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

Member
Joined 2011
Paid Member
Output protection / SOA

Attached a SOA plot of 8, 4, 2 R load. Short circuit starts too late at about 1.7R load.
Sorry - my fault :(:
I changed the emitter resistor values the last days from 0.22R to 0.47R and adapted the short circuit values not correctly.
Thanks michaelkiwanuka for reporting this to me.
Give me some time to correct these values - or feel free to help here.

BR,

Toni
 

Attachments

  • 2ef_soa_simulation.png
    2ef_soa_simulation.png
    136.5 KB · Views: 2,489
Member
Joined 2011
Paid Member
where is R79?
The output stage and the feeds to the protection are hidden.

Ratio of R22 to R56 is liable to entering triggering while still handling a valid audio signal into a valid audio load. R56 > R22*2 and preferably 3times to 4times. 27>12*2
If you split R22 into R22a+r22b=R22 and tap of the junction, you can experiment with different R22a:R56 ratios.
R22a < 27/3 or < 27/4 will delay the onset of triggering and still protect the VAS from overheating.
 
Last edited:
Member
Joined 2011
Paid Member
where is R79?
The output stage and the feeds to the protection are hidden.

Ratio of R22 to R56 is liable to entering triggering while still handling a valid audio signal into a valid audio load. R56 > R22*2 and preferably 3times to 4times. 27>12*2
If you split R22 into R22a+r22b=R22 and tap of the junction, you can experiment with different R22a:R56 ratios.
R22a < 27/3 or < 27/4 will delay the onset of triggering and still protect the VAS from overheating.

If you can't open the asc file: R79 in ltspice simulation file is R53 in the schematic picture in post #1.

Thx for checking VAS load.

Will check your findings at weekend.

BR, Toni
 
check operation of Q37.
By the time you apply 400mVbe there will be some triggering effect building up.
That why I said 3 to 4 times for that resistor ratio.

It might be something else.
Check the output protection. Again 400mVbe could be approached at that particular point in the waveform.
 
Member
Joined 2011
Paid Member
check operation of Q37.
By the time you apply 400mVbe there will be some triggering effect building up.
That why I said 3 to 4 times for that resistor ratio.

It might be something else.
Check the output protection. Again 400mVbe could be approached at that particular point in the waveform.

You are correct - the output protection is the source of this trigger spike. Sim load was 4R.

We have to do some rework to meet the SOA of the 2 EF stages to drive 4R loads. A lower supply voltage is required for safe 4R usage.

BR, Toni
 
Member
Joined 2011
Paid Member
Is this true?

Or do you mean that the transient pass capability of the protection threshold needs to be increased to closer match the output device capability for short term transients.
The DC SOAR is very different from the transient SOAR.
The protection locus should take account of this.

I mean power supply Voltage reduction due to SOA problem mainly at the pre driver stage 2SC5171 and 2SA1930. Maybe I have forgotten to recheck SOA after reducing R17 from 150R to 33R (to reduce distortion at high output levels).

So finetuning is needed:

  • pre driver stage: using other driver bjt's with better SOA or 4 instead of 2 bjt's
  • VAS: split R22 to avoid too early trigger of VAS over current protection
  • recalculate protection threshold and also recalculate transient pass capability of output short circuit protection
With current predrivers SOA the title of the project should be better corrected to "2stageEF high performance class AB power amp / 200W8R" as it is currently not save to drive 4R loads at full power. :headbash:

BR, Toni
 
The simulated THD performance is about 0.007% at 20kHz@200W@8R.
The real life THD performance is about 0.004% at 20kHz@200W@8R (80kHz bandwidth).
Hi Toni,

really impressive, obviously you have managed to get every smallest thing right.

Did you evaluate THD20k at low output power, e.g. at the point where the OPS leaves class A operation? My impression from listening experience is that this figure is quite important, if it is within the region of actual listening level.

Best regards,
Matthias
 
Member
Joined 2011
Paid Member
Hi Toni,

really impressive, obviously you have managed to get every smallest thing right.

Did you evaluate THD20k at low output power, e.g. at the point where the OPS leaves class A operation? My impression from listening experience is that this figure is quite important, if it is within the region of actual listening level.

Best regards,
Matthias

Do you mean something like the attached plot? Leaves class A between 10 and 20W output power but no extremely rising of THD viewable ...
This plot made with homebrew GPIB software controlling Levear VP7723D THD Analyzer. Amplifier was running only at +/- 30V supply.
Note: need to measure this again at full supply voltage.

