I wholly agree, that would be quite irrational.
Is it? Is that why you give no component values in #266?
I have had bad experiences with soi-disant JLH filters. For example his so-called 'bootstrap filter' is not a filter at all, but a bodge-up with an ultimate slope of only 6dB/oct.
No.
Yet I could post dozens of "hi-end" pre amps that do. All across the board, AYRE, John Curl, and down through Sony and Yamaha. Below is one Walt did for LT, IIRC Walt Jung did not recommended cap coupling on the input.
Attachments
Using one of the modern opamps such as the LME4970 will only produce typical input current values of less than +/- 10 nA.
In a recent project it was interesting to see that several types of the opamps with a gain of Av =100 and DC coupled all the way had an output offset of less than 2 mV and many were 1 or less. (A bit of cheating as the input was 60 ohms.) Older opamps such as the NE5534 were trimmed and still had difficulty reaching and staying that low.
Matching and laser trimming are at scary levels these days. I had a part with a 3uV mean offset due to the input pins being a few 1/10's of a degree different temperature after warmup, the distribution of the trimmed offset was so tight it was obvious.
I worked out one example using algebra circa 1978/9 or maybe even earlier This was before TDA1034 (we were Beta testers for this secret chip). Alas, I no longer have values.
The experience was so traumatic that I went away and wrote my own Linear Circuit Analysis package to avoid ever having to do it again. 😱
Yes. JLH was certainly a pseudo guru and spouted a LOT of liquid BS.
But some ... only some .. of his stuff was solid. His 12dB/8ve RIAA filter was one of them.
Mark Johnson, as Douglas Self has declined, could we ask you to spend 20min of your time to do the honours? It would be nice to have a formal expression and hopefully a few design eqns & a simple design method too for those with 1 brain cell like me.
Guru Wurcer, please dun mek fawn orf piple hu kunt reed en rite and neber deed "high school maths"
___________________
Thank you Guru Wurcer.Scott Wurcer said:
In the Jurassic era, I spent a LOT of time investigating what happens if you put the EVIL high input bias of a 5534/2 through your MM or high output MC cartridge.
I can pontificate at length on the electro-mechanical AND electrical EVILS of doing this. They are different for the 2 types of cartridge
But to cut to the chase, I was unable to find ANY effect which wasn't COMPLETELY swamped by other factors. I challenge anyone here to show ANY effect.
So I will claim that the effects are unmeasurable including via DBLTs ... to at least the rigour of Self's "unmeasurable hum" 😎
________________
Don't use ANY OPA with Ib cancelling.
A RIAA preamp must be the worst case to highlight this EVIL. IIRC, there's a couple of AD & NS application notes that explain this.
Wayne Kirkwood at Flat Phono Preamp Based on John's P10 and 2SK389 measures some modern OPAs with 'real life' MM cartridges.
NJM2068 comes out well and so does 5534 (of course).
For Instrumentation Amps, Guru Wurcer recommends AD524 which on paper, appears ideal for MM.
Last edited:
Right now I'm listening to a duet of pipes and didgeridoo also something one should probably not do.
Sorry! Should be Don't use ANY OPA with Ib cancelling.
Only acceptable with a large glass of Burgundy 🙂
I almost forgot to mention one of the interesting but virtually unknown characteristics of capacitors. Although leakage current is well known, what I found way way back is that the leakage current seems to stay constant with changes in voltage across the capacitor. The other handy number was back then the Mylar capacitors leaked about 100 uA per microfarad.
So it is quite possible that a modern opamp will source less current into a phono cartridge than an older circuit with a blocking capacitor. What is probably not so obvious is that using an input capacitor with a modern chip may actually not change the current!
Good point Ed! My '70-ies vintage SONY TA-E88 has the cartridge coupled to the pre through a 1000uF (!) electrolytic. But the voltage across the electrolytic will be minuscule, and even if the voltage does not in general effect leakage, I would expect that at some point the leakage would disappear. Maybe I can measure it.
BTW When you say that Mylars leak 100uA per uF, did you mean electrolytics? Seems huge for Mylars.
Jan
At identically zero volts any leakage would have to be driven by a chemical reaction or would eventually go to zero.
Jan,
I found those results on a Mylar capacitor in a project I did in 1969. Safe to assume there are better capacitors today. Or a slighter possibility, my recall could be off.
Scott,
Soakage will pretty much insure that with an asymmetric waveform there will still be an effective DC tail. (Asymmetric with regard to peak not average voltage.)
The last time I got into the definition of capacitance everyone got hot and bothered. But the first derivative still is CdV/dT + VdC/dT = i. V is not periodic for music unless you count the time of the entire signal in which case T is extremely large not the range of .1 second normally used.
I found those results on a Mylar capacitor in a project I did in 1969. Safe to assume there are better capacitors today. Or a slighter possibility, my recall could be off.
Scott,
Soakage will pretty much insure that with an asymmetric waveform there will still be an effective DC tail. (Asymmetric with regard to peak not average voltage.)
The last time I got into the definition of capacitance everyone got hot and bothered. But the first derivative still is CdV/dT + VdC/dT = i. V is not periodic for music unless you count the time of the entire signal in which case T is extremely large not the range of .1 second normally used.
Last edited:
I felt moved to run this filter through the simulator.
It works well, it is Butterworthy, and has the stages suitably arranged to prevent internal gain eroding the headroom.
I have not however looked at the component sensitivities.
This asymmetric waveform stuff leads to too many misconceptions I'll leave it at that rather than discuss it again. I also thought Jan was talking about t -> infinity.
Last edited:
It's questionable if these apparently-clever filters really are. You wind up with a couple of 470nF caps on the input instead of say 100uF, which is likely to make the effects of current noise worse at low frequencies.
... one reason why I prefer to connect the MM cartridge directly to 5534 +ve input pins ... even with their huge evil bias currents [*] 🙂
You gonna work out at least 1 JLH type 12dB/8ve filter which allows this? 😀
[*] but not as evil as OPAs with truly EVIL Ib cancelling schemes 😱
Last edited:
I agree, but i was thinking of other uses rather than just RIAA. But i thought the concept was clever/unique. Anyway, as the LF is 20db down before correction, the HPF doesn't need to be at the input.
I made an addition to that filter, to create a new type in here http://www.diyaudio.com/forums/solid-state/303292-new-type-filter.html
kgrlee kindly commented on it in there. Have a look & see what you think
But the stage we are talking about does give the 20dB boost at 20 Hz, so the very front end is the place to put a subsonic filter if you can. What they call a roofing filter. By my calculations a 100uF input cap increases the noise at 20 Hz by 0.1 dB, falling above that. A 470 nF input capacitor will obviously be much worse.
That's not a filter.
As I just said, by my calculations a 100uF input cap increases the noise at 20 Hz by 0.1 dB, falling above that. If that is too much, use a bigger cap.
However, if you insist on running DC currents through cartridges that are patently not intended to handle them, well, I have much to do and I am not going to try to persuade you.
FWIW, I intentionally tried and measured/listened to results for passing small amounts of DC, up to a few mA, in the coils for a number of MM cartridges.
I was curious as to whether any bias in the B-H curve altered things, plus it is jolly convenient indeed in a number of circuits if it is tolerable.
I found dc up to a few mA to have little, if any, effect on measurements or listening tests.
There we are, typically one can use an MM cartridge as the cathode resistor in a grounded grid or cascode valve/fet stage, IME. Taboo, but in practice no notable effect I found, and opens loads of doors for dc coupling.
I shall fetch my own coat.........😉
LD