Added AUDIONICS CC-2 BY BOB SICKLER 1977-78 to the first post.
They say that this was a very reliable design.
They say that this was a very reliable design.
Hi Pete,
You are correct about the heat sinks being too small for that amplifier. I measured one of the four 2-finned heat sinks by itself today. It measured 1 C/watt. With 2 such heat sinks per channel, a channel has 0.5 C/watt of heat sink. That does not take into account a small amount of heat sinking added by the chassis of the amplifier.
The stock heat sinks were under-sized for the FTC 1/3 power for 1 hour preconditioning requirement that was issued in 1974 after the PL700 was in production for a few years. I believe that this shortcoming was subsequently addressed by PL with augmented heat sinks in a re-design. In the year 2000 the preconditioning requirement was relaxed to 1/8 rated power, which would be 44 watts. A class AB amplifier will theoretically dissipate about 30% of its rated power when operated at 1/8 rated power. That comes to about 105 watts power dissipation for each channel of the PL77. At 0.5 C/watt for a channel, that means a rise of 53 C, putting the top temperature at about 76 C with a 23 C ambient. So it doesn't appear likely to pass the 1/8 power preconditioning. The thermal cut-out is probably set to 70 C. These are just rough calculations that are subject to both exacerbating and mitigating realities.
My plan to bias each MOSFET at 150 mA to reduce transconductance droop in the crossover region and to provide a good-sized class A region thus appears too optimistic. Each heat spreader will house 2 pair of MOSFETs, with each MOSFET dissipating 15 watts at idle with a 100-V rail if I bias them at 150 mA. Each heat spreader would thus dissipate 60 watts into a 1 C/watt heat sink. If all of these estimates here are correct, that is a non-starter.
Cheers,
Bob
Nice work Bob.
Just looked at a picture and everytime I see that amp I think that's how NOT to make a heat sink.
There should be a picture of it with the caption, Don't DO THIS! Looks about right for a 60W amp.
IMO, they have to be replaced, these are not big enough but maybe something along these lines:
https://www.amazon.com/Wisecoco-Aluminium-Heatsink-Accessory-100x50x50mm/dp/B0C6JRYZ2B
I'd make them .5-1" short to leave room for fans on top, and have them start at 60 deg C or
something like that.
Shouldn't we use 110 deg F as ambient for the hottest day in the hottest climate?
Or something like this, my dad (a thermo expert) explained that the fine pitch is designed for fans:
https://www.amazon.com/LuoQiuFa-Alu...-2-spons&sp_csd=d2lkZ2V0TmFtZT1zcF9hdGY&psc=1
Just looked at a picture and everytime I see that amp I think that's how NOT to make a heat sink.
There should be a picture of it with the caption, Don't DO THIS! Looks about right for a 60W amp.
IMO, they have to be replaced, these are not big enough but maybe something along these lines:
https://www.amazon.com/Wisecoco-Aluminium-Heatsink-Accessory-100x50x50mm/dp/B0C6JRYZ2B
I'd make them .5-1" short to leave room for fans on top, and have them start at 60 deg C or
something like that.
Shouldn't we use 110 deg F as ambient for the hottest day in the hottest climate?
Or something like this, my dad (a thermo expert) explained that the fine pitch is designed for fans:
https://www.amazon.com/LuoQiuFa-Alu...-2-spons&sp_csd=d2lkZ2V0TmFtZT1zcF9hdGY&psc=1
Regarding stability of Quasi-Comp outputs, KSTR, wrote this in post #163:
https://www.diyaudio.com/community/threads/lm3886-pcb-vs-point-to-point-with-data.252436/page-9
ST AN2653, very nice stability paper: https://www.st.com/resource/en/appl...ading-applied-to-ts507-stmicroelectronics.pdf
Not sure if KSTR is correct but I do understand that he is very sharp, I welcome further comments:
"In my view and experience, with chips amps the local output snubber does two main things, in order of importance :
https://www.diyaudio.com/community/threads/lm3886-pcb-vs-point-to-point-with-data.252436/page-9
ST AN2653, very nice stability paper: https://www.st.com/resource/en/appl...ading-applied-to-ts507-stmicroelectronics.pdf
Not sure if KSTR is correct but I do understand that he is very sharp, I welcome further comments:
"In my view and experience, with chips amps the local output snubber does two main things, in order of importance :
- prevent/damp spurious oscillations of the output stage (way) above the closed-loop bandwidth by loading the output down. I'm completely with JohnW here. Chip amps typically have either quasi-complementary NPN output (LM3886) or a buffer with local feedback (TDA7293) for the negative half of output current swing and those are prone to oscillation. For that reason the snubber should go shortest path from output pin to the neg supply pin rather than ground and the decoupling at that frequencies should go rail-to-rail rathern than making a detour through ground.
- help phase margin by introducing a properly spaced pole-zero pair below closed-loop bandwidth, as given in consort_ee_um's link."
Thinking over the last post where KSTR says that the snubber (Zobel) should go from the output
to the negative rail. I assume this is because the negative, CFP side is more prone to oscillation,
as others have noted here, and that connection provides the shortest lead length to the output
device. Now, what to do in a complementary amp where both sides are CFP? Perhaps two Zobels
placed right at each output device to provide the shortest path - thoughts?
Also, I wonder given that the Tiger amps have long wires (4-6") from the output devices to the
board, should the pair of Zobels be placed right at each output device to bypass the lead inductance?
to the negative rail. I assume this is because the negative, CFP side is more prone to oscillation,
as others have noted here, and that connection provides the shortest lead length to the output
device. Now, what to do in a complementary amp where both sides are CFP? Perhaps two Zobels
placed right at each output device to provide the shortest path - thoughts?
Also, I wonder given that the Tiger amps have long wires (4-6") from the output devices to the
board, should the pair of Zobels be placed right at each output device to bypass the lead inductance?
When my shop gets back in service, I may revisit this. I had a CFP amp board that I set aside and gave up on due to oscillations I couldn’t fix. Not crazy voltage, just +/-25V. I hadn’t tried zobels to both rails, maybe it can be resurrected. This also gives me a couple more rabbit holes to look down concerning QC class H. The four boards in that unit got rewired for full comp, but I’ve still got 6 more unmodified boards in my stash. I could try returning zobels to the two negative rails locally on the board and see what happens. I even remember any of this in a year or so. I only fully explored supply bypassing, but the zobel was always tied to ground.
I think that it is connected with the work of the load, and not with one output arm of the amplifier. An article from a French magazine (Le Haut-Parleur N°1338 13-01-1972 p. 214-215) where the author says that the Zobel chain is used for load shunting. Also, most circuits of the late 60s and early 70s were built on a quasi-complementary output and unipolar power supply.
That is a single supply amp, and with the Zobel going to ground that is the same as if
it went to the negative supply in a dual rail design. Both rails are the same as AC ground
because of the caps across them, all we're saying is that it is a shorter path to the output
devices by connecting to the rails.
it went to the negative supply in a dual rail design. Both rails are the same as AC ground
because of the caps across them, all we're saying is that it is a shorter path to the output
devices by connecting to the rails.
I often use Zobels or the equivalent in multiple places in my amplifiers. I often put a Zobel before and after the output relay (if there is one). The first Zobel is very close to the output transistor(s). The second one is right at the output terminals of the amplifier. The latter helps to damp the output at high frequencies to absorb some EMI at high frequencies, since the speaker cables act as antennas. Also, when the relay opens if there is a fault, the output Zobel can mitigate some of the EMF spike that may occur across the relay contacts.
In output stages with multiple pairs I will often put a smaller Zobel right at the output of each pair. Dissipation-wise, I prefer not to use wirewound non-inductive sand resistors, and can use metal oxide film (MOF) resistors in multiple Zobels to get the dissipation down low enough to use MOFs.
I also use effectively Zobels in each power supply rail at the output transistors by putting a low-impedance LPF in the rail between the output transistors and the driver transistors. The series LPF resistor at that point might be only 2.2 ohms, so it both gives some HF rail filtering and the action of a Zobel. This can also help reduce parasitic oscillation tendencies by breaking the path between the output transistor and driver transistors at high frequencies. I also use an effective Zobel at the amplifier input, using a small-value capacitor, perhaps a couple hundred pf shunt preceded by a 68-ohm input series resistor. This forms an LPF at very high frequencies but also terminates the input interconnect from the preamp in a compromise characteristic impedance at very high frequencies, all helping to reduce EMI susceptibility at very high frequencies.
I admit that some of this may be overkill, but Zobels are cheap.
Cheers,
Bob
In output stages with multiple pairs I will often put a smaller Zobel right at the output of each pair. Dissipation-wise, I prefer not to use wirewound non-inductive sand resistors, and can use metal oxide film (MOF) resistors in multiple Zobels to get the dissipation down low enough to use MOFs.
I also use effectively Zobels in each power supply rail at the output transistors by putting a low-impedance LPF in the rail between the output transistors and the driver transistors. The series LPF resistor at that point might be only 2.2 ohms, so it both gives some HF rail filtering and the action of a Zobel. This can also help reduce parasitic oscillation tendencies by breaking the path between the output transistor and driver transistors at high frequencies. I also use an effective Zobel at the amplifier input, using a small-value capacitor, perhaps a couple hundred pf shunt preceded by a 68-ohm input series resistor. This forms an LPF at very high frequencies but also terminates the input interconnect from the preamp in a compromise characteristic impedance at very high frequencies, all helping to reduce EMI susceptibility at very high frequencies.
I admit that some of this may be overkill, but Zobels are cheap.
Cheers,
Bob
Version 3 of my QuantAsylum QA403 is now up on my website. The QA403 is a great, high-performance audio analyzer for only $600.
Cheers,
Bob
Cheers,
Bob
This is the animal called Zobel in German and apparently Sobol in Russian:
https://upload.wikimedia.org/wikipedia/commons/7/72/Sobol-bur1_(cropped).jpg
It is a close relative of the European pine marten.
https://upload.wikimedia.org/wikipedia/commons/7/72/Sobol-bur1_(cropped).jpg
It is a close relative of the European pine marten.
I think this is borne-out in other ways - such as an audiophile myth I've long railed-against - that all caps must have 'low-esr'
Absolutely, not true.
ESR IS useful - when the applications are one, you speak to. A little inherent damping, is a very, very useful thing.
It is also why the humble tantalum cap remains with-us, and often recommended say in datasheets for fast opamps, line-drivers and similar. These need some small bulk local capacitance, but teh intrinsic ohm sometimes more - makes for a very graceful characteristic when the inevitable few nF of parallel (film/ceramic) VHF bypass is in parallel, and when the result is fed from - say - a regulated rail (where the regulator inevitably has a synthetic output inductance that increases as its internal f/back drops-away - predictable for jellybean 3-pin regs, but a wildcard most seem unaware of in tweaky aftermarket parts & homebrew 'high-performance' regs - unless Diyer's set out to characterise their own latest 3am brainchild. which is obvious, most do not.)
I'll take tedious integral stability, over peaky performance, everyday, thanks.
And - thanks again for your books, qnd your continued support of this community!
Bob, your methods are certainly NOT overkill. Tracing and curing instability is always difficult. Often the very act of probing for it changes conditions sufficiently, that it goes away, and/or happens somewhere else.
If I come across anything cheap that enhances stability, like Zobels or correct & judicious decoupling of rails, I ALWAYS design it in.
The converse is also true. Very often, Golden Pinnae stuff, like Golden Pinnae caps across decoupling caps, make things worse. At best, the Golden Pinnae stuff doesn't make things worse ... but don't leave out the good cheapo stuff 😊