AndrewT said:
Heathsinks run 55 deg @1.5A bias (21 deg ambient). At 2.6A they are 70+. Spreader plate runs about 2 deg above heathsink (they are quite thick and have profile designed just for that - transfer heat efficiently). TR's cases run less than a degree above that. Yup, those white things are Beryllium oxide washers and they work like magic.
isolation ...
The best way to isolate the output (back EMF and RF getting its way into feedback network) is to cut the feedback line from output altogether. If you match the output transistors well, they can run outside feedback loop and still not contribute much to distortion at all (and what they do is very low order harmonics - almost good
)
After experimenting a bit with the amount of feedback taken from the output (from 100% all the way to 0%) I found that I preferred running it overall feedback free.
Try it, you may like it !
LineSource said:
The best way to isolate the output (back EMF and RF getting its way into feedback network) is to cut the feedback line from output altogether. If you match the output transistors well, they can run outside feedback loop and still not contribute much to distortion at all (and what they do is very low order harmonics - almost good
)After experimenting a bit with the amount of feedback taken from the output (from 100% all the way to 0%) I found that I preferred running it overall feedback free.
Try it, you may like it !
I was wondering if/when the feedback issue was going to be raised. Next step could also be DC coupling of the feedback as mentioned by AndrewT - who needs a Black Gates if you can have no gate A DC servo could perhaps be used for those serious about offset issues, but with proper care offset should be OK.
A DC servo could perhaps be used for those serious about offset issues, but with proper care offset should be OK.
it's easier to use an opamp for a servo. However, I prefer to just leave out the servo if everything runs stable without it. Some systems have also been known to sound worse when a servo was added even though it wasn't supposed to make any difference.
Rotel used an interesting strategy on their 98x and 99x series of amps. All of them used almost EXACTLY the same circuit diagram, with very small changes. The only changes between the smaller and larger ones were higher voltages and more output devices. Their flagship "balanced" input amps actually just had an AD711-based converter stuck in series before feeding it to the single-ended input amp
What was interesting though was that its RB981 and RB985 5channel THX amp used a complex two-stage servo on each channel, which was dropped on all the future models (still keeping the same basic diagram). It was nice though to measure the DC offset after 7 years' abuse and it sits on 0.01mV!
Of course that was with a full-feedback design such as the KSA series, so the servo wasn't really necessary.
Rotel used an interesting strategy on their 98x and 99x series of amps. All of them used almost EXACTLY the same circuit diagram, with very small changes. The only changes between the smaller and larger ones were higher voltages and more output devices. Their flagship "balanced" input amps actually just had an AD711-based converter stuck in series before feeding it to the single-ended input amp
What was interesting though was that its RB981 and RB985 5channel THX amp used a complex two-stage servo on each channel, which was dropped on all the future models (still keeping the same basic diagram). It was nice though to measure the DC offset after 7 years' abuse and it sits on 0.01mV!
Of course that was with a full-feedback design such as the KSA series, so the servo wasn't really necessary.
Mark,
How does it compare to Aleph 1.2 (or 2 , don't remember which one you were building) or maybe Aleph 3 if you have heard it?
I mean bass aside (I suppose there is no comparision i that departament). How is the imaging, mids and highs comapred to Aleph? Is Aleph better in any departament in your opinion?
I know the sound of diffrent Alephs very well, but have no clue how the Kreall (KSA 50 or KSA 100) may sound.
Bartek
PWatts said:
Has anyone been successful in getting a DC-offset servo to work which adjusts the constant current source bias instead of summing the DC offset signal with the feedback into the input of the DIFF amps?
I've attached a simple example that I've been playing with, but cannot get a complete 0V null. I've seen this type of circut work on a headphone amp, but not on a power amp.
Does anyone have a working circuit we could Spice on the Krell KSA100?
Attachments
feedback
There is nothing wrong with feedback per se, but class A amps are particularly convenient for this type of experimentation as fully biased output stage contributes very little to distortion. Nelson discusses this in part 2 of A75 article.
It costs literally nothing (OK one resistor maybe if you want to reduce the amount of overall feedback instead of cutting it out altogether), so can't see why shouldn't one try. As I said, you may like it - I do !
PWatts said:
There is nothing wrong with feedback per se, but class A amps are particularly convenient for this type of experimentation as fully biased output stage contributes very little to distortion. Nelson discusses this in part 2 of A75 article.
It costs literally nothing (OK one resistor maybe if you want to reduce the amount of overall feedback instead of cutting it out altogether), so can't see why shouldn't one try. As I said, you may like it - I do !
Hi,
Pwatts can stand up for himself, but I don't think he was critisising your for suggesting the global feedback modification.
Simply that taking some discussion on board before the PCB is finalised may be advantageous to the final design since no-one else had brought it up.
Even if only by providing 0.1inch pin pairs for connecting links across the PCB for alternative routes.
I think he is willing to consider many improvements provided they FIT the mk1 & mk2 footprint.
First impressions appear to be that the extra stage (fet) and extra voltage are yielding significant advantage over the 50 Klone.
Pwatts can stand up for himself, but I don't think he was critisising your for suggesting the global feedback modification.
Simply that taking some discussion on board before the PCB is finalised may be advantageous to the final design since no-one else had brought it up.
Even if only by providing 0.1inch pin pairs for connecting links across the PCB for alternative routes.
I think he is willing to consider many improvements provided they FIT the mk1 & mk2 footprint.
First impressions appear to be that the extra stage (fet) and extra voltage are yielding significant advantage over the 50 Klone.
AndrewT said:
Can't really comment on footprint (mine is in fact clone of KMA200) but typically output stage is on a separate board connected by wires, so no need for board modification at all. Just don't solder the NFB link ! Or have an in-line resistor (if you want partial feedback).
KSA50 is in fact praised by many knowlegeable people (including Mr Pass himself) as the best sounding Krell. There are undeniable advantages of having higher heardroom etc., but sonically it seems like simpler could indeed be better.
Just something I read on this forum, so don't shoot the messenger !
(I need more than 50W so there was no option for me. Speakers I'm dealing with (ATC100 and VAFi66) simply need more mumbo than KSA50 can provide. So I had to settle for de-tuned KMA200)
Actually all the Berylium oxide insulators I've seen and used were much thinner and somewhat greyish in color, they also have a tab at one end to handle them by. Yours are probably the later Ceramic/Alumnium oxide type... they look just like the ones I buy surplus for 50 cents each. They work almost as well as Beryllium but are not toxic. They crack easily and absolutely flat mounting surfaces are essential.
They're very different amplifiers and both are great designs in their own right. It is more a case of the Krell being more suited to my Dynaudios.... Of the Alephs the A-3 that I built for Luckylyndy is still my favorite Aleph.... The best imaging of them all at least on my speakers. I say build both the Krell and a similar size Aleph and decide for yourself which you like best!
What difference did you hear?
Mark
Bratislav said:
KSA50 is in fact praised by many knowlegeable people (including Mr Pass himself) as the best sounding Krell. There are undeniable advantages of having higher heardroom etc., but sonically it seems like simpler could indeed be better.
I wonder which version they were refering too?