Lateral mosfet w Normal or Diamond buffer?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.

fab

Member
Joined 2004
Paid Member
Hi to all

This time I would like to get your opinion on the differences between the Normal emitter follower buffer and the Diamond buffer to drive a Lateral Mosfet power output stage.

I will attached 2 amp schematics:
- one with Normal buffer
- other with Diamond buffer (same one as APEX PA52 device)

Except for the buffer to drive the output stage the rest of the amp is the same. According to the basic measurements I do with Microcap 6 simulator I can not see really the differences between both topologies.

For the diamond buffer, the low VGS off the output lateral mosfets, makes me use a simple VR for bias current (but with regulated supply) instead of the Vbe multiplier. Maybe I could use a low VGSoff lateral mosfet for this but I think that I would need to select especially low VGS off devices for that purpose...

I have read some other threads on the diamond buffers but thy use a current source in place of Q17 and Q46...

first the Normal buffer...

Please comment.
 

Attachments

  • amp_fab_normal_buffer_simple_rev_0.cir.jpg
    amp_fab_normal_buffer_simple_rev_0.cir.jpg
    77.1 KB · Views: 3,264

fab

Member
Joined 2004
Paid Member
darkfenriz said:
Maybe there is little difference, but you could adjust R130/R131 and R40/R42 in your second schematic assymetrically (smps-styled) to prevent any current spikes in output mosfets at fast slew or transient.

Can you explain your reasoning or what criteria to use to adjust assymetrically these resistors?
Are you talking about cross conduction issue?

If there is little difference why APEX selling its Amplifier IC PA52(A) at such a high price? or is it the rest of its circuit or the overall?
 

fab

Member
Joined 2004
Paid Member
darkfenriz said:
Maybe there is little difference, but you could adjust R130/R131 and R40/R42 in your second schematic assymetrically (smps-styled) to prevent any current spikes in output mosfets at fast slew or transient.


darkfenriz said:
Yes, I mean cross conduction at transients. Aside from better discharging of gates and probably lower cross-conduction and slightly higher slew rate I don't know why diamond output should be pricier.

Darkfenriz,

I am not following you here. First you say that the second schematics (diamond buffer) may have cross conduction but secondly you say that the diamond buffer is better for lower cross conduction...
Also, what do you suggest as resistor values for R130/R131 and R40/R42?
 
fab said:

Lateral mosfet w Normal or Diamond buffer?

This time I would like to get your opinion on the differences between the Normal emitter follower buffer and the Diamond buffer to drive a Lateral Mosfet power output stage.

I will attached 2 amp schematics:
- one with Normal buffer
- other with Diamond buffer (same one as APEX PA52 device)

---------------
Please comment.


Good question you put here, fab.
Same question can be asked for Bipolar Output stages.


I have began to use more and more Diamond!
I my simulations ... with great results, in compare to 'normal' Darlington outputs.

This is because of super results, when I did my clone of AD797 hifi Op-Amp:
See details in:
AD797 Clone people! Who can make a better one.
http://www.diyaudio.com/forums/showthread.php?threadid=92094
... and when I did my follow up, with 'an improved' Hifi Op-Amp
also using Diamond Output, like AD797:
New JFET Hifi Op-Amp release!
http://www.diyaudio.com/forums/showthread.php?threadid=94557

======================================


If this way of output stage, Diamond, works in preamps and Op-Amps
of course it will work great in some Power Amplifier applications, too.
And so it does.
There are people that have done excellent power amps this way before!

My thought:
- diamond output stage, with 2 drivers and 2 output power transistors
can not beat the gain of a triple darlington. Like in Leach Super Amp, for example.
But in some amplifiers we do not need as much gain.
And here I can not see any drawback using Diamond.
Not when compared to a usual 'Double Darlington' output. ( 2drivers + 2 output T )


Here are my current testings, fab
See Attachment Circuit.
I am designing a Class A amplifier with 100 Watt idle supply.
+/- 25 VDC and 2.0 Ampere current.

Results are very promising.
My first verson here, 1.0, is simulated with MJ15024/25.

I will try 'the better' MJL21193/94 and MJL3281A/1302A.
I have no doubt my figures will be improved. Those better devices usually lowers THD figures considerably!
I have no doubt!


Regards, lineup
Lineup Audio Lab
 

Attachments

  • lineup-diamond-classa-1.0.png
    lineup-diamond-classa-1.0.png
    17.7 KB · Views: 2,759

fab

Member
Joined 2004
Paid Member
lineup said:
...
My thought:
- diamond output stage, with 2 drivers and 2 output power transistors
can not beat the gain of a triple darlington. Like in Leach Super Amp, for example.
But in some amplifiers we do not need as much gain.
And here I can not see any drawback using Diamond.
Not when compared to a usual 'Double Darlington' output. ( 2drivers + 2 output T )
[/B]

What about diamond buffer at 20 ma and a pair or more of lateral mosfet power transistors?...

lineup said:
Here are my current testings, fab
See Attachment Circuit.
I am designing a Class A amplifier with 100 Watt idle supply.
+/- 25 VDC and 2.0 Ampere current.

