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

[USMC Spike]

if , for example f1 is say, 14kHz (in the signal band) and f2 is 30 kHz (ultrasonic) then you can get, for example, a signal at 30-28kHz= 2kHz! As these HF areas are where the amps are least linear it is something to be careful with. There are 2nd order products as well, f1+f2, f1-f2, 2f2, 2f1, but these are rarely as pathological.

Now that you mentioned it in that example it makes sense now, how these ultrasonic issues influence the audio band. Probably also way we want
to elimiate RFI from entering our phono amp/pre amp. Where signals are smallest....

Some preamps have large overload capabilities and interstage RF filtering so supposedly they can handle it, but it's certainly not universal.

I see the important of having the large "overload" capability. Very good.

I didn't want to focus on the Nuemann thing, but wanted to be sure that I understood the concept in general. Yes, not a standard,
because some like their tea hot, some with milk, some with sugar, cream etc.

Interesting about the lathe cutting also. One of the lathe mastering guys contacted me about amp. I don't know all the mechanics
of it, but from what he told me, something in the cutting heads transformer fried, where ever that is. He fixed it by taking a number
of turns out of it and he couldn't recall the final impedance of it. What he then mentioned surprised me when he said that he
then used a good tube amplifier to drive the head cutting lathe...and when he did that, he mentioned the resulting
masters and pressings sounded better then they had ever been. He couldn't recall exactly the details but was happy
with the result.

North Texas has become a hot spot for new vinyl pressing. There are a few companies here including Josey who purchased
Stan Getz A&R record plant (that never stopped pressing vinyl), read about it here: LINK
Josey Record MFG: LINK
Hand Drawn Pressing: LINK
There is also another new record pressing plant going up in Austin, Texas also.

We are discussing phono pre/amp here so it's good to know that vinyl pressing is growing.

 

[USMC Spike]

Oddly enough it's in my response ;-)

We are living in the "Golden Age" of information. So much is accessible now, google can be our friend, sometimes.
What was once buried in journals, syllabi, institutions of learning and higher education along with corporations putting
data all accessible on the internet is great.

While I typed "What was once buried" the song, "What Were Once Vices Are Now Habits" - Doobie Brothers came across me.

 

[wyn palmer]

Ok. I'll run some sims for a few cases. In each instance, just to simplify things. I'll assume that we are primarily interested in the +15db case with the a single pole cutter filter at 42KHz. Please ask for others if you want.
1. The everything at the input- no extra RF rejection, non inverting amp. (output- V(outallinone))
2. A reduced fixed gain block with the AD797MCPRE output RIAA stage- a "mini" AD797MCpre intended for a MM amp . (output= V(outmmorignorffilt))
3. As 2, but with the RF rejection filter added. (output= V(outmmorig))
4. The mixed amp. (output= V(outmixed1))
a. Sweep from 10 Hz to 20MHz


As far as the vinyl is concerned only what happens to 40kHz or below is important- but as far as IM is concerned the issue is more complex as any EMI that is picked up, or for that matter output from cartridges that have noise peaks or other pathological issues above 40kHz, can become problematic if the gain is not rolled off.
You can see that the two designs with the extra RF roll off ( both 2 opamps) have a lot more margin at frequencies above 100kHz or so without having a degraded RIAA compliance (the cutter pole is a different matter …), and about 1dB better margin at 40kHz.
b. Same input but a 20-100kHz sweep and looking at the output of the input opamp to see how they fare.

Note that there is almost 10dB more margin for overload at frequencies 20kHz and above for the mixed design.

The conclusion- assuming that noise is acceptable and that there is not some nebulous "golden ears" concept that I can't assess then the mixed design is the best of the bunch.
Next it's time to look at noise.

 

[nj pheonix]

Am i correct in assuming the schematics from post #13 Is what we're discussing now.
(The second one?)
And the 4th is an improved MM version?

