Diff amp input cap for improved phase margin - Page 2 - diyAudio
Go Back   Home > Forums > Amplifiers > Solid State

Solid State Talk all about solid state amplification.

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 21st June 2007, 11:46 AM   #11
AndrewT is offline AndrewT  Scotland
diyAudio Member
 
Join Date: Jul 2004
Location: Scottish Borders
Quote:
Originally posted by john_ellis
Hi pooge

Often in power amps, the effects of low impedances are overlooked. People tend to think of a diff. input pair as being optimum but then feed them with moderate impedances (for example, on the feedback side perhaps a 10k feedback resistor is used with about 330 ohm feedback ratio resistor). You can see the effects of a 330 ohm versus 33 ohm resistor on the frequency response using a simulator. (Of course you can run this simulation without a complete amplifier; and if you use 33 ohms you'll need to change the feedback reistor as well)

Therefore keeping the input stage operating from a low input impedance (which is given by the shunt input capacitor) actually has the effect of increasing the bandwidth of the input stage because the impedance is low. Of course there is the loading effect on any series resistance, which acts to filter the upper frequency response if there is a resistor present. You can achieve similar results by using low input impedances (e.g. 50 ohms) but generally preamps aren't designed to drive such low impedances.
Hi John,
if the input grounding resistor were 20k, setting Zin to this value and the source impedance (with or without a DC blocking capacitor) were <=50ohms, would the source provide effective loading of the LTP? Then only low frequencies would present a high Rs when the reactance of the blocking cap takes on a bigger effect, but this still allows the majority of the audio spectrum and above to operate well.
__________________
regards Andrew T.
  Reply With Quote
Old 21st June 2007, 01:07 PM   #12
GK is offline GK  Australia
Account disabled at member's request
 
Join Date: Jan 2006
Quote:
Originally posted by john_ellis
Hi pooge

Often in power amps, the effects of low impedances are overlooked. People tend to think of a diff. input pair as being optimum but then feed them with moderate impedances (for example, on the feedback side perhaps a 10k feedback resistor is used with about 330 ohm feedback ratio resistor). You can see the effects of a 330 ohm versus 33 ohm resistor on the frequency response using a simulator. (Of course you can run this simulation without a complete amplifier; and if you use 33 ohms you'll need to change the feedback reistor as well)

Therefore keeping the input stage operating from a low input impedance (which is given by the shunt input capacitor) actually has the effect of increasing the bandwidth of the input stage because the impedance is low. Of course there is the loading effect on any series resistance, which acts to filter the upper frequency response if there is a resistor present. You can achieve similar results by using low input impedances (e.g. 50 ohms) but generally preamps aren't designed to drive such low impedances.

Problem is that when you start going much lower than 330 ohms for the feedback resistor to ground, the DC coupling capacitor required (to prevent the LTP voltage offset being multiplied by the closed loop gain) becomes rather large; as does the current forced through it. This can be a problem if you are worried about capacitor distortion.

A better option woud be to cascode the LTP.

Cheers,
Glen
  Reply With Quote
Old 21st June 2007, 01:17 PM   #13
pooge is offline pooge  United States
diyAudio Member
 
Join Date: May 2003
Location: Northern Va.
I was just considering that the situation might arise due to Miller effect, the Miller capacitance forming a LP filter with the source resistance? (If so, then a cascode input could reduce the Miller effect.)

A shunt capacitor at the input would bypass the input resistance as a source, lowering the source impedance at high frequencies. But this would shunt the input signal away from input, so I maybe this isn't the right thinking about improvement of differential bandwidth vs. input filtering.
  Reply With Quote
Old 21st June 2007, 01:26 PM   #14
diyAudio Member
 
Join Date: Sep 2002
Location: Sweden
Draw a small-signal diagram, and it should become apparent. Don't forget to include Cbe of the transistors. You may omit Cbc, though. That is probably not what matters here.

The input capacitor will form a capacitive voltage divider toghether with the Cbe's to help keeping up the voltage over the transistor b-e junctions at high frequencies.
  Reply With Quote
Old 21st June 2007, 01:38 PM   #15
GK is offline GK  Australia
Account disabled at member's request
 
Join Date: Jan 2006
Quote:
Originally posted by pooge
I was just considering that the situation might arise due to Miller effect, the Miller capacitance forming a LP filter with the source resistance? (If so, then a cascode input could reduce the Miller effect.)

Miller capacitance is part of it; much more so if very low noise BJT's or JFETs are used for the LTP differential pair, as such devices typically have quite high collector-base and drain-gate capacitance.
  Reply With Quote
Old 21st June 2007, 01:40 PM   #16
jcx is offline jcx  United States
diyAudio Member
 
Join Date: Feb 2003
Location: ..
spice is only as good as the models, but the principle seems to be easily seen in this sim

R1 is the source input R, easliy 10K for a 50K volume pot
R3 at 1K is more typical of feedback network R
V2 AC source shows the AC gain for the feedback signal through the diff pair to the input of the VAS ( at R2, V(out) )

Click the image to open in full size.

the optimum for these Q models seems to be ~46 pF (white)

~46 pF input gnd/bypass gives the flatest reponse with 48 degrees extra phase margin at 2.6 MHz

but the 2nd order slope gives more phase shift beyond 15 MHz for the peaked response vs the 1p "0" C1 trace (green)
  Reply With Quote
Old 21st June 2007, 06:05 PM   #17
diyAudio Member
 
Join Date: Sep 2006
Hi Andrew T

Using a 20k input resistor is not a problem if the preamp output impedance is low. As you say, the preamp impedance is important on the input side.

There is a way to trick the feedback side with low impedances while keeping the dc path at 10 or 20k too. Use parallel RC networks which keep the ac gain balanced but shift the impedances down with frequency.

For example if you have a 10k and 330 ohm feedback network a 1k resistor in series with a 1 nF capacitor in parallel with the 10k needs a 33 ohm in series with a 30.3 nF capacitor in parallel with the 330 ohm to maintain a flat frequency response. This will improve the diff amp bandwidth over the range 15.9 kHz-159 kHz.

The capacitor and resistor values need to be highly accurate to maintain performance: 5% tolerance or better is needed, which is the main problem.

Some amplifiers have had stability problems in the past because of this change in frequency response with loading. If your design needed a low impedance to be stable, if may have been fine when connected to a preamp but unstable if the preamp were unplugged!
Or the other way round.

cheers
John
  Reply With Quote
Old 21st June 2007, 06:12 PM   #18
diyAudio Member
 
Join Date: Sep 2006
Hi Glen

Some of my best designs have been completely D.C.
They sound very good and as they have no capacitors in the audio path at low frequencies, low impedances aren't a problem. I'm a little surprised that more fully D.C. designs aren't seen, although I agree that the offset voltage may be a little worse.

cheers
John
  Reply With Quote
Old 21st June 2007, 11:57 PM   #19
GK is offline GK  Australia
Account disabled at member's request
 
Join Date: Jan 2006
Quote:
Originally posted by john_ellis
Hi Glen

Some of my best designs have been completely D.C.
They sound very good and as they have no capacitors in the audio path at low frequencies, low impedances aren't a problem. I'm a little surprised that more fully D.C. designs aren't seen, although I agree that the offset voltage may be a little worse.

cheers
John

Hi John.

I agree that complete DC coupling in the feedback path is the best way to go. If your input devices are well matched, the offset voltage at the output can be kept small. However, they do have to be quite well matched (or trimmed). Suppose you have a closed loop gain of 30. An input offset voltage of only 33mV will result in 1V at the output.

33mV isn't a lot, and if you are using BJT's, input bias current needs to be taken into consideration also. Suppose you have an LTP with a tail current of 4mA and your BJT's have a beta of 200. This gives an input bias current of 10uA.
If the non-inverting input is returned to ground via a 10k resistor to give a reasonable input impedance, there will be a voltage of 100mV developed across the 10k input resistor. A 330 ohm resistor on the inverting side to ground will only drop 10uA*330R = 3.3mV, giving a differential offset voltage of ~97mV, just due to input bias currents alone - not taking into account Vbe/hfe variations or tollerances of emitter degeneration resistors (if used).
With a closed loop gain of 30, the output offset voltage will be approximately 3V.

Really tightly matched JFET's for the LTP or a DC servo can be a solution here, but they don't come without other issues also.

Cheers,
Glen
  Reply With Quote
Old 22nd June 2007, 07:09 AM   #20
AndrewT is offline AndrewT  Scotland
diyAudio Member
 
Join Date: Jul 2004
Location: Scottish Borders
Quote:
Originally posted by john_ellis
There is a way to trick the feedback side with low impedances while keeping the dc path at 10 or 20k too. Use parallel RC networks which keep the ac gain balanced but shift the impedances down with frequency.

For example if you have a 10k and 330 ohm feedback network a 1k resistor in series with a 1 nF capacitor in parallel with the 10k needs a 33 ohm in series with a 30.3 nF capacitor in parallel with the 330 ohm to maintain a flat frequency response. This will improve the diff amp bandwidth over the range 15.9 kHz-159 kHz.

The capacitor and resistor values need to be highly accurate to maintain performance: 5% tolerance or better is needed, which is the main problem.

Some amplifiers have had stability problems in the past because of this change in frequency response with loading. If your design needed a low impedance to be stable, if may have been fine when connected to a preamp but unstable if the preamp were unplugged!
Or the other way round.
Hi John,
your description of the parallel RC seems close to what my Sugden power amp has. It keeps blowing a rail fuse (when the wrong size was fitted) when the input RCA is disturbed and sends full rail voltage to the speaker terminal. I have been lucky to discover this when testing and never lost a driver. You have got me wondering now!
__________________
regards Andrew T.
  Reply With Quote

Reply


Hide 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
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
How to measure phase margin of an amplifier? ojg Solid State 96 22nd October 2010 09:38 PM
Leach150 phase margin problem!!! mclarenpingu Solid State 27 3rd March 2008 08:44 AM
Measuring Phase Margin davidallancole Solid State 4 19th December 2007 06:18 AM


New To Site? Need Help?

All times are GMT. The time now is 07:32 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2