Converting the AD797 MC Phono Pre to a MM Phono Pre - A Mixed Design.

No, not yet. I'll have to figure that out. I'll go back and see if I can't figure
out how to insert the Aweighting network.
To use the weighting network you wire the preamp output to it (through a noiseless R- the value is unimportant- just copy the 6.04k noiseless R8) and specify the measurement node to be the output of the weighting network (V(AWEIGHTOUT)) which was configured to be unity gain at 1kHz.
So, this is what you do.
Copy R8 (6.04k noiseless) and drop it onto the schematic above the A weighting network. Draw a short line to either end. Copy the AWEIGHTIN pin and attach it to one of the lines. Copy the outmixed1 pin and attach it to the other line. Change the noise simulation output to be v(AWEIGHTOUT). Run the noise simulation and click on the AWEIGHTOUT pin. Measure the noise result as before.
 
I

The unweighted one is often called Z weighting (Zero weighting) and the most common one is the A weighting.

So, to use the weighting network you wire the preamp output to it (through a noiseless R- the value is unimportant- just copy the 6.04k noiseless R) and specify the measurement node to be the output of the weighting network (V(AWEIGHTOUT)) which was configured to be unity gain at 1kHz.
In this case, the Aweighted output is 5-6dB lower (about 110uvrms) than the Z weighted one and the amp has about an 80dB S/N ratio.


1st attempt:
From this schematic and output:
upload_2019-1-11_7-2-15.png
 
Last edited:
Got your new instructions. Have to get my little girl to school.
"I shall return" -- Gen. McArthur.
"Me too" -- Spike
 
For some unknown reason you changed the name of the output pin from the opamp from outmixed1.
I thought I was supposed to rename it 6.04k_noiseless?
Recall I'm a rookie at this. I'll get there.

I'm trying to document the schematic changes so I can revert if necessary.
I revert the output pin to "outmixed1" as before.
Then, I think I got what you instructed. I'll edit this post and attached when finished...right or wrong.
or is it better if I make a new post?

Also, note, I've gone back and edited out some of the longer graphics
or changed full sized to thumbnails in my prior posts to streamline navigating within the thread.

Here is the new failure.
I thought I'd rename the output and run it
once to see if it worked before making any other changes.
analysis failed.JPG


Let me erase the resistor I floating above "A-weighting network." with nothing attached.
I'll try and run this sim again.

Good Godfrey, who'd of thunk it.
The simple act of, there is no copy function, I found the term
down near the bottom of the Edit pull down menu:

"Duplicate" or F6. Eureka.

Now for all you other LTSpice rookies out there,
components not connected to anything give the Matrix error shown above.
Delete the component that is sitting around not doing anything and your
sim should run.

Here is the not run:

and in the next post I'll have the result.
 
Last edited:
I thought I was supposed to rename it 6.04k_noiseless?
Recall I'm a rookie at this. I'll get there.

I'm trying to document the schematic changes so I can revert if necessary.
I revert the output pin to "outmixed1" as before.
Then, I think I got what you instructed. I'll edit this post and attached when finished...right or wrong.
or is it better if I make a new post?

Also, note, I've gone back and edited out some of the longer graphics
or changed full sized to thumbnails in my prior posts to streamline navigating within the thread.
aweight1.PNG aweight2.PNG
I've moved things a bit so I can snip the details of the circuit.
I've also included the A weighted result for the 2.5mv sensitivity circuit.
I've also included the noise sim panel.

aweight3.PNG
 
Wyn, thanks. I included and intermediate step which didn't work.
I new see, the position of the tools you made for analysis can be moved
from their location to another.

I didn't know if the following:
Board boundary
|
|
|
|


Were limits set by LTSpice, you, or both
and that noting out side those boundary's would work.
or if something was added to it, it wouldn't work any longer.

AND

As I was documenting what I've done, you condensed everything
and put it where it is simpler to see it and get it to work.
Thank you for your patience, it can be trying for us both. :)
But then, that is how we learn, and demonstrate mastery.

My grandfather used to say, the hardest thing to do is to
write clearly so that other people can understand what
you are saying. You already know this and I recognized it
when you wrote early on that you needed to think about
how to proceed in order to explain it to me. (paraphrased).

Let me move mine around and when I get the same result as you,
we both know were on the same page.


aweighted w 3V noise.jpg
 
Last edited:
The "board boundary" is just that. It's where the PC board circuit ends. Everything outside is simulation stuff.
 
I thought I was supposed to rename it 6.04k_noiseless?
Recall I'm a rookie at this. I'll get there.

I'm trying to document the schematic changes so I can revert if necessary.
I revert the output pin to "outmixed1" as before.
Then, I think I got what you instructed. I'll edit this post and attached when finished...right or wrong.
or is it better if I make a new post?

Also, note, I've gone back and edited out some of the longer graphics
or changed full sized to thumbnails in my prior posts to streamline navigating within the thread.

Here is the new failure.
I thought I'd rename the output and run it
once to see if it worked before making any other changes.
View attachment 1382758


Let me erase the resistor I floating above "A-weighting network." with nothing attached.
I'll try and run this sim again.

Good Godfrey, who'd of thunk it.
The simple act of, there is no copy function, I found the term
down near the bottom of the Edit pull down menu:

"Duplicate" or F6. Eureka.

Now for all you other LTSpice rookies out there,
components not connected to anything give the Matrix error shown above.
Delete the component that is sitting around not doing anything and your
sim should run.

Here is the not run:

and in the next post I'll have the result.
The pull down edit menu is mostly the same as the tool bar at the top. The copy (duplicate) icon (two pages adjacent and overlapping) is right next to the snip (scissors) and a bit to the left of the search string (binoculars) command. I've never bothered to learn the bind keys.
It's unexpected that this has, in essence, turned into a tutorial on the use of LTspice- but it is what it is.
I wonder if everyone else has grown bored and wandered off, so to speak.
 