 

[wyn palmer]

Am i correct in assuming the schematics from post #13 Is what we're discussing now.
(The second one?)
And the 4th is an improved MM version?

Not quite- the original post showed the evolution from the original MC pre to the MM I suggested.
This is a bit different- this includes an all in one stage design that I will not post the schematic of as, frankly, it's a bit of a Kluge with bits of circuits bypassed and scaled so that I don't have to start from first principles, together with three two opamp designs- one like the original MCpre without the extra RF pole, and one with the RFpole, but scaled for MM. The two opamp designs I will show at the end (just wait for it ;-))

 

[nj pheonix]

Thanks for the clarification
No rush
Carry on

 

[wyn palmer]

Ok. I'll run some sims for a few cases. In each instance, just to simplify things. I'll assume that we are primarily interested in the +15db case with the a single pole cutter filter at 42KHz. Please ask for others if you want.
1. The everything at the input- no extra RF rejection, non inverting amp. (output- V(outallinone))
2. A reduced fixed gain block with the AD797MCPRE output RIAA stage- a "mini" AD797MCpre intended for a MM amp . (output= V(outmmorignorffilt))
3. As 2, but with the RF rejection filter added. (output= V(outmmorig))
4. The mixed amp. (output= V(outmixed1))
a. Sweep from 10 Hz to 20MHz
View attachment 1375898

As far as the vinyl is concerned only what happens to 40kHz or above is important- but as far as IM is concerned the issue is more complex as any EMI that is picked up, or for that matter output from cartridges that have noise peaks or other pathological issues above 40kHz can become problematic if the gain is not rolled off.
You can see that the two designs with the extra RF roll off ( both 2 opamps) have a lot more margin at frequencies above 100kHz or so without having a degraded RIAA compliance (the cutter pole is a different matter …), and about 1dB better margin at 40kHz.
b. Same input but a 20-100kHz sweep and looking at the output of the input opamp to see how they fare.
View attachment 1375907
Note that there is almost 10dB more margin for overload at frequencies 20kHz and above for the mixed design.

The conclusion- assuming that noise is acceptable and that there is not some nebulous "golden ears" concept that I can't assess then the mixed design is the best of the bunch.
Next it's time to look at noise.

And now for the noise.
The mixed design comes in at 107uv rms , unweighted, 20-20kHz which gives a S/N ratio relative to a 0dBv output of 79dB.
The all in one comes in at 80uv, or 82dB.
The original MM derived from the MCpre comes in at 78uv- also 82dB. The RF pole makes no difference.
So why is the mixed worse (although who really cares). It's mostly because I chose to use fairly high value resistors in the input stage (1.5k, 7.5k vs. 820 and 2k) and output stage (320k vs a bit less than 200k) to make the choice of feedback cap simple at 10n. I had some well matched polyprops of that value on hand. The noise could be improved by using larger caps and rescaling appropriately. The other design uses c. 20n caps and the Rs are reduced accordingly.
I'll post the schematics in a bit as I'm off to the airport.

 

[wyn palmer]

And now for the noise.
The mixed design comes in at 107uv rms , unweighted, 20-20kHz which gives a S/N ratio relative to a 0dBv output of 79dB.
The all in one comes in at 80uv, or 82dB.
The original MM derived from the MCpre comes in at 78uv- also 82dB. The RF pole makes no difference.
So why is the mixed worse (although who really cares). It's mostly because I chose to use fairly high value resistors in the input stage (1.5k, 7.5k vs. 820 and 2k) and output stage (320k vs a bit less than 200k) to make the choice of feedback cap simple at 10n. I had some well matched polyprops of that value on hand. The noise could be improved by using larger caps and rescaling appropriately. The other design uses c. 20n caps and the Rs are reduced accordingly.
I'll post the schematics in a bit as I'm off to the airport.

The two relevant schematics were in fact posted earlier- MMnew and Mixed1, but I've attached a single channel .asc for each design and the schematics are shown below.

