Available XPL909's and RCA410's

[Gepetto]

Hi Mark

In an attempt to provide some additional insight.

Many, but not all PL amps exhibit the purple heatsink phenomenon on the negative side, 2nd and 4th heat sink from the transformer. Parasitic oscillation is going on in the quasi construct leading to extra heat on these negative legs, leading to more heat in the heat sinks leading to purpling of these two stacks. The collector to ground (heat sink) capacitance in these legs is one of the contributing factor since the collectors are swinging with the audio signal on the negative half. This does not matter on the positive half because the collectors on this leg are tied to B+ and this extra collector to heat sink capacitance just contributes to extra B+ decoupling capacitance. This is a don't care situation as opposed to the quasi-comp negative half.

This high frequency oscillation also made the Zoebel work harder since these frequencies are typically above the turnover frequency of the Zoebel network. This is driving power from the negative output stage into the Zoebel as unnecessary dissipation.

Full comp fixes this condition as both positive and negative halves have the collectors at the B+ and B- potentials respectively. A superior arrangement. Also having each half as an emitter follower construct also contributes to better (lower) source impedance at the output.

Use a slow, low gain transistor (low gain-bandwidth) like the RCA410 in this critical quasi position to avoid this parasitic oscillation or better yet, convert your amp to full comp.

 

[Geegz]

Hi Mark

In an attempt to provide some additional insight.

Many, but not all PL amps exhibit the purple heatsink phenomenon on the negative side, 2nd and 4th heat sink from the transformer. Parasitic oscillation is going on in the quasi construct leading to extra heat on these negative legs, leading to more heat in the heat sinks leading to purpling of these two stacks. The collector to ground (heat sink) capacitance in these legs is one of the contributing factor since the collectors are swinging with the audio signal on the negative half. This does not matter on the positive half because the collectors on this leg are tied to B+ and this extra collector to heat sink capacitance just contributes to extra B+ decoupling capacitance. This is a don't care situation as opposed to the negative half.

This high frequency oscillation also made the Zoebel work harder since these frequencies were above the crossover frequency of the Zoebel network. This was driving power from the negative output stage into the Zoebel as unnecessary dissipation.

Full comp fixes this condition as both positive and negative halves have the collectors at the B+ and B- potentials respectively. A superior arrangement. Also having each half as an emitter follower construct contributes to better (lower) source impedance at the output.

Use a slow, low gain transistor (low gain-bandwidth) like the RCA410 in this critical quasi position to avoid this.

Joe,

I believe I have read this before that the RCA410 is only needed in the 2nd and 4th rows, do we know why PL used them originally in all 4 bottom row locations? Also to confirm when upgrading outputs in an original configuration amp can the 1st and 3rd row, bottom transistor location be replaced with the modern(new) ON Semi without any further modifications to the stock (factory) layout?

 

[Gepetto]

Joe,

I believe I have read this before that the RCA410 is only needed in the 2nd and 4th rows, do we know why PL used them originally in all 4 bottom row locations? Also to confirm when upgrading outputs in an original configuration amp can the 1st and 3rd row, bottom transistor location be replaced with the modern(new) ON Semi without any further modifications to the stock (factory) layout?

On the first question, I just do not know. I also don't know why many things were done in these amps originally and probably will always be a mystery to me. A guess is that the RCA410s may have been lower cost than the PL909 devices but that is just a guess. They used them in the less demanding locations and saved some $$. That is a total perhaps...

On the second question, yes. The dominant pole of the positive half original output stage is the PL909 device. The modern ON Semi device is much faster than the original RCA410 device as is the TO-5 2N3439 transistor mounted on the control board. The PL909 will dominate the turnover frequency. The upper stage is unity voltage gain, it is an emitter follower to the class A amplifier stage in the original design. That ClassA stage is stable by design.

 

[FredR]

Hi Mark

In an attempt to provide some additional insight.

Many, but not all PL amps exhibit the purple heatsink phenomenon on the negative side, 2nd and 4th heat sink from the transformer. Parasitic oscillation is going on in the quasi construct leading to extra heat on these negative legs, leading to more heat in the heat sinks leading to purpling of these two stacks. The collector to ground (heat sink) capacitance in these legs is one of the contributing factor since the collectors are swinging with the audio signal on the negative half. This does not matter on the positive half because the collectors on this leg are tied to B+ and this extra collector to heat sink capacitance just contributes to extra B+ decoupling capacitance. This is a don't care situation as opposed to the quasi-comp negative half.

This high frequency oscillation also made the Zoebel work harder since these frequencies are typically above the turnover frequency of the Zoebel network. This is driving power from the negative output stage into the Zoebel as unnecessary dissipation.

Full comp fixes this condition as both positive and negative halves have the collectors at the B+ and B- potentials respectively. A superior arrangement. Also having each half as an emitter follower construct also contributes to better (lower) source impedance at the output.

Use a slow, low gain transistor (low gain-bandwidth) like the RCA410 in this critical quasi position to avoid this parasitic oscillation or better yet, convert your amp to full comp.

Seen the purple many times. Truth.

 

[mlucitt]

Joe,
Thanks for the reply (post #21). I have never seen the dreaded purple heatsinks, maybe I am just fortunate. And I understand now, the term "low gain bandwidth" is shorthand for low gain-low bandwidth (ideally gain around 10 hFE and Ft <2 MHz?) in the RCA 410/66546. I try and convert my customers to full complementary, but some want to stay "original" or they have a limited budget, like me. I assume the benefits of full comp are more critical in the 700 Watt amps than in the 400 Watt amps, and that also relates to the higher cost for more output devices in the 700s.

