The LM1875 for best

[kokoriantz]

To function as parafeed inverter, AD8008 is a must. The power supply of 42V is high for all low distortion opamp, forcibly it needs extra regulated supplies. I am not troubling myself for cheap stuff but what best can be pulled out from LM1875 and best components it will get. It looks Bob Cordell's choice of LM4562, adopted by FF, is very good. I am thinking to put either the inverter as single AD8007 and the input opamp on socket to let the user choose his opamp, or have the input as AD8008 but straps or switches allows the user to disconnect internal opamp and plug in his choice.
The OPA2134 is out of the list, too slow.
Each LM1875 cost minimum 2$ if bought by 10pcs. the AD8008 is about 1$. The most cost is the power supply comparatively. Instead of having an armada of audio expensive capacitors, the low impedance supply is cheaper and higher performance.

[kokoriantz]

This is the spectrum of AD8008 with 1.5V input unity gain 1khz.

[kokoriantz]

I already liked it, now I am in love with AD826. The opamp has OLBW of 20khz instead of few as others. What I discover by measuring its unadvertised distortion, 0.00001% at 1khz and 0.0001% at 10khz 1.5V. here is the spectrum.

[mjona]

Originally Posted by kokoriantz

To function as parafeed inverter, AD8008 is a must. The power supply of 42V is high for all low distortion opamp, forcibly it needs extra regulated supplies. I am not troubling myself for cheap stuff but what best can be pulled out from LM1875 and best components it will get. It looks Bob Cordell's choice of LM4562, adopted by FF, is very good. I am thinking to put either the inverter as single AD8007 and the input opamp on socket to let the user choose his opamp, or have the input as AD8008 but straps or switches allows the user to disconnect internal opamp and plug in his choice.
The OPA2134 is out of the list, too slow.
Each LM1875 cost minimum 2$ if bought by 10pcs. the AD8008 is about 1$. The most cost is the power supply comparatively. Instead of having an armada of audio expensive capacitors, the low impedance supply is cheaper and higher performance.

Don't use AD8008 it will slew (1000V/uS) to the rail voltage before the LM1875 can respond. The limit for the latter is 20V/uS which is within the capacity of AD711/712 as in the composite article linked in post 27 and OPA2134 and LM4562.

You seem to have bypassed the composite article.

[kokoriantz]

If I make the amp to receive other opamps than the user decides what to use. I never walk on roads that lead to Rome, I always find myself my way. The slew rate for me must be highest possible to correct in time domain the output stage. Of course in saturation things go wrong but a pair of zeners as FF find the solution is enough.
See FF review LM1875 in parallel configuration and used in a composite amplifier.

[kokoriantz]

I was amazed by seeing that the LM1875 with gain of 9.2 can still receive extra NFB without oscillating and reach 5Mhz. I looked upon the transfer function with jaw fallen. Have a look.


Did you see it, incredible, I could get first order transfer function only with open loop, feedbackless amplifiers. The cut frequency is 700khz but instead of cutting with -90°shift it is -45° and goes all the way to 5Mhz. This why on contrary to other composites where the amp is at its max phase margin, the opamp than needs to be much slower to carry the dominant pole. Here it is the amp is carrying the dominant pole, and opamp should be the fastest possible to let the amp louse alone its phase margin at 5Mhz. Now, that I understand the very particular character of this Lady, I will treat her with still more perseveringly.

[kokoriantz]

Do you remember on post 33 I tested FF's version using inverter mode? It had a strange behavior at the transient, the negative input for short period behaving as positive input.

Well, I wanted to take advantage of this defect and apply both inputs at short period, here is the miracle,


Now, the dominant pole of 700khz is extended to above 100Mhzzzz!?#@%$&¶~€.

[mjona]

With a parallel feed involving a power amplifier block and a high speed op.amp feeding into a common load ask yourself what of levels of current are passed by each of these and what individual voltages will in the nfb signal if the power amplifier block is able to deliver 4 A and the high speed op.amp around 100 m.a. or less.

[kokoriantz]

Good point of view. I will consider measuring when I am fixed with the type of composite. The current multiplier had an issue at 5Mhz, resolved by 50 ohms resistor across the inputs.

[kokoriantz]

This gives with very wide band amplifier driving the power. I don't say any more opamp. The frequency response is over 10Mhz and what an ultra fast transient!
The distortion should be slightly higher but I am in great certainty, this is ultra high definition amplifier.

It is not easy to understand if you don't know how CFA works. In short, the impedance of the negative input determines the open loop gain.
This is the input amplifier's transfer function.
compoalon.jpg
The scope graph transient is 20ns scale. I have never seen such.
20ns.JPG