Wow! So many years, and the thread is still here!
Still did not figure out that it is better to limit a frequency band before any active components than to fight for a slew rate above the sky? ;-)
Not any more about SR-limiting or slew induced distortion.... It appears we can expect certain other important characteristics to follow from a high SR design -- such as might be found in a CFA design....... flat, constant, low output Z vs freq, for example.
THx-RNMarsh
Talking from the perspective of a loudspeaker driver, low output Z is only relevant at the bottom end because of the sharp impedance rise around resonance. At higher frequencies, drivers typically show a bit less distortion with current drive, i.e. a higher output Z. I even made an amp with a split feedback loop that provides for a low Z at the low end, and a high Z at the top end, just for this reason, but I found the advantage to be too small to bother. But it was there. Bruno Putzeys uses a similar scheme in his Kii. For any effect to show up, the Zout had to vary from well below .1 Ohm to many 10's of Ohms. Way bigger than any Zout variation/F you might experience in any decent amplifier.
In other words, if your amplifier is used to drive loudspeakers, the kind of rising impedance one might see because of lower loop gain at higher frequencies is of no consequence at all, and if there is any effect, it would only be beneficial.
yes... so. look at a typical high gnfb (typical VFA) power amp output Z and then add series Ls to stabilize it. Speed that amp up and the fall-out is a wider bandwidth and flat Z to higher freq. Sure it simplification but you can fill in the blanks and design for not needing Ls. I havent found any stability issues in my test system with Damir's CFA amp when not using Ls. Others?
A higher Z out is not the same as a changing Zo vs freq. Include phase. But also I havent used dynamic drivers in many years until a good set came along (JBL). So what i notice may not apply to many others.
What would be needed to design stable amp without effective or real Ls. ?? CFA is easier IMO.
THx-RNMarsh
Last edited:
The main reason to have almost constant distortion throughout whole audio frequencies and for all power level is the compensation used in this amp. Output impedance is also result of used compensation (and it's getting flat LG up to 20 kHz). This is not because CFA was used, but somehow I found simpler to introduce it in CFA with excellent phase and gain margin.
Damir
It is always simpler with CFA is my long experience with topology. Not needing an Ls always as part of stability with load helps over-all, also. Would you talk about constant Zo vs freq. using CFA, pls. Do you find it easier, harder or same to have constant (flat line) Zo vs Freq with CFA?
THx-RNMarsh
Last edited:
This is simulated output impedance of the 200W CFA.
It is also result of relatively high and flat Loop Gain.
Damir
Attachments
Looks good. It is easier IMO to get wide constant OLG vs freq with CFA.
I am having two more built BTW (three total) in next couple weeks. Fine tuning the mechanical layout and ventilation, air flow to keep inside chassis lower temp.
THx-RNMarsh
My short term goal is to build 20 of them. I have an investor put up the money..... he is actually a friend and managing director of the company (and 1/4 owner). Damir can get some money and we get a pair for 'free' --- just my time and connections. I dont count the cost to travel as i would be in Bangkok anyway for other reasons (like my family in Nepal). After the 20 are gone, company might make more depending on how fast they sold etc.
Three are needed to make minor changes and test/evaluate and be sure of consistency unit to unit and to go out to local reviewers to help get the first batch sold. I am facilitating it all plus input on operations, marketing, testing, direction (like a CEO would do). I have no salary from this company nor investment other than my time and energy for my friends.
-RNM
Last edited:
I only have qa400 and not 401 here in Bangkok.... if that works, I can try it while in Asia. THx-RNM
George,
Thanks, as well to Pavel, hadn't seen these pages. I also did a bit of measuring. To that end, I made an amplifier with variable output Z to determine if an optimum point exists. It does.
Some conclusions:
- the effect is never large (and loudspeaker distortion measurements outside an anechoic room have to be taken with a grain of salt).
- the effect is frequency dependent. So, the same driver may require predominantly current drive in one area, and predominantly voltage drive in another.
- all drivers are different. Wildly so.
Hence my earlier expressed opinion that a driver and its amplification should be seen as a single subunit to be optimized.
If interest exist in the schematic of the little variable Zout amp based on an LM3886 I will dig it up. The toy itself is at LAHQ I believe, so can't post pictures.
Thanks, as well to Pavel, hadn't seen these pages. I also did a bit of measuring. To that end, I made an amplifier with variable output Z to determine if an optimum point exists. It does.
Some conclusions:
- the effect is never large (and loudspeaker distortion measurements outside an anechoic room have to be taken with a grain of salt).
- the effect is frequency dependent. So, the same driver may require predominantly current drive in one area, and predominantly voltage drive in another.
- all drivers are different. Wildly so.
Hence my earlier expressed opinion that a driver and its amplification should be seen as a single subunit to be optimized.
If interest exist in the schematic of the little variable Zout amp based on an LM3886 I will dig it up. The toy itself is at LAHQ I believe, so can't post pictures.
All you need is a PC w/ sound (pretty much any PC made in the last 20 years). No measuring tools needed. Just a willingness to test with some music you know and learn.
Demian
In fact, good headphones are ideal for this application, even if they're not normally your cup of tea. Removes a lot of confounding factors (modicum of isolation, consistent local environment as long as you're carful to position the phones on your ear in the same place, tend towards lower distortion inherently) so you can tell where you sit/stand with regards to the introduced distortion levels.
For me, songs with simple arrangements and clearly separated vocals highlight the issue better than more complex music (masking?). Fortunately, I like a lot of folk music, and that tends to fit the bill.
Thanks for the link, by the way, Demian.
I asked Matt about adding different transfer curves but no response yet. This is definitely a demo of how audible crossover distortion can be. There are a lot of permutations to explore.
George, PMA more power to you not a lot of interest in real engineering here these days just fairy tales.
Dan, read my post, no general conclusions can be drawn from measurements of a specific driver, they are all different. For one thing, aluminum voice coils respond much better to current drive than non-conductive vc's. This may explain the difference in the measurements between bass, mid and tweeter Pavel made.
- Status
- Not open for further replies.