MultiLoop Amp Output impedance

Status
Not open for further replies.
Jump to Latest
#3
A 100 ohm resistor between the buffer and op amp will have no bearing on output impedance, regardless of feedback configuration. It is an anti-parasitic scheme, intended to isolate the output stage of the op amp from induced signals being introduced into the op amp. And it works; and it is a good idea.

If I have a long circuit board trace between the op amp and buffer, I split the resistor (47+ 47 ohm) and put one resistor right on the op amp output, and the other resistor right on the buffer input. This effectively isolates both parts of the amplifier from induced signals.

Likewise the 10 ohm output impedance of the 634 will be greatly reduced by the global feedback. If your circuit uses large loop gain (heavy feedback) it will be significantly under 1 ohm.
 
#6
R7 could be replaced with a Zobel instead with similar aid to stability

its not really isolating the buf634 input C but rather damping any squirrely Z from the buf634's 100% internal feedback that can give a region of negative input Z way above the op amp GBW

you can look up EF negative input Z and base stopper - which can also sometimes be functionally replaced with Zobel to gnd (or collector/drain rail for a shorter loop)
 
#12
I guess no harm in going to 10k pot as far as your signal source is happy with that.
Changing R2 to 100k is possible, but it will affect your lower cut off frequency. You can increase proportionally C1 to keep it similar.
Not sure what was the exact purpose of R1 in this case. May be some sort of protection of opamp input in case of heavy input overload. See first of all if this sort of situation possible in your case.
 
#13
I believe R1 is a ill thought out attempt to even up source Z to the op amp +,- inputs

clearly misses the fact the volume pot Z varies by much more


the R5,R6 "inner loop" is also a conceit inspired by Otala's "flat open loop gain over the full audio range" prescription
also a logical fail, doesn't work as intended in this circuit even if Otala was correct - which he wasn't

I quite agree that the ccs pull down is not likely helpful when driving the BUF634 input


Tangent and Headwize did introduce many to DIY, lots of documentation, encouragement, support - but I'm afraid some "cargo cult" design features became popular despite poor EE justification
 
Last edited:
#15
jcx said:
I believe R1 is a ill thought out attempt to even up source Z to the op amp +,- inputs

clearly misses the fact the volume pot Z varies by much more


the R5,R6 "inner loop" is also a conceit inspired by Otala's "flat open loop gain over the full audio range" prescription
also a logical fail, doesn't work as intended in this circuit even if Otala was correct - which he wasn't

I quite agree that the ccs pull down is not likely helpful when driving the BUF634 input


Tangent and Headwize did introduce many to DIY, lots of documentation, encouragement, support - but I'm afraid some "cargo cult" design features became popular despite poor EE justification
Click to expand...

The purpose of the inner loop is to reduce the op amp bandwidth to avoid oscillations.

Regards.
Alfredo Mendiola Loyola
Lima, Peru
 
#18
It is roughly correct, except that I see a different gain. Rather 26dB, or round about it.

Don't forget about the difference between a bandwidth and a gain-bandwidth. While they are related, they are not the same 🙂

Added:

Here is example how R6 affects loop gain stepping from 1e8 Ohm to 499kOhm. Used voltage sources instead of opamap.
 

Attachments

  • loop_gain_1.png
    loop_gain_1.png
    48.3 KB · Views: 148
Last edited:
#19
amendiola said:
I have seen several multiloop amp configuration where there is a 100 ohm resistor between the op amp and the buffer.
Click to expand...

Curious about this and just did some datasheet diving. 🙂 Along with the other reasons there is an internal diode clamp of about 3V from input to output in the buf634. Turns out the LME49600 has exactly the same thing. I've missed that about both chips.

If for some reason the buffer output can't follow the input - which could happen with op-amp drive if someone doesn't get a solder joint right on a DIY build - the buffer input will conduct hard. Both buffer chips have an internal 200R input resistor to help mitigate the current flow, but if the rails are 12Vdc and the drive op-amp shoots up to the rail due to a broken feedback loop, that would be 12V/200R = 60mA, enough to possibly fry the op-amp's output stage if the op-amp is one without short circuit protection.

Looks like having a "safety" 750R-or-so resistor in series with the buffer input might not be a bad idea. That would limit build accidents to 16mA on the op amp output. I don't see any spec for the input capacitance on either the buf634 or LME49600 datasheets, but guessing 100pF that would put the corner freq of the filter formed at 1.5MHz, way out of audio range, so it shouldn't matter.
 
Last edited:
Status
Not open for further replies.
Top Bottom