B
Previously i have challenged cfa experts to modify ska gb150 into cfa. I found the cfp output cfa is hard to stabilize.
There are 2 examples as a starting point. Latfet version is from Ruwe's work in ssa thread. Another is the Audio Labor Snell.
ADD:
To stabilize faster device in a cfp output, the feedback resistance can be made bigger. But this is I believe against the cfa spirit. I tried to use no more than 2k2.
Previously i have challenged cfa experts to modify ska gb150 into cfa. I found the cfp output cfa is hard to stabilize.
There are 2 examples as a starting point. Latfet version is from Ruwe's work in ssa thread. Another is the Audio Labor Snell.
ADD:
To stabilize faster device in a cfp output, the feedback resistance can be made bigger. But this is I believe against the cfa spirit. I tried to use no more than 2k2.
Last edited:
Richard, don't you think that feedback present its main interest from an industrial point of view, to avoid this expensive and tricky step, allowing the use of out of the box components, while keeping the results satisfactory enough ?
While DIYers don't have enough samples or equipment (or at an expensive price) to get a chance to match active devices ?
Last edited:
To me, this seems as important as the topology, but i am not as advanced as many here. Simply start with lowest distortion in simple settings and add only as needed. Not much else is needed in that simple example. at least you cant argue with the numbers.
Cfp is known to sound enjoyable. SKA is measured very well.
This is what i dont understand from you experts, trying to compare vfa versus cfa. It is similar like comparing sealed box with vented box. What secrets are there to be revealed?
Imo it is just part of design decision based on the design objective. The problem is usually people cannot relate numbers with sound perception.
Cfa for example will easily cover the high end. Yes this is imo critical than most people might think. But vfa also can do sufficiently. What is "sufficient" thats the question. Minimum thd20 also must be set for vfa.
Cfa especially with lateral tend to have weaker bass than vfa. So set a minimum standard bass performance to be achieved. Imo the weak bass is caused by high noise floor. So use good power supply for front end, shunt type if necessary.
Cfa also has worse thd compared to vfa. So set a minimum standard thd. 0.02% For example. Or based on percentage of each order of the distortion. Of course cascode must be compared with cascode, a no-no for me.
Jay, I would expect this.
You are feeding a CFP that has a pole and lots of phase shift from a very wide bandwidth CFA front end. The result is instability.
You would need to transition the feedback loop from the output to the TIS stage to overcome this problem - something similar to TMC.
Last edited:
Jay, i think i was not clear enough in my fist point. You have to keep the impedance value of the feedback resistance low enough to minimize the distortion due to the non linear parasitic capacitance of the emitter. Means far enough from the open loop bandwitch limit. If you have to reduce the bandwidth at this precise point, for some obscure reason i cannot imagine, you better add a parallel linear capacitance, instead of increase the feedback resistance value.
On the second point, i don't understand your point neither. When you are about to design an enclosure, you have to know the behaviors of each type of charge (baffle plane, closed box, bass reflex, horn, 1/4 wave etc...) in order to reach better what you want to achieve, efficiency, linearity, damping, response curve... If you are obliged to use a given speaker, it may reduce your choice depending of its Small and Thiele parameters.
Same thing with VFA vs CFA (or error cancellation, or error correction or no GFB)
I believe that, if i should prefer a VFA amplifier (or a class D one) for the basses of a multy way system and CFA for the trebles. It is a decision we take before beginning the design.
All the interest of this thread is to better understand the behavior of CFA (and VFA) in order to know the advantages of each. In other words, in in what areas they are better. HD distortions, IM, TIM, Slew rate, Bandwidths, etc. And, if we can, how those numbers affect our listening experience.
Last edited:
Yes i'm aware of such solution but am not qualified to produce one that is functioning (I still dont understand how things work. Thats why i expect the experts to try their best at this cfa version of ska. At least this is to be fair to the vfa guys, instead of modding crescendo to cfa or modding vssa to vfa.
I don't understand your point. Comparing any amp two version is the best way to point the differences. VSSA, as it is very simple seems a perfect candidate.
I asked Dadod to offer us two versions of the VSSA, one, the original CFA he had yet simulated, and an other VFA to compare.
I wonder why no answer !
Could-you do that ? Here are the links:
http://www.diyaudio.com/forums/solid-state/240712-cfa-topology-audio-amplifiers-2.html#post3595853
http://www.diyaudio.com/forums/solid-state/240712-cfa-topology-audio-amplifiers-6.html#post3595756
http://www.diyaudio.com/forums/vend...-lateral-mosfet-amplifier-15.html#post3314786
Last edited:
I don't think that this is good way to compare CFA with VFA. To transwer one to another is not good, you have to start from what is best for each of those technologies and you end up with completely different configuration, if you know what I think. There could be a good way to compare, let's say, with similar number of active components?
And I am new to CFA, not yet built any, so I am learning here.
I'm thinking this too. That why I proposed the complementary LTP input for comparisson as this would be most similar topology wise, but such VFA by nature is sub optimal, especially when it comes to applying feedback methods.
Edit: My proposition was more to take a reference CFA and see what I could do with it to crank out as few THD as possible, rather than comparing.
Last edited:
Dadod, the purpose was not, in my mind, to get an answer of what is BEST but to better understand the differences brought by both topologies. For that, i believe we better keep the two versions as close as possible.
Not a contest of who has the biggest, but where each one bring advantages like distortions, bandwidths, slew rate, phases margins etc.
It can. This amp is simple, and reported to sound very nice by all the builders, hundreds to come, on the way to achieve their own.
Yes the game, as i see-it, can consist to add two transistors in the input in LTP, change the feedback path to the -input and tune the currents and gains of each stage to get the best of this VFA version.
Then, we can see how each behave in simulation. And if all together, we try to improve each versions by all the ways possible, i believe we will have learned a lot.
Last edited:
If we're to do a combined effort maybe we should start from scratch. Collectively decide each part of the topology. For example, front end, which VAS(TIS) type, output stage etc... Then we could compare different compensation schemes (which is where in my inexperienced view most of the gains are to be had).
We already have several examples of very good VFA style amps which we can aim at.
We already have several examples of very good VFA style amps which we can aim at.
Since it's still in a theoretical phase I would suggest ideal current sources rather than bloating the individual circuits with real implementations, so the focus is purly on the parts that make up the actual amplifier.
That is shared by many contributors, here, as VFA amps with LTP are majority in the market.
Using an existing CFA, yet known as very good as a base only ask VFA expertize to built the comparison.
And bring some advantages to the VFA version, with the two LTP added transistors, witch can cancel some distortions.
Why not ?
PSRR is part of the CFA's evils. Yet addressed in VSSA. Fair play.
LC reported a listening difference between SMPS and linear, in favor to SMPS. A consequence of this ?
Last edited: