Question for ThorstenL about phono EQ

[gurevise]

Thorsten,
I have simulated several phono EQ circuits you've suggested at various threads here and they all have about 1.5 dB error from 20hz to 20kHz.
Signal at 20kHz is off by almost 2 dB against 20Hz.

I used HagTech schematic to implement reverse EQ circuit:
Hagerman Technology LLC: iRIAA Filter for Phonostage DIY Kit
I run several other phono circuits through the same simulator and they have much lower error.
Here is the simulated schematic:

http://dl.dropbox.com/u/17742412/Phono sch 2.jpg

EQ error:
http://dl.dropbox.com/u/17742412/EQ Curve.jpg
Pay no attention to absolute gain values on EQ error chart as V/I stage gain is very low.

Any ideas of why I get this much error? Are you using a different EQ from standard iRIAA?

Thanks
Sergy

 

[gurevise]

Re posing massage with visible images attached:
----------------------------------------------
Thorsten,
I have simulated several phono EQ circuits you've suggested at various threads here and they all have about 1.5 dB error from 20hz to 20kHz.
Signal at 20kHz is off by almost 2 dB against 20Hz.

I used HagTech schematic to implement reverse EQ circuit:
Hagerman Technology LLC: iRIAA Filter for Phonostage DIY Kit

I run several other phono circuits through the same simulator and they have much lower error.
Here is the simulated schematic:

An externally hosted image should be here but it was not working when we last tested it.

EQ error:

An externally hosted image should be here but it was not working when we last tested it.

Pay no attention to absolute gain values on EQ error chart as V/I stage gain is very low.

Any ideas of why I get error this big? Are you using a different EQ from standard iRIAA?

Thanks
Sergy

 

[ThorstenL]

Hi,

Thorsten,
I have simulated several phono EQ circuits you've suggested at various threads here and they all have about 1.5 dB error from 20hz to 20kHz.
Signal at 20kHz is off by almost 2 dB against 20Hz.

You need to simulate the whole circuit, not just the EQ. This includes rolloffs in coupling capacitors, source impedances and so on.

Ciao T

 

[ThorstenL]

Hi,

I used HagTech schematic to implement reverse EQ circuit:
Hagerman Technology LLC: iRIAA Filter for Phonostage DIY Kit
I run several other phono circuits through the same simulator and they have much lower error.
Here is the simulated schematic:

http://dl.dropbox.com/u/17742412/Phono sch 2.jpg

Your schematic loads the iRIAA circuit appreciably. The LF EQ you have is actually very accurate, you can find it identically in John Curl's Vendetta Phono which measures ruler flat. Actually, I just checked, make R17 7.5K, to see if this helps...

Also try converting the circuit to "generic" that is make the first amp a normal non-inverting one and then use a pair of 1K5 resistors with the 100nF cap to ground from the junction, instead of the 750 Ohm and Howland current pump.

This will remove the various interactions you are seeing...

Ciao T

 

[gurevise]

Thorsten,
Here is the new schematic attached. Front end is INA103 like configuration.
7.5k resistor reduced error from 20Hz to 20kHz to exact 0.6 dB.
Honestly, I do not understand why :-(
I'd think 6.8k (R13) gives more accurate EQ time constant.

An externally hosted image should be here but it was not working when we last tested it.

An externally hosted image should be here but it was not working when we last tested it.

Thanks
Sergy

 

[gurevise]

Here is the response simulation from 10Hz to 100kHz
Only 1.3 dB error. Very good

An externally hosted image should be here but it was not working when we last tested it.

 

[ThorstenL]

Hi,

Here is the new schematic attached. Front end is INA103 like configuration.

Can you use a INA103 Model?

This particular Phono design I originally simmed in Tina TI which includes all the Op-Amp models etc.

7.5k resistor reduced error from 20Hz to 20kHz to exact 0.6 dB. Honestly, I do not understand why :-(
I'd think 6.8k (R13) gives more accurate EQ time constant.

For this timeconstant (318uS) it appears in parallel with the 68K.

The remaining rolloff is still too much, with the 33R resistor in series with the 100nF Cap I'd expect to see some HF Boost.

Could it be that the fairly slow Op-Amp's in your simulation contribute to the early HF rolloff?

Ciao T

 

[ThorstenL]

Hi,

I had a look for the original work I did on this, but it seems that this was one of the few areas of my laptop that was not backed up anywhere when it was stolen a few weeks ago.

So I re-created the circuit I used as best as I could from memory, it should be mostly accurate (see attached). There is a small correction to the 750R resistor in the EQ, as the implementation of the HCP I used has a finite (but well defined) output impedance, so we need to increase the 750R to compensate.

As shown gain is 66dB. Noise around -76dB re. 0.5mV.

The HF rolloff in the simulator is, I think, an artefacts of the Hagermann network vs. others. The actual unit build measured better than +/-0.1dB from flat response and equal levels of channel imbalance without any selection using the AP2 at work. It seems AP disagrees with JH on what the RIAA curve is...

The actual unit build used NOS East German Mil-Spec 0.5% Polystyrene & Tinfoil Cap's (plastic encased) and 0.1% metal film resistors. Also not shown, the offset trim is manual via the Offset pins on Op-Amp's, but I would probably expend an OPA627 on a servo instead if I where to make another one like this.

Not shown are the power supplies, overkill dual mono, with 4pcs 120VA Torroid transformers, 10 * 10,000uF per rail in CRC filters followed by a Darlington Capacitor multiplier followed by 80uF/35V Film Cap's plus the rest of the usual decoupling...

Ciao T

 

[gk7]

Hagermann advocates the non standard "Neuman Pole" I believe, this might explain the error at high frequencies. (Which isn´t an error actually).

 

[ThorstenL]

Hi,

Hagermann advocates the non standard "Neuman Pole" I believe, this might explain the error at high frequencies. (Which isn´t an error actually).

Ah, yes, the shape attained with his iRiaa is indeed precisely the inverse of what I'd expect applying the 50KHz extra turnover to a system using the classic Lipshitz or equivalent Network.

So, one mystery less.

Ciao T

PS, I neither per se advocate or detract from the "Neumann" EQ, I do prefer to listen with it included though, as do many others we tried this with (including blind)

 

[gurevise]

Thorsten,
Why would film 80uF local PS bypass capacitors are better than 1000uF electrolytics? Is capacitor's DF or DA are that important/audible in OpAmp power supply bypass application? I'd think that electrolytic cap would be better for having much lower impedance.

Thanks
Sergy

 

[ThorstenL]

Hi,

Why would film 80uF local PS bypass capacitors are better than 1000uF electrolytics?

I have no idea why, but if concept/budget/parts availability I find avoiding any electrolytic capacitors in the parts of the circuits that directly connect to audio circuits beneficial for sound quality.

We can speculate that the electrochemical nature of the electrolytic capacitors plays a role, micro-phonics are also a possibility. If i figured it out I'd probably be at least up for an AES Fellowship or such.

I should also mention that the capacitor multipliers are pre-loaded at a few 100mA, this not only drastically lowers the output impedance of the follower (300mA pre-load drop it to around 0.1 Ohm) but also helps the PSU to absorb any currents send back into the rail.

The PSU for my implementation will sound like a gross excess of over-engineering to any who would run such a circuit on a small wall wart. Again I lack any reliable information or guess why the improvement from such extreme power supply implementations any more than I can answer the Film Cap part in any way that offers a technical explanation.

Ciao T