What is the ideal conventional rumble filter?- Douglas Self

Status
Not open for further replies.
Jump to Latest
#181
rjm said:
Yes, but like weight-loss advertisements in magazines, they may not tell the truth!

Anyone can draw a line perfectly following over a bump and say "us!" and a wiggly line going everywhere and say "them!". Nice advertising. Of the engineering, I am skeptical.
Click to expand...

When in doubt, the simple proof of the pudding is to simulate exactly the same situation in the electric domain.
Take a second order 10Hz L-C filter with a Q of resp 8 and 1 to represent the cart.
Feed both with a 4Hz "warp" signal, subtract in and output voltage and compare both situations.
You could use a C=1000uF, L=0.25H and R=126 and R= 16 for a Q of resp 8 and 1.
That will prove if your scepsis is with justice or not.


Hans
 
#182
Here is the simulation for a Cart having a Fres = 10Hz, and cart/arm damping of resp 8 and 1.
The Voltage representing the LP surface, has a warp for 1/4 of the LP circumference.
The horizontal axis takes 2 seconds, the time for 1 full revolution.

LP-Warp.jpg

Red line is the input "warp", Blue is the deviation from the record surface for a Q of 1 and Green represents a Q of 8.
The max deviation for a Q = 8 is approx. 2 times as much, but much worse is that the oscillation goes on till the end of the 2 seconds time frame.

The non linearity of the stylus suspension will transform this resonance into IM distortion, as shown in posting #100.

Hans
 
#184
#185
mosfetDemon said:
There was something in the back of my mind, I mentioned in the other thread,and seeing the Nelson Jones thread just now, triggered off an ancient memory.
I was thinking it was JLH, it wasn't, it was Laurence Nelson-Jones.

Hope this link works,
http://www.americanradiohistory.com...rld-1971-06-OCR-Page-0095.pdf#search="nelson"

If not, its:-
Rumble Filter, by Laurence Nelson-Jones, WW June, 1971.

American Radio History: Documenting the History of Radio TV and FM broadcasting

HTH.
Click to expand...

It does -interesting history too.
 
#188
I wanted to share with everyone the Wireless World article that was mentioned several posts ago: "Differential Rumble Filter," J.P.Macaulay, Wireless World, Sept 1979 pg 75.

Pro Audio Design Forum • View topic - Differential Rumble Filter (Elliptic EQ) Wireless World

Differential_Rumble_Filter_Wireless_World.jpg


Let me begin by explaining what an "Elliptic EQ" is. It is not an elliptic, or Cauer filter. The similarities in names cause confusion.

Elliptic EQ is a vinyl mastering term for a bass to mono filter to reduce vertical modulation. Virtually every LP mastered, if it contains high level stereo bass, will have an elliptic filter applied in mastering.

The stereo width of a signal, when displayed in X-Y, forms a ellipse. In mastering, a bass to mono "filter" is applied to reduce the LF spread of the ellipse. It is therefore called an elliptic filter or elliptic EQ despite having no relationship to a Cauer filter.

An example of elliptic EQ (the bass to mono kind) can be found in the Nuemann EE-70 and EE-77: Pro Audio Design Forum • View topic - Using the Precision MS Matrix for Mono Crossover LF Blending The EE-70 uses an inductor to shunt Left and Right, the EE-77 (shown later in the thread) uses a resistor working against series caps.

The later Neumann VAB-84 subtracts the LF side information from the opposing channel to create bass to mono and is very similar to the "Differential Rumble Filter" shown by Macaulay.

Many have noted here that the dominate vector for rumble and warp is vertical modulation. Since vertical modulation at low frequencies is avoided during mastering (with certain exceptions) removing it generally does not remove information in the original recording. Because of this the vertical channel can be aggressively filtered.

There are a couple of other ways to do this in addition to Macaulay.

One is to break the signal down into Mid Side, high pass filter the side, and the re-encode back into left and right. A commercially made board I offer for mastering can easily be made into a rumble filter using MS and an external HP filter. The MS board - which contains a lot of circuitry simply for balanced I/O - serves as an example of how simple MS is to accomplish using differential line receivers: Pro Audio Design Forum • View topic - Mid Side M-S Matrix Construction Information

The MS encoder/decoder are IC3/4 and IC10/11.

It is doubtful that a high pass filter in the Side as high as 100 Hz could be heard given that little, if any, information was recorded in the lower octaves.

Another method is to subtract Side by adding an inverted copy of itself to the opposite channel. This is similar to the VAB-84. To provide elliptic EQ mono to bass a LP filter is inserted into Side. The Stereo Width controller is a modified VAB-84 design which eliminates the sliding filter: Pro Audio Design Forum • View topic - Stereo Width Controller Construction Information

Again most of the circuit bits are simply for balanced I/O - the actual processor is about 5 ICs.

