Best Preamp IC

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I want to build a line stage preamp and an RIAA preamp using the very best (sonically) ICs available. This is for comparison against my 18 year old preamp (with RIAA) that I designed and built using 417A tubes.

I would like 6-10dB gain. The preamp will be battery powered. I have 14 batteries each 12V/15AH. So voltage should not be an issue. Which IC should I consider?

Are there any JFET ICs, I mean not just at the input but all the way through?

Are there any ICs without local or global feedback?

I would suggest just using sockets. Then you can try them all out! Personally I like the cheap Burr-Brown OPA-134 series. JFET inputs with good specs. Built for audio.

I don't know of any pure JFET opamps. But many of the low power "5V" opamps are all CMOS. Stay away from rail-rail types.

As far as I know, all opamp ICs use no feedback. That is applied externally. You can always use fixed gain devices or instrumentation amps if you want something different.
SS Line Preamp

Hi Mohan,

I am presently developing a SS/tube hybrid line preamp, and have just completed both blocks, which will be separated by a volume control.

Contact me by phone if you would like to assist with development/auditioning; it sounds mighty good with wonderful sound stage, uses no ICs at all, and teams with a tube for output buffering.


I was going to say 6H30.:D

I have not used a lot lately, but I am a fan of the AD797, which is low noise and has good drive capabilities. The other is the INA103 which I used as a mic preamp and it didn't sound to bad.

Other suggestions from people who's opinions I trust are

LT 1354
LT 1115
LM 6172 (bit finicky)
AD 825
OPA 627

If you use an opamp, you are going to be using feedback, thats how they are designed.

If you are not deadset on opamps, perhaps look at the Pass Pearl phono and one of the line stages. Run off batteries I find ss gear generally to improve greatly in terms of noise and especially reduction in grain and that grey/white haze (Krell sounds) that affects many ss/opamp designs.

Preamp IC's

I will second the notion of the AD797, although the price is rather breath-taking (or at least it was the last time I checked.) Audio Amateur carried a pre-preamp design using the AD797 a couple years ago. The ones I have are SMT, but the DIP flavor is available for $9 or so from Newark Electronics.
Audio Amateur Magazine had a nice pre amp construction project by Grayson King. The design uses a combination of the ad744 and the ad811. Walt Jung had first written about this combination in Audio Amatuer 3/1992. The pre amp article was in 1-1994.

I think you can order the boards from Audio Amateur Magazine (now called audio express) The boards come with copies of the construction article. I use the AD811's in Norm Tracy's X-DAC and they sound quite good.
Great response and great help. No need to re-invent the wheel with all of you at hand.

Using sockets is a great tip for someone like me who has not had much to do with ICs. I plan to make up a board using different sockets. Thanks Hagtech. What do you mean by rail-rail types. Should I consider only single supply rail?

Thanks Hugh, my services are at your disposal as far as tubes and listening are concerned. I am a bit green on the cold emission side.

Brett, I shall try all of your recommended ICs. I want to have as quiet a noise floor as possible. Hence my decision to go with battery power. I have tried batteries with tubes. Greatest benefits are low noise and a marked improvement in low frequency reproduction. I used a choke input power supply with three stages of LC filtering for comparison.

I have high sensitivity (104dB 1W/1M) speakers with no crossovers for my reference. They need only 60mW at average listening levels. I like dynamics. From the IC specs, is it possible to determine the headroom available or is it simply a function of its implementation that determines the headroom?

Thanks again to Jackinnj and Audionut. I think I have those Audio Amateur magazines. I shall dig them up.

By the way, why do most ICs employ differential topologies? I know that in the discrete world, no two devices are alike and so the sum and differencing process yields a little less than perfection. Are there any Single Ended ICs? Pardon my ignorance.

When you say IC, you do mean opamp, right? Just making sure.

While it ain't a JFET part I'd sudjest the LM6172 (or LM7171 if you want single). I tried swapping different parts and I really like how those sound compared to everything else I tried. The only thing is people say it's prone to oscillation, but I just took precautions and it's never given me any problems what so ever.
Mohan Varkey said:
What do you mean by rail-rail types. Should I consider only single supply rail?

For op amps, "rail-to-rail" generally means chips designed for low supply voltages (~ 5V or less), which can accept inputs to both supply rails, and which will deliver output voltages approaching both supply rails to within a fraction of a volt. The problem is that no one input circuit can cover the whole rail-to-rail range. So most rail-to-rail op amps use dual complementary input stages that hand off as the rails are approached, causing input crossover distortion.

By the way, why do most ICs employ differential topologies? I know that in the discrete world, no two devices are alike and so the sum and differencing process yields a little less than perfection.

Transistors on ICs tend to match much more closely than discretes can ever hope to match. It's a complete turnaround from the discrete world; in IC design, closely matched transistors are cheap, decent resistors are extremely expensive, caps more so, and forget about inductors altogether!
I've built a very nice sounding phono stage using an INA103 on the input, followed by a passive RIAA leading to an AD8610 on the output. This setup is used for both MM and MC; for MC I run a pair of Jensen 346 transformers into the INA103.

