[robzy]

Hey guys,

Just finished drawing up my szekeres with a cmoy hacked on. It will be powered off a 15v PSU.

Any comments/suggestions? Anythings that will obviously not work? (Im not 100% sure if i have configured the opamp for a single voltage PSU properly)

(I have both a resistor and CCS, only one will be populated on the board though)

Rob.

(I just realised there is a schematic just like this in the Project Addendum for the szekeres... grrrr.....)

 

[dsavitsk]

Even better, read through this thread for tips on really making this design shine. http://www6.head-fi.org/forums/showthread.php?t=10266

 

[xxie24]

There are some issues I can spot from your design.

1) The opamp input bias in your design is made for NE5532 similar devices. The PSU noise has direct connection to the input of opamp in single supply. For OPA132 I suggest you change to the standard single supply opamp bias circuit (like Fig2 in this note) or at least increase the C1 to a much larger range.
2) Put a 20uF something tantalum in serial with R2
3) change the R6 from 220k to 100-220ohm range.
4) R10 compromise the LM317 CSS. get rid of it.

 

[ericj]

Why not build a Tori amp and forego the output capacitor?

Edit: Described here -- http://headwize.com/ubb/showpost.php...=19965&fpage=2

 

[robzy]

dsavitsk, thanks for that link, i some how managed to miss it when i searched through headfi for "Szekeres" before.

xxie24: Exactly the type of info i was after

Thanks a lot!

Anyways, what i realised is that there is a circuit practically exactly the same in the Project Adendum:

Anyway, this is what i came up with:

R6R will turn into ~5k because i forgot to add in the current limiting resistor (R1 from headwize design).

There is no 4.7k resistor from opamp output to ground because that is there for class A biasing, and i want to come up with a better solution for that. (What is the input impedance of the "szekeres" buffer by the way?)

I have both a CCS and a resistor because i want to be able to choose which to use, so i can try out both. (Both will _not_ be used simulataniously).

So, any other errors i need to look into?

ericj, thanks for linkage, but at the moment i think i will stick with the szekeres.

Rob.

 

[dsavitsk]

Quote:

Originally Posted by robzy (What is the input impedance of the "szekeres" buffer by the way?)

I think the issue is the gate capacitance (I say that like I know what it means, which I assure I don't.) Anyway, here is someone saying it doesn't matter in this application: http://www.geofex.com/Article_Folder...osfetfolly.htm

 

[robzy]

Quote:

Originally Posted by dsavitsk I think the issue is the gate capacitance (I say that like I know what it means, which I assure I don't.) Anyway, here is someone saying it doesn't matter in this application: http://www.geofex.com/Article_Folder...osfetfolly.htm

That was an interesting read! Much much thanks for linkage.

I wanted to know the input impedance of the szekeres, though, so i would know how much i have to pull through the opamp to drag it into Class A. Anyone know?

Rob.

 

[Tomo]

Hi,

I am staring at my old EE homework on AC circuit diagram. I think if the capacitance 1uF is considered pretty big, we can say the impedance is 100 kohm || 220 kohms = 69 kohms.

For lower frequencies, this value will be larger.

Ciao,

Tomo

 

[robzy]

Quote:

Originally Posted by Tomo I am staring at my old EE homework on AC circuit diagram. I think if the capacitance 1uF is considered pretty big, we can say the impedance is 100 kohm || 220 kohms = 69 kohms.

Hey, thanks for the reply, but im a bit confused;

Where exactly did you get the 100kohm figure from? And did you get the 220kohm value from R6? If so, it is actually meant to be a 5k resistor.

And assuming you are talking about R6, isnt it in serial, not paralel?

Thanks,
Rob.

 

[Tomo]

Hi,

What I did is pretty darn difficult. ... I just understood 5 years after actually taking the darn course. ... I am a uber-slow learner.

Anyways, I am talking about R4 and R5 resistors which forms a voltage divider. R6 is less important because MOSFET gate does not connect anywhere. You see MOSFET gate is "floating" over Silicon over Source and Drain. You get pretty small capacitance across the gate and this Silcon area. So effectively this is a broken circuit in AC domain. (unless for uber-high frequency or uber-high gate cap) ... I know this is kinda hard to swallow, too. ... I understood this 2 years after the other darn course.

Anyways, so

R4 || R5 ~ input impedance (APPROXIMATELY!)

Input cap does matter but complicates the problems. You just want estimates, right?

Tomo

 

[robzy]

Its time's like these i am really glad that i chose an amp design i was not familiar with, learning about it makes it all worth it.

Esentially the gate of the MOSFET as infinite resistance, right?

Thank you for your answer, only thing is - though - that i was actually refering to the 2nd schematic where R4 and R5 have been removed.

Does that mean that the "szekeres" parth of the amp (the opamp load) would esentially have infinite resistance? And thus only a tiny amount of current would be needed to pull it into Class A?

Rob.

 

[Tomo]

Hi,

I am talking about the first Szekeres amp schematics you posted.

Well infinite resistance is a wrong word. ... There is no current that flows from Gate to Source or Drain. Gate draw no (DC) current.

This seemingly lead to your conclusion about MOSFET requiring very little "hustle" to drive. However, this is FAR from the truth.

