Has anyone tried the OPA828 as part of a RIAA preamp?
Having FET input and being reasonably low noise, it looks as a good candidate.
Having FET input and being reasonably low noise, it looks as a good candidate.
Excellent candidate - probably needs 100nF ceramic cap between V+ and V- for decoupling as its a fast opamp. Expensive and only available SMT though.
My idea had been to try it as a replacement for AD745 on Gary Galo's RIAA preamp.
Where can I find an OPA828 LTSpice model for me to try on the simulation I already have with the AD745?
Where can I find an OPA828 LTSpice model for me to try on the simulation I already have with the AD745?
Last edited:
Where did you get 80 dB of open-loop gain for the OPA828? Every value listed in the spec table is well over 110dB; the typical open-loop response graph shows 140dB at LF.
Don't think there's any opamp has that kind of low gain! Figure plucked from where?
These are three simulations I did with Gary Galo mods suggested for one of the Adcom's RIAA preamps.
The original preamp was proposed by Linear Technologies on the LT1115 datasheet.
You must excuse the added opamps present on the simulations, but these simulations were made for my private use only, and I found it useful to make quick comparisons between opamps. You may try that too, if you want to.
In fact I would be grateful if these simulations, made months ago, should suffer any changes to make results realistic.
Even better should be if someone replaces the AD745 on the sim for an OPA828, and see what changes. Please do upload that sim.
Many people have told me that it's difficult to simulate RIAA preamps, particularly for noise or for AC response. Suggestions on how to make these simulations meaningful are very much welcome.
In case someone suggests why I don't do with the OPA828 the same I did with the AD745J model, back then I did what I was suggested here, and I don't remember what it was, the steps to follow.
The original preamp was proposed by Linear Technologies on the LT1115 datasheet.
You must excuse the added opamps present on the simulations, but these simulations were made for my private use only, and I found it useful to make quick comparisons between opamps. You may try that too, if you want to.
In fact I would be grateful if these simulations, made months ago, should suffer any changes to make results realistic.
Even better should be if someone replaces the AD745 on the sim for an OPA828, and see what changes. Please do upload that sim.
Many people have told me that it's difficult to simulate RIAA preamps, particularly for noise or for AC response. Suggestions on how to make these simulations meaningful are very much welcome.
In case someone suggests why I don't do with the OPA828 the same I did with the AD745J model, back then I did what I was suggested here, and I don't remember what it was, the steps to follow.
Subsonic is 140dB (claimed). 130dB at 20Hz. GBW curve is smooth. Plotting a 40dB@1KHz RIAA on the OPA828 GBW graph, we see the 'critical' point is 2KHz-20KHz, where we have 46dB gain margin. (58+dB in bass.)
That should be really ample for any need.
Attachments
Why use 1 opamp? Walt Jung has shown a 2 opamp version has better noise and overload performance
My intention is to build several RIAA preamps, at least five, and listen to them fed by a Thorens TD-150 turntable, with Linn Basik II tonearm and Shure V15 cartridge.
It's quite likely there won't be any MC cartridges used on this test. So those interested on MC pre-pres are on their own on this project. Sorry.
Two of the projects will be discrete: one with dual FET input, the other with bipolars.
Three of the projects will be based on opamps, certainly trying different ICs to listen what changes.
My attitude and findings towards subjective tests, which I usually do with a friend who builds extremely high quality speakers, is that things do change when you change passive or active parts.
The question is no to "expect" a change, which is the difficult part of it. As we are both professionals, he building speakers and me being a retired audio engineer, that worked for film & TV location and mixing audio recordings, we got to train our ears and mind to avoid such expectation and identify what changes.
Then we decide whether we like these changes or not, though rarely do we differ on preference.
As you see we try to avoid the concepts of "better" or "worst", and prefer to go for more "real life" response, particularly on acoustic instruments or human voice. With them we have a reference from having listened to many actual presentations or concerts, where room acoustics had been essential to be as neutral as possible. I guess you understand what I mean.
The listening room is medium size, amplifying is with a Luxman 5M21 power amp, and the 3-way speakers used are the ones my friend makes.
I will design the pcbs myself, dual sided, and build the preamps. I have done both things quite a lot.
