Tír Na HiFi

In another thread viewtopic.php?f=8&t=4077, I mentioned that rediscovering my old 1985 Sanyo amp to access its Phono input left me decidedly unimpressed, and I was considering building a simple phono stage. Frydryck replied that he had built the design I was considering a few years ago and he was kind enough to loan me it. Well it was a complete revelation - the best records that I have sound quite magical through his preamplifier. It is a very different experience from CD and I would be pushed to say which I prefer. Voices in particular sound quite remarkable to me now on vinyl. With some (not all) recordings, there's a presence that CD lacks. So I attempted the build, having all the components lying around at home.

Now I know measurements aren't everything, but that I believe is because we haven't fully correlated the measurements to what the ear is sensitive to. In any case, if something sounds like shit (1985 Sanyo) and something sounds wonderful (Frydryck's build) then I'd hope this can be quantified. A colleague loaned me an Audio Precision analyser, and using this I could work out the amplifier responses.

First up, the Sanyo . Here, as a function of frequency, I plot the difference in decibels between the Sanyo gain and the RIAA response, for a 4mv input signal. All curves are normalised to the response at 1kHz. Hardly a stellar performance, with deviations of one to two dB at the highest and lowest frequencies. More to the point are the noise numbers. The data sheet for the Sanyo advertises a signal to noise of 95dB - but that's through the line-level input. They are curiously silent about the phono performance. I can see why. Referenced to the output of a 4mv signal at 1kHz, I measured the unweighted noise between 22Hz and 22kHz to be 31dB, while the A-weighted noise (corresponding more closely to what the ear is sensitive to) is 55dB.

Now, Frydryck's DIY based on http://phonoclone.com/diy-pho5.html. The components are mounted on a PCB bought from the website with some very nice upgraded capacitors. Frydryck mounted the power supply in a separate box and it's fully dual mono: toroidal transformers with Schottky diodes bypassed with capacitors. The build quality is superb. Here's the RIAA response (note I had to change the scale on the y-axis!) . Wow! Less than 0.2dB across most of the range and perfect matching between left and right channels. I also did a SPICE simulation for the circuit and you can see that the deviation from the RIAA curve at the highest frequencies is due to the circuit design. The signal-to-noise values are 44dB unweighted and 70dB A-weighted. I think that's well below the intrinsic noise of most turntables.

So finally to my build. This just uses what I had lying around: so a wall-wart transformer with the rectification, smoothing and regulation in the same box as the phono circuit, which is just soldered on a breadboard. The capacitors are only ceramics. Here's the RIAA response. . There's a mismatch of 0.4dB between the channels at high frequencies while the signal-to-noise values are 42dB unweighted and 68dB weighted. So, not quite as good as Frydrycks but a heluva lot better than the old Sanyo. There are some obvious improvements I can do - making a proper PCB and possibly upgrading the capacitors, but I suspect I'm already below the turntable noise floor, and more importantly, the vinyl I'm spinning is sounding wonderful. I guess some of the hype is right after all.

That are very nice measurements. Would you be able to measure the THD of the circuits at 1Khz?

looks very simple, was thinking of building one for a friend,

but would like to incorporate volume control, so probably followed by a potentiometer
is that viable?
whats the minimum voltage supply required?

From what I know about this circuit, the PCB ground layout is absolutely critical as well as placement of the decoupling caps.
In terms of the voltage the circuit will happily run on anything between 10-15V, but the performance would depend on the opamp used. I do not know about the volume pot. I presume that there is nothing against it, but I think the type of opamp used will have an influence on this.

hmmn
wasnt planning on putting much effort into it,
i was thinking if it could be recalculated to work with 2 lipo cells 6.6 v and output pot
all done point to point fitted and into a small box
ive no opamps so that can be whatever is easily available
maybe im just making work for myself

do you remember what the gain is?

I was not precise enough. The phono stage requires symmetrical power, that is +/- 15V. Two lipo cells won't do. You need eight of them. If I remember correctly the gain will be about 40dB.

F

frd1996 wrote:I was not precise enough. The phono stage requires symmetrical power, that is +/- 15V. Two lipo cells won't do. You need eight of them. If I remember correctly the gain will be about 40dB.

F

looking at opa134 datasheet says +/- 2.5v up to 18 vdc
also says their testing was done at +/- 15vdc but results but results were more or less consistant in above voltage range
considering having upmost sq aint that important for this app
+/- 3.3v dc could work?
only fear begin that gain vout would exceed ps capability and clip?
in which case gain could be reduced by lowering the feedback loop resistor value to gnd?
or does that bugger up the Allen Wright mod?

edit... looks ok i see this on richard m's site about that resistor to gnd

"This resistor sets the gain. Lower values result in more the gain. Use 2.2k for 30dB, 680R for 40dB, and 220R for 50dB."

frd1996 wrote:That are very nice measurements. Would you be able to measure the THD of the circuits at 1Khz?

The measurements were performed by feeding the phono amplifier with a 1kHz signal. The output was analysed using a notch filter to remove the source, so this gives the total noise. To get the harmonic distortion from a 1kHz signal would require scanning the output as a function of frequency, and the Audio Precision kit didn't have this feature.
What I tried doing, was to send the amplifier output back into my computer and taking the Fourier Transform. The result is in the plot. Major caveats: the ADC in the computer is probably rather poor, and the power supply spuria are almost certainly due to the computer and not the amp. I can't see any harmonics on the 1kHz signal, but this is not a precise result.

nige2000 wrote: +/- 3.3v dc could work?
only fear begin that gain vout would exceed ps capability and clip?
in which case gain could be reduced by lowering the feedback loop resistor value to gnd?
or does that bugger up the Allen Wright mod?

edit... looks ok i see this on richard m's site about that resistor to gnd

"This resistor sets the gain. Lower values result in more the gain. Use 2.2k for 30dB, 680R for 40dB, and 220R for 50dB."

I think the gain quoted is for a signal at 1kHz. The lowest frequencies have an additional 20dB boost from the RIAA. Therefore if you put in a 680R resistor, a 4mV signal gets amplified to 4V. You'd want to make sure the power supply was well in excess of that to prevent clipping. Also some MM cartridges will output more than 4mV)

Another point from the web site is that if you change the feedback resistor you also have to change the 'Allen Wright' resistor, but that's not a big deal.

http://www.sengpielaudio.com/calculator-gainloss.htm

The gain calculator here states 40db is vin x 100
In this case 3.6mv x 100 = 0.36volts meaning there's still loads of head room at 3.3v

The diagram from the website seems to suggest that r2 isn't part of the Allen right MOD and changing its value doesn't affect this adjustment
And that r2 value changes is a way to adjust the gain