Follow Up: Lounge Audio LCR Mk III measurements

[Ecaroh]

Yes but this is an internal current limit spec and the opamp will be very non-linear way before that. Doug Self has some measurements on it in SSAD, page 146 (my 2nd Ed.). 10mA peak should be about the max in linear operation, but that still is huge compared to the usual output current requirements.

Thank you, Jan. The Self measurements are great to have.

 

[Ecaroh]

LT1028 is good, AD797 even better, especially if you use the distortion cancellation feature (although many designers don't even bother, the distortion is already far below the other non-linearities in a phono chain).

Any advantage in an MM-only design to be reaped from AD797 vs. 5534? Other than a chance to spend $11/chip ;-) ? Specs are great.

 

[LTig]

Thanks for this, SIY. I wouldn't be using the MC input, so nothing to address there...

The MM part looks very close to my DIY preamp, except that I have much more gain in the first stage (MC).

Just by a coarse look at the gain setting networks I think that the gain for MM is less than 40 dB at 1 kHz (~16 dB for Z1/101 and ~20 dB for Z2/102) which is a little low for my taste. A typical MM pickup with 5 mV output results in just 310 mV. But this can be fixed by increasing R3/R103 or decreasing R4/104 in the first stage, there is a lot of reserve in stage 1.

 

[SIY]

Any advantage in an MM-only design to be reaped from AD797 vs. 5534? Other than a chance to spend $11/chip ;-) ? Specs are great.

No, the 797 is excellent for MC, but too much current noise for MM. The guy who designed the 797 uses a FET diff amp for his own phono stage.

edit: Here's a nice trick You can calculate the optimum source resistance (i.e., minimum noise figure) for an opamp from the formula Ropt= en/in. When you plug the numbers in for the 797, you get 450 ohms. That's fine at low frequencies, but as the inductance kicks in in the midrange and treble (where you hear the hiss), the 797 doesn't look as good.

 

[LTig]

Any advantage in an MM-only design to be reaped from AD797 vs. 5534? Other than a chance to spend $11/chip ;-) ? Specs are great.

As SIY said the AD797 is not suited to MM (~ 1kOhm source impedance) but very well suited to MC (~5..20 Ohm source impedance). If you really want to burn money on MM try the OPA227 (unity gain stable) or OPA228 (gain higher than 5, should work in your case).

 

[SIY]

MM source Z is actually higher than 1k as you get into the midrange and treble- remember that there's like a Henry of inductance, so less of the 47k load is shunted.

 

[stelios]

Using this method and a 10 mV 1kHz square wave generated by the APx525, I determined that the input resistance and capacitance here were close to spec at slightly over 50k ohm and 110 pF, respectively.

Errr.. after taking into account the 10pf of the x10 probe, guess what's in front of the op-amp in the circuit (of the basic version that is)...

At least one top engineer I know bought one of these units and tried measuring the inductance of the equalization coils and failed for exactly this reason- the inductance was a strong function of applied voltage.

Could the inductors be the weakest link in this implementation? They do not look anything premium or even a level below that to me. That could explain some things given the cost of these parts.

Excellent post. So much detail on measuring these devices. This would be a must read for prospective builders or those interested in measuring performance.

 

[SIY]

after taking into account the 10pf of the x10 probe

That number takes this into account.

Could the inductors be the weakest link in this implementation?

It’s one of them- it’s pretty much all weak links. Poor power supply, poor layout, poor topology, poor inductor choice, poor opamp choice. Take your pick.

Thanks for the kind words.

 

[stelios]

That number takes this into account

Ah.. still pretty close then: only 10% off...

 

[stelios]

Here's a first attempt at a simulation of what I can I tell the basic circuit is.

Important DISCLAIMER:
This is based on what I think the basic circuit is and my ability to simulate this in LTSpice. Any errors in figuring out the circuit or doing a good job at simulating it are of course my own.