Results are very promising.
My first verson here, 1.0, is simulated with MJ15024/25.

I will try 'the better' MJL21193/94 and MJL3281A/1302A.
I have no doubt my figures will be improved. Those better devices usually lowers THD figures considerably!
I have no doubt!
[/B]

Thanks lineup it is good to know that some peoples are interested in the diamond buffer...
You may have notice that the diamond buffer before the output transistors I am using is the same one as APEX (my revision 1A circuit) and not the Walt Jung one with current source(same as yours). It would be intersting to compare both diamond variations...

Your circuit looks quite interesting. I like jfet too for performance but they are always complicated to match or even get the proper IDSS current compared to bjt where only HFE parameter is used for optimization. I found that the "low memory" input stage with bjt I used is very linear too (simulation and real measurements too).

Please keep us informed of your simulation and prototype results.
:smash:
On my side I have a version (based on my revision 1A circuit) on breadboard powered by only 28 Vdc with limited current. THD is about 0.1 % in 16 ohms which is high for me but I think this is a breadboard/power supply/grounding scheme problem. I am working out a pcb layout in paralel. I plan to maybe connect a more higher current capability power supply at 40 Vddc to see if my current bias for each stage is sufficient....
:rolleyes::)
 
fab said:


What about diamond buffer at 20 ma and a pair or more of lateral mosfet power transistors?...

Thanks lineup it is good to know that some peoples are interested in the diamond buffer...
Please keep us informed of your simulation and prototype results.
---------------------------------
:smash:
On my side I have a version (based on my revision 1A circuit) on breadboard
powered by only 28 Vdc with limited current.
THD is about 0.1 % in 16 ohms which is high for me but I think this is a breadboard/power supply/grounding scheme problem.

I am working out a pcb layout in paralel.
I plan to maybe connect a more higher current capability power supply at 40 Vddc
to see if my current bias for each stage is sufficient....
:rolleyes::)


I don't know how you think 'diamond buff + lateral' would be setup ..
-----------------------------------------

Actually I was a bit wrong, that you can not use
Diamond at 3 stages, to match triple darlington gais.
You just make, driver a sziklai pair. Or make output use sziklai config.
But this is only if we need high gain and low load, burdon, onto VAS stage.
-----------------------------------------

Your 0.1% breadboard might be something! With some tweaking.

Here I quote Hugh AKSA Dean:
'An acceptable level of distortion for my power amplifiers is THD 0.05% or less.'

And what I know, there is no valid listening tests
proving that humans can hear lower distortions, than 0.1% THD.
So I say, AKSA is well in line with reality here :cool:

------------------------------------------

You may have a look at my latest :cool: Diamond Follower, 500mA Class A.
I designed it today and did virtual simulation tests to refine it.

In the Pass Labs GRollins topic:
MOSFET follower headphone amplifier
http://www.diyaudio.com/forums/showthread.php?s=&threadid=95841

In my own HUMBLE Opinion:

;) Post #44 is the most interesting and innovative in that Topic, so far ;)
... MOSFET single device followers, there are 20 by the dozen around :D already


Regards
lineup
 

fab

Member
Joined 2004
Paid Member
lineup said:


I don't know how you think 'diamond buff + lateral' would be setup ..


simply like this config attached. The VGS of the lateral must be >= 0.7V for about 100 ma bias current (where the temp curve change slope).
This is tha case for the SK134/J49 pairs I have.
 

Attachments

  • lateraloutput_diamond_buffer_comp_eftype.jpg
    lateraloutput_diamond_buffer_comp_eftype.jpg
    43.8 KB · Views: 1,513

fab

Member
Joined 2004
Paid Member
Vad circuit

Hi vad,

I am not trying to steal anything from anybody.... The APEX is disclosing his schematics for diamond buffer and Walt Jung too. This is not new stuff...since other diy forum members have disclosed their stuff too...

I can barely see your circuit attached image. Is it only a buffer or a voltage amplifier? Do you have a better resolution so I can fully appreciate it?
Have you built it?
What is your appreciation?

Thanks:)
 
vAD said:
..... it is not necessary to steal the Russian works .....

.

:D :D :D

vAD
I have seen 1 or 2 Russian Copy
of good amplifiers schematics.
At some Russian web sites.
And in Chinese website, they sell copy Nelson Pass amps!!!!.
So I am not sure who steal ... most ;)
East or West?

Thanks anyway, vAD :)
for your schematic ( can you make better quality = bigger Image, please )
Interesting!
I will Save & Have a look.

Regards, lineup
 
fab said:

simply like this config attached. The VGS of the lateral must be >= 0.7V for about 100 ma bias current (where the temp curve change slope).
This is the case for the SK134/J49 pairs I have.

.


you have never thought about USING SCHOTTKY DIODES
instead of resistors in Diamond

Like I do.

1. You try to find a schottky with the VOLTAGE DROP you need.
At approximate current you want to use in DRIVERS.
Schottky Diodes can have VF 0.200 - 0.650 Volt

2. You make Current sources ADJUSTABLE to fine Tune the Idle Current in output stage.


:cool:
Again:
Study my circuit!

