Designing a Phonostage with the S&B 10K LCR RIAA Module

[Kuei Yang Wang]

Konnichiwa,

I noticed in several cases that despite previous notes and the application notes on the S&B Website for the LCR people struggle to work out what can be done and what not. So, I'll try to cover here the principles.

Please note that said principles apply equally to the 600 Ohm RIAA Modules, the S&B 600R and 10K RIAA Modules use EXACTLY the same Circuit, merely the component values have been adjusted to give the desired impedance.

The first thing to understand is that these networks are constant impedance T-Networks. In other words, they operate in a constant impedance envoironment and themselves present a constant impedance load at their nominal impedance (600R/10K).

This means in a "classical" situation the RIAA would be terminated with it's nominal impedance on either side, in other words the source impedance of the receeding stage would be made equal to the RIAA Modules and the RIAA Module would be loaded with it's nominal impedance.

In practice one of the terminations can be omitted, but NOT both. This means you either operate the RIAA Module with a near infinite load impedance and from a source impedance equal to it's nominal impedance.

OR you load the RIAA with it's nominal impedance and then you are free to select the source impedance with impunity.

In fact, I used this feature in my E810F/D3a Phonostage with the 600R RIAA Module shown here:



Here I use a Penthode stage which hase around 8KOhm output impedance and a gain of around 50db. This stage is then loaded with the 600 Ohm RIAA which results in a loss of gain of 23db brining the gain into the RIAA down to around 27db.

The bottom line, you need either a load or source that is equal to the nominal impedance of the RIAA within a few %. In any case, the stage driving the RIAA Module will be loaded by an impedance equal to the RIAA Modules Impedance, plus the effect of the Anode load, which appears in parallel with the RIAA Module.

The easiest way to GUARANTEE that the RIAA performs accuratly with total disregard to valve aging is to load the RIAA Module with it's nominal impedance and to insert a coupling capacitor of suitable value in front of the RIAA Module. The minimum capacitor that works with a 10K RIAA Module is around 2.2uF and for the 600R RIAA it is 39uF.

The next thing is to have a first stage in the Phonostage that is able to drive the load of the RIAA Module without massive distortion and at the levels required plus it is required to stiull develop usefull gain into the load presented by the RIAA Module.

While it may be tempting to use an ECC83 as first stage, this stage would show an output impedance in the region of 50KOhm or more and would distort heavily into a 10K load.

Therefore we need a Valve with low anode impedance and reasonably high gain. Among the common small signal valves the ECC88/6DJ8/6922 will answer the purpose reasonably.

Better choices will be valves with more gain and higher transconductanace such as the WE 417A, WE 437A, russian 6S45PE, 6S4P, 6S3P or triode wired high transconductance pentodes such as the E180F/6688/6Z9P, C3g or E810F/7788 and D3a/7721. Some of these may even be used directly to drive the 600R RIAA Module.

For our example however I will stick to the 6DJ8 and the 10K RIAA as it is readily available and known. The calculation of the other examples shall be left to the genteele reader as an excercise in applied math.

Let us first understand what sort of signal comes from the cartridge. Most cartridges are rated for an output in a certain mV amount for a deflection of 5cm/s. A typhical range for Moving Magnet Cartridges is an output voltage of 2.5 - 10mV @ 5cm/s.

Conveniently so-called low-output moving coil cartrdiges have outputs usually in the region of 0.25 - 1mV @ 5cm/s, making the use of a stepup device with a 1:10 stepup (transformer) easy to get the levels to the same range as the Moving Magnet Pickups.

Now is 5cm/s the maximum level on a record? Nope, records can be cut with up to 14..16db MORE level than that, at least for a range of 100Hz - 5KHz (appx.), this is called headroom. Many well recorded and well cut LP's will have levels as high as that.

In order to accomodate all possible LP's and pickups we therefore must expect operational levels on the grid of the first stage of around 14db above 10mV or in other words 50mV. We should also expect levels CONSIDERABLY HIGHER at high frequencies caused by pop's and clicks, however these only appear with Moving Coil cartridges, with MM Cartridges the high frequencies roll off following a 2nd order function somewhere at the top of the traditional audio band, depending on capacitive loading.

So, our Grid of our first stage will be exposed to around 50mV operational signal and see transient pops and clicks at maybe 100 - 200mV. And it will drive a 10K RIAA using a valve with a nominal gain of 33 (ECC88). So the maximum operational signal applied to our RIAA Module will be around 1.5 - 2V, less in reality due to gain losses into the 10k load.

Now for low distortion it is desirable to have only 1/10 of the actual anode current varied by the signal (rule of thumb), so we need to find the peak current into 10K for say 2V peak voltage, we get 0.2mA, so our first stage shall have at the very least an anode current of 2mA plus whatever is needed to account for the current variation in the anode resistor.

My 6DJ8 Datasheets charaterise the Valve at 90V Anode