Circuit simplifications used:

1. no power supply circuitry included in the circuit
2. both rails assumed to be perfect 12V dc sources without any internal resistance specified (whatever LTSpice uses as default) so:
   1. no power supply local decoupling capacitors were included
   2. no local regulators were included
3. all passive components are perfect, no tolerances, parasitic elements or other properties specified other than the component's value
4. smd capacitor between the legs of the first op-amp assumed to be 1pF or 10pF - it didn't seem to make any difference - as I cannot tell what the component value is
5. variable resistor in the grounded leg of the second T filter of the LCR circuit assumed to be 0 Ω

Assumptions used in the simulation:

1. 10k output load is a simple resistor
2. 5mV perfect input voltage source without any internal resistance specified (whatever LTSpice uses as default)

Now that I think of it, I should have used some more realistic loads and source.
I also need to check what the power supply rails' voltage is.

Anyway, I'm not sure whether this is of any use. to anyone. I haven't checked to see if response complies to RIAA. I am trying to figure out how to obtain a spectrum analysis from LTSpice. Apparently it should be quite easy. So is spelling apparently but I always get it wrong.

If anyone has any comments on how to improve this to make it more realistic (or even of any value if this is utterly useless), please bring them on.

 

[stelios]

This what it looks like with a source having 1kΩ internal resistance and with an 1H inductor connected in series, to simulate a cartridge. I think it now looks a lot better.

Also the coupling capacitor on the output of the first op-amp before the LCR filter was doubled. This did not seem to make much of a difference.

 

[stelios]

It wasn't hard to obtain the FFTs - assuming you don't care much about details that is.

Please note that I didn't take particular care of obtaining these, I just fired LTSpice at three frequencies, 100Hz, 1kHz and 10kHz for 2s without any other setting and recording set to start from the beginning (0s). As no settling time was taken into consideration, there will probably be transient artefacts in the FFT but I guess they should be minor.

If you have any particular advice on how best to do these in LTSpice, please comment.

Same disclaimer and notes apply to these as to the last FR above.

 

[SIY]

The problem starts with the basic concept: instead of using resistors and capacitors, which are close to ideal, they used inductors, which are far from ideal. That likely has more impact than the poor choice of opamp. Ditto the poor layout, construction, and grounding.

 

[stelios]

The problem starts with the basic concept: instead of using resistors and capacitors, which are close to ideal, they used inductors, which are far from ideal. That likely has more impact than the poor choice of opamp. Ditto the poor layout, construction, and grounding.

Using reasonably good inductors, given there are 4 of them, would make the cost shoot up by at least 50USD, just for the parts alone - without taking import duties into account, i.e. assuming these inductors could be procured domestically. If premium brands were selected, the inductors' price alone would eclipse the current price of the whole device by a couple of times at least.
On the other hand, the rest of the premium resistors and capacitors used must cost a lot.. So if the parts cost were balanced a bit, I guess they could obtain better inductors. Then, of course, the whole selecting and matching business comes in and the cost sky-rockets again!

How can one simulate for poor grounding scheme and layout in LTSpice? By adding parasitic capacitors and inductors judiciously? Quite a risky affair...

I wonder what the measurements would show if the power supply rails were fed by two batteries. Of course that would require a minor surgery as the wall transformer feeds both rails out of a single secondary.

 

[stelios]

Here's an example of a premium LCR module (the whole filter):

https://www.sowter.co.uk/acatalog/E_A_Sowter_Ltd_PHONO_CARTRIDGE_TRANSFORMERS_30.html

It is called 1280 RIAA 600 Ohm EQ Unit after the Tango EQ-600P LCR unit it is emulating.
The technical report is here:
http://www.sowter.co.uk/pdf/LCR-RIAA Ahlswede.pdf

At 117GBP per unit, a pair easily exceeds the price of the Lounge audio RIAA preamplifier.. The LCR filter of the Lounge Audio preamplifier is also modelled after the Tango unit and seems to be a slight variation of it.

 

[stelios]

The 1kHz FFT uploaded.
I uploaded the 100kHz FFT twice by mistake.