For clarification, I'm asking

That is not a hard and fast LTSpice rule/feature etc?

It was you creating that arbitrarily, so that when you or
other's go back and look at things, it's a visual cue to know
this is what I've designed the board, and over here on the other side
I put my various sim tools that I use for analysis.

I didn't mean for it to become an LTSpice tutorial.

Just imagine trying to do this in NI MultiSim?
 
Yes, I'm compared the pull down and the Icons across the top.
Most folks I would assume are at work at the moment and would more
that likely join or view in the evening or over the weekend. If questions
they would probably ask then.

Moving along, I'm comparing the result of the two V(onoise) plots.
They are quite different.
For the one that I simmed, the peak (could I use pole interchangeably)
is approx 2.3kHz. I didn't think that I was using that kind of signal.

When I've seen this type of plot before, I'd then feed it through a
DUT, followed by a null filter to a distortion analyzer for measurement.
or use an FFT which calculates the THD or THD + N.

I'd reduce the fundamental peak to let's say -40dB so i could easily see
the harmonics, which would rise above the noise artifacts or noise floor (grass).

The plot that you posted has a different more rounded peak about 900Hz to 1kHz.
It appears very broad also and asymmetric. Oh but so much less noise from peaks
under 2.0uV/Hz1/2 trough approx .09uV/Hz1/2.

How'd yours get magnitudes less noise?
 
Last edited:
For clarification, I'm asking

That is not a hard and fast LTSpice rule/feature etc?

It was you creating that arbitrarily, so that when you or
other's go back and look at things, it's a visual cue to know
this is what I've designed the board, and over here on the other side
I put my various sim tools that I use for analysis.

I didn't mean for it to become an LTSpice tutorial.

Just imagine trying to do this in NI MultiSim?
No. It's just to make things more transparent. I'll often annotate schematics with things of interest.
 
Last edited:
I initially used the national instruments tool.
I moved to ltspice after a while as I found it more convenient. Both are ok for free tools. The full scale purchased spice simulators are superior, but beggars can't be choosers.
 
Yes, I'm compared the pull down and the Icons across the top.
Most folks I would assume are at work at the moment and would more
that likely join or view in the evening or over the weekend. If questions
they would probably ask then.

Moving along, I'm comparing the result of the two V(onoise) plots.
They are quite different.
For the one that I simmed, the peak (could I use pole interchangeably)
is approx 2.3kHz. I didn't think that I was using that kind of signal.

When I've seen this type of plot before, I'd then feed it through a
DUT, followed by a null filter to a distortion analyzer for measurement.
or use an FFT which calculates the THD or THD + N.

I'd reduce the fundamental peak to let's say -40dB so i could easily see
the harmonics, which would rise above the noise artifacts or noise floor (grass).

The plot that you posted has a different more rounded peak about 900Hz to 1kHz.
It appears very broad also and asymmetric. Oh but so much less noise from peaks
under 2.0uV/Hz1/2 trough approx .09uV/Hz1/2.

How'd yours get magnitudes less noise?
I have difficulty in keeping track of what you are doing- however;
The A weighting filter essentially puts a funky band pass filter in series with the noise, in an attempt to match the hearing sensitivity of the ear.
https://en.wikipedia.org/wiki/A-weighting
At 20 Hz it's about 50dB down and at 20kHz about 10dB down. It peaks in the 1-6k region.
The filter that I provided realizes the A weighting frequency response with the gain normalized at 1kHz as defined in the standard.
 
For some reason I'm trying to wrap my brain around the schematic and sim function.

Do I need to draw a line from outmixed1 down to aweightin--6.04K ?
Then see the output from 6.04K--aweightout?
 
For some reason I'm trying to wrap my brain around the schematic and sim function.

Do I need to draw a line from outmixed1 down to aweightin--6.04K ?
Then see the output from 6.04K--aweightout?
I'm losing track of what you are doing.
The pin names are also labels that identify a node. By copying the pin and attaching it to another node (line) you are telling the simulator that they are the same point in the circuit. By doing as I described you are connecting the amp output node to the A weighting network input node by way of a 6.04k ideal, noiseless, resistor- which won't matter as the input to the network has a 100Meg input resistance. There is no need to relabel anything or add extra wires. Just follow my instructions and illustrations without deviation and you'll be fine.
 
I've got my ARC SP3-A1 on my bench.
I've got a low noise oscillator from my HP339A,
I can set at 1kHz at 1Vrms to start.
I'll fee that into the phono circuit of the SP3-A1
I'll take the phono output and run it through
my distortion analyzer. And notate the result.

I can try a couple of different test frequencies
lower, higher. Measure them.

At least I'll have a baseline from which to compare
to when my boards arrive and I build them (longer time frame)
compare them to your known measurements of your AD797 phono (short time frame).

Other's might be interested in following along and it would be measuring
prior to the board I make.

Later when the board is finished, I'll do the same tests
using the same equipment. I can also take pics of the
set up etc.
 
I'm losing track of what you are doing.
The pin names are also labels that identify a node. By copying the pin and attaching it to another node (line) you are telling the simulator that they are the same point in the circuit. By doing as I described you are connecting the amp output node to the A weighting network input node by way of a 6.04k ideal, noiseless, resistor- which won't matter as the input to the network is a 100Meg input resistance. There is no need to relabel anything or add extra wires. Just follow my instructions and illustrations without deviation and you'll be fine.
I think I did. You did see my V(onoise) plot in post #227, yes? maybe refresh browser then
check.
 
Back
Top Bottom