ostripper said:
This did start out as a DIY project here. I started at the earliest point in the design that gave acceptable results. I had intended, as I stated, to walk the builder through the evolution of the design over it's 25 year life. Many changes have been made in that time. The circuit became more complex and the performance improved. The resistors you refer to are one of the changes.
I felt that presenting my design in it's latest and most complex embodiment would be, perhaps, a little intimidating for someone that did not work is this field. There was no point in presenting a design here that virtually no one would attempt to build.
megajocke said:
More current does not flow in the bias string because the current in the string is determined by the gain of the bias transistor pair and the current in the CCSs. When current flows in the base of a drive transistor, less total current is available to flow in the bias pair. As a result, less current flows in the diode string. It is the voltage from the emitter to collector that increases on one of the bias transistors. This happens because there is less base current for that transistor. Less base current means less emitter current and the voltage drop from collector to emitter increases as the transistor tries to turn off.
Well you have 13, Carlos likes it, and you have caught the
interest of many with your novel topology. I don't think it is
fair for people to DEMAND of you to PROVE it's validity..
they should just build it and see !
This reminds me of hugh dean and his amps , people were
DEMANDING he give up his IP in the forum 😕
Not many have built my amps.. so on the next ones I might
just print up the boards and sell the whole deal as a kit.
I dont know how I will use these amps yet, but I will use them somewhere. (I have 3 HT systems). Don't fret ,people are the
same everywhere.. picky , critical , frivolous..etc. keep up the
good work.
OS
Steve Dunlap said:
MS Word's infuriating auto typo corrector inserted the wrong word.
I promised myself I wouldn't comment on this any more, but here is one last post anyway.
I don't recall requesting you to provide "proof" of much, especially such as comprehensive measurements of a working unit - which you have made clear you are not in a position to provide.
The most I think I asked for (following a post of yours in the Blowtorch thread) was an example of a <0.005% THD open loop power amplifier (50-100W).
Following that I have studied the Krill circuit and I simply do not accept that it operates entirely as claimed or that it would meet the claimed distortion spec. I've explained why to some degree, but that is as far I care to take it. We'll just have to agree to disagree.
As far your measurements go, I am not accusing you of lying by questioning their validity.
It is generous for you to offer your workshop facilities. In case somebody with the required technical skills and resources takes you up on your offer it should be noted that there have been numerous good suggestions (amongst the feuding) made here regarding what tests to run - which could probably be taken as constructive input.
Cheers,
Glen
Hi Steve, maybe we are saying the same thing.
As both transistors have the same base current, different emitter/collector currents and are both conducting the base current must be enough to keep the transistor with highest emitter/collector current on.
The transistor on the side where output stage draws current will then have more base current than needed to just keep it on and collector-emitter voltage for that one will decrease while collector-emitter voltage for the non-conducting output stage side will increase because it has higher emitter/collector current for the same base current.
The voltage between emitters of bias transistors is still the same though, set by the bias string and vbe of transistors.
I'm unsure if you are claming that voltage between emitters of bias transistors vary or that it does not. That is, the voltage over the capacitor which is also trying to keep it constant. Others have asked this too without getting a clear answer.
As both transistors have the same base current, different emitter/collector currents and are both conducting the base current must be enough to keep the transistor with highest emitter/collector current on.
The transistor on the side where output stage draws current will then have more base current than needed to just keep it on and collector-emitter voltage for that one will decrease while collector-emitter voltage for the non-conducting output stage side will increase because it has higher emitter/collector current for the same base current.
The voltage between emitters of bias transistors is still the same though, set by the bias string and vbe of transistors.
I'm unsure if you are claming that voltage between emitters of bias transistors vary or that it does not. That is, the voltage over the capacitor which is also trying to keep it constant. Others have asked this too without getting a clear answer.
ostripper said:
Yes, I agree ... completely.
Steve, please continue and thank you, very much appreciated.
🙂
PS: Nice comments by Charles Hansen 😎
traderbam said:
That sums up my first impression, but after seeing and reading some results I think that some people are out of a certain comfort zone. I think several people pointed out that component substitution with "better" devices i.e. higher hfe actually made things worse. A class A/B output stage still has a crossover discontinuity even if beta is infinite, that is no change in base current drive.
Steve - When Charles Hansen called it a triple it was OK?
megajocke said:
In reply to your first paragraph:
With no input signal, the base current in each of the bias pair is equal as is the emitter current and collector current. This assumes the CCSs are equal.