 

[USMC Spike]

It will be interesting to see how long his round trip to the airport takes and differs from ours, depending on where we are.

Oh wow, that was fast. Not LaGuardia, LAX, etc.

Let me check before I leave.

 

[wyn palmer]

Greensboro, NC. 10 mins drive, parking 100 ft from the carousel. No taxi time, flightaware, no problem.

 

[USMC Spike]

Nuts, they don't work. They are missing the data too them.
wyn, do you have a generic Value file for lack of a better word.
That is the placement of the A Sign, 3th to the right, pencil inclusive?
only data is seen is for ipopampout2.

Okay, we are just looking at the schematics where.
Non working as I can tell. but the devices and component valued are IDed and are filled in.

Once you know what your various frequencies are, etc you can then
just crank out the values...I think you posted the filter part component finder thing.
somewhere in an earlier post. that or it's in the other AD797 thread.

In the mean time I think I can get em going.

 

[wyn palmer]

I'll check them out in a while.

 

[wyn palmer]

Nuts, they don't work. They are missing the data too them.
wyn, do you have a generic Value file for lack of a better word.
That is the placement of the A Sign, 3th to the right, pencil inclusive?
only data is seen is for ipopampout2.

Okay, we are just looking at the schematics where.
Non working as I can tell. but the devices and component valued are IDed and are filled in.

Once you know what your various frequencies are, etc you can then
just crank out the values...I think you posted the filter part component finder thing.
somewhere in an earlier post. that or it's in the other AD797 thread.

In the mean time I think I can get em going.

Yes, the schematics are not simulatable as is- they're intended to be used as the basis for a board design so they lack sources etc.
But- if you check them out they're the same as the prior ones.
A comment- the .10pf cap can be deleted if all you want is to use an LME49710, or kept as a placeholder for a 10pF cap if you want to use an LT1115.

 

[USMC Spike]

Yes, the schematics are not simulatable as is- they're intended to be used as the basis for a board design so they lack sources etc.

Yes, I went back and checked them. Then looked again.

So essentially we go with the mixedriaanew schematic for the board and go from there.
I'm glad that you changed the schematic layout so its easily identifiable.

3. As 2, but with the RF rejection filter added. (output= V(outmmorig) Has the vertical network (stacked nodes)
4. The mixed amp. (output= V(outmixed1) Has horizontal network (nodes) and afterhfpole.

So those caps/resistor changes in the input and output are responsible for 27dB of noise
(that should say 27uv increase in the noise output.)
And it is still okay to listen to? This just for the MM cartiridge yes?

Now I'm brain dead. I need run errands and will return. It will sink in over time
and re reading playing with the sim on my own etc, while I try to get other things
accomplished.

Good stuff.

 

[USMC Spike]

Greensboro, NC. 10 mins drive, parking 100 ft from the carousel. No taxi time, tripaware, no problem

I gotta tell ya, as I'm sure you know, it doesn't get much better than that.

 

[wyn palmer]

Yes, I went back and checked them. Then looked again.

So essentially we go with the mixedriaanew schematic for the board and go from there.
I'm glad that you changed the schematic layout so its easily identifiable.

3. As 2, but with the RF rejection filter added. (output= V(outmmorig) Has the vertical network (stacked nodes)
4. The mixed amp. (output= V(outmixed1) Has horizontal network (nodes) and afterhfpole.

So those caps/resistor changes in the input and output are responsible for 27dB of noise
(that should say 27uv increase in the noise output.)
And it is still okay to listen to? This just for the MM cartiridge yes?

Now I'm brain dead. I need run errands and will return. It will sink in over time
and re reading playing with the sim on my own etc, while I try to get other things
accomplished.

Good stuff.