You understand a transistor (whether BJT or MOSFET) is a electrocity operated switch. For MOSFETs, the switch is operated by charges in the gate. ... Yes, charges as in sweaters making cracking sound in winter time. (for the ladies, skirt sticks to your leggs in the winter time. Very annoying, I hear.) Electrocity is composed of those sparks.

As I have mentioned, Gate is floating. So no electrocity conducts through gate. So how in the hell is a MOSFET do its thing??

Well electrocity is composed of charges. And even though electrocity don't conduct, you can "pump" charges onto the gate. This is that which makes MOSFET work! ... (You just finished EE2XX in college!)

You: So what, old geezer?

Well have you ever pumped car-tires with hand-operated pump? Wow wee I must have lost 10lb doing that. Pumping charges into MOSFET gate is similarly hard. And, the opamp has to do this hard ordeal.

This is usually indicated by "driving capacitive load" section in opamp specification sheets.

Class-A business is a whole another story. You pull a opamp output stage into Class-A ... that is what you heard. This require "steady" current. And, we were talking about AC current. (+- changes and stuff)

Tomo

 

[Tomo]

Another thing,

I hate to burst your bubble, but you might want to avoid "pulling" your opamp to Class-A.

I am not accustomed to "pulling" opamps, but that would mean your MOSFET gate have relative direct path B+ (V+). Gate is very sensitive. You can destroy it with a small spark, let alone brief moment of high voltage.

You might want to check out Tori-Amp. This requires dual PSU, but uses a very fascinating method. You might use similar tactics to operate an opamp in Class-A. (This is also done with SDS amp)

Me? I am a conservative, so I avoid "pulling" opamps altogether. Instead, I go for opamps operating naturally class-A. (And blame it on manufacturers) However, since JFET CCS, I am tickled to try "pulling."

Tomo

P.S. Just for a note, it has been our experience but using CCS to replace R9R in your schematics does not improve the performance. You see my endevers on the addendum (yeah I am the weird last name guy). Rick helped me out a lot doing them, but we came up ... rather unsatisfactory. You can ask Rick. He's wiser.

P.P.S. Your LM317T / U2R is not connected correctly. You should check out the LM317 specs for exact CCS instructions.

 

[robzy]

Ahh... i always kind of assumed that you were the Tomo from "Tomo's CCS" and (now that ive had proper looks at it) the headwize article.

Thank you a lot for the explanations on MOSFETs, in all honesty i did not have much idea at all on how they work or exactly what they are.

Thanks for noticing i had hooked up the LM317 CCS wrong, if you notice in the first schematic it was correct, but for some reason i managed to screw it up in the second one. All fixed now though

(From what i have read most people tend to swing towards the resistor instead of the CCS, but i am kind of curious to hear both)

I am very confused though, and i think that what you are saying contradicts what i understand from others (Oh the wonderfull world of audio

)

About "pulling" (or what ever the correct word is) opamps into class a, the idea of it is to keep the output stage turned on the entire time in order to increase linearity. (And also activate on the PNP/NPN, i forget which). To do this you work out the amount of current the opamp will be dilivering and then add a current source that draws greater than that amount.

I figured the best way to work that out would be to use the impedance of the MOSTFET (as the opamp saw it) to work out the max current draw and then choose a nice CCS to draw current from the opamp. But according to what you are saying - there is no way to do that?

In the headwize addendum (on the schematic i posted above my second schematic) the guy has added a 4.7k resistor to ground after the opamp output in order to draw the opamp into class A. I do not like the idea of a resistor as a current source (after reading tangent's article on biasing opamps into class A) and was wondering which Current Regulating Diode to replace it with.

But according to what you are saying - there is basically no way to do this?

Also, you say:

Quote:

Originally Posted by Tomo but that would mean your MOSFET gate have relative direct path B+ (V+). Gate is very sensitive. You can destroy it with a small spark, let alone brief moment of high voltage.

The MOSFET gate would only have a relative direct path to GND, not to the +15v. All i would be doing is replacing the 4.7k resistor with a CCS instead.

Rob.

 

[Tomo]

Hey, Rob.

Sorry, I got misunderstood you the first time.

Quote:

In the headwize addendum (on the schematic i posted above my second schematic) the guy has added a 4.7k resistor to ground after the opamp output in order to draw the opamp into class A. I do not like the idea of a resistor as a current source (after reading tangent's article on biasing opamps into class A) and was wondering which Current Regulating Diode to replace it with.

I don't think 4.7K is there for that reason. I believe it is there to allow a current drain while turning off/on. As you probably know, the opamp is biased by the voltage divider. The other input is floating due toa a capacitor 4.7uF. This is a large capacitance.

Every time you turn on, this capacitor must charge up to 6~8V. Every time you turn off, this capacitor must discharge to 0V. This is a law of UNIVERSE thing. (Curse the divine(s) for inflexibility.

)

This charging and discharging require instanteneous current ("surge"). This cause a "thump." I believe 4.7Kohm is there to divert this "surge" away from Opamp and MOSFET. It is NOT intentionally pulling the opamp; pulling opamp would be the side-effect. (Dude, this took me 1 whole year to understand.)

So be advised, smack in CCS in place of 4.7kohm may not be a good idea.

Tomo