I would concur with that assessment. Amazing, isn't it, that old design has lasted so long! AND, it's still a bargain at ~ $1.09! It's also available in DIP-8. The only opamp that comes close without being ridiculously expensive is the LME49721---it has less THD and about the same noise; still only ~ $1.59, but only SOIC-8.
Ahem... ever heard of bias current cancellation? The +/- in the LT1115's input bias current spec should have been a clue. The 5534 features no such luxury, making any comparison a mixed fruit bag type affair.
Going by >2 kHz current noise density (a little under 0.9 pA/√(Hz)), LT1115 effective input bias current has to be something like 1200 nA. The 5534A can't have more than 300 nA if it is to meet its 0.4 pA/√(Hz) spec.
The bias current cancellation rears its ugly head if you look at the "Total Noise vs. Unmatched Source Resistance" graph in the datasheet: At 1 kHz, expected total noise density for Rs = 10k would be 16 nV/√(Hz) - the graph shows about twice that! (That's an effective 3 pA/√(Hz) instead of 1.) Noise from cancellation current is correlated so will cancel out if impedances are matched at +/- inputs, but with any imbalance it will show across that. This is well-documented especially for older schemes. That's what Gofer was talking about.
Now an MM RIAA application has massively unmatched impedances at both input pins - the feedback side never really goes above the ground leg resistor of a few hundred ohms, the input side reaches over 20 kOhms around 10 kHz due to the L.
The results are not pretty. High-frequency noise would be up to 10 dB worse than with basically any run of the mill (moderately-)low-noise part (5534, 5532, LM833, MC33078, NJM4580, NJM2068, ...). Which is what you found out the hard way years ago.
Last edited:
LT1115 current noise 1.2pA, NE5534A current noise 0.4pA, so no you are wrong. Current noise (shot noise) is not fundamental, its a consequence of charge-carriers being forced to cross a barrier, rather than drift in concert, so
its possible for current noise to be greater or less than the classic formula for shot-noise. In particular metal conductors exhibit almost no shot-noise at all.
However lets look at the actual typical bias currents:
LT1115 +/-50nA
NE5534A 500nA
Clearly the LT1115 has bias-current cancellation as the sign of the bias is undefined. Thus the actual currents flowing consist of two unknown currents, substantially larger than 50nA, flowing through different input devices on the chip into the input pin, mainly cancelling out at that junction.
Alas noise from separate transistors does not cancel, both transistors generate current noise that combines such that the power adds, making the noise worse.
500nA corresponds theoretically to 0.4pA/√Hz, agreeing nicely with the NE5534A observation (you expect a pn-junction to show shot-noise behaviour as its a potential barrier). 50nA ought to give 0.13pA/√Hz, but the 1115 shows ten times that. Persumable the 1115 has about ~1µA input currents cancelling to within 5%, so the current noise is basically about that of 2µA or so.
Some opamps have a clever arrangement where the noise-cancelling current for both inputs is correlated, so that it cancels if the inputs see the same impedance from the outside circuit.
Alas in a high gain amp (needed in a low signal, noise-sensitive amp like a phono amp) this can only help if the circuit is balanced. For
an RIAA preamp that would require two sets of filter components - this isn't normally done and the circuit is asymmetric for the opamp inputs - thus bias-cancellation makes noise considerable worse, for no benefit (ultra low offset currents as such are little benefit to ac amplifiers).
Last edited:
Exactly.
BTW, while we are generally in agreement (as usual), I think you're missing a factor 2 in your shot noise calculation. Ib is a per-input spec after all, at least I am pretty sure it is, so you'd have to go with 2*Ib for both. That's how I've implemented it anyway.
The ib comp issue was covered here several times. This might be of interest, https://authors.library.caltech.edu/53119/1/00261888.pdf
All you guys can chatter on about the NE5534 as long as you like. The AD745s that fixed the noise issue have been in the preamp that sparked this thread since 2004, and they're staying (thanks, Scott).
The new design I recently shared will be going into the KFJC master studio - all discrete, not an opamp in sight. The test results shown are for the real deal, not a simulation. All that remains is for the modules to be mounted into a 2U rack case with the output transformers.
The new design I recently shared will be going into the KFJC master studio - all discrete, not an opamp in sight. The test results shown are for the real deal, not a simulation. All that remains is for the modules to be mounted into a 2U rack case with the output transformers.
I have taken Sy’s spread sheet that calculates cartridge noise and modified it so you can enter up to 5 opamps and compare.
The LM4xxx devices come out at about 60 dB SN with a typical cart, the 5534A at 67 dB and a high performance JFET opamp like the AD745 at a shade over 70 dB.
However, the Stanton you talk about has a very high self inductance, so it may well exacerbate the perceived ‘hiss’ whenthe needle is up, mandating a FET input opamp. This still should not detract however from the 5534A - a true gem. BTW I can get these from RS for about £0.30 here in the UK (TI product).
I’ll publish the spread sheet tomorrow (late here now).
The LM4xxx devices come out at about 60 dB SN with a typical cart, the 5534A at 67 dB and a high performance JFET opamp like the AD745 at a shade over 70 dB.
However, the Stanton you talk about has a very high self inductance, so it may well exacerbate the perceived ‘hiss’ whenthe needle is up, mandating a FET input opamp. This still should not detract however from the 5534A - a true gem. BTW I can get these from RS for about £0.30 here in the UK (TI product).
I’ll publish the spread sheet tomorrow (late here now).
Last edited:
Here is a link to the spread sheet
So Just How Quiet is Your Phono Stage?
As always, if I've made any errors - apologies. Let me know and I'll make corrections as required.
Important note: the amplifier and cartridge data is DUMMY data - enter your own data into the GREEN cells to get the results which are shown in the top RHS of the spread sheet.
So Just How Quiet is Your Phono Stage?
As always, if I've made any errors - apologies. Let me know and I'll make corrections as required.
Important note: the amplifier and cartridge data is DUMMY data - enter your own data into the GREEN cells to get the results which are shown in the top RHS of the spread sheet.
Stanton cartridges for pro use (in radio stations, etc) that are able to survive misuse/abuse are well known for it's bright sound. These cartridges are not suitable for home listening. One need to understand that they are produced for FM stations and FM is bandwidth limited to 15kHz. Stanton 500 series is purpose made to add some brightness, so when used for intended purpose they have just about the right sound.
We here at KFJC got rid of the Stanton 500AL cartridges (kinda the equivalent of a board with a nail in it) around the late 90s in favor of Stanton 680s. Since Stanton went belly-up, we have switched to the Ortofon OM Pro S. The styli look delicate, but they are well -protected and hold up under 24/7 abuse from "sensitive artists".
We had a very brief flirtation with Shure cartridges, but the cantilevers aren't protected and the test cartridges didn't last 24 hrs before the cantilevers get bent all to hell.
We had a very brief flirtation with Shure cartridges, but the cantilevers aren't protected and the test cartridges didn't last 24 hrs before the cantilevers get bent all to hell.
When I was young for some time I used Stanton 680EE. I really liked it. Stanton was cartridge of choice for Thorens turntables. Then I switched to Ortofon OM series. For a while I also used Audio Technica and Denon high output MC. I think that may be Audio Technica has something that is sturdy enough for pro work but good sounding. Audio Technica is a good brand for cartridges.
Does KFJC use US turntables like Russco and Gates?
Does KFJC use US turntables like Russco and Gates?
Here in Serbia the court would require you to prove you suffered a substantial financial loss because somebody copied your circuit, otherwise they would not accept the claim. In the market (over)saturated with products it would be impossible to prove substantial financial loss due to several diy-ers, or even a small company, copied some schematic. It's impossible in Serbia to be prosecuted for stealing a pencil, because the value of stolen good is negligible.
But in Japan, old people, with insufficient pensions, sometime deliberately steal something small from the supermarket with the intention to go to prison. Japanese state is notorious for disproportional sentence for small crime, so old man go two years in prison for stealing something worth two dollars. One year in prison costs a Japanese state over 40 thousands dollars, so it's better to be in the prison with decent food and in the warm place than dying of hunger poor and alone.
I got to thinking about legacy issues and whether anyone would be able to fix one of my weird discrete designs 10-15 years down the line. I ended up doing an opamp-based design that is physically a lot smaller than the discrete design and would be a little easier to diagnose and fix. The design is very similar to the one that started this thread, with OPA227 for input, TLE2074 for second stage, and DRV134 for single-ended to balanced converter. When I can find my AD745s, I'll consider slotting them into the input stage to replace the OPA227s. All the opamps are socketed, so if there are any issues, replacement is simple. However, by the time this thing eventually needs fixing, I suspect that the leaded versions of the parts I used would no longer be available (sigh). Hopefully, all the servicing needed would be periodic recapping., something we are busy doing with a lot of the equipment that has been hanging around for 10-20 years. Our powered monitor speakers seem to be especially hard on caps...
Until this day I don't understand why the different noise behavior of various op amps in RIAA stages matters at all - go to
https://www.diyaudio.com/community/...e-noise-i-want-to-know-all-approaches.296358/
From my point of view, this only does matter with microphone pre-amplifiers, where every db less SNR is important. And possibly for the special case where an RIAA stage is fed from a DAC via an anti-RIAA network but actually not in real life with LP's on record players.
For me, the sound character of the op-amp is much more important when playing a record, which is determined to a large extent by the kind of distortion (you would see in a diagram of residual distortion with fundamental notched out) and not by the amount of distortion - low THD+N means not automatically good sound resp. sonic character - the sound character of the NE5534 is clear evidence of this.
If one had the diagrams from post #2 under
https://www.diyaudio.com/community/...st-possible-thd-n-really-the-best-way.367692/
for the AD745 (or the AD817/AD825) and the NE5534, everyone would certainly understand why the NE5534 should not be the first choice for a RIAA stage.
Unfortunately such diagrams are not published for op amps in general - they would be very helpful in type selection .
check out also this threads:
https://www.diyaudio.com/community/threads/im-wondering-about-ad745.67124/
https://www.diyaudio.com/community/threads/the-greatist-ad745-psrr-versus-ad797.145053/
https://www.sciencedirect.com/topics/engineering/conventional-op-amp
A very long time ago, about 40 years, when I needed to make a RIAA preamp, I was most successful with an inverting amplifier, in spite of the theoretical noise disadvantages. With an inverting amp, any cart impedance improves noise, not increases it, and an open input is dead silent. And there are no headroom issues caused by the EQ curve. In those days, I was using, I think, a 4136, which is a quad 4558 with a strange pin-out. I have to admit that I was not very picky because any phono preamp was going to be noisy. You just need to use the RIAA curve to advantage instead of fighting with it. I had some extra gain with an extra stage, perfect for a quad chip. It was a very long time ago and involved a lot of tweaking, so I don't remember the details.