Ultimate preamp IC ?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Now I'm convinced by power chip amps, I'm willing to add a buffer to my bridged LM4780 amp.

The bridged mode has a quite low input impedance, <1kohm on one of my amps, and I feel like my soundcard doesn't give its best through such a load.(it is a **** laptop soundcard, but I will replace it with a TransitUSB)

Anyway, I was wondering : what would be the best ICs (semiconductors, not valves) to drive my LM4780 ?

I have found the Analog Devices AD797 to be an excellent choice :
- 0,9nV/Hz input noise
- 25µV offset, allowing a output-capacitor-less design
- -120dB distorsion through 1kohm@1kHz

But I didn't see any well documented DIY preamp using it, so I'm wondering it there was anything better to drive chip amps.

Is there anything better than it ? I know carlosfm likes the AD815, and a french shop sells a preamp based on the AD825.
There are also opamps from Burr-Brown like the OPA2604, which seems to be liked very much also.


Any comment ?
 
Try the LF353 in a unity gain buffer. The specs are nearly identical to the LM3875 chip (except power dissipation).

I like then as they sound very 'natural, detailed but un-hifi ... and cost next to nothing!

Use one chip perchannel - just dont connect the remaining opamp to the circuit (the LF353 is a dual opamp) - or at a push use a single chip for both channels, but imaging suffers a little..

Owen
 
I guess I am in a little different camp -- I have used Analog Devices and Linear Tech -- the AudioXpress mod of the Adcom GFP565 used ADI devices and sounds wonderful. I have used BB(TI), but just have more stuff from the former 2 vendors.

It may be deemed "heretical" but I think that in the realm of high perf audio opamps the world flattens out and 1) the quality of the supply rails and 2) good layout dominate performance.

the AD797 is a great chip for low noise applications, but it can be a bit tricky to work with -- and it is quite expensive. You can get really great performance from a less expensive chip if you use 2sk170/2sj74 as a pre-pre.
 
You mean I could achieve greater performance with a discrete design ?

Nuuk seems to have a big offset voltage which I'm quite afraid of. I'd like a capacitor-less signal path if possible.

Have to study PSU bypassing. I will surely build a double regulated PSU.

Got some NE5534 arround to play with, but no regs and no support to hold the opamp. They are said to be unstable at unity gain.
 
do you need gain or simply need impedance conversion?

dual jfet buffers (i.e. discrete) work quite well and can be made with excellent offset performance fairly easily.

mlloyd1

youyoung21147 said:
You mean I could achieve greater performance with a discrete design ?

Nuuk seems to have a big offset voltage which I'm quite afraid of. I'd like a capacitor-less signal path if possible.

Have to study PSU bypassing. I will surely build a double regulated PSU.

Got some NE5534 arround to play with, but no regs and no support to hold the opamp. They are said to be unstable at unity gain.
 
youyoung21147 said:
I have found the Analog Devices AD797 to be an excellent choice :
- 0,9nV/Hz input noise
- 25µV offset, allowing a output-capacitor-less design
- -120dB distorsion through 1kohm@1kHz
I must remind that you can't achieve so good performance as some parameters suggest. You must know how your signal source is in order to choose the right opamp.

Noise is in most cases uninteresinting. Most modern opamps won't add any noise if we talk line level signals.

My Gainclones use AD8620 or OPA2134 but there are many many more to choose from. I recommend an opamp with JFET inputs.

NE5534, much hifi for the price
OPA134/2134
OPA627 (a bit expensive)
AD8610/8260 (only SMD)
AD825 (only SMD)
AD745 (only SMD)
many many more.
 
Noise is in most cases uninteresinting. Most modern opamps won't add any noise if we talk line level signals.

What noise treshold can be considered "correct" and not causing "hiss" on the speakers ?

AD745 (only SMD)

I thought of this one too, it is more adapted to 1->10k input impedance sources, allowing a pot before the opamp.

What where your criterias for choosing your buffer opamps ?
 
Compare the noise from a CD. If you consider it "noise free" you want 80-85 dB S/N or more. If you have a line level amp it is impossible to get less than 90-100 dB S/N. If you want 110-130 dB S/N you might study the noise parameter more. So what I'm trying to say is that the noise is unimportant in most cases. It's only important in RIAA and mic amps.

If you choose a JFET input opamp you are much more free to choose resistor values and will get less output offset due to input offset currents.

The criteria when choosing buffer opamps is that the properties when the gain is 1 or slightly more. Some opamps aren't stable for gains less than 5 or 3. Check the datasheet for this. OPA637 for emxaple can't be used for gains less than 5.
 
OK, thanks !

I have checked the noise from my c r a p laptop soundcard, and it has -73dB SNR. Noise is clearly audible in headphones, and quite noticeable when listening at normal levels, especially on quiet passages or when the soundcard is idle. It's really boring, and that's maybe why I go a bit "paranoiac" when talking about noise :D


If you choose a JFET input opamp you are much more free to choose resistor values and will get less output offset due to input offset currents.

I have read this in datasheets, and that's one of the reasons I where interested in the AD797 : when used on a low-impedance source, offset may be as low as 0.1mV !

