it switched on.

Looking good there baron! You've everything under control. Looking forward to see more progress pics!
 

sonictrance said:

  Looking good there baron! You've everything under control. Looking forward to see more progress pics!

Click to expand...

 Thanks . Still one or two things on order which I couldn't get locally but it's getting critical now with the soldering/desoldering stage coming up. Should be interesting......... I don't know one end of a soldering iron from the other!

baronbeehive said:

 

sonictrance said:

  Looking good there baron! You've everything under control. Looking forward to see more progress pics!

Click to expand...

 Thanks

. Still one or two things on order which I couldn't get locally but it's getting critical now with the soldering/desoldering stage coming up. Should be interesting......... I don't know one end of a soldering iron from the other!

Click to expand...

Soldering is easy, especially with the Cardas solder! Do some practise on the side first and you'll be OK. 

baronbeehive said:

...
I don't know one end of a soldering iron from the other!

Click to expand...

Sure you do.
Make sure you have smallest tip possible.
You can get a file to clean the tip too.

Keep an eye on the angle of the iron, so you don't end up touching other parts and making melted marks on them.
It's not easy but if you melt a bit off your caps don't worry.

maxx134 said:

It turns out that the first mod tried in the other thread is much more important than ever contemplated. ..

I am talking about the power resistors (330ohm in MK6) for the plate/anode.

Even so much as a slight variance in resistance will affect sound, depending what headphone you ate using!

Why?

Because this resistor directly affects the output impedance operational points. ..
And the resistance change is an exponential one, not a linear progression at all..
A few ohm difference can make a tenfold difference in the circuit/tube operation for headphone impedance setting..

So I was correct all along to insist on using tight tolerance, non-inductive wirewound MILLS resistors. ..

I have to credit to :
http://www.soniccraft.com/index.php

As they went above and beyond the already tight tolerance of these resistors,
And offered an amazing "precision matching",
Down to level "L3":
L3 = 1/4 factory tolerance 1/4% matching.


So all my resistors for the next mod (headphone impedance matching mod),
are down to tolerance levels beyond what test equipment use!

Click to expand...

 
This is very exciting, getting down to experimental territory, this could be very interesting, well done Maxx and co. Maths isn't my strong point but I just about get where you're coming from.

I don't know what is a "servo design", but the Zo of a WCF is calculated like that 

Click to expand...

Our WCF uses a DC servo to null the DC at the output. The only time the DC is zero at the output is when both tubes have a matching impedance, therefore the servo maintains an equal impedance between them. Changing the plate resistor value will screw up the impedance divider of the tubes and will probably cause a required uneven impedance between the tubes to maintain a 0v output and it will alter the intended load line operating point of 75ma/75v on the MK6 which is what the amp was designed for given its voltage swing (which is close to cutoff to begin with which is why the NFB is needed).
 
So I question whether the cavalli paper refers to an OCL WCF at all.

coinmaster said:

Maxx, I cannot grasp how ...

Click to expand...

If you were thinking about the opamp control in relation to the plate resistor,
Remember that opamp control is between two tube circuits, not between two triodes within a tube.

My testing has been verified long ago.
The currently "in progress" mod is the selector switch with setrings for various headphone impedances.


Oh yeah, I just read your last post..

All plate resistors are changed to be even more precisely & exactly matched the same..

And with new ability for all plate resistors to be "switched" to new values, all evenly at the same time.

The new design is also "fool proof" against switch failure, so it is impossible to cause damage.

So your point is "moot", or does not apply.

But your points were interesting of how you are understanding things.

I edited the end of this post for clarity.

Edit:
Another point you forgot is the amp protection circuit which will trigger the output relay instantly against this type of "DC offset" fault.

I know because in my tests I have triggered it this way, by simulating this condition by turning to maximum, one of the four tiny trimpots for the opamp.

Anyways there is your answer.

Coin, this is how I see this thing working.
 
 

Our WCF uses a DC servo to null the DC at the output. 

Click to expand...

 
A "DC servo" does not "nullify the DC at the output", only output transformers or ouput capacitors can do that, in my book. And we don't have any of those in either the MK6 or MK8. More of that below.
My understanding is that this "DC servo" is only monitoring the output. I confess I didn't pay much attention to it as it is not supposed to be in the signal path.
Whatever, when the "what it is monitoting" (voltage or current) exceeds the programmed value, it triggers the relay and cut the output.
Plain and simple.
 
 

The only time the DC is zero at the output is when both tubes have a matching impedance, therefore the servo maintains an equal impedance between them. 

Click to expand...

 
You should be more precise about what is your reference voltage. Here it really matters as it differs between "SE operation" and "symetrical".
 
The output voltage of an OCL WCF is roughly half the rail voltage, minus the voltage drops due to the Ra and Rk. Basically,
100V for a 200V power rail. The voltage reference is taken at the bottom rail, below the cathode.
And basically the only thing that can change this 100V voltage would be to have a significant unbalance between the internal resistance of the two triode of the tube.
Whatever tube you put, intil they have the same internal resistance, they will output the same 100V. 
 
So what ?
 
When running symetrical, both tubes of a channel are running at 100V. The "difference in potential" (not sure of the trad. here) which is the true definition of a voltage by the way is 100V - 100V => 0V.
So, we don't get electrocuted ! Nice !
 
When running SE, it's a bit more tricky ! Only one tube is being used and its WCF is still outputing 100V. But we need to have a "difference in potential" of 0V at the headphone if we want to not fry everything ...
That's why the "ground" of the SE plug is not linked to the rail ground but to the "mid-point" of the power PSU where the potential is also at 100V above the one of the "rail ground".
And then TA DA !!! we have our equilibrium/balance between potential as they are seen by the headphone : 100V - 100V = 0V.
 
 
 

Changing the plate resistor value [...] will probably cause a required uneven impedance between the tubes to maintain a 0v output

Click to expand...

 
I don't see why we would end up with an "uneven impedance between the tubes" when we have applied the same Ra for both tubes of a channel.
 
 

... and it will alter the intended load line operating point of 75ma/75v on the MK6 which is what the amp was designed for given its voltage swing (which is close to cutoff to begin with which is why the NFB is needed).

Click to expand...

 
 
You're right in saying that it will change the "operating point". That's a good point. 
 
And it would be bothering that by lowering the Rplate we increase the plate current above the "max Ia specs". We'll have a look into that.

My understanding is that this "DC servo" is only monitoring the output. I confess I didn't pay much attention to it as it is not supposed to be in the signal path.
Whatever, when the "what it is monitoting" (voltage or current) exceeds the programmed value, it triggers the relay and cut the output.
Plain and simple.
 

Click to expand...

Not at all, the WCF is like a resistor divider, If you have equal and opposite voltages feeding two equal value resistors you are going to have 0v between them and tubes are not perfect resistors. Each tube of the same type is going to vary from one another and shift with time. The DC servo (opamps) vary their output voltage (the grid bias voltage of the top tube) based on their input (the output voltage of the amplifier), an opamp will keep changing it's output until its input is 0v. Without the opamp you cannot have an OCL WCF.
 
The "flat" IC in the center of the amp is the thing that cuts the output when the DC offset is higher than the threshhold value. Otherwise if something fails it's your headphones that will go up in smoke since they are not designed for any amount of dissipation.