Go Back   Home > Forums > >
Home Forums Rules Articles diyAudio Store Blogs Gallery Wiki Register Donations FAQ Calendar Search Today's Posts Mark Forums Read

Solid State Talk all about solid state amplification.

My attempts at a design of a 3 stage amplifier
My attempts at a design of a 3 stage amplifier
Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 10th August 2018, 09:22 AM   #21
edbarx is offline edbarx  Malta
diyAudio Member
 
Join Date: May 2018
Quote:
Originally Posted by BesPav View Post
Oh, no, man, it couldn’t be just added... You must understand precise balance of this system, gain margin and phase margin, influence to the overshoot and rise speed...
That would be exultant for me, but understanding down to minute details is very very time consuming. I am a person, who due to wanting to understand down to details, takes far longer than others to understand.

My knowledge of electronics still lacks the quantitative aspect of filters and, it seems, amplifier stages. The latter, due to their intrinsic capacitance influence phase and gain as frequency is changed. However, the mathematics is extremely cumbersome if one goes the 'from basic principles' route. A simple RC filter becomes a daunting complex number calculation as I will briefly illustrate:

Let us assume we have an RC network fed by V at f Hertz.
The impedance of the whole network is:

Z = R + Xc

Let us assume we are measuring the filtered signal across C.

So, the quotient Q = Xc/(R + Xc)

Let us now introduce complex numbers in the expression:
Using p to stand for pi = 3.142 approx

Q = [ -j/(2pfC) ] / [ R - j/(2pfC) ]

To get rid of the imaginary part in the denominator, multiply both numerator and denominator by the complex conjugate of the denominator.

Q = { [ j/(2pfC) ]*[ R + j/(2pfC) ] / [ R^2 + 1/ ((2pfC)^2) ]

For clarity let us represent the denominator by D.

So, Q = (1/D)*[ jR/(2pfC) - 1/((2pfC)^2) ]

Most circuits require more complex LCR networks giving rise to longer and more complex formulae.

The required calculation complexity is what is keeping me back from developing the skills necessary for successful circuit design.

P.S.
I installed LTSpice on my Windows partition but I need to ask how I can make spice recognize schematics made with KiCad?

Thanks for everyone's patience.
  Reply With Quote
Old 10th August 2018, 09:51 AM   #22
suzyj is offline suzyj  Australia
diyAudio Member
 
suzyj's Avatar
 
Join Date: Jan 2006
Location: Western Australia
The rapidly increasing complexity of the math is exactly why we use simulation.

There may be a way to export a net list from KiCAD that’s useable in LTspice, but I generally just draw the schematic twice. The spice version can be simplified some, omitting filter caps and such forth.
__________________
http://www.suzyj.net
  Reply With Quote
Old 10th August 2018, 09:54 AM   #23
jazbo8 is offline jazbo8
diyAudio Moderator
 
jazbo8's Avatar
 
Join Date: Jan 2011
Location: In Transient
My attempts at a design of a 3 stage amplifier
Quote:
Originally Posted by edbarx View Post
I installed LTSpice on my Windows partition but I need to ask how I can make spice recognize schematics made with KiCad?
Don't waste your time...
  Reply With Quote
Old 10th August 2018, 11:23 AM   #24
Mooly is offline Mooly  United Kingdom
diyAudio Moderator
 
Mooly's Avatar
 
Join Date: Sep 2007
My attempts at a design of a 3 stage amplifier
Quote:
Originally Posted by edbarx View Post
P.S.
I installed LTSpice on my Windows partition but I need to ask how I can make spice recognize schematics made with KiCad?

Thanks for everyone's patience.

Here is the simulation file in case you want to have a play. I changed a few things to get a reasonable performance as there were severe performance limitations when squarewave testing.

You will need to rename the text file of models and add it to the simulation in place of the one I used.

Good luck
Attached Images
File Type: jpg Amp 7.JPG (253.8 KB, 94 views)
File Type: jpg Amp 8.JPG (261.9 KB, 92 views)
Attached Files
File Type: asc Draft1.asc (10.7 KB, 7 views)
File Type: zip discrete models.zip (86.4 KB, 7 views)
  Reply With Quote
Old 10th August 2018, 02:15 PM   #25
edbarx is offline edbarx  Malta
diyAudio Member
 
Join Date: May 2018
I succeeded to load the schematic in LTSpice and to run a simulation but I only got a list of voltages and currents. I couldn't find how to produce graphs. Nevertheless, this is the teething problem stage of using complex software for the first time.
  Reply With Quote
Old 10th August 2018, 03:08 PM   #26
edbarx is offline edbarx  Malta
diyAudio Member
 
Join Date: May 2018
Finally, I got the square wave plot.

From the previous simulation, we found that the frequency response extends well into ultrasonic frequencies to several hundred kilohertz. The square wave distortion suggests some harmonics were not amplified, or better, that the bandwidth is not wide enough. For an audio amplifier this cannot be what is causing the distortion of the square wave. I think, it is a poor dV/dt, slew rate. So, I must redesign the VAS with aim of improving the slew rate. The Miller Capacitance is too low to be the cause. I am tempted to think, the MJE340 and MJE350 are the cause. Hm, if I use a cascode to prevent both Base-Collector junction capacitances from experiencing voltage changes, I might get a better dV/dt (slew rate).

One itching question is: is a square waveform with T = 40 micro-seconds of the right frequency? Such a period belongs to a frequency of 25kHz. So, with 'our' huge bandwidth, that should be space enough for up to the 16th harmonic (assuming a bandwidth of 400kHz). Sixteen harmonics give rise to a better approximation for the square wave. So, I conclude it must be the slew rate that is degrading performance.
  Reply With Quote
Old 10th August 2018, 05:52 PM   #27
Mooly is offline Mooly  United Kingdom
diyAudio Moderator
 
Mooly's Avatar
 
Join Date: Sep 2007
My attempts at a design of a 3 stage amplifier
The sim file I attached has the front MJE devices replaced with smaller and faster ones.

As well as looking at the output voltage, also look at the input voltage across the filter. That has the (wanted) effect of reducing slew rate. I also added a small cap across the feedback resistor to introduce a small amount of phase lead compensation.

If we go all out and remove the above caps we appear to get a better square wave response but I would caution against trying to squeeze the last bit of rise time this way as it will be at the expense of stability.

So here is the slew rate under these conditions. I've also added an 8 ohm resistive load.

So over 1uS the amp slews at approximately 22v/us which is pretty decent in practice.

The last image is with the input filter cap back in place and looking at the voltage across this cap (the input to the amplifier.
Attached Images
File Type: jpg Amp 8.JPG (248.9 KB, 82 views)
File Type: jpg Amp 9.JPG (178.2 KB, 76 views)
File Type: jpg Amp 10.JPG (250.3 KB, 73 views)
  Reply With Quote
Old 11th August 2018, 07:30 AM   #28
edbarx is offline edbarx  Malta
diyAudio Member
 
Join Date: May 2018
Thanks for helping me. I am excited seeing my amplifier circuit evolving.

Till now, there is no overload current protection. I am adding two transistors to divert the driver current away from the output transistor's bases. I have seen this circuit online. Adding a light emitting diode indicator would inform users that their amplifier is reaching overload point.

Another idea is to use an automatic gain reducer when the output nears maximum preventing overload.

I think, it may be done using an opto-coupler. The diode sensor section would be connected across one 0.22 Ohm, those found on the emitters of the output transistors, while the output transistor would shunt the biasing input resistor at the base of the differential pair.

Additionally, a clip detector is also to be added. Mathematically, a clip is when dV/dt = 0 and stays like for a period instead of assuming zero value instantaneously. A small capacitor feeking the base of a transistor is sensitive to dV/dt since Q = CV and hence, dQ = CdV which leads to dQ/dt = C*dV/dt. The latter is what we want. This means, the output of a transistor connected in that way goes high when the output clips. To sense non-zero durations a capacitor connected across the output resistor, would allow the output to rise with a delay. This arrangement, should also sense clips that do not take place at the outputs maximum. Such clips are also fatal for loadspeaker coils as there is no reactive components at clips irrespective of whether they take place at maximum output voltage or not.
  Reply With Quote
Old 11th August 2018, 08:21 AM   #29
Mooly is offline Mooly  United Kingdom
diyAudio Moderator
 
Mooly's Avatar
 
Join Date: Sep 2007
My attempts at a design of a 3 stage amplifier
The easiest way to add current limiting would be to include a low value resistor in the collector path of the output devices and sense volt drop across those. Using an opto and sensing across the 0.22 ohm is possible but would add to the complexity. An LED in an optocoupler needs around 1.7 volts before it will conduct. That implies a current of around 7 amps, and you have three pairs... you could add sophisticated electronics to derive an LED control voltage that was referenced to each rail (as the 0.22 ohms are) but the circuitry would be as complex as the main amp.

Detecting clipping can be done with an opamp monitoring the voltage difference between the - and + inputs of the amplifier. When clipping occurs the amp losses control and a big voltage difference will occur between the two inputs. An opamp can detect that and light an LED but you need extra circuitry to make the LED light for a reasonable time. Clipping for 1ms would be barely visible from the LED, so you need a pulse stretcher such as a monostable to light the LED for say 2 seconds each time a clip event is detected.
  Reply With Quote
Old 11th August 2018, 05:21 PM   #30
Pingrs is offline Pingrs  United Kingdom
diyAudio Member
 
Join Date: Jul 2005
Location: NW England
Check Bonsai's site for Michael Kiwanuka's article on VI limiting.

Brian
  Reply With Quote

Reply


My attempts at a design of a 3 stage amplifierHide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
Amplifier design: how to connect a stage with the next one? Bricolo Solid State 11 10th August 2018 07:34 AM
Pre-amplifier circuit not working despite attempts to fix it Rudi1441 Analog Line Level 5 18th July 2016 02:35 PM
Phono stage design considerations part 1: choosing 1st stage tube Joshua_G Tubes / Valves 207 5th March 2016 12:07 PM
LF Line Level Stage design to preceed class D amplifier undefinedza Analog Line Level 13 18th September 2014 08:26 AM
how to design mosfet amplifier stage? pzung Solid State 2 22nd January 2010 06:58 PM


New To Site? Need Help?

All times are GMT. The time now is 04:47 AM.


Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 14.29%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Copyright ©1999-2018 diyAudio
Wiki