• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Openamp MM phono preamplifier

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,777
Likes
11,175
Location
Prague
This is a review and measurements of my Openamp phono preamplifier. The project was released in 2012 as an open project.


Openamp is a preamplifer for MM phono cartridge. It uses operational amplifiers, a monolithic buffer and a feedback RIAA correction. DC servo is used to remove DC voltage from the output. Output stage/buffer of the Openamp works at 20mA idle current. It remains in class A for all amplitudes of output voltage for loads 600 ohm and more.
The preamplifier is supplied from 2 x 15V to 2 x 18V regulated power supply.
Gain is 40dB

openamp_pcbassy.jpg


Openamp Measurements

Let's start with the output noise plot, with the input shorted by 50 ohm. Our sample uses OPA134 opamps, which are not the parts with the lowest noise, however the JFET input opamps are better for the MM phono preamp, one of the reasons is the absence of the input bias current flowing through the MM cartridge.

Openamp_noise_E1DA_2.png


We can see that the unweighted output noise over 16Hz – 45kHz band is 86uV, this goes down to 78uV over 22Hz-22kHz unweighted and 45uV when A-weighted. So the 1kHz SINAD at 5mV/1kHz input (500mV output) cannot be better than 76dB (22Hz-22kHz). Let's see.

Now the usual measurement with 5mV/1kHz input:

Openamp_0.5V_E1DA_2.png


We get THD = -106dB (excellent) and SINAD = 73.9dB (not so good, noise limited. OPA627 as input opamp is about 5dB better in noise).

Let's investigate distortion near clipping point:

Openamp_8.4V_E1DA_2.png


At 8.41V output, which means 84.1mV input, we get THD = -97.7dB and SINAD = 94.5dB. This is a very nice overload margin.

“Most important measurement in a phono stage is frequency response”, as @amirm used to say, so here we go:

Openamp_FR_E1DA_2.png



Now the THD and THD+N vs. output voltage plots

at 1kHz

Openamp_FR_THDN_1kHz_2.png


We can see the distortion “knee” at 9V.

And at 5kHz

Openamp_FR_THDN_5kHz_2.png


Unfortunately my DAC + output divider does not yield enough noiseless voltage at 5kHz (remember RIAA curve makes lower gain at higher frequencies) and if I omit the output divider, I get too much DAC noise. However, oscilloscope shows there is no problem with clipping level at high frequencies, it remains the same as at 1kHz.

Conclusions

This project was given to the public 10 years ago, together with schematics and Gerber files. I believe it made a very good MM preamp, with very high output drive capability due to output LT1010 buffer used. One may load it with 100 ohm without problem, with high overload margin.
 
Last edited:
Let me show a less conventional look at distortion plots:

Openamp_THDN_dBFS.png


It is the same plot as THD and THD+N vs. voltage plot at 1kHz as shown in post #1, but now in dBFS scale.

The top trace shows rising amplitude of fundamental frequency, we go from -40dBFS to 0dBFS. Below, from top, is a THD+N trace and it reflects just noise, THD+N is here defined just by noise.

Below the THD+N trace is the THD trace. This is distortion only, without preamp noise component. But, we can see the system and FFT "noise floor", which is somewhere at -130dB. So the real distortion can be seen just above 1V and we can see it rises from -130dB at 1V to -100dB at 8V.
 
20kHz voltage swing. Again, there is no limited swing at this high frequency. Show me the second phono preamp with such behaviour :). This sample had 2x17V power supply, thus the 30Vpp swing. With 2x15V supply, the swing is 26Vpp.

1665172923520.gif
 
Last edited:
Hi PMA,
You may recall I mentioned previously at "the other place" how much I liked my build of your preamp.
It's nice to see impressive measurements supporting the quality I've heard.
:cool:
mlloyd1
 
Some more measurements, as measuring phono preamplifiers is always a challenge:

openamp_noise.png



THD and THD+N with 1kHz/5mV input
openamp_thdn_5mV.png


1kHz 4V output
openamp_thdn_4Vout.png


22Hz 4V output
openamp_22Hz_4Vout.png


10kHz 1.1V ouput
openamp_10kHz_1Vout.png


10kHz 4V output
openamp_10kHz_4Vout.png



THD and THD+N vs. output voltage at 1kHz up to 4V
openamp_thdn_level_1k.png
 
Last edited:
Looks like really excellent performance. From the last plot, it looks like noise dominates over the entire output range, at least at LF (where RIAA boost is the most, natch) and distortion is very low. Looking earlier, overload performance is also excellent, one of the areas some phono preamps falter (if not fail dramatically).

