DIY-AMP <ZAS> Low THD & IM high bandwith

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

I see no problem to post the full schematic of my amp. the amplifier topology is not new, i build the first prototyps in 1990 with the good old NE5534A operational amplifier. Using an R&S audio-analyser in the last 2 years, I optimized distortion and bandwith by changing some values and placing two additional current mirros. i also spent a lot of time and money with secondary problems like qiescent current stability and semiconductor death by oszillation.
For best results, a good pcb- and power supply design is very important. i think i found it.....

FEEDBACK WELCOME !!
 

Attachments

  • zas.pdf
    43.1 KB · Views: 419
I had a 200W PA design, with the topology something like that.
I had never build my design... Maybe I will...

Regards,

PS:: Schema of the last version attached.
 

Attachments

  • edl_amp_10.gif
    edl_amp_10.gif
    17.1 KB · Views: 969
Re: DIY-AMP <ZAS> Low THD & IM high bandwith

murphy said:
Is someone interessed in building an fast amplifier with very high bandwith and really low distortion (THD and IM) at moderable costs ?

apparently not as many people as you might think



This amp is based on real measurements , it is no an result of computer simulations !!!

of course not, few op amp macromodels get supply terminal currents right - you can find my comments on the ad8610 model in another thread
 
djk said:
edl , I like your design.

The TL071 will suffer phase reversal when the amplifier is driven into clipping. Newer opamps generally don't do this.

Hello,

Thanks!
You're right about the OPAMP.
In the prototype I wanted to use NE5534, with about 22-33pF compensation. (In the schematic I forgot to change the TL072 to NE5534...)
There is one more "fault" in my drawing, a paralel RC-network is mising in the output of the OPAMP.

I will attach the new schema!

Regards,

(Hey everybody, if there is indication of interest about my design, I assemble a prototype! I think it would be adequate as HIFI-amp, not only as PA-amp.)
 
Updated schema atteched.

The design is only an idea, I've only simulated it in my head and with paper+pencil, no software used...

Murphy, sorry for using your thread, a new one will be opened in matter of interest.
 

Attachments

  • edl_amp_11.gif
    edl_amp_11.gif
    16.9 KB · Views: 704
Something like this...

(About caps: "F" means foil, "K" means ceramic, "E" means elko, "BPE" means bipolar elko. Of course I suggest to use foil types instead of the creamics too, but they are not easy to get in every electro-shop here...)
 

Attachments

  • edl_amp_12.gif
    edl_amp_12.gif
    17 KB · Views: 664
Hi Murphy,

I'm always interested in high-bandwidth designs! Your amp seems to measure extremely well.

I hope you won't mind answering a few dumb questions though, I'm having a bit of difficulty getting my head around how it actually works!

Obviously, I can see that the current mirrors reflect the current drawn by the op-amp to the Vbe multiplier and output stage, but as far as AC analysis of the circuit goes, I don't really know where to start :(

I have a few other questions:

1. What are the benefits of using a Sziklai connected Vbe multiplier?

2. Purpose of C7?

3. Purpose of D2 and D3?

4. The feedback path has no DC blocking capacitor. How do you ensure a small DC offset at the output?
 
thanh said:
HI edl ! your amp seem to be a VFB amp while murphy'amp is a CFB amp . Your amp have more stage than murphy' amp


Hello dear Thanh,

Yeah, you're right!
My amp design and Murphy's amp aren't same.
As I wrote:
I had a 200W PA design, with the topology something like that.

Murphy, again I apologise for the threadstealing, I will open a new one, when I will start to assemble a module of my design.
 
my differences to other concepts

Ok, i try to explain what is the small difference to other concepts of amplifiers.
Everybody knows VGA and CFA designs. this is a mixture between them, it is 90% current feedback and the rest is voltage feedback.
Why this ?
Current feedback path is realised by R17 , C6 , R10, R12 with the main advantage of large bandwith and high slewrate.
But in this case an additional voltage feedback path is inserted by R13,R14 R16 (0,1% restistors should be used for best channel matching) which limits the gain of the amp to a value of 19,1 dB.
During the last year, i tested many relations between Current feedback and Voltage Feedback, measured distortion spectrum and heard a lot of natural instruments. i think that i have found the maximum quality point. it is not the point of best measurement values, of course you could improve them a little bit in the frequency region at 20kHz, but only at cost of phase reserve which means you have more risk of high frequency oszillations. The sziclai pair is a little bit more sensitive for HF-oszillation.
nevertheless i prefer the sziclai pair because of the very low output resistance of this stage. that results in lower distortion especially with real loads. With this values i choosed, the amp is in this layout stable with capazitive loads, even when you connect high capazitive and low inductive cables without speaker. in many cases that causes HF-oszillation greater than 1MHz. At my opinion this is the worst case for an amplifier with large bandwith. Surely this amp is stable too at normal labor load (2.2uF paralleled with 8 ohms). No problems occured even at 20kHz.
 
Hi harry dymond !

You had some additional question:

I didn´t want to use an integrator to compensate the offset voltage or an Cap in the feedback path.
My solution to prevent DC voltage at the output is using 0,1% resistors in the ciurrent mirror (R2,R3,R23,R24). Their function is to improve matching of the current mirrors by degeneration. This was very important for the DC-offset. Additional i used AD8610 which has low offset and the input resistant of the circuit is low too. By doing so i had in all cases offset below 10mV. If you match

D2 and D3 are implemented for the case of overdriving/clipping to result fast recovery.

C7 is a trim-cap to adjust frequence response/phase reserve. i adjusted it by viewing 100kHz rectangular signals to a response without overshot. Alternative 2 paralleled 33pF 1% do this job.
 
Hi Murphy,

Thanks for taking the time to reply to my questions.

Could you possibly suggest a good source of information that explains the basic working principle of this topology?

One thing that confuses me about dual current feedback and voltage feedback designs is this:

You provide a reference signal to your amplifier. The control loop will now try and replicate this reference signal in your load, but amplified. In the case of a current feedback loop, the amplifier is trying to replicate the reference signal as a current in the load; in the case of a voltage feedback loop, the amplifier is trying to replicate the reference signal as a voltage across the load.

For a dual-feedback system driving a resistive load, this is no problem, as the current and voltage are in phase with each other. Surely when you switch to a reactive load, the current and voltage are out of phase and the two feedback loops are fighting against each other?
 
Feedback design

@djk: thanks for the links about patents concerning current feedback design. i build this design , it worked well , but the damping factor was not realy high enough and the IGBTs had to be selected to get low distortions. the main difference i see is that Mark Alexander used a 100% current feedback topology and my one is a mixture between voltage feedback and CF.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.