The Phonoclone and VSPS PCB Help Desk

one of my all time favourite tracks. but there is something weird going on. it sounds like there is a swamping of the lower registers and the midband still sounds chesty. base guitar has a nice tune to it though even if it is light. i feel there is something going on elsewhere. there was other stuff i picked up on but one of the main ones was notes were being chopped for some strange reason.

i actually plugged the DT770 PRO's in to have a listen.

don't take it as criticism as it still sounds mighty fine.
 
Could be audacity screwing up some conversion (source is 24bit - wav is 16bit) - that's one explanation for chopped off notes. Conversion from 32bit might be better.. Yes, that's the characteristic of the Grado. A new stylus (better one - MCZ) is planned, that should make it a little brighter. Used to have an Ortofon 2M red, which is good on treble, however, it's extremely sensitive to ssssssshhibilance :D

Oh and I'm not taking it up in a bad way. Not at all :) we're here to learn :)


Wirehead, got the message, thanks. The Pixies in stereo sounds much better. Is your VSPS a 300 or stereo model?
It's a regular stereo model, diy - no pre-made pcb :)
 
Last edited:
Is the joke on us?

A small fee paid to one of the repositories on the internet and I was able to download the service manual for the Marantz SR220, the 1970's receiver that provided the comparison for the VSPS300 listening test earlier.

Irony? Well, the SR220 is a VSPS, more or less.

TL4558 op amp, single stage active RIAA. The op amp (dual u741 essentially) is pretty nasty by today's standards, and the power supply cuts corners (the positive rail is a zener with a pass transistor, the negative rail is just a zener), but the circuit itself is almost the same. The simulated response shows the same high frequency peaking noted in the tests, but no nearly as significant in magnitude. It was not designed to be +2 dB at 10 kHz, either the components aged with time, or tolerances were not significantly tight.
 

Attachments

  • SR220 RIAA.png
    SR220 RIAA.png
    271 KB · Views: 195
  • fig4.png
    fig4.png
    49.7 KB · Views: 193
  • fig1.png
    fig1.png
    13.5 KB · Views: 181
  • fig2.png
    fig2.png
    12.3 KB · Views: 179
Resistors and film capacitors are marked 5% in the component list. The specified RIAA frequency response error is +/- 2 dB, and the phono section S/N is given as 70 dB.

The coupling capacitors are 4.7 uF electrolytic, and are not marked bi-polar. Its fair to assume regular polar types are used. This seems to be quite typical. I can only imagine that the DC offsets were so small that a reverse bias situation would cause no harm...
 
I think that we should not trust the simulation, sometimes do not give logical results..

See the value of the network and see the RIAA curve, some values ​​are not correct..

After all, it is impossible that RIAA curve is correct with the values ​​of resistors 750k and 110k, whatever LTSPICE showed..

I think it's only right phono A simulation..
 

Attachments

  • phono A.png
    phono A.png
    33.8 KB · Views: 175
  • RIAA phono A.png
    RIAA phono A.png
    31.9 KB · Views: 59
  • phono B.png
    phono B.png
    22.1 KB · Views: 53
  • RIAA phono B.png
    RIAA phono B.png
    29.9 KB · Views: 50
  • phono C.png
    phono C.png
    23.1 KB · Views: 52
  • RIAA phono C.png
    RIAA phono C.png
    29.9 KB · Views: 46
The larger of the two resistors in the network adjusts the bass frequencies, and is often compensated for by roll-off from any of the coupling capacitors elsewhere in the circuit. It's the least sensitive of the RIAA components. Elsewhere, you should find that you can move any pair of R,C values a good way away from the "mathematically correct" values with comparatively little effect, if the deviations tends to cancel rather than reinforce.

In any event, its really impossible to see anything on the response chart unless you plot the deviation from idea, not just the absolute signal magnitude vs. frequency.

LTSpice is reliable for these simple network calculations. I used to do them in MathCAD or occasionally even Excel (pro tip: its a pain, don't bother) for the same result, but these days I'm too lazy and its so much easier to just run it up in LTSPice...
 
The larger of the two resistors in the network adjusts the bass frequencies, and is often compensated for by roll-off from any of the coupling capacitors elsewhere in the circuit. It's the least sensitive of the RIAA components. Elsewhere, you should find that you can move any pair of R,C values a good way away from the "mathematically correct" values with comparatively little effect, if the deviations tends to cancel rather than reinforce.

In any event, its really impossible to see anything on the response chart unless you plot the deviation from idea, not just the absolute signal magnitude vs. frequency.

LTSpice is reliable for these simple network calculations. I used to do them in MathCAD or occasionally even Excel (pro tip: its a pain, don't bother) for the same result, but these days I'm too lazy and its so much easier to just run it up in LTSPice...
 
The larger of the two resistors in the network adjusts the bass frequencies, and is often compensated for by roll-off from any of the coupling capacitors elsewhere in the circuit. It's the least sensitive of the RIAA components. Elsewhere, you should find that you can move any pair of R,C values a good way away from the "mathematically correct" values with comparatively little effect, if the deviations tends to cancel rather than reinforce.

In any event, its really impossible to see anything on the response chart unless you plot the deviation from idea, not just the absolute signal magnitude vs. frequency.

Relation of the two resistors determines the curve, 750/110 can not possibly give a 20dB gain, no matter what the simulations showed..

I think it's easy to see everything on the response charts..
 

rjm

Member
Joined 2004
Paid Member
After all, it is impossible that RIAA curve is correct with the values ​​of resistors 750k and 110k, whatever LTSPICE showed.

LTSpice is accurately modeling the frequency response of the RIAA network.

Perhaps you are thinking of the conventional RIAA network where the resistor and capacitor are connected as (R1||C1)+(R2||C2) and R1~10xR2. (see the input inverse-RIAA I'm using in the sims below for an example) In the VSPS/Phonoclone it is R1||C1||(R2+C2). The ratio of the resistance values are different.

/R
 

Attachments

  • vulejov results.png
    vulejov results.png
    55.9 KB · Views: 185
Perhaps you are thinking of the conventional RIAA network where the resistor and capacitor are connected as (R1||C1)+(R2||C2) and R1~10xR2. (see the input inverse-RIAA I'm using in the sims below for an example) In the VSPS/Phonoclone it is R1||C1||(R2+C2). The ratio of the resistance values are different.

Look again the results for a phono A and phono B in post 2287, it's the same RIAA network but very different curve..
 
What exactly is different about the two curves?

Gain, but that's not my point..
Correct question is what exactly is different abot the two RIAA networks.. nothing!

It's possible the transistor circuit is different. The network values are the same, but the impedance it works against may not be constant over the audio band.

Again, some simulations are wrong..
 
Last edited: