2stageEF high performance class AB power amp / 200W8R / 400W4R

It does seem that regulations and guidance in this arena are very opaque.
This is typical of all regulations. One needs to be trained in the discipline to then make use of the regulations.
Much of my early training was simply to understand what the regulations required of me as a Designer. Then they changed the regulations. About 50% of the department went back to college to be retained for the new regs.
I can recall that happening twice in my 28year career as a Civil Engineer.

Going on a bit further.
The reason we can't manufacture equipment to meet the ClassII standard for double insulated mains equipment is because no one here has attempted to train us to use, or comply with, the regulations.
WE can't design and build and test and guarantee, if we don't know what ALL the various clauses require us to do.
 
Last edited:
Member
Joined 2011
Paid Member
...
The reason we can't manufacture equipment to meet the ClassII standard for double insulated mains equipment is because no one here has attempted to train us to use, or comply with, the regulations.
WE can't design and build and test and guarantee, if we don't know what ALL the various clauses require us to do.

That's as true as true can...

The documents are very expensive and without expensive (in the sense of money and time) training it would be impossible to fulfill all requirements... :(

BR, Toni
 
Member
Joined 2011
Paid Member
SA2014/SA2015/SA2016 power supply ...

0.45mm clearance/creepage minimum ...
 

Attachments

  • sa2014_power_supply_2pin_diodes_big_pcb_top.jpg
    sa2014_power_supply_2pin_diodes_big_pcb_top.jpg
    892.5 KB · Views: 932
  • sa2014_power_supply_3pin_diodes_small_pcb_bottom.png
    sa2014_power_supply_3pin_diodes_small_pcb_bottom.png
    613.9 KB · Views: 170
  • sa2014_power_supply_2pin_diodes_small_pcb_bottom.png
    sa2014_power_supply_2pin_diodes_small_pcb_bottom.png
    613.2 KB · Views: 175
  • sa2014_power_supply_3pin_diodes_big_pcb_bottom.jpg
    sa2014_power_supply_3pin_diodes_big_pcb_bottom.jpg
    465.8 KB · Views: 173
  • sa2014_power_supply_2pin_diodes_big_pcb_bottom.jpg
    sa2014_power_supply_2pin_diodes_big_pcb_bottom.jpg
    462.4 KB · Views: 668
  • sa2014_power_supply_3pin_diodes_small_pcb_top.png
    sa2014_power_supply_3pin_diodes_small_pcb_top.png
    683.8 KB · Views: 703
  • sa2014_power_supply_2pin_diodes_small_pcb_top.png
    sa2014_power_supply_2pin_diodes_small_pcb_top.png
    682.4 KB · Views: 735
  • sa2014_power_supply_3pin_diodes_big_pcb_top.jpg
    sa2014_power_supply_3pin_diodes_big_pcb_top.jpg
    894.2 KB · Views: 863
Last edited:
Member
Joined 2011
Paid Member
SA2014/SA2015/SA2016 power supply ...

Note: Gerbers are free to use only for non commercial DIY projects.
 

Attachments

  • sa2014_power_supply_3pin_diodes_big_pcb.pdf
    875 KB · Views: 139
  • sa2014_power_supply_3pin_diodes_small_pcb.pdf
    683 KB · Views: 138
  • sa2014_power_supply_2pin_diodes_big_pcb.pdf
    906.4 KB · Views: 135
  • sa2014_power_supply_2pin_diodes_small_pcb.pdf
    710.5 KB · Views: 146
  • sa2014_power_supply_3pin_diodes_big_pcb.zip
    349.5 KB · Views: 111
  • sa2014_power_supply_3pin_diodes_small_pcb.zip
    262.2 KB · Views: 130
  • sa2014_power_supply_2pin_diodes_big_pcb.zip
    371.9 KB · Views: 113
  • sa2014_power_supply_2pin_diodes_small_pcb.zip
    280.6 KB · Views: 127
Member
Joined 2011
Paid Member
SA2014 IPS reloaded ...