Finally concerning PSU, I was thinking about batteries : is there anything purer than a battery PSU ? I have some 12V/ 25Ah lying arround, so it wouldn't cost me much doing the PSU this way.

In this case, is any decoupling next to the IC necessary ?
 
Ex-Moderator
Joined 2002
youyoung21147 said:
In this case, is any decoupling next to the IC necessary ?

Decoupling is always a good thing, and may even be vital for stability. Batteries are not an ideal source, they have internal resistance that affects their ability to provide current quickly, so PSU caps do improve matters. Howver they can generally be significantly smaller than AC supply equivalents.
 
youyoung21147 said:
I have checked the noise from my c r a p laptop soundcard, and it has -73dB SNR. Noise is clearly audible in headphones, and quite noticeable when listening at normal levels....
That means anything will do. :nod: 73 dB is a cassette deck with dolby!
youyoung21147 said:
Finally concerning PSU, I was thinking about batteries : is there anything purer than a battery PSU ? I have some 12V/ 25Ah lying arround, so it wouldn't cost me much doing the PSU this way.
A battery is massive overkill. Also here almost anything will do, like a LM317/337, 7815/7915 based supply.

youyoung21147 said:
In this case, is any decoupling next to the IC necessary ?
Nothing to think about, I mean, not to have. You just have it as a basic rule.
 
youyoung21147 said:

Finally concerning PSU, I was thinking about batteries : is there anything purer than a battery PSU ? I have some 12V/ 25Ah lying arround, so it wouldn't cost me much doing the PSU this way.
You can filter the battery.

If you don't want to go to the extreme of building the super-regulator consider using the very low noise regulators from Linear like the LT1763 and LT1964 -- at least 2 orders of magnitude better performance than the LM317/LM337.

bypassing is a good idea, I use 100nF.
 
That battery sounds fine - remember to bypass the chip on the board, and you'll have a ball.

The current limits on battery delivery are very similar to a regulated supply, so no worries there.

I have a number of BB and AD opamps - and find them to be very 'hi-fi' sounding - a bit brittle and etched - the humble 353 is a dual differential Jfet input

================================================
LF353 Description
Dual General-Purpose JFET-input Operational Amplifier
LF353 Vendor
Texas Instruments
LF353 Features

* Low Input Bias Current...50 pA Typ
* Low Input Noise Current 0.01 pA/ Hz\ Typ
* Low Input Noise Voltage...18 nV/ Hz\ Typ
* Low Supply Current...3.6 mA Typ
* High Input Impedance...1012 Typ
* Internally Trimmed Offset Voltage
* Gain Bandwidth...3 MHz Typ
* High Slew Rate...13 V/us Typ

Parameter Value
Number of Channels 2
Available Channels D
Vs (Min) (V) 7
Vs (Max) (V) 36
Iq per channel (Max) (mA) 3.25
GBW (Typ) (MHz) 3
Slew Rate (Typ) (V/us) 13
VIO (25 deg C) (Max) (mV) 10
Offset Drift (Typ) (uV/C) 10
IIB (Max) (pA) 200
Pin/Package 8PDIP,8SOIC
CMRR (Min) (dB) 70
Vn at 1kHz (Typ) (nV/rtHz) 18

===============================================

LM3875

Gain Bandwidth (MHz) 8
Slew Rate (Volts/usec) 11
SupplyCurrent Per Channel (mA) 30
Supply Min (Volt) 20
Supply Max (Volt) 84
Offset Voltage (mV) 10
Max Input Bias Current (nA) 1000
Output Current (mA) 6000
Shut down No
Special Features AvCl>10

================================================

As the '353 can be picked up for $0.20 or so, add $0.20 of resistors, using your batteries as a supply, you can try it for yourself...

Owen
 
Yes, maybe 25Ah batteries may be overkill here :D

Anyway, I'll surely give it a try.

Concerning regulators : thanks Jackinnj ! These devices look very pretty.

I think that the PSU will be filtered through two stages of regulation, for example LM317+LT1763 or even LM317+LT1763+7815, to have battery-like DC.

As usual, I see that the biggest work is the PSU :smash:
 
You gain very, very, very little using regulation on a battery supply - the LF353 runs nicely on +/- 12V !

Beware... you may even 'worsen' the situation adding regulators - this is why I used the '353... a simple brute force 12V supply with a couple of thousand uF close to the chips themselves sounds by far the best to my ears.... and damned close to a thoriated filament DHT pre I built (infact it was significantly better in some ways..!)

I made the judgement call in favour of the LF353 with the LM3875 - just seems to do more right!

If you start simple, and then build up, you can more easily compare notes as complexity increases,a nd decide what is best for you...

Have fun!


Owen
 
owen said:
Beware... you may even 'worsen' the situation adding regulators - this is why I used the '353... a simple brute force 12V supply with a couple of thousand uF close to the chips themselves sounds by far the best to my ears.... and damned close to a thoriated filament DHT pre I built (infact it was significantly better in some ways..!)
I am glad you used a subjunctive verb.

Why would you use a chip with inferior PSRR, CMRR to drive the LM3875, LM3886 or LM4780?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.