Almost makes me want to get my old TT out of storage...
 
  • Like
Reactions: pma
Looks like really excellent performance. From the last plot, it looks like noise dominates over the entire output range,

Yes, it does. The version measured is with OPA134 input opamp. Its voltage noise defines to overall noise. If this opamp is changed to OPA627, the noise goes down of 4dB. With OPA827, we should get even lower.

JFET input opamps are important part of the RIAA MM preamp design. Though BJT opamps may have lower voltage noise, their input current noise creates across cartridge inductance and resistance additional noise voltage and the resulting preamp output noise (with cartridge connected) is then worse than with good JFET opamps.
 
Yes, it does. The version measured is with OPA134 input opamp. Its voltage noise defines to overall noise. If this opamp is changed to OPA627, the noise goes down of 4dB. With OPA827, we should get even lower.

JFET input opamps are important part of the RIAA MM preamp design. Though BJT opamps may have lower voltage noise, their input current noise creates across cartridge inductance and resistance additional noise voltage and the resulting preamp output noise (with cartridge connected) is then worse than with good JFET opamps.

YES! I have many times in the past noted that folk must pay attention to the details when swapping op amps, especially in phono preamps. A bipolar input, especially compensated for input current, will have very high noise due to base current (and the compensation current injection if present) as you say when driven by a high-impedance source (like a MM cartridge). That makes a lot of classic "low-noise" opamps unsuitable for phono stages. A MOSFET input will also have very high noise due to 1/f and similar low-frequency noise. Chopper-stabilized versions have that plus noise from the chopper circuit. JFETs are the transistors of choice for high impedance and low noise phono preamps.

One cautionary tale I always relate is a friend, actually a good circuit designer, who decided to use low-noise microwave GaAs (or InP, I forget) FETs in a discrete phono preamp as we had plenty of samples around. The catch, as he soon discovered, was that a 100+ GHz GaAs FET has a 1/f noise corner around 100 MHz, just a bit above the audio band... ;)
 
With minor apologies for bumping a slightly old thread...

I've been reading Douglas Self's "Electronics for Vinyl" in which he compares the theoretical noise of various choices of op amp in phono stages. He makes a point, inspired by an article by Marcel van de Gevel in Electronics World (October 2003), that the cartridge inductance means that current noise is more problematic than you think. Nevertheless his modelling seems to show that 5534 op amps result in the best noise behaviour.

The ven de Gevel article suggests using a figure of 12k for the effective impedance of an MM cartridge. Taking that at face value, the noise contribution of a 5534s input current noise (0.6pA/rtHz) is equivalent to 7.2nV/rtHz . So the OPA627 specified in the OpenAmp documents would outperform this, having noise voltage density of 5.2nV/rtHz and essentially no noise current (down in the fA region). However the OPA134 used in the measured sample with 8nV/rtHz would be a bit noisier than a 5534.

Nowadays, at least when components are available, we have access to cheap and quiet devices like the OPA167x. It would be interesting to see whether the predictions come true: is it true that a 5534 is comparable to the OPA134 in the above circuit? And how much quieter is the OPA1678? Of course this requires measuring when connected to a cartridge.

Self makes the point that searching for very low noise in this application seems like a waste of time, given that groove noise will overpower the noise of the phone stage anyway, but claims it is still worth doing because it is easy and cheap to achieve. The OPA627 doesn't qualify as cheap! But OPA1678 certainly does.
 
Last edited:
best way to find out is to buy some op amps, drop them into your openamp and see which you like best.
i started with opa627, tried a couple of other jfet input parts and went back to opa627. it was quietest and had most pleasant sound for me.
my noise levels were not measured; i just listened to hiss with my catridge attached.
maybe if you ask nicely, you can convince PMA to try some other parts you are curious about and present actual measurements.
as i recall, he did measure noise for some alternate parts, just not the latest Burr Brown/TI parts you asked about.
but I'm one of those weird folks that think that even though the opa627 and opa637 are old, they are still great parts in the right application. i have old stock - my parts have the Burr Brown logo; the original mfr before TI swallowed them.

good luck!
mlloyd1
 
Back
Top Bottom