I run the INA103 at +/- 22VDC and the AD8610 at +/- 13VDC. The INA103 is set at 47 dB gain, the passive RIAA loses 20 dB, and the AD8610 has 15 dB of gain (at 1kHz). This set-up maximizes overload and minimizes noise. Total gain at 1kHz is thus 42 dB (or another +11 dB or +21 dB with the Jensen ahead of the INA103). I use the INA103 in balanced mode: 24K resistors on each input leg to ground with 150 pF between the inputs for MM. I switch out this input and use the Jensen recommended impedances for the MC transformer inputs.

By running the INA103 at 22 volts, it can generate 18V max output. Since the gain is 222x (refer to the datasheets for details), the max input signal at 1kHz is 18V/222 , or about 80 mV. Since most MM don't put out more than 3 mV (eg, Shure V15), this gives about 28 dB of headroom to prevent overload even on the worse of transients (I have never seen a cartridge that tracks more than about 16 dB over quoted output).

The 18V max output is then reduced 10x by the passive RIAA to 1.8V. Since the AD8610 is running at the max voltage (+/-13VDC), it can output 11.8 V maximum. With a gain set at about 15 dB (5.75x), the max input voltage for the AD8610 is 11.8/5.75, or about 2.0 V. Therefore, as long as the input is not overloaded, neither will the second stage be overloaded with 1.8V.

I've experimented with all active RIAA, active-passive and all passive. I definitely don't like the all active. The upper pole is much better passively derived. I believe with the proper set-up (avoiding overloading of stages and minimizing noise), the all passive RIAA sounds better. This phono stage is very quiet and wonderfully musical.
rljone phono amp

Sounds like you have a great design here. The INA103 is a very low noise part with great distortion specs. 47dB is a lot of gain for one stage, but should be ok here. The bipolar inputs may be prone to demodulating RF signals, but your 150pF cap should save the day.

You might want to post a schematic here, it may be of interest to many readers.

I've done a similar design (but not with an IA), see schematic below. I don't get quite the headroom you do because of lower supply voltages. And I agree with your 16dB findings.


  • bugleschem.gif
    16.3 KB · Views: 2,889
To remove grudge, I placed in series with each input to the INA103 a 1.2 microhenry, non-magnetic inductor, in parallel with a 50 ohm resistor. I also use a series 3.5 microhenry inductor on the output in parallel with a 39 ohm resistor. On the output side, since I run a balanced stereo system, I uses another Jensen transformer (PC mount).

As for schematics, I draw things by hand on paper which I can barely read. However, I do make PCB on a laptop PC (I normally Macs) and I can upload the board datafiles for ExpressPCB ( if there is interest. The problem is that most parts are surface mount and are kinda tricky to solder unless you have a steady hand, are very myopic (or have a magnifier), and have a particularly fine-tipped iron. The only non-surface mount devices are the INA103, two '431 precision zeners, two series transistors for the +/- 13V for the AD8610 op amp, and 5 electrolytic caps. (People can order the boards from them; $60 for 3 boards.) The boards have few markings once made, but the file can be printed and shows where the parts go. There is also a 12V relay to allow switching between the MM and MC, and this has it's own 7812 regulator on board; I spec'd out a very small relay so the whole circuit board is about 2 x 3 inches per channel, including the pcb mounted output transformer.

On other issue I didn't mention was matter of DC offset. It was too much coming out of the INA103, so I had to block it, using a 8 microfard cap paralleled with a 1 microfarad cap, both in series between the INA103 and and the RIAA circuit, with a 100K resistor to ground (f3 < 0.5 Hz). I've also made one version using a servo op-amp (with no series caps) to remove DC from the INA103, but have not yet listened to it. The DC offset was too much without it.

Power supply has 6 x 22,000 microfarad caps in a C-L-C-R-C arrangement, generating the +/- 22VDC supply. L is 20 mH and R is 2 ohms. These are present in both rails; the inductors are dual units toroids having one half in the ground as well as the high voltage rail. This power supply on a scope looks as clean as a 12V battery. (I can also upload this pcb information too.)

The two phono boards including parts, plus the power supply board and its parts (transformer, caps) and the MC input and ouput transformers, and chassis cost me about $900. (Each MC transformer is $250, so this is half the cost, so figure around $400 for only MM.) I spent more time calculating and measuring this phono section than just about any other single project. But is sounds REALLY good.
Reporting on progress. I built two line stages using AD797 and LM6172 (standard configurations) and listened to them for over a month. My preference continues to be with tube line stage. This does not mean that I have given up on ICs. I shall return to these from time to time. Thanks everyone for all the help and advice.

Mohan, the AD797 would be a good choice for the phono stage as it has very high gain with low distortion, but for line level stages the AD8610 or AD8620 (dual version) is far superior to the opamps you are using.

The LM6172 is a high current bipolar opamp, a really poor choice for your application. Some people use it for headphone amps because of the high output current, but it is a finicky, cranky chip, and does not sound as good as the high end Analog Devices chips.

The AD797 is not a great choice for line level signals either. It shines for extremely low level signals such as phono cartridges, tape heads, and microphones.

Sorry I missed this thread when you originally posted. If it is not too late you might want to search HeadWize and Head-Fi, those forums are more opamp oriented.
Thanks Morsel. I am glad that you found this thread. I shall try AD8620 and report.

Seeing that you are into opamps, can you advise whether you have any clear preferences in the topology of these devices as far as audio is concerned. Are there any opamps with single ended input or output? Am I dreaming? You never never know if you never never ask!!

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.