Breaking the signal into Mid and Side and only processing Side is the least obtrusive way to eliminate the effects of unwanted vertical modulation because its focusing on where the rumble/warp actually is and where the mastering engineer avoids putting signal anyway.

For those of you wanting to roll up your sleeves an actually try it be my guest.

If you happen to be looking for a fully-balanced differential preamp and Phono Transfer system you'll find one used by mastering engineers here: Pro Audio Design Forum • View topic - Phono Transfer System
 
#190
Thanks Jan.

I think I actually have a pair of OPA2134s in my Phono Transfer System flat preamp at the moment. The preamp can also use a 5532, 2114, OPA1612 (on adapter.) Any good low-noise non-bias-compensated dual.

The NJM2068 seems to have both a combination of low voltage noise and - for a bipolar - low current noise. In the noise department it out-performs all of the aforementioned slightly when the source is a real cart. The OPA2134 is a close second.

The most expensive part, the OPA1612 is also the noisiest. It performs very well with a resistive source but the bias-current compensated input's current noise combined with the carts inductance gives it a severe noise penalty. In other applications the OPA1612 is fantastic.

I couldn't find a single LME49720 or LME49860 from multiple small lots that did not have unacceptably high burst noise. I stopped using them before they were discontinued.
 
#191
@ mediatechnology et al

Nice links 🙂

I don't know why the OP27 or OP2227 are often overlooked ? From the OP27's data sheet

Magnetic Tape Head
<1500 O Low is very important to reduce self-magnetization problems when direct coupling is used. OP27 IB can be neglected.

Magnetic Phonograph Cartridges
<1500 O Similar need for low IB in direct coupled applications. OP27 does not introduce any self-magnetization problems.
Click to expand...

Plus lots of other Very good specs !

My phono stage i designed & built nearly 30 years ago with OP27's is hard to beat !
 
Last edited:
#192
Thanks for the comments.

@Zero D I think you may mean the OP270 dual OP27.

This has kinda veered off the differential rumble filter. But, since there was interest in the preamp link I might want to add that that the fully-differential phono input, which allows a floating, balanced cart connection is relatively immune to op amp bias current because only the difference in the bias currents, the Ios, is seen by the cart. Even with very high bias current op amps such as the NJM2114/5532 the Ios, and thus the cart current, is typically less than 10 nA.

I recall someone here looking at cart performance with large amounts of bias current and not being able to measure any change in THD with currents in the 10s or 100s of uA. I've always wondered about that.
 
#193
mediatechnology said:
Thanks for the comments.

@Zero D I think you may mean the OP270 dual OP27.

This has kinda veered off the differential rumble filter. But, since there was interest in the preamp link I might want to add that that the fully-differential phono input, which allows a floating, balanced cart connection is relatively immune to op amp bias current because only the difference in the bias currents, the Ios, is seen by the cart. Even with very high bias current op amps such as the NJM2114/5532 the Ios, and thus the cart current, is typically less than 10 nA.

I recall someone here looking at cart performance with large amounts of bias current and not being able to measure any change in THD with currents in the 10s or 100s of uA. I've always wondered about that.
Click to expand...
You are right that for some obscure reason, max allowable DC current flowing into a cart is one of the secrets of this industry.
But that DC current leads to a force on the needle, thereby moving the needle a bit from its neutral position is obvious.
When suspension and motor where strictly linear, it would not result in added distortion, unfortunately this is not the case.

However an MM cart with an internal R of 600ohm, producing 5mV@1kHz, will see a peak current of +/- 12uA, so it seems to be safe to keep quite a bit below this value, say at 1% or 120nA.
This value will not shift the needle too much out of its neutral comfort zone.
It is just an assumption, but it gives some guide.

Coming to the 5532 in a balanced setup, there is no mentioning in the specs how the two amps compare.
Theoretically number one could have an Ibias of 800nA and the other of 200nA.
When both have an Ioffset of 10nA, this means an Ioffset of ca. 800 - 200 = 600nA between both plus inputs.
This would result in 300nA current flowing into the Cart, a bit more then wanted but probably probably still acceptable.

Hans
 
#194
B.t.w. , there is a very simple and elegant way even in the example above to almost eliminate the DC current flowing into the Cart .
Use a small ceramic 100k pot instead of the two 50k resistors and connect a (10k) resistor from the taper of the pot to ground, that's all.
By alternative short cicuiting the input an leaving it open, the output voltages of the two 5532 op-amps should remain the same after having adjusted the pot properly.
This can only be the case when there is no voltage change at the input, so in both cases it will be 0V differential.
With 0V no current will flow into the Cart, but the termination resistance for the cart will remain unchanged 100k or 47k depending on the input load switch.
It will even increase CMRR for a resistance value greater than 0 Ohm.
What remains is the difference in temp dependance, but both amps in one housing will have a tight temp coupling so this effect will be minimal.