I always put an oscilloscope on the outputs of the finished amplifier with the dummy load attached and have never seen oscillations up to the limit of my scope which is 100MHz (Tektronix 2235). Perhaps the Zobel network/Boucherot cell is damping those oscillations effectively. Is there a bandwidth I should be concentrating on and a voltage level to set on the scope (I use 10V/div)? And are the parasitic oscillations more apparent at low output or full power output?
Thanks

 

[Gepetto]

Joe,
Thanks for the reply (post #21). I have never seen the dreaded purple heatsinks, maybe I am just fortunate. And I understand now, the term "low gain bandwidth" is shorthand for low gain-low bandwidth (ideally gain around 10 hFE and Ft <2 MHz?) in the RCA 410/66546. I try and convert my customers to full complementary, but some want to stay "original" or they have a limited budget, like me. I assume the benefits of full comp are more critical in the 700 Watt amps than in the 400 Watt amps, and that also relates to the higher cost for more output devices in the 700s.

I always put an oscilloscope on the outputs of the finished amplifier with the dummy load attached and have never seen oscillations up to the limit of my scope which is 100MHz (Tektronix 2235). Perhaps the Zobel network/Boucherot cell is damping those oscillations effectively. Is there a bandwidth I should be concentrating on and a voltage level to set on the scope (I use 10V/div)? And are the parasitic oscillations more apparent at low output or full power output?
Thanks

Hi Mark
I cannot answer your question with any hard answers, because the info I have comes in from customers that don't do parametric analysis for me. They just ask for help to fix an observed problem with their amp. A decent guess is 200-500 KHz range, not more than that because that is where the loop gain rolls off to unity. I attach an example picture and would venture a guess that the oscillation is occurring in that range but it is just a blur at 100usec/div setting provided so I cannot give you a hard frequency.

Full comp benefits both PL400 and 700s. The amps sound sweeter in the full comp mode whether 400 or 700. Technically a much better solution too.

 

[nakdoc]

If you look at gepetto's screen, the negative peak is "fatter" than the rest of the sine wave. Oscillation at low amplitude compared to the audio sine wave is occurring and added to the fundamental frequency. Generally oscillations that cause heating are much greater amplitude that you see here.

 

[Gepetto]

If you look at gepetto's screen, the negative peak is "fatter" than the rest of the sine wave. Oscillation at low amplitude compared to the audio sine wave is occurring and added to the fundamental frequency. Generally oscillations that cause heating are much greater amplitude that you see here.

Agree, this is a telltale sign into a resistive load that may bloom into something else with a reactive load and heat.

 

[mlucitt]

Thanks Joe and doc, this answered my question. I know forensic analysis of these old amps is more art than science and your answers were helpful in my understanding of the oscillation problem. I am familiar with the 'ringing' phenomenon in which the oscillation is 'riding' on the lower frequency sine or acoustic waveform; I have just never seen it before. Joe, thanks for the picture, it is worth 1000 words. I will use it to show customers what they cannot hear, but may be experiencing in the quasi-complementary amplifiers. Can you post the picture in the files section, I cannot capture it in the thread.

 

[laatsch55]

What scope ya got there Joe?

 

[Gepetto]

What scope ya got there Joe?

Dunno Lee, I have a TEK TDS2024B, that was some customer scope shot of his amp misbehaving so I cannot tell you for sure what scope it was.

 

[Gepetto]

Thanks Joe and doc, this answered my question. I know forensic analysis of these old amps is more art than science and your answers were helpful in my understanding of the oscillation problem. I am familiar with the 'ringing' phenomenon in which the oscillation is 'riding' on the lower frequency sine or acoustic waveform; I have just never seen it before. Joe, thanks for the picture, it is worth 1000 words. I will use it to show customers what they cannot hear, but may be experiencing in the quasi-complementary amplifiers. Can you post the picture in the files section, I cannot capture it in the thread.

It is a pretty healthy 1KHz output swing that the oscillation is occurring on so I would guesstimate a 2-3Vp-p oscillation going on by eye balling it. The large scale factor throws off how large the oscillation is.

 

[Gepetto]

I never got scope shots on this instance but here is some of the back and forth from many years ago...

Shop: The unit has a nasty oscillation on the lower half of the wave on both channels. His problem is somewhat similar to oscillations we would occasionally see on original 400's for which Phase instructed us to install a cap over the drivers to remedy. Note that he is using 66546 drivers and 2SD555 outputs across the board. The insulators are mica. I suspect that you have encountered this problem before. Have you a recommendation for a remedy?

Me: I have seen marginal stability the lower half waveforms when users have tried MJ15024 as both the pre-driver and final drivers which is never our recommendation.
We have never recommended the 2SD555 and never will.
Recommendation is RCA410 or SJ2741 as pre-driver and MJ15024 or MJ21196 as final drivers. The local feedback loop formed by Q12 and Q14-Q18 is absolutely critical to proper operation of the PL400 and PL700 amp. The pre-driver should be a slower device than the final driver in order to ensure stability in this local loop.

Shop: You certainly didn't let the grass grow under your feet before answering and I appreciate it. Yesterday, I pulled the 2SD555's and put in the first row of MJ15024 (the ones above the row of 66546 drivers) and presto, no oscillation. I have used MJ15024 in the past with good results, but I am curious as to which you would prefer in this case. I have not used the 21196, but am open to it if you recommend it.

This scenario has occurred in multiple cases. The construct is sensitive to both driver and output device speed (GBW) and care should be taken to avoid upsetting the original conditions. It happens, and perhaps much of it goes undetected because of the ultrasonic nature of it. Full comp is a much better setup, avoids this sensitivity altogether.