A small mod of the input stage helps to stabilize the CCS currents for a wide temperature range:
Replaced the 2 x 1N4148 for the CCS sources by one green LED (D1=LED; D2=wire).
R7=180R and R27=68R emitter resistors at CCS sources instead of
R7=82R and R27=27R

Note: part numbers refer to schematic post 1145.

Attached a stability simulation result.
Since some months my 4 Channel SA2014 runs without any problems.

BR, Toni
 

Attachments

  • sa2014_ips_mod.png
    sa2014_ips_mod.png
    118.7 KB · Views: 632
Hi Toni. I was curious as to how you selected the size of the base stopper resistors. From SA2014 to SA2015 some went from 47R to 22R even with the same transistors. I thought base stoppers resistor sizing essentially transistor dependent.

Also, have you ever looked at how dependent output quality is on the 12V bias supply for the differential pair? (I should have a fiddle with your .asc myself...)
 
Last edited:
The "Tian probe" is one method of plotting loop gain for phase and gain margin assessment. Bob Cordell provides another 'method' of breaking the loop. The method of choice for the LTspice forum gurus (and Linear Technologies in their tutorials) is even simpler to implement. I once asked Helmut Sennewald (the keeper of the Yahoo LTspice forum) why "their" method was different from Cordell's.** He aligned both versions of the analysis for the regulator circuit I provided. Annoyingly I lost the analysis on my local computer and I suspect the circuits are long-gone from the LTspice forum's temp files section. I've not seen the Tian probe method discussed outside of these forums - not that this means anything, it merely prompts the question I posed above.

EDIT: ** if I recall correctly the differences in the plots between the two methods were well below the areas of interest and it was not necessary to break the loop with a 1GN inductor and inject via a 1GF cap.
 
Last edited:
Member
Joined 2011
Paid Member
A sophisticated .asc
How do we use the .txt to get the various simulations to run?

... and you need this functions in your "plot.defs" file too:

Code:
.func TianA1() {-1/(1-1/(2*(I(Vi)@1  *V(x)@2  -V(x)@1  *I(Vi)@2  )+V(x)@1  +I(Vi)@2  ))}
.func TianB1() {-1/(1-1/(2*(I(Vj)@3  *V(y)@4  -V(y)@3  *I(Vj)@4  )+V(y)@3  +I(Vj)@4  ))}
.func TianA2() {-1/(1-1/(2*(I(Vi)@5  *V(x)@6  -V(x)@5  *I(Vi)@6  )+V(x)@5  +I(Vi)@6  ))}
.func TianB2() {-1/(1-1/(2*(I(Vj)@7  *V(y)@8  -V(y)@7  *I(Vj)@8  )+V(y)@7  +I(Vj)@8  ))}
.func TianA3() {-1/(1-1/(2*(I(Vi)@9  *V(x)@10 -V(x)@9  *I(Vi)@10 )+V(x)@9  +I(Vi)@10 ))}
.func TianB3() {-1/(1-1/(2*(I(Vj)@11 *V(y)@12 -V(y)@11 *I(Vj)@12 )+V(y)@11 +I(Vj)@12 ))}
.func TianA4() {-1/(1-1/(2*(I(Vi)@13 *V(x)@14 -V(x)@13 *I(Vi)@14 )+V(x)@13 +I(Vi)@14 ))}
.func TianB4() {-1/(1-1/(2*(I(Vj)@15 *V(y)@16 -V(y)@15 *I(Vj)@16 )+V(y)@15 +I(Vj)@16 ))}
With this functions you can do a parameter stepping for 4 values for one parmeter.
TianA is the overall feedback loop
TianB is to check the IPS local loop

Vsimclip: is to set the DC positition to the extremes to simulate behaviour near clipping. Especially have a look at gain/phase margins also in the local loop.