Hans
 
#195
You are right that for some obscure reason, max allowable DC current flowing into a cart is one of the secrets of this industry.
But that DC current leads to a force on the needle, thereby moving the needle a bit from its neutral position is obvious.
When suspension and motor where strictly linear, it would not result in added distortion, unfortunately this is not the case.
Click to expand...

That was my original line of thinking as well. However it has been pointed out elsewhere (Richard Lee aka ricardo on the proaudiodesignforum) that any small current moving the needle from its neutral position is de-minimus relative to vertical tracking force.

I do see difficulty setting up enough cart current to provide deflections equal to VTF measured in grams. I think the THD tests, showing no effect, were posted here. My jury is out on this one.

Coming to the 5532 in a balanced setup, there is no mentioning in the specs how the two amps compare.
Click to expand...

Correct. The Ios I refer to are between sections, not inputs, and they are not a datasheet specification. I measure it on every one I build and it turns out they are quite consistent when the sections are on the same IC. I've never had a part measure more than 10 nA so I've never had to change one for that reason.

Theoretically number one could have an Ibias of 800nA and the other of 200nA. When both have an Ioffset of 10nA, this means an Ioffset of ca. 800 - 200 = 600nA between both plus inputs. This would result in 300nA current flowing into the Cart
Click to expand...

Also correct. But the bias current distributions between 5532/2214 sections are very close in practice. What I typically see are matchings that are within 10 nA. Its very difficult for me to measure currents lower than that and 10 nA is about the most I've ever seen between 5532/2114 sections. Not saying there aren't parts and production runs that aren't good just never seen them not be very close.

So to use an example one 5532 section may have an Ib of 700 nA and the other 710 nA. The net current, the Ios of the INA is 10 nA. This is not specified but it has been found to be very consistent and very low. It also should be noted that in the case of non bias compensated op amps the signs are always the same.

Someone concerned WRT to Ios could just use an OPA2134 in that socket and be done with it.
 
#196
Responding to your second comment I haven't seen a need for a trim. Less than the 10 nA I can measure is good enough for my clients.

However you do point out the benefits of having a common mode impedance >0R. I've experimented with this by inserting an Rcm instead of the on-board jumper.

On mic preamps I sometimes recommend a higher Rcm >0 to increase LF CMRR due to capacitor mis-match. (T-bias.) THAT's 151X datasheet was changed at my request to reflect that line of thinking. (figure 5).

There's very little, if any, noise penalty from having a higher Rcm because the op amp noise currents developed across it appear in common mode.

In some instances a high Rcm may benefit a phono preamp. I've compared the two and in my setup both are equally quiet so I don't use it. The only magic is in the fully-balanced cart connection.
 
#197
mediatechnology said:
Responding to your second comment I haven't seen a need for a trim. Less than the 10 nA I can measure is good enough for my clients.

However you do point out the benefits of having a common mode impedance >0R. I've experimented with this by inserting an Rcm instead of the on-board jumper.

On mic preamps I sometimes recommend a higher Rcm >0 to increase LF CMRR due to capacitor mis-match. (T-bias.) THAT's 151X datasheet was changed at my request to reflect that line of thinking. (figure 5).

There's very little, if any, noise penalty from having a higher Rcm because the op amp noise currents developed across it appear in common mode.

In some instances a high Rcm may benefit a phono preamp. I've compared the two and in my setup both are equally quiet so I don't use it. The only magic is in the fully-balanced cart connection.
Click to expand...
I agree, that when Ibias is so impressively close between 5532 amps, there is no need for further tuning.
Have you ever considered using the much newer OPA1642, it has 0.8fA/rtHz and 5.5nV/rtHz. It seems to me a more optimal choice for a MM application.

Hans
 
#198
I need to try the OPA1642. I've tried the bipolar version, the OPA1612, in a number of circuits and its a fantastic dual. The FET version, the 1642, looks very similar after the front end sharing what appears to be an identical OP stage.

Brown Dog has made me some OPA1612 adapters and are now stocking them. I should see if they have 1642s and, if not, encourage them to stock them. The OPA1642 should work very well in the preamp's balanced front end.
 
#200
Douglas Self will be presenting his Devinyliser concept to the Audio Engineering Society in London at 6:30pm on Tuesday 11th October at Kings College, London. This will be a longer version of the short presentation he gave to the Paris AES convention in June this year.

The event is free to all, and we hope to see some of you there.
 
Status
Not open for further replies.
Top Bottom