Another very fast CFA concept

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi miro1360,

This looks really good, I like the concept, beautifully simple and fully symmetric. But I think this will be some effort to match the transistors well to minimize offset. Also I think thermal tracking and BIASing may not be as trivial as the circuit suggest since we would have to use independent components on a PCB vs. everything being on one chip. A short circuit protection would need to be added as well.
I guess with decent transistors this could perform very if not exceptionally well as your simulation suggests. Perhaps if you can, repeat the sim with a square wave, this will be most telling how the edges will look like. Eventually some frequency compensation needs to be added depending on selected transistors and perhaps layout parasitics.

On another thread there was just various CCS discussed which could be used here.

Cheers
 
hi kct,

perfect to hear from somebody experienced that this can work well ... you heard how TPA6120 sounds (in headphones)? I heard it and build simple buffer for speakers and sound of it is amazing (it is why I was looking for internal schematic of TPA long time and finally found something close to it).

Well, with LtSpice and advanced simulations, better if somebody more experienced look into that (in attachment is simple asc).

From this simple schematic without CCS

I did 1kHz 1V square:
https://ctrlv.cz/shots/2017/09/22/fXD6.png

Or 1kHz 27V:
https://ctrlv.cz/shots/2017/09/22/WZ0D.png

100kHz 1V:
https://ctrlv.cz/shots/2017/09/22/sOme.png

100kHz about 20V:
https://ctrlv.cz/shots/2017/09/22/buS1.png


some AC analysis:
https://ctrlv.cz/shots/2017/09/22/v0qc.png

these all pictures are with +-63V and high gain about 28 ...

most important is to build something and try how it sounds (at least for me is sound most important) :D probably start with smaller power out ...
 

Attachments

  • TC1_01_HV.asc
    11.6 KB · Views: 155
  • TC1_01.asc
    11.6 KB · Views: 102
Hi miro1360,

yes, I heard the TPA chip performance in the headphone amp. Combined with a lovely Denon AH-D5000 headphones the sound is beautifully crystal clear and very defined, incredible stereo image as well. No hint of a hiss or noise.

I might have some limited experience and I can do some stuff but by no means I would want to claim to be an amplifier designer.;)

A while ago I did a power amp project that yielded the same crispness and clarity, I got used to with the headphones, now with my vintage pioneer DDS speakers and if needed cranked to 100W+. Altho turns out mostly I listen to power levels at or below 1W ! At 10W it is starting to get uncomfortably loud.:D
When watching action movies it might peak briefly at about 20W which shakes and resonates all sorts of things in my house. For me that power level suffices, others need a 1kW sub-woofer to reach they're comfort level... :D

How much power do you feel you would want to go for with this proposed design ?

How much money do you feel the amp circuit should cost ? (excluding power supply and heatsink)

How much money spend do you feel would acceptable for additional components that must be bought for matching purpose ?

From the square wave simulation images you did, I am guesstimating a rise/fall time of about 100ns. This is exceptionally good in my books. If such a rise/fall time is truly achieved with a build it would yield certainly the very crisp and clear sound, very comparable to the TPA chip performance. I would expect this power amplifier to measure very very well THD and TIM. Maybe not quite as good as a TPA chip but very very good for any power amp.
However, it takes two to tango, the amp is one thing but the speaker connected to it is another. I lean towards speakers equipped with ribbon tweeters hence I acquired vintage pioneer speakers. Of course most people would rater want something newer than vintage 80s stuff, so I would recommend ADAM audio passives or CANTON passives. Of course there are higher end speakers with diamond dome tweeters / mids and other cool trimmings if money is not really a problem...

In a few days, I might have lots of time at my hands, perhaps I might play with my sim a bit....

But let's get a few more opinions before starting to solder anything.

Cheers
 
This circuit should work well. Have done extensive sims on a circuit that is based on the AD844. Only real difference being the front end. Will be very fast. You can utilise both shunt and miller compensation but have had no success with two pole comp. Maybe due to low ish open loop gain.

Paul
 
kct:

I tried TPA only with AKG K701 and BD DT880 and exactly as you said.
I mounted it also here as headphones amp:
easyDAC = NOS R-2R AD1862 DAC + Headphone Amp

How much power. I think for me is 10W enough - my another thread where I gave buffer after TPA have this power and it is enough for speakers. Here is most simplified version which I am going test:
https://ctrlv.cz/shots/2017/09/22/h7fO.png (but this is another thread)
How much power speakers need can be dependent also on how much sensitivity speakers have.