Just to put this in perspective, a very good LP A-weighted S/N is about 70dB, which is a few dB better than the A weighted S/N of a 15IPS 1/4" 2T professional analog tape deck (think Studer or Otari). This phono stage is about 10dB better than that unweighted- so 15/20db better weighted. You won't notice it. You could just increase the Cs by some proportion and lower the Rs by the same amount to get a few dBs of improvement, but the opamps are , as Hypnotoad explained, already overkill as far as noise is concerned, and the improved resistor noise would just be more of the same. It wouldn't be OK for a MC- but MC phono stages usually have an unweighted S/N in the region of 66dB for 26dB or so lower signal, so there's really no comparison.

 

[USMC Spike]

So the AD797 excels for use with the MC cartridge but is too much.

So the LME49710 or LT1115 phono stage is 10dB better unweighted!
and weighted..15/20dB. So we are way good already. Now I got to think
about Identifying components that I already have and those that I need.

 

[USMC Spike]

I found this, the NAB RIAA Reproduction characteristic.
Interestingly in a differing section it discussed the voltage frequency curve
and time constants.
T1 for those was as listed here 3180us, while T2 was 50us to 90us.
3180us was constant while T2 changes with recording speed.
I guess they lump it into two speeds. 15 & 7.5 IPS, 3.75 &1.875 IPS.

Here is the first reference. And the second reference for the V/f curve.
.....................................................(except where modified by eq time constants)
..............................

Tremaine, H. (1969). Audio Cyclopedia. Howard W. Sames & co. Inc., New York, New York. U.S.A.

 

[wyn palmer]

I found this, the NAB RIAA Reproduction characteristic.
Interestingly in a differing section it discusses the voltage frequency curve and time constants.

T1 for those was as listed here 3180us, while T2 was 50us to 90us.
3180us was constant while T2 changes with recording speed.
I guess they lump it into two speeds. 15 & 7.5 IPS, 3.75 &1.875 IPS.

Here is the first reference. And the second reference for the V/f curve.
.....................................................(except where modified by eq time constants)
..........View attachment 1376612....................View attachment 1376621

Tremaine, H. (1969). Audio Cyclopedia. Howard W. Sames & co. Inc., New York, New York. U.S.A.

https://en.wikipedia.org/wiki/RIAA_equalization
This includes an interesting section on the "Neumann pole", and the unintentional (and unwanted) additional zero that both designs with the additional pole eliminate that many phono stages do not. Much of my discussion on the design revolves around why this zero is unwanted and the negative consequences of having it.
The RIAA time constants are 75us, 3180us and 318us.
If you look at the mixed design the RIAA 75us TC is implemented in the first stage (7.5k R, 0.01u cap= 7500*10^-8=75*10-6=75us), the 3180us is in the output stage (318k, 0.01u) as is the "318us". The "318us" does not appear to be accurate (the RC product is 321us) but this is deliberate.
This is adjusted- in simulation- to compensate for the errors introduced by various things- the fact that the opamps are non-inverting, that they have finite gain, particularly at HF, and that there is an extra pole added for supersonic rejection. These deviations from "ideality" if you will, are the reasons for there being a small amount of ripple in the frequency response curve. If you look back at the response curves you will see that the phase curves for the phono amps are extremely good for the compensated ones with the extra HF pole, and not so good for the uncompensated ones with the unintended zero.

 

[wyn palmer]

I found this, the NAB RIAA Reproduction characteristic.
Interestingly in a differing section it discussed the voltage frequency curve
and time constants.
T1 for those was as listed here 3180us, while T2 was 50us to 90us.
3180us was constant while T2 changes with recording speed.
I guess they lump it into two speeds. 15 & 7.5 IPS, 3.75 &1.875 IPS.

Here is the first reference. And the second reference for the V/f curve.
.....................................................(except where modified by eq time constants)
..........View attachment 1376612....................View attachment 1376621

Tremaine, H. (1969). Audio Cyclopedia. Howard W. Sames & co. Inc., New York, New York. U.S.A.

I designed a variant of the mixed MM stage for a tape deck playback unit that someone asked for. It was able to support both NAB (IEC2) and CCIR (IEC1) playback at all the standard speeds. I've no idea what he did with the design.