BR, Toni
 

Attachments

  • SA2013_thd_20_to_20khz_60V_supply.png
    SA2013_thd_20_to_20khz_60V_supply.png
    65.8 KB · Views: 1,840
Dear keantoken, manso, AndrewT

interesting discussion about current mirrors. During learning LTSPice I have simulated "simple current mirror", "EFA current mirror", "wilson current mirror" and "4 transistor improved current mirror". Douglas Self has made a comparison between these 4 CM types (APAD Handbook, Fifth Edition, page 85) .
One of the conclusion I made, that if using bjt's with beta > 200 the error of the simple current mirror between current in/out is very low in comparison to the "better CM" variants. So I decided to use a simple current mirror as the simulation showed only a few ppm THD difference.
As there are many empty input stage pcb's left (got it cheap from overproduction) I will try to test different bjt's and degeneration values during the next days and will give a feedback

  • if 2N5401 is really a mistake in this case or
  • more or less degeneration will improve or worsen distortion measurements
  • produces more or less harmonics. (currently only H2 and H3 are visible above 100dB / measurement noise floor of HP3585A)
Keep on going to discuss the 2stageEF amplifier so one day an improved schematic/pcb revision will see the real world.
Btw.: the same input/vas stage exists assembled with matched 2SK170 and about 150R degeneration. Will add ltspice asc file the next days.

Note: one can never have enough good amplifiers!

BR,

Toni

Hi Toni

Keantoken is correct to suggest the use of low Vce Sat transistors especially since only the simple mirror is used. Youre right in assuming higher beta gives more accuracy although ideally beta should be even higher than 220, more around 500 mark. Youll find that given LTP transistor matching is done lower offset can be achieved.
Parts like the 2N cannot be said to be a mistake, as they will work but they are not optimal. Higher current mirror degeneration will lower the loop gain slightly so a very small difference should be seen regarding THD but you will get as a big bonus on noise reduction.
 
Do you mean something like the attached plot? Leaves class A between 10 and 20W output power but no extremely rising of THD viewable ...
This plot made with homebrew GPIB software controlling Levear VP7723D THD Analyzer. Amplifier was running only at +/- 30V supply.
Note: need to measure this again at full supply voltage.

BR, Toni
Thanks for the fast reply.

I thought of pure THD obtained from spectral analysis (simulated or even measured). The problem with THD+N is that the 'real' noise is uncorrelated to the signal whereas the distortion products are correlated. Our hearing system seems to do a brilliant job in detecting correlated signals, even if they are well below the 'real' noise floor.

Best regards,
Matthias
 
8pr of 150W devices is a total of 2400W of output device.
Divide by 5 or 6 for BJTs gives a predicted maximum output power of around 400W to 500W.
The 5200/1943 are not good at high supply rail use, so I would not try to get 400W into 8r0, but the supply rails for 400W into ohms are just about low enough to be a sensible target.

I don't see why you have a problem when checking your temperature de-rated SOAR for 400W into 4ohms.
You should also find that the cold SOAR can easily drive 2r0 and possibly drive 1r3.

How are you modeling the temp de-rated SOAR?
 
Member
Joined 2011
Paid Member
Output protection / SOA

Attached a SOA simulation for Rload 8, 4, 3, 2.5 and 2 Ohm stepped.
Assumed that cooling is good TTC5200 and TTA1943 are in the safe SOA region down to Rload of 3 Ohm.

Reworked short circuit protection. Removed C46 and C47.
NOTE: It is not possible to keep the output bjt's in safe SOA part during a short circuit situation. If power supply fuses are fast enough the outputstage may survive a short circuit!
Next amp version will get a multislope VI limiter...

Using the SPICE models of keantoken from post #22

  • removed the mysterious simulation spikes seen/reported in post #27. The output protection wasn't the source of spike so there is currently no need to split R22. Q37/R22 limits VAS current to about 56mA. There is enough distance to normal VAS operation (~ 20mA).
  • synchronized simulated and real life THD values.
Thanks goes to keantoken providing these bjt models.


Updates to pre-drivers are in work due to very bad SOA of 2SC5171 and 2SA1930.



BR, Toni
 

Attachments

  • soa_5200_1943_rl_8_4_3_2.5_2.png
    soa_5200_1943_rl_8_4_3_2.5_2.png
    129.4 KB · Views: 1,784
............output protection wasn't the source of spike so there is currently no need to split R22.
The VAS protection is there to protect the VAS. There is no correlation between an output spike and the R22a value.
Q37/R22 limits VAS current to about 56mA. There is enough distance to normal VAS operation (~ 20mA)...........
are you sure?
what IC current is passing the protection transistor when 56mA of VAS current is flowing.
How low does the VAS current need to go to turn off the protection transistor.
I think very many builders get this wrong and then blame the protection for sounding wrong, when It is possibly early triggering that is affecting the sound.