The more power amp have, the more it cost money, because case is bigger, heatsink, supply, caps for higher voltage. Somebody needs 200W, I need funny 10W :D ... this CFA amp can be in range 10-100W

About money, this depends on tuning :D because in most simplified version it can be on 5x5cm board, but with tuned version with all compensations, DC protection on board, 16x10cm board can be good :D but this price goes higher rapidly. For basic test 5x5cm board can be enough (some DC protection externally).

Matching components, this can be serious problem, because these meters cost tons of money. This is why I was thinking about simplest DC offset reduction with multitrim and voltage divider :D

Speakers, this is another dimension which cost also a lot of money. My speakers are very simple: GrabCAD - CAD library .... but sometimes in future I do something bigger and better.
-----------------------
mcd99:
AD844 is also very fast thing
-----------------------
Havenwood:
It depends, how it is configured, because this circuit can have both, non-inverting input and inverting input.
Schematic is now configured as "positive/non-inverting input" and from theory, it can have high input impedance. I tested it in simulation with 10k and 100k potentiometers and I have not found differences.
Alternatively it can have inverting input for relatively low impedance.
-----------------------
chalky:
I think, A40/A33 have different input stages and are more complicated for me, especially in PCB design, they are massive
 
CFA amplifier

Hi miro1360,

I used virtually identical complementary current mirrors on the first stage and second stage of a power amplifier design during the mid 1990s. The output stage was configured with Hitachi complementary lateral mosfets. Offset drift was considerable and very difficult to minimise due to the use of discrete transistors in all the current mirrors. I had intended to sort out the offset issue but other more pressing matters overtook my life at the time.

A 250W version of the power amplifier was used at the HiFi News HiFi show in 1996 at the Ramada Inn adjacent to Heathrow Airport. It was certainly a very fast and transparent amplifier and it drove the Analysis ribbon speakers very well. I also used a lower power 100W version for an acoustically transparent PA system for folk musicians playing various venues in Cumbria, UK. The musicians were very pleased with the sound quality and the clarity of voice and instruments using the system.

It is safe to say that this circuit configuration has potential and is worth pursuing. Good luck with your endeavours.

Regards,
Paul
 
Hi miro1360,

well your DAC looks great. I did something very similar you can find some images of my build here. I used the TPA6120A2.

kct Gallery - My Photo Gallery

Well, I was bored today and messed around with my sim, this is what I am getting. Altho the amp easily performs well into the 150kHz range I am not overly impressed with THD. I don't have the proper spice models for the driver and end stage transistors so I used something generic.
 

Attachments

  • VFA_4_Rev.8.jpg
    VFA_4_Rev.8.jpg
    348.3 KB · Views: 1,032
  • VFA_3.jpg
    VFA_3.jpg
    482.9 KB · Views: 1,017
Hi miro,

well, I messed around some more and achieved a slight improvement with THD. :)
But my simulation result are not nearly as good as what you get. :(
I guess my transistor models are perhaps not good enough.
 

Attachments

  • VFA_4_Rev.9.jpg
    VFA_4_Rev.9.jpg
    382.7 KB · Views: 834
possible differential input for this CFA
50W/8R, modification for SMD input transistors (BC856/46)

This means your input impedance approx 2.2kohms, too low.
Impedance should be at least 22k. Peferablly 68k-100k

EDIT: You'd also want to change gain of tarnsistors, and current and mismatch the amp on purpourse in circuit simulator.
If the operating point changes allot with small gain changes or current changes. That indicates that your amp may not work when partically built

I had considered building this circuit, hower this circuit seems to have a low impeance and has the input on emitter. Therfore I gave up on the idea in building this circuit.
I also considered using its vas stage. But I don't like current mirrors in amps.

at q7 q8. you can chagne the diodes to a vbe multiplier or install any type of current gain stage you want. It would work. If the gain stage operates properly
Presonally I can think of a very practicall way of adjusting speaker dc voltage to zero. without using your complicated method having a+-1v at the nfb part. However I do not want to give it away, I do not want to see any one compete with the amp I sell If I ever do sell one.
 
Last edited:
possible differential input for this CFA
50W/8R, modification for SMD input transistors (BC856/46)
Guys if you want my method for dc voltage adjustment. make r16 r14 adjustible with pot.
Or use a adjustible constant current source in place of r16 r14
The orignal circuit is designed for high frequency, I do not see it being hifi at all.
 
Last edited:
Hi Sajti,

well it seem to give better THD with higher values.
I will try with lower values again and see if this gets better yet.
This should be becasue of more emitter feedback, emitter resisance larger means the gain changes less when transistor internal impedance changes due to change in transistor emitter current.

Increaseing bias current has a similar effect. internal resistance is 25mv/emitter current.
If current higher impedance internal small, changes in internal resistance caused by change in current is less
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.