C3g Aikido Phono

[Andrew]

It depends what you want to study?

If you want to view what the RIAA correction looks like, then the wire that goes into the grid stopper on the first valve of the second stage is where I would start, use a directive something like '.ac oct 16 20 20k'.

[Nick]

notwithstanding Nick's comments about importance of overall RIAA accuracy.

Well, I find it important, but having measured commercial stages, many are all over the place, so YMMV, but I certainly think if you want to compare caps, then you need to be sure just what you are comparing.

[simon]

That makes a lot of sense Nick.

Right, I've had a play and assuming I have a grasp of what's going on (dangerous!) it's been very interesting. Starting with the phono as it is currently:

RIAA existing.png (72.72 KiB) Viewed 3017 times

Two things that strike me:

First, there's approx 0.18dB peak to peak difference in signal level between 20Hz and c850Hz. Pretty impressive I think.

Second, above 1kHz there's a bit of a dip centred around 8kHz, perhaps not ideal?

[simon]

Next sim is with a 15k series resistor before the RIAA, and the RIAA values changed as discussed previously.

RIAA new 1.png (66.86 KiB) Viewed 3017 times

Now there appears to be 1.6dB difference, but it looks nicer above 1kHz

[simon]

Next I played with the series resistor, this is with 20k.

RIAA new 2.png (74.37 KiB) Viewed 3017 times

The voltage difference is now down to ~0.35dB, but it doesn't look quite so sweet above 1kHz.

[simon]

The sim seems very sensitive to small changes in component values in the RIAA, or perhaps this just reinforces Nick's point about accuracy. This is with a 3% reduction in the value of the resistors.

RIAA new 3.png (77.57 KiB) Viewed 3016 times

There's about 0.33dB difference but it's going a bit rollercoaster above about 200Hz

Edit - I thought I had another plot with the series R changed back to 15k which had a 1.7dB difference but looked smoother, but I don't

[simon]

Finally I tries the existing configuration with a 100nF coupling cap instead of the 56nF which shows slightly less difference at 0.12dB rather than 0.18dB, but just a little more lumpy.

RIAA existing 100nf coupling cap.png (73.35 KiB) Viewed 3015 times

And the 56nF coupling cap for ease of reference

RIAA existing 100nf coupling cap.png (73.35 KiB) Viewed 3015 times

None of which tells me what might actually sound better. Within reason I wonder if a smoother plot might be more important than smaller voltage difference? But it rather looks like Andrew did a very nice job in the first place?

[Nick]

Now there appears to be 1.6dB difference, but it looks nicer above 1kHz

Be aware that as the scale changes to fit the larger LF dip, the HF will look flatter. It may actually be the same.

Once you start increasing the series resistor (from 0) you will have to readjust the caps and resistors to match.

[Nick]

Try inserting a 1R series resistor and measure the current through that against frequency, then set the signal level to represent what actually happens and see what current is being pulled.

[simon]

Now there appears to be 1.6dB difference, but it looks nicer above 1kHz

Be aware that as the scale changes to fit the larger LF dip, the HF will look flatter. It may actually be the same.

Fair point, I wasn't thinking clearly. For completeness the peak to peak difference for the bit higher than@1kHz is:
existing RIAA - 0.03dB
existing RIAA with 100nF coupling cap - 0.04dB
15k series resistor - difficult to specify but could argue it's 0.1dB or greater even
20k series resistor - 0.03dB
20k series resistor RIAA resistors 3% lower - 0.07dB

Once you start increasing the series resistor (from 0) you will have to readjust the caps and resistors to match.

I thought that's what I'd done, but wonder if I've misunderstood?

I need to think about your second email, not sure I follow it at first reading.

[Nick]

I thought that's what I'd done, but wonder if I've misunderstood?

So you did, sorry, I didn't notice.

When using spice and RIAA, I find setting the value of a component to a parameter, and then sweeping that parameter across a range of variables to make it simpler to pick the optimum value.

https://www.analog.com/en/technical-art ... lysis.html

[Andrew]

That makes a lot of sense Nick.

Right, I've had a play and assuming I have a grasp of what's going on (dangerous!) it's been very interesting. Starting with the phono as it is currently:

RIAA existing.png

Two things that strike me:

First, there's approx 0.18dB peak to peak difference in signal level between 20Hz and c850Hz. Pretty impressive I think.

Second, above 1kHz there's a bit of a dip centred around 8kHz, perhaps not ideal?

That's not a lot of a dip, as you say, look at the scale.

[Andrew]

Next sim is with a 15k series resistor before the RIAA, and the RIAA values changed as discussed previously.
RIAA new 1.png
Now there appears to be 1.6dB difference, but it looks nicer above 1kHz

By adding the series resistance you have made it an easier load but the bass now rolls off, doesn't it?

[Andrew]

Try inserting a 1R series resistor and measure the current through that against frequency, then set the signal level to represent what actually happens and see what current is being pulled.

Real or imaginary, or both?

[simon]

Try inserting a 1R series resistor and measure the current through that against frequency, then set the signal level to represent what actually happens and see what current is being pulled.

When using spice and RIAA, I find setting the value of a component to a parameter, and then sweeping that parameter across a range of variables to make it simpler to pick the optimum value.

https://www.analog.com/en/technical-art ... lysis.html

Okay, I've added a 1R after the 56nF coupling cap but can't make the link with how to set the signal level. The current pulled is variable so how is the current set?

1R series resistance.png (135.87 KiB) Viewed 2841 times