Sorry, did I miss where you said how you measure PIM, and how you decide whether a design passes or fails on that metric? I assume you must measure it since you said it is important, and you try to control it. If you don't measure it, how do you know if it is controlled?
Last edited:
Hi John, what open loop figures have you achieved and what are the closed loop numbers for such a design. One open loop amplifier example I have heard extensively was intrinsically low distortion tube amp with multi-sectioned output transformer (Sowter custom) that was switchable OL/CL. The CL operation was subjectively clean and "accurate' sounding and pleasing 'hi-fi', however the OL operation took on another degree of expression of sound. It sounded very slightly harmonics 'dirtier', very slightly less 'controlled' in the bass/mids and overall sounded ever so slightly 'looser' but in a very nicely pleasing way, sounding more like an unnatural 'control' over the sound/signatureis removed or lessened. In my experience the result is overall subjectively 'friendlier' sound with more believable sense of dynamics and fine detail.....an underlying noisy 'busyness' is removed.
As I understand it this 'busy' excessive excess noise is due to recirculated system (input diff pair type NFB) intrinsic excess noise. CFB done right goes some way to reducing this CL noise behaviour by direct cancellation at or near to the input of the amplification stage instead of input/nfb differenceing stage that dynamically can never perfectly balance whilst using active devices. Matched pair transistors (large dimensions) closely on the same substrate have some chance of reduction/synchronicity of excess noise behaviours, discrete transistors will never have adequate noise 'sameness' to properly do the job.
Solid State open loop amplifiers can also sound especially good in my experience, but the way to do it in transistor OPS is to use multi paralleled pairs to keep distortion and load dependencies down, and this is expensive in materials and power consumption.....unfortunately.
Dan.
Well John,
You obviously either had a defective bunch of 500's, or you haven't a clue about them. I repaired them for Superscope as I had the high rating level for service depots (I was allowed to work on everything). I also did overflow work for them. I kinda think I know those very well, and that doesn't match up with what you are claiming. Fast? The 250 was faster so much so that you could hear it in the audio range? Naw, that isn't a true statement at all John. Do I detect some heavy biases against an excellent product there by chance?
John, that product line was not designed to be good. It was designed to sell at a low price point while making a good profit (a really good profit actually). Compare it to any quality amplifier that uses a set of KT-88's. You will see very quickly where corners were cut (everywhere). Even the output transformers are too small for the power output. A core can only support so much of a flux density before it begins to saturate. The worst part of this is that you are well aware of this and are posting as if this wasn't a problem.
To tell you the truth, the biggest problems I have had with all amps is how poorly they have been serviced. The 500 had this problem in spades as people generally did not take the care and attention to detail that this design needs to perform at it's best. This is true of all power amps, but serviced properly, they are amazing amplifiers. Amplifiers
serviced by travelling technicians were the worst I've seen, save the ones "fixed" by DJs working on their own gear. The only way I would trust field service is if they were only changing modules that plug in.
John, if you know of any of these amplifiers, please send me one. I'll pay for the shipping without complaint as long as you send it USPS, 4 days or longer. I know I have to strip it down and replace a lot of parts, but you would make me a very happy man. I'll even promise not to sell it, because I want it for myself. Heck, send me three and I'll use an active crossover and be in bliss. That is excessive though. One would do.
-Chris
You obviously either had a defective bunch of 500's, or you haven't a clue about them. I repaired them for Superscope as I had the high rating level for service depots (I was allowed to work on everything). I also did overflow work for them. I kinda think I know those very well, and that doesn't match up with what you are claiming. Fast? The 250 was faster so much so that you could hear it in the audio range? Naw, that isn't a true statement at all John. Do I detect some heavy biases against an excellent product there by chance?
I actually don't care. They sounded awful and were designed to be built cheaply for their power class. Current value has nothing to do with the quality of the thing. There's a lot of lore built up around Dynaco tube products that is undeserved. If you gave me one, I would happily sell it so I could buy something that sounded good. I have a box of KT-88's that I can match if need be. Or I could order a new matched set. No problem.
John, that product line was not designed to be good. It was designed to sell at a low price point while making a good profit (a really good profit actually). Compare it to any quality amplifier that uses a set of KT-88's. You will see very quickly where corners were cut (everywhere). Even the output transformers are too small for the power output. A core can only support so much of a flux density before it begins to saturate. The worst part of this is that you are well aware of this and are posting as if this wasn't a problem.
To tell you the truth, the biggest problems I have had with all amps is how poorly they have been serviced. The 500 had this problem in spades as people generally did not take the care and attention to detail that this design needs to perform at it's best. This is true of all power amps, but serviced properly, they are amazing amplifiers. Amplifiers
serviced by travelling technicians were the worst I've seen, save the ones "fixed" by DJs working on their own gear. The only way I would trust field service is if they were only changing modules that plug in.
John, if you know of any of these amplifiers, please send me one. I'll pay for the shipping without complaint as long as you send it USPS, 4 days or longer. I know I have to strip it down and replace a lot of parts, but you would make me a very happy man. I'll even promise not to sell it, because I want it for myself. Heck, send me three and I'll use an active crossover and be in bliss. That is excessive though. One would do.
I don't have to look it up. I am familiar with it as well and it is a very good amplifier. A rebuilt 500 is better though, sorry.
-Chris
Huh?
John, would you please explain how you deal with PIM that you say is important? Is it that you don't think there is a reliable to measure it, so you just have to try to minimize the chances of it being a problem? What would you say?
Yes, and you didn't mention the terrible "quality" of the stock radio; it was a Volkswagen I was driving at that time. The difference between a BT, a 320 bps MP3 and a CD was very obvious. I actually cross-checked this on my cheap computer speakers at home with the same outcome.
"There are 50 mio. people who don't care in the US alone", so what?
Appears to be a bad or broken link.
EDIT: It doesn't work in Chrome, but does work in Edge.
Don't see applicability to audio as practiced here. They are discussing effects seen in RF amplifiers operating in the GHz range.
Last edited:
John, regarding Re value. You are right, of course, that smaller would be better, from the point of view of crossover distortion and output stage transfer function linearity. The best Re = zero Re, however this is impossible due to thermal runaway. I find even Re = 0.1 quite dangerous regarding reliability, but still manageable. I use 0R22 as a compromise. 0R47 is too high speaking about output stage transfer function linearity. Your 0R15 might be an optimal compromise, if you can handle thermal runaway. I have designed a different thermal Vbe multiplier, NPN-PNP which works very well, for me. It could be seen yesterday in one of my posts. It is able to keep very stable Iq value, with temperature.
No, not impossible. Only need a few more parts, added complexity and cost. Both design technique and technology have progressed and numerous solutions have been made available.
There is this thread 1W - 7 Components American Beauty Buffer adaptable to bipolars. Pass Labs XA25 use no source resistor which is also easily adaptable to bipolars.
There are also Hall Effect and other current sensor solutions available for decades usable to realize zero Re amplifiers. As an example, the ACS723 has 0.65 mΩ resistance for current detection which can even be positioned at the collector. A few guys at Pass forum have successfully built power amplifiers using Hall current sensors to regulate bias in Babelfish M25, AKA M2 on steroids, AKA M2-XA25 bstrd AKA M2 gone Berserk.
If "the best Re = zero Re" then the point to gain by maintaining decades obsolete "zero Re impossible" is to avoid having the best. Or perhaps there are inherent flaws with the various design technique and low resistance current sensor solutions when used in power amps?
> The best Re = zero Re, however this is impossible due to thermal runaway.
You can use laterals with 0R (-ve tempco).
I know it is a very expensive solution ......
Cheers,
Patrick
You can use laterals with 0R (-ve tempco).
I know it is a very expensive solution ......
Cheers,
Patrick
Deezer, Tidal, Quobuz, and HighResAudio all offer music streaming with CD resolution now, the last two even higher resolution, albeit with a smaller catalogue. There is no reason - other than cost - to restrict yourself to MP3 streaming.
John, when you first mentioned PIM I thought you'd added the P by mistake, "Passive" right? Is it of any consequence in audio power amps, how does it differ from IM?
Also Renardson
Phase Intermodulation
Glad we have some audio amplifier topics, this is closer to thread name/orientation.
Phase Intermodulation
Glad we have some audio amplifier topics, this is closer to thread name/orientation.
To me, better and more predictable linearity with bipolars, good bipolars. Also lower Rout, even with Re added.
- Status
- Not open for further replies.