Second paragraph:
When there is a signal present, the bias transistor on the side of the output stage that is conducting will have less current flowing in the emitter. This is because more of that constant current is now flowing in the base of the driver transistor.
Third and fourth paragraph:
The AC voltage between the emitters of the bias pair remains constant (well very close). I thought I had answered your last question. I may have done that in a thread started on another DIY site. There is not an AC voltage drop across that cap. If you look at either side of the cap, you will see an exact replica of the input signal throughout the entire cycle. If you look at both sides of the cap differentially, there is no signal.
scott wurcer said:
Out of a certain comfort zone? This is a DIY site. There are some here that are not as deeply into EE as you are. Perhaps you are referring to me. I really don't have a comfort zone. I'm equally bad at many things.
I believe they pointed out that in simulation it made things worse to use better devices. As far as I know, I am the only one here to build units with many different transistors and measure the results. The newer better transistors do test better.
If you want to call it a triple as Charles did, that's OK by me. It would be a triple with no stability issues and no need for a feedback loop. Since others had suggested it be made into a triple, I appear not to be the only one that doesn't see it as a triple. Perhaps just semantics. I don't consider it an issue.
Well, I substituted "better" transistors in my simulation, where better is defined as "can actually be bought at retail in 2009." All of these devices are high beta and high Ft. The most obvious problem was oscillation; it was necessary to add a 68pF dominant pole capacitor on the voltage gain stage. I was surprised Dunlap's schematic lacked C(dom), but perhaps this is part of the 25-year evolution of the circuit.
Steve Dunlap said:
I do also refer to output devices.
I believe it was an earlier schematic.
From pdf document properties.
Title: 50W DIY.DWG
Author: Stephen E. Dunlap
Created: 11/12/2008 3:07:30AM
megajocke said:
D5-D7 is not biased correctly. R12 doesn't make sense. It diverts current around D6 and D7 and turns them off. As it is Q8 and Q11 are acting merely as saturated switches in the circuit and are redundant and superfluous if the bias generator is placed more optimally. That part of the circuit is not very well defined and will not handle varied voltage bias conditions well if different output stages are used.
More importantly the cap you refer to C6 AC clamps the supposed dynamic action of the OP bias section D5-D7, Q8 and Q11.
The OP stage voltage bias generator Q8,Q11,D5-D7 could be better implemented by replacing it with a more common vbe multipler or something similar in the Q7 and Q10 base circuit. Again with b-e resistors on Q7 and Q10 to bias the vbe circuit optimally.
Steve Dunlap said:
Okay, then we agree - I first thought you were suggesting the voltage over the capacitor did contain a signal component.
But weren't you suggesting the output transistors didn't switch off?
CraigBuckingham said:
I said almost the same thing on Page 2 of this thread. IMO the circuit behaviour is too dependent on hfe and other parameters to be very useful. Especially the current through the diode string is dependent on the hfe of bias transistors.
R12 makes sense though, it only diverts some current when correctly adjusted (this is the bias trimmer). The less current that flows through the diodes the lower the voltage drop. It only looks like a knee because it is an exponential relation. The circuit is still too dependent on device parameters for my taste. Double the hfe of the bias transistors and current through diode string halves for example.
I do not like circuits that depend on the exact value of hfe - it's a too uncontrolled parameter. I will only make sure to use parts with enough of it. Higher than needed shouldn't effect circuit operation in any way. Your last paragraph is exactly the way I would do it if I wanted something similar.
I'm not saying it is non-working or anything like that - just that it's behaviour is a bit too undefined and hfe-dependent for my taste.
megajocke said:
Yes, I saw you comment earlier. I only had time to quickly read up to page 5 and then jump to page 31 to see if anyone had addressed the designs issues.
I am just adding to one of the flavours of consensus.
All of the talk about the OP bias section D5-D7, Q8, Q11 dynamic action is misplaced as C6 will clamp most of it.
megajocke said:
I was suggesting that. Some here agree and some do not. Either way, the measurements are good and many that have heard the amps like the sound.
megajocke said:
The circuit does not in any way depend on the exact value of hfe. That is why there is an adjustable resistor there. As you say, higher hfe will have little effect. Lower hfe will also have little effect. What little effect is encountered is easily adjusted out with R27.
I do match the transistors in the bias pair.
- Status
- Not open for further replies.