Lineup Audio LAB - CLASS A Diamond 20Watt RMS

What has lower AC impedance?

Resistor or Schottky diodes.
Say at +/-2 mA Volt AC across 10 Ohm resistor. DC Current level = 20 mA.
You will have a Voltage drop variation of 10% !!!! for your Diamond Buffer.
From a follower we want to it to follow, as good as possible.


It is not often something revolutionary new in a Lineup Audio amp circuit.

But it is Those Smart Combinations, of known ways
.. and that Little Detail changed or added :cool:

This is the true Benefit and Greatness of a Typical Lineup amplifier.
They are UNIQUE ... ALWAYS ... in some details!



Regards, lineup
 
vAD said:
Hi, Fab, Lineup!
Lineup. Badly looked!!!?Look here http://www.vegalab.ru/forum/forumdisplay.php?f=42
the Scheme of management Mosfets has come from power electronics, and driver Mosfet is now discussed. Small temperature losses, high speed ïåðåçàðÿäà capacitance Gate-Source !
ï.ñ. fab, did not collect, but potential at an output, excellent.


Very good, vAD

Where Schematic in russia forum?
Link Topic Post Image schematic image?
Thanks!
 

fab

Member
Joined 2004
Paid Member
New rev 4A

lineup said:



you have never thought about USING SCHOTTKY DIODES
instead of resistors in Diamond

Like I do.

1. You try to find a schottky with the VOLTAGE DROP you need.
At approximate current you want to use in DRIVERS.
Schottky Diodes can have VF 0.200 - 0.650 Volt

2. You make Current sources ADJUSTABLE to fine Tune the Idle Current in output stage.


:cool:
Again:
Study my circuit!

Lineup Audio LAB - CLASS A Diamond 20Watt RMS

What has lower AC impedance?

Resistor or Schottky diodes.
Say at +/-2 mA Volt AC across 10 Ohm resistor. DC Current level = 20 mA.
You will have a Voltage drop variation of 10% !!!! for your Diamond Buffer.
From a follower we want to it to follow, as good as possible.

Your last circuit did not show Schottky diode but regular diode (based on used symbol). :devilr:
Anyway, 0.2 V (thus 0.2 + 0.6) is still too much for my lateral power mosfets in class AB. So I got back to diamond buffer with current sources and removed emitter resistors but used a pot across the bases of the diamond input to adjust mosfet bias current...

see attached...for comments... :confused:

thanks
 

Attachments

  • amp_fab_diamond_buffer_simple_rev_4a.cir.jpg
    amp_fab_diamond_buffer_simple_rev_4a.cir.jpg
    70.9 KB · Views: 1,181
Re: New rev 4A

fab said:


1. Your last circuit did not show Schottky diode but regular diode (based on used symbol). :devilr:

2. Anyway, 0.2 V (thus 0.2 + 0.6) is still too much for my lateral power mosfets in class AB.
.

1. The library I used (MultiSim) had placed these Schottky I used
in ordinary Diodes category. And so the symbol is like that.

2. The voltage you need is dependent on what idle bias you use in output stage Laterals.
Of course there are Schottky Diodes with also lower Vf than 0.200 V.
At rather low current for a higher current rated Schottky.

I have no idea what Voltage you need, to remove resistors.
Removing resistors will reduce output impedance, of course.
And this is my point.

There are quite many nice New Schottky Diodes.
Most of them are new package = SMT.
Even the old library I use, have several Schottky Diodes that are suitable for my purposes.


If you look at my schematic:
http://www.diyaudio.com/forums/attachment.php?s=&postid=1130867&stamp=1171019146

2mA is very low level to drive a Class A working at 2 Ampere.
( +/- 3600 mA output )
But anyway I tried this circuit.

At first input 2SK170GR were biased only 1 mA per JFET.
It worked! But not very best, when max output into 4 Ohm.

As the Diamond at full Ouput ( +/- 3.6A into 4 Ohm )
needs a current drive at Diamond input, MJE15030/31 bases,
like +/- 0.2 mA peak
it is good to raise to 2mA in each 2SK170GR.

I am really surprised that only 2+2 mA can drive MJE15030/31 input so good here.
But this only shows how great drivers MJE15030 and MJE15031 are!



2SK170GR, in my model, has got max current (at VGS=0) like 4.80mA,
so is not good to increase current to more than 2 mA.

My own rule of the thumb (based on looking at many Curves of JFETs) is:
:cool: Use maximal like 40% of the Top Current (VGS=0 value) of a JFET :cool:
This is where we can still get a good linearity in transfer
and also have good drive capacity = output current level.

2.0mA/4.8mA = 41.7%


Regards, lineup
 
Re: New rev 4A

fab said:

Anyway, 0.2 V (thus 0.2 + 0.6) is still too much for my lateral power mosfets in class AB.
So I got back to diamond buffer with current sources and removed emitter resistors
but used a pot across the bases of the diamond input to adjust mosfet bias current...

see attached...for comments... :confused:

thanks


I am sure it will be good your way, fab
:)
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.