The TianProbe() is from the LTSpice examples directory LoopGain2.asc:

Code:
~/.wine/drive_c/Programme/LTC/LTspiceIV/examples/Educational> cat LoopGain2.asc|grep -i -e tian -e '(x'
TEXT -8 -344 Left 2 ;Here the open loop gain is determined from the closed loop system[1].\nThe open loop gain can be plotted by plotting the quantity:\n \n-1/(1-1/(2*(I(Vi)@1*V(x)@2-V(x)@1*I(Vi)@2)+V(x)@1+I(Vi)@2))\n \nAlternatively, you add the following line to your plot.defs file:\n.func T.et.al() -1/(1-1/(2*(I(Vi)@1*V(x)@2-V(x)@1*I(Vi)@2)+V(x)@1+I(Vi)@2))\nAnd then plot simply T.et.al()\n \nThis is an improvement over the technique shown in LoopGain.asc \nbecause it (i) accounts for reverse feedback(it doesn't even\nmatter if you reverse the direction of the probe -- you still compute\nthe same open loop response) and (ii) the inserted probe elements\nresult in a smaller, sparser circuit matrix.
TEXT -8 456 Left 2 ;[1] Michael Tian, V. Visvanathan, Jeffrey Hantgan, and Kenneth Kundert,\n    "Striving for Small-Signal Stability", IEEE Circuits and Devices Magazine,\n     vol. 17, no. 1, pp. 31-41, January 2001.
BR, Toni
 
Member
Joined 2011
Paid Member
Hi Toni. I was curious as to how you selected the size of the base stopper resistors. From SA2014 to SA2015 some went from 47R to 22R even with the same transistors. I thought base stoppers resistor sizing essentially transistor dependent.

Also, have you ever looked at how dependent output quality is on the 12V bias supply for the differential pair? (I should have a fiddle with your .asc myself...)

I have choosen the values in dependence of base current, simulation results (square wave), examples and gut feeling

12V bias: AFAIR yes. The results are shown in my schematics. The 12V Bias is fixed in SA2014 and floats with input signal in SA2015/SA2016. The difference is minimal and maybe not worth the extra transistor.

BR, Toni
 
Thanks Toni.

For those who work on a Mac for which LTspice doesn't seem to install the example files, the educational files can be found here also:

Index of /files/LTspiceIV/examples/LTspiceWorldTour2011

Compare the files Loopgain and Loopgain2. The method principally 'taught/suggested' by those at the Yahoo LTspice forum is in the former, i.e. Middlebrook's older method. Namely, plot loop gain (open loop gain less closed loop gain) directly by plotting -V(x)/V(y) and then measure or observe phase and gain margin where the loop gain is 1 (0dB) and phase is 180 degrees, respectively. (Compare with the discussion in Cordell's book where the loop is broken, closed with an inductor and the stimulus injected via a capacitor.)

Here is the personal website for the guy who created the examples. It has more background/discussion and also has links to the examples.

https://sites.google.com/site/frankwiedmann/loopgain

It is interesting to note:

"Middlebrook developed this [older] method using a simplified model which did not take into account backward transmission through the loop. This limitation was later removed by the other two methods. Nevertheless, the method usually provides pretty accurate results and, like the other two methods, will always tell you correctly whether a circuit is stable for small pertubations or not"
 
Last edited:
I have choosen the values in dependence of base current, simulation results (square wave), examples and gut feeling

Thanks. Interesting as I understood the base stopper was used to combat parasitics not often modelled.

12V bias: AFAIR yes. The results are shown in my schematics. The 12V Bias is fixed in SA2014 and floats with input signal in SA2015/SA2016. The difference is minimal and maybe not worth the extra transistor.

Ah, I was wondering also why Q1 was added in SA2015. I hadn't appreciated that it might be better to have the bias float with the signal and was more focused on whether a better regulator than a Zener would be beneficial. I'll have to chew on why...
 
Member
Joined 2011
Paid Member

Thanks for the (here well known) links.
I can't exactly remember what the decision was to use the Tian probe. Maybe because the Middlebrook method wasn't direction independent and the GFT probe needed 3 simulation runs instead of 2 simulation runs compared to Tian's method. As you know I found a method for a simple parameter stepping during using of the tian which makes it more versatile. ;)

BR, Toni