Intact Audio

Hello,

during my studies concerning the usage of the differential amplifier for LR or LCR RIAA I want to present some results to you.

The results based on an D3a differential amplifier which was designed totally balanced so that both tubes are in series with respect to the signal. So they are forced to swing the same signal.

The distortion spectra is dependent on the load here which is mostly dominated with 3rd order distortion for higher levels. The 3rd order distortion is getting higher if the load is more heavy. For practical reasons I shunted the resistive load with an 1nF capacitor. We always have to deal with capacitances in the real world.

You can see that the 2nd distortion stays below 10**(-5) while the 3rd seems to have an minimum for lower output levels but the raises very steep.

The overall distortion is very low but what may missing here would be the single ended sound with more second order distortion.

From this point of view the load should not be so heavy if you prefer low 3rd order distion.

Best regards,
Bernd

Hello,

but I found that you can use a trick which was inspired by Linn Olson and Stephie.

Just shunt the middle of the output transformer to the cathodes and feed the amplifer with a slight unbalanced signal. The capacitor only comes into the game if the output signal is not symmetric.

In this configuration you can tune the second order distortion while the 3rd order distortion remians fairly constant compared to the balanced amplifier. The idea behind all this is that you have a single ended amplifier that runs several dB below the push pull amplifer. How far the SE amp runs behind the PP depends on how symmetric is the input signal feeded.

Best regards,
Bernd

Hi,

I would like to show the effect of unbalancing the inputs for a fixed load of 600 Ohms parallel with 1nF.

The sum of both resistors always equals 47k. if r1=22k then r2 is 25k.

As you can see from the graph the 3rd order distortion is quite the same ofr all scenarios but second order can be trimmed to become dominant. So, a slight difference of +- 1% already prefers the 2nd order distortion. If you use an poti you have something like an harmonic tuner. 4th and 5th order is not shown here as they are far below.

Best regards,
docali

Hello,

here you see an example of an fully differential LCR that works quite well with the class A push pull differential D3a stage.

Best regards,
docali

Wow, this is killer. This is very close to the phono stage I wish to build. I would like to modify the EQ RIAA to eliminate the resistance, I think.

Question: what software are you using for the simulation and the distortion analysis?

VB, Maturin

Hi Maturin,

I used LTSpice for the simulations but the graphical software is Ptplot.

What resistance do you want to eliminate? If you go to much down with the inductances you loose much gain.

I opened also an other thread for diff amp with LR RIAA.

Best regards,
docali

Thanks. I didn't know about Ptplot, I'll look it up.

I'll also look at the LR thread; now that you mention it, I'd rather get rid of caps LOL.

VB, Maturin

Hi,

there is a nice script from Dmitry Nizhegorodov which extracts the values for PtPlot from the spice output.

Yes, NoC area
But many people say you should use an LCR because it is constant T-network which presents a nearly constant load independent from frequency to the source. I do not know if Stephie did this into consideration as he designed his LR-phono?

Best regards,
docali

Which brings up an interesting question:

the inverse RIAA that is used to cut the master with, what about phase shift?

It seems to me that a phono preamp should "undo" the phase shift built into the cutting EQ as well as the the frequency EQ's, no?

VB, Maturin

Hi,

so far as I know there was no standard what inverse RIAA was used. I think long time ago they used indeed inductive components but in newer times capacitors only. The phase shift should be different with different inverse RIAA.

What sems more important for us is what load sees the source tube by an inverse RIAA and the phase shift should be small over the whole frequency range.

brgs,
docali

You're probably right; I do remember reading studies that suggested that phase shift wasn't very important.

I suppose I was hoping that the majority of cuttings were made on one brand of machine with one "standard" reverse RIAA with predictable phase shifting; without that, I guess the only thing we can do is minimize further phase shifting.

VB, Maturin

Hi,

if you take a look at my LR-RIAA the phase shift is only about 1-2 degress at 20 KHz. This looks fastastic in my eyes. The high frequencies should be more important here.

If someone has time it would also be interesting to see the difference of the impedance with frequency compared for LR and LCR.

Brgs,
docali

Hi,

I did some interesting simulations to show you the difference of LCR and LR.

The input impedance of the LCR-RIAA is really very stable and nearly the same over the whole audio frequency range while the LR-RIAA has 10 times more deviations.

Maybe this is the what makes the magic of the sound of an LCR. It has higher parts count but is much ore easy to drive from the preceeding stage and distortions should be equal through the frequency range.

Brgs,
docali

That's very interesting. The other thing I see are the affected frequencies; for the LR, there are major changes in impedance from 100 Hz to 1 KHz - a crucial area of the spectrum - the LCR has already done with changing impedance by 100 Hz; 100 Hz and below is not at all a crucially important part of the spectrum for phase or distortion (within limits of course

Huh...listening tests will only tell the final tale, but according to this I guess I'll have to think about which caps in an LCR would be the least offensive LOL. With caps I think we have two problems: the distortion of the dielectric and microphonics - the microphonics can be tamed by potting, but then you have massive amounts of dielectric coloring the sound. Polypropylene, anyone? And maybe we should be isolating our caps against vibration just like some do for the tubes, especially at the low signal levels.

VB, Maturin

Does that mean that if the resistors were a minimum i.e. both outputs tied to Vcc that the gain would be at a maximum?

Generally speaking, you can only use a transformer input for a phono cartridge if that cartridge is a moving coil type. Moving magnet types have a source impedance of some 1500 ohms, and it's a reactive source as well, which makes it basically impossible to use a transformer with them. Thus, if you want "noise free gain" with a MC cartridge, then you need to use a step up transformer, which then couples to your differential vacuum tube stage.