pa300 elektor amp

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
PA300/600

Hello, Sandra.
Thank you for the circuit. I will look closely.
I know this pattern, I already saw it somewhere?
Just a question. AOP is really my generation.
For me, a LM318. Offset on the negative. See example
Other image: LM318 on amplifier 150 watts / 8 or 300/4, typical of the 1980s.
Thank you very much, if I have any questions, I'll come back to you.
Jacky
 

Attachments

  • Cfg_Aop.jpg
    Cfg_Aop.jpg
    136.4 KB · Views: 493
  • LM318_600w.jpg
    LM318_600w.jpg
    230.8 KB · Views: 533
583/5000
Dear Kay.
You are talking about the Pa.400 / 600 project. I have so much project going on that I am a little lost. laughs!
You have to know that it's tests. We are not super specialists. So we try all kinds of solutions.
There are no limitations, just recalculate the Zener resistance couple to power the Aop. We saw the Pa 600 Olivier, then starting from Pa.300 we made the changes. Then we will test and see what it gives.
For now: Pa.300 +/- 60 volts
Pa.400 +/- 75 volts
And according to Olivier Pa.600 +/- 90 volts
Good evening, Kay.
Jacky
 
Hi Jacky,
no, now I was talking about your 2nd pic in #41. Both outpul halves consist of a Sziklai each with quadruple power devices. As a Sziklai behaves similar to a Darlington with the polarity of the 1st transistor, we have a NPN emitter follower in the positive rail and a PNP in the negative one. Hence, output swing is limited to the opamp's supply voltages of about +/- 17.4 V, minus two diode drops, minus the 1st transistor's Vbe, minus the power devices' saturation voltage. So this amplifier is in about the 20 watts range, independingly of the outer rail voltages.
The Elektor PA 300 is quite another thing, as there's a level translator between the opamp and the power stage, which also is the reason why the GNB loop refers to the opamp's non inverting input and the input to the inverting one, without changing the signal's polarity.
Similar designs can be found in RCA'S Power Transistor Manuals from the early 1970ies.
Best regards!
 
Btw, I'd really be very curious in how this old PA 300 design would behave with more recent power devices, such as MJL1302A/3281A's or even MJL4281A/4302A's, wich are beefier, faster, more linear and having a larger hFE than the old MJ15003/15004's - and are much cheaper on top.
Best regards!
 
Your 2nd design's rail voltages are limited to the opamp's ratings. So, why that many output devices?
Best regards!

Any transistor having a load in the collector connection is a common emitter amplifier having voltage gain = the voltage drop across the load.

The output of such a stage is inverted with respect to the input.

Two such stages are necessary for the output to be in the same phase as the + input allowing the nfb divider network connection to be the - input.

As I see it the op.amp has to increase the base emitter voltage of the BD devices sufficiently to draw current through the power device emitter to base connections to increase conduction - a case of causing two diode junctions to be made more forward biased.

This is a bit like the local water supply in a tank on top of a tower or on a hill so when one turns on the tap to fill the washing machine or make a coffee there will always be enough pressure in the pipes - so it depends on how far the tap is turned on.


.
 

Attachments

  • Power Amp.PNG
    Power Amp.PNG
    304.8 KB · Views: 354

PRR

Member
Joined 2003
Paid Member
Your 2nd design's rail voltages are limited to the opamp's ratings. So, why that many output devices?...!

Fooled me too for a while.

Note the 820r and 33r. The output compound works at nominal gain of 25! If we believe this, 10V swing from the chip allows 250V rails! Actually I bet loaded gain is nearer "10", but that's still a lot of leverage from chip-Volts.
 

Attachments

  • why_that_many.gif
    why_that_many.gif
    15.1 KB · Views: 315
Fooled me too for a while.

Note the 820r and 33r. The output compound works at nominal gain of 25! If we believe this, 10V swing from the chip allows 250V rails! Actually I bet loaded gain is nearer "10", but that's still a lot of leverage from chip-Volts.

It would be interesting to simulate this circuit just for the precise degree - there are some other aspects that could be of interest too.

The op.amp and the BD transistors remind me of an op.amp based headphone amplifier designed by John Linsley-Hood that operated largely in Class A push-pull.

In the present instance the combination drives the 33R and 820R divider set up.

The 33R can source or sink current - a ploy sometimes used to speed up three terminal regulator responses - the 820R is in series with the output providing some small feed forward correction signal voltage at the point output transistors cross-over.

Looking at the CFP elements the predominant collector load on the output transistors is the speaker.

That on the collectors of the BD's is the input resistance of the output transistors in combination in parallel with 1k.

Low collector load values suggest a lowish value of open loop gain around the discrete component CFP stage - putting a limit on the amount of closed loop gain. That would be above unity - but it would not need to be greater than low single numbers.
 

PRR

Member
Joined 2003
Paid Member
> the predominant collector load on the output transistors is the speaker.

We would hope so. (the speaker is the "paying work" of the amplifier.)

> ...the collectors of the BD's is the input resistance of the output transistors in combination in parallel with 1k.

All those bases make a real low impedance, such that the 1k is negligible and the BD collectors more negligibler. The BDs work at less than unity voltage gain.

The direct solution is to see the BD emitter sees about 50 Ohms. The power transistors work as current mirrors with gain like hFE. While hFE will vary a LOT over an class AB swing, let's try "55", or even "50". So the 50r at BD emitter mirrors-around as a 1 Ohm transconductance. With 8r load the voltage gain is 8.

If output stage current falls as low as 10mA, while device hFE stays high, we "could" have voltage gain near 80, in which case the 820r would prevent a gain-peak through low current transition. Such a peak could exasperate stability trouble, and you can't hide instability at idle.
 

Attachments

  • why_that_many_2.gif
    why_that_many_2.gif
    6.2 KB · Views: 276
Re stability question

On the last point firstly there is a dc feedback path from the op.amp + BD buffer combination via the 820R resistor leading to the 330k feedback resistor, and, secondly an effectively parallel a.c. one from the output of the op.amp to the inverting input of this via a 15 pF capacitor - ignoring the 1Meg resistor in parallel to keep things simple - the frequency point where the impedance of the capacitor is equal in impedance to the 330k resistor is about 32 kHz. At that point the voltage gain will be -3 dB down ( to 70.7%).

That does not tell you if the amplifier will be stable or not but indicates a point in the circuit where changes to the compensation might be applied.

If I was building this I would make a SPICE file and run a Tian plot simulation to look at the gain bandwidth product graph to see where -3 dB point actually lies, to check for any peaking, and to see what the gain and phase margins are. It would be wise to do this if substitute transistors are used.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.