How much Feedback on active RIAA design?

[Stefanoo]

Hi guys,

i have a doubt i have been trying to figure out in the past days but without too much success.

I aw working on a phonostage.
I started designing the amplifier first and thinking then on fitting an appropriate RIAA network.
My plan is to split the 75us time constant (possibly along with the newmn constant) as passive filtering and the other two: 138us and 3180us as active.

Speaking of the active RIAA:

how much Open loop gain/feedback is needed for the active to actually take effect?
I will explain my self a bit further:

I have designed an amplifier with 1MHz of open loop bandwidth - 4dB of feedback - 1.75MHz close loop bandwidth.

After have designing the network and have done the proper calculations i connected it to the amp but it didn't work as expected.

I thought right away that this is because open loop bandwidth is not high enough.
The closed loop gain is about 42dB so open loop is around 46dB.

I know that when the amplifier has a low Open loop gain the closed loop gain can't be calculated by the simple ratio of the feedback network.
Thus i thought that the same might be happening on my circuit, I don't know if i' m wrong though.
If what I have said above is correct indeed: does this mean that an active RIAA configuration can only be applied with high feedback/OL circuits?


Another question aside: do you guys think that the newman constant (3.18us) is important and necessary?
I saw different circuits around without it.

Thanks for the attention.


Best,

 

[h_a]

Another question aside: do you guys think that the newman constant (3.18us) is important and necessary?

The Neumann-constant was used to avoid damaging the cutterhead that would resonate at 50kHz.

As far as I know this protection was de facto standard, so I would include it.

Have fun, Hannes

 

[Giaime]

The closed loop gain is about 42dB so open loop is around 46dB.

40dB is basically what you need at 1kHz for a decent phono stage for MM cartridges. At 100Hz (a decade before), with the approximation of a first order system (20dB/dec), you'll need 60dB of gain: how are you going to get it, if the open loop gain is already lower than that?

To make the closed loop system response almost exactly equal to the theoretical gain (or noise gain, that's the inverse of the feedback ratio), the loop gain (difference between open loop and noise gain) should be sufficient, for example at least 20dB or more. For sufficient precision in the time constants, you either have to make the loop gain higher, or you have to "tweak" the component values with the loop closed to obtain the desired response.

An amplifier with 46dB of OL gain is sufficient to make a phono stage for 15-20dB of closed loop gain at 1kHz, and this is too little gain for MM cartridges, frankly speaking.

 

[AndrewT]

the Low Pass filter @ 2122Hz can be passive and I think it offers the possibility of better performance from the whole RIAA amp.
But, do you fit it before the gain stage and accept the extra noise
or
fit it after the gain stage and make the stage work with the high level high frequencies that the stage must pass?
The remaining LF EQ is much simpler to create around the active stage or can be passive between two gain stages. The two gain stages with all passive filtering is more likely to give very high performance than combining everything around one stage.

Q.) is the Neuman a Low Pass filter or a shelf to reduce the signal going to the cutter?

 

[Stefanoo]

the Low Pass filter @ 2122Hz can be passive and I think it offers the possibility of better performance from the whole RIAA amp.
But, do you fit it before the gain stage and accept the extra noise
or
fit it after the gain stage and make the stage work with the high level high frequencies that the stage must pass?
The remaining LF EQ is much simpler to create around the active stage or can be passive between two gain stages. The two gain stages with all passive filtering is more likely to give very high performance than combining everything around one stage.

Q.) is the Neuman a Low Pass filter or a shelf to reduce the signal going to the cutter?

Alright....I have 2 gain stages the first one bursts 43dB and the second around 40dB (without equalization).
The 75us constant will be placed between the 2 stages and contrary to what you said i supposed it should lower the high frequency noise although thermal noise and low frequency noise should rise a little bit.
But since it is a phono stage i don't think that the latter should constituite a problem (just my 2 cents).
Besides the fact I totally agree with you that moving that 75uS time constant allows the RIAA to a more frexible degree of choice, component's value speaking, to match the right time constant i.e. less dB error on the equalization.

speaking of the Newman constant, i have read that it was placed to not burn the cutter (as you guys has highlithed) but since it is at 50KHz i was wondering if you guy had never a chance to evaluate if the increased RIAA network complexity due to the Newmann time constant was worth in term of the sound or not.
In the sense of: the network becomes a little more complex and in my configuration if i group the 75uS and 3.18uS as passive since the forst stage also needs a Low pass filter for DC coupling, the low freq response worsen a lot (i will try to simulate different values to see if i can find a fittable network) so i was evaluating whether it would be worth or not.....

mmm...i don't know if i explained my self clear enough....