For the power supplies I plan to build Jung super regulators, separate for each channel. Though other simpler IC supplies will be tried to. It's a good chance to do that.
It's unlikely that I will try Salas regulators, because FET picking for them is critical and I don't want to get into that.
These are the general ruling for this project.
It's quite likely there won't be any MC cartridges used on this test. So those interested on MC pre-pres are on their own on this project. Sorry.
Two of the projects will be discrete: one with dual FET input, the other with bipolars.
Three of the projects will be based on opamps, certainly trying different ICs to listen what changes.
My attitude and findings towards subjective tests, which I usually do with a friend who builds extremely high quality speakers, is that things do change when you change passive or active parts.
The question is no to "expect" a change, which is the difficult part of it. As we are both professionals, he building speakers and me being a retired audio engineer, that worked for film & TV location and mixing audio recordings, we got to train our ears and mind to avoid such expectation and identify what changes.
Then we decide whether we like these changes or not, though rarely do we differ on preference.
As you see we try to avoid the concepts of "better" or "worst", and prefer to go for more "real life" response, particularly on acoustic instruments or human voice. With them we have a reference from having listened to many actual presentations or concerts, where room acoustics had been essential to be as neutral as possible. I guess you understand what I mean.
The listening room is medium size, amplifying is with a Luxman 5M21 power amp, and the 3-way speakers used are the ones my friend makes.
I will design the pcbs myself, dual sided, and build the preamps. I have done both things quite a lot.
For the power supplies I plan to build Jung super regulators, separate for each channel. Though other simpler IC supplies will be tried to. It's a good chance to do that.
It's unlikely that I will try Salas regulators, because FET picking for them is critical and I don't want to get into that.
These are the general ruling for this project.
Now let's list the RIAA preamps I want to try.
1) Linear Technology LT1115 datasheet RIAA preamp.
https://www.analog.com/media/en/technical-documentation/data-sheets/lt1115fa.pdf
Based on this I will replace the LT1115 with an OPA828 following what Gary Galo did on the Adcom GFP-565.
Adcom s GFP-565 preamplifier - PDF
Even if many will disagree, I do not intend to try the AD745J IC, as it is a quite expensive part and tricky to adapt, for me at least.
My comparison will be between the LT1115 (instead of the original LT1028, to which it is related) and the OPA828. It's quite likely I will try other affordable low noise FET chips.
2) Walt Jung dual chip passive RIAA preamp.
3) Datasheet preamp from LM4562.
For the discrete preamps, I will assemble:
4) Luxman 5C50 updated version. Original transistors are not available, so it's not a clone. Dual FET at the input.
5) Super Low Noise. This is a version of a design published in Elektor mag in the '80s, which was later reborn by Elektor with AD797s in parallel. Simulation with much modern low noise transistors is very promising. THAT manufactures some N and P pairs that would be interesting to try to. Increasing power supply voltage to +/- 30v and adding a DC servo improved THD results considerably in the simulation. But listening to a prototype would be the real test.
1) Linear Technology LT1115 datasheet RIAA preamp.
https://www.analog.com/media/en/technical-documentation/data-sheets/lt1115fa.pdf
Based on this I will replace the LT1115 with an OPA828 following what Gary Galo did on the Adcom GFP-565.
Adcom s GFP-565 preamplifier - PDF
Even if many will disagree, I do not intend to try the AD745J IC, as it is a quite expensive part and tricky to adapt, for me at least.
My comparison will be between the LT1115 (instead of the original LT1028, to which it is related) and the OPA828. It's quite likely I will try other affordable low noise FET chips.
2) Walt Jung dual chip passive RIAA preamp.
3) Datasheet preamp from LM4562.
For the discrete preamps, I will assemble:
4) Luxman 5C50 updated version. Original transistors are not available, so it's not a clone. Dual FET at the input.
5) Super Low Noise. This is a version of a design published in Elektor mag in the '80s, which was later reborn by Elektor with AD797s in parallel. Simulation with much modern low noise transistors is very promising. THAT manufactures some N and P pairs that would be interesting to try to. Increasing power supply voltage to +/- 30v and adding a DC servo improved THD results considerably in the simulation. But listening to a prototype would be the real test.
> Super Low Noise. ... with AD797s in parallel.
Parallel devices make sense for LOW impedance sources. Under 1k.
