Jer's PLWO to be... (Jer vs the 700, Round 2... it's "Personal" this time)

[Gepetto]

Catching up after a couple of days in Woodstock. Jer do you mean a printed circuit board?

 

[jbeckva]

Catching up after a couple of days in Woodstock. Jer do you mean a printed circuit board?

Yes sir. Basically thinking I'll do a light board for the series I since I don't see anyone already offering one. I read up a little, and I THINK I know where I'm going wrong...

I've been doing the "immersion" method where you set the PCB face down in the etching solution. I've read tho, that the "sponge" method works better (put a good portion of the solution on a sponge and wipe the PCB for a minute or two until the traces are left). I'll have to try that. I wish I had a laser printer tho - heard that is really the "way" to transfer the design from CAD to PCB.

 

[jbeckva]

Ahh.. now THAT is indeed BETTER (at least to me earz...)

I like the MJ15024's for drivers. Just got done swappin' out the RCA's...

Listening to some "Gold Dust Woman"... vocals are right there, in your face. More defined highs..hmm..

Disclaimer - these were the original RCA's that came with the 700, not the ones Lee sent. So they coulda just been old, but in any case..

ME = STOKED!!

 

[jbeckva]

Well don't bring the bubbly out just yet. Hey Joe - I put a scope on the output today. With 15Khz, definitely seeing a crossover notch "knee" between 180 and 270 degrees - both channels too??. I can get rid of it, but only by biasing very high i.e 550mV. Have any ideas on where to start looking? Should I (for example) take those 4uF orange caps out that are on the +/- 100 rails?

 

[Gepetto]

With what load Jer?

 

[jbeckva]

8 ohms.. nothing special?

 

[Gepetto]

speaker or resistor Jer? on both rising and falling edge or just one edge? Often it is the zoebel network that causes this. Is the zoebel right on the output terminals?

 

[jbeckva]

Just one edge. I removed the original Zoebel because with Don's DCP it is included (I think??)

 

[jbeckva]

With speaker too. Retested same setup with different amp, with no notch.

 

[laatsch55]

It should be on the trailing edge of the positive half. I have had 3 amps that exhibited this behavior. It's not audible and believe it's an artifact of the zoebel network.

 

[laatsch55]

Yes, the Zoebel network is included on Don's board./ For shit's and giggles, bypass Don's board and see if it's still there .

 

[jbeckva]

Yes, the Zoebel network is included on Don's board./ For shit's and giggles, bypass Don's board and see if it's still there .

How about I just unsolder the caps for the Z net on Don's board? That should take 'em out of the circuit eh?

 

[laatsch55]

That'd do it, but when it looks the same, would you just jumper around it?? Please?

 

[jbeckva]

I put pure resistive loads on it with the Z disconnected by pulling the caps. Same basic thing going on... the only way I get rid of that notch is if the bias is turned up to something ridiculous.

Now hmmm... one thing I did note - the +20 volt's reading a good .7 volts lower than the -20 volt. Would that make enough of a difference? We want the bias to be right around cutoff but not cutoff... correct?

 

[jbeckva]

FWIW - The notch appears when going from the 90 positive peak to the 270 negative peak. Going positive from 270 up to 90 looks fine. So when I'm going "negative", that is where the notches appear - and they do appear on both positive and negative sides of zero.

Do I need to take a pic??

 

[jbeckva]

Hmm... another finding.. those 150 ohm resistors across the MJ15024 drivers? 3 out of the 4 are reading a good 10 ohms high. Sooo.. if I wanted to keep "away" from cutoff .. ie. MORE away.. would reducing the resistance B-E across the drivers get me closer to that goal or less so?

Should I also try looking at the drive signal coming out from Q7/Q10, to see if it has the notch or if it's the drivers on the chassis?

 

[laatsch55]

Yeah, you could start scope chasing it from the driver in. But something is naggin at the back of my head. we are either missaing something so simple we're not considering it or we're chasing bad factory components???

 

[Gepetto]

If I get some time this week I will try to model what you are seeing a bit further. For reference Jer, what output amplitude into 8 ohms were you experimenting with so that I can drive the model similarly? I would describe what you are seeing as discontinuity blips in the waveform at 2 points on the falling slope.

Since there is global feedback, you will see the artifacts everywhere in the signal path (the amp responding to correct for it) so you will be chasing your tail. The bias transistor gets driven harder in the negative slope portion of the waveform than the positive causing the bias voltage to depress slightly on the negative slope and increase slightly on the positive slope. This is inherent in the original PL amp design.

To answer your question about bias levels:

For the PL400:
0.38V bias setting produces approx 5.3W per channel of idle power dissipation
0.55V bias setting produces approx 7.7W per channel of idle power dissipation

For the PL700:
0.38V bias setting produces approx 7.6W per channel of idle power dissipation
0.38V bias setting produces approx 11W per channel of idle power dissipation

At .55V you are still 0.1V away from turning on the main outputs so you will still see tiny artifacts but they will be less. Remember these are class AB amps and will have some residual artifacts.

BTW, you should see less of this as the volume goes up (amp slews faster through this small transition region)

 

[Gepetto]

BTW Jer
As an experiment, you can try setting one channel R38/39 value to 15 ohms instead of 10 ohms and then set your bias to 0.55V. This will produce approximately the same idle power dissipation (7.3W) per channel as the original 10 ohm R38/39 value with a bias setting of 0.38V

The precision of R36 (150 ohms) matters very little except for channel to channel consistency.

 

[jbeckva]

I guess a pic is in order here. This is at 15K. 2V per division.. Note where it's driving negative like you were saying. Is this acceptable performance compared to what you have seen?