 

[Stefanoo]

Re: Re: How much Feedback on active RIAA design?

40dB is basically what you need at 1kHz for a decent phono stage for MM cartridges. At 100Hz (a decade before), with the approximation of a first order system (20dB/dec), you'll need 60dB of gain: how are you going to get it, if the open loop gain is already lower than that?

To make the closed loop system response almost exactly equal to the theoretical gain (or noise gain, that's the inverse of the feedback ratio), the loop gain (difference between open loop and noise gain) should be sufficient, for example at least 20dB or more. For sufficient precision in the time constants, you either have to make the loop gain higher, or you have to "tweak" the component values with the loop closed to obtain the desired response.

An amplifier with 46dB of OL gain is sufficient to make a phono stage for 15-20dB of closed loop gain at 1kHz, and this is too little gain for MM cartridges, frankly speaking.

thank you for your informative answer.
i figured the first part.
I didn't know the second

How would you calculate the feedback of ative RIAA network since it varies with frequency?
Let's say that OL gain is 60dB and CLG at 1KHz is 30db and at 100 one decade it is 50db and lets say that goes to its maximum at 20Hz equal 56dB (i am just guessing some numbers): would you say that the amount of feedback is equal to the OLG - the maximum CLG, which it is 56dB on the example stated above, i.e. 4dB?

 

[Giaime]

Re: Re: Re: How much Feedback on active RIAA design?

How would you calculate the feedback of ative RIAA network since it varies with frequency?

Learn about Bode plots, it's easy. And it let you understand much more than doing Spice AC analisys.

Let's say that OL gain is 60dB and CLG at 1KHz is 30db and at 100 one decade it is 50db and lets say that goes to its maximum at 20Hz equal 56dB (i am just guessing some numbers): would you say that the amount of feedback is equal to the OLG - the maximum CLG, which it is 56dB on the example stated above, i.e. 4dB?

The loop gain at 20Hz would be 4dB, it will be 30dB at 1kHz and 10dB at 100Hz, considering your numbers. The loop gain is not constant with frequency if either (or both) the OLG and the noise gain are non-costant...

One of the advantages of opamp based phonos (with feedback equalization) is that the OL gain goes down with frequency by about the same ratio as the desired CL response (the noise gain), so doing active equalization should come to almost flat loop gain across the entire audio band...

 

[Stefanoo]

Re: Re: Re: Re: How much Feedback on active RIAA design?

Learn about Bode plots, it's easy. And it let you understand much more than doing Spice AC analisys.



The loop gain at 20Hz would be 4dB, it will be 30dB at 1kHz and 10dB at 100Hz, considering your numbers. The loop gain is not constant with frequency if either (or both) the OLG and the noise gain are non-costant...

One of the advantages of opamp based phonos (with feedback equalization) is that the OL gain goes down with frequency by about the same ratio as the desired CL response (the noise gain), so doing active equalization should come to almost flat loop gain across the entire audio band...

Giaime....we are both engineer i guess vecchio ordinamento since i got 110 i suppose i know bode plot...che pensi? eheheh....

i was throwing numbers and since i was short with time I didn't really explain my self very well:

what i was trying to say is: since the feedback varies along with frequency....just because CLG varies with frequency....yet for convention, you still define a CLG @1KHz.
Thus, i was wondering if the same concept applies as short definition to feedback value for Active RIAA eq.
I don't know if i explained my self a little better this time.

I didn't know that thing about OP amp based RIAA...it's interesting...i still don't understand how come the OL varied .....

unfortunately I don't personally like......matter if preference and personal taste....op amp audio stuff but thanks the thought was very informative anyways.

 

[Giaime]

Re: Re: Re: Re: Re: How much Feedback on active RIAA design?

I didn't know that thing about OP amp based RIAA...it's interesting...i still don't understand how come the OL varied .....

The opamp is internally compensated with a dominant pole at 10-100Hz (newer opamp can have this higher in frequency), so the open loop gain goes down by 20dB/dec.

About the other parts of your reply, I have to think more about it, I don't think I understood your question.

 

[Stefanoo]

Alright....

Just for the folks:
i made some listening tests, specifically comparing a RIAA with and without the Newmann constant.
Although i would like to run more tests/measurements to confirm my opinion the latter one was superior, just more natural and more dynamic.

now....i have a question:


i have read on Jung's papers and on other sources that phono stage start to increase its distortion and loose dynamic by going up with frequency.
This statement confuses me a little bit...and i sincerely don't understand what connection there might be with decreasing of dynamic and higher distortion at high frequency and how it is possible to control this parameters.

Thank you guys for your attention.


Best.