MM needles barely make 1k at low frequencies and are rising to 5k by 2KHz where we hear hiss best. At these impedance levels several "ordinary" opamps do fine. '5532. AD797. "Fine" in the sense that opamp hiss is low relative to the resistive hiss of the cartridge (which is far lower than the surface hiss of good vinyl).
I'm not saying it is wrong to try. But unless your "only" goal is low hiss with a low-Z MM cart when the needle is not in the groove, it may have no real benefit.
The THAT parts are sharp for low Z, but selected 2N5089 working 20uA-50uA can hit similar hiss level with MM sources.
Parallel devices make sense for LOW impedance sources. Under 1k.
MM needles barely make 1k at low frequencies and are rising to 5k by 2KHz where we hear hiss best. At these impedance levels several "ordinary" opamps do fine. '5532. AD797. "Fine" in the sense that opamp hiss is low relative to the resistive hiss of the cartridge (which is far lower than the surface hiss of good vinyl).
I'm not saying it is wrong to try. But unless your "only" goal is low hiss with a low-Z MM cart when the needle is not in the groove, it may have no real benefit.
The THAT parts are sharp for low Z, but selected 2N5089 working 20uA-50uA can hit similar hiss level with MM sources.
I'm not interested in the SLN with the AD797s in parallel.
The one I did build 30 years ago was the original Elektor one, with BC550/560 in parallel. Unfortunately I was quite inexperienced back then, and I couldn't cure a hum problem I had with it.
There's a thread on DIYAudio with people that recently built several versions of the SLN.
The one I did build 30 years ago was the original Elektor one, with BC550/560 in parallel. Unfortunately I was quite inexperienced back then, and I couldn't cure a hum problem I had with it.
There's a thread on DIYAudio with people that recently built several versions of the SLN.
Do keep in mind that frequency response is pretty much dropping at 20 dB per decade, so what's going on at high frequencies is not going to matter nearly as much.
That said, you're correct in that few MM cartridges are going much below 500 ohms even at DC - that's at least 2.8 nV/√(Hz) right there. I still believe that the NE5534(A) was pretty much tailored to such an RIAA application. Low enough noise to reach 3 dB from theoretical limit - check. Drives a few hundred ohms (as needed towards the high end) - check. Pretty much about as good as it really needs to be.
Being the cheapskate that I am, I would probably go for an NJM2068 (or two in parallel if needed, they're cheap enough and a dual anyway) plus a discrete emitter follower buffer with, I don't know, BC639/640 or something (or the BCP/BCX53/56 surface mount jobs) at a decent current. Sort of the low-tech version of what the GFP-565 had (LT1028 w/Class A bias + LT1010). The '2068 apparently is just a hair noisier than an OPA2227 (a 3 nV/√(Hz) part) at low Z, so would seem to be at least a match for a 5534A purely on voltage noise.
Mind you, with BJTs there's that whole coupling capacitor business, which gets to be a bit of a nuisance on the feedback network with potentially thousands of µF (though I don't see why you couldn't also implement a DC servo there, much like it was done for the AD745 FET input version, ultimately sort of combining both circuits).
How can I make RIAA simulations more effective and realistic?
Do THD simulations really mean what they show, for instance? What about noise?
Just to play with, I took the LM4562 simulation I had from the datasheet, which didn't provide interesting THD results, and replaced it on the sim with a FET low noise chip listed on LTSpice, reasonably priced.
Has anyone anything to say about this sim, for or against?
Do THD simulations really mean what they show, for instance? What about noise?
Just to play with, I took the LM4562 simulation I had from the datasheet, which didn't provide interesting THD results, and replaced it on the sim with a FET low noise chip listed on LTSpice, reasonably priced.
Has anyone anything to say about this sim, for or against?
Attachments
Last edited:
most op-amps do not have distortion modeled. You can often open the model in your simulator and the comments will tell what is modeled. How well any particular spec is modeled is yet another question.
OTOH the source impedance is rising near 20dB/dec. An amp with significant input hiss current will follow that curve.
Decades ago one of the big semi makers published an analysis of phono hiss including source effect, EQ, and ear curve. IIRC the conclusions were not conclusive but HF hiss is significant despite the discount of the RIAA curve.
Doug Self may have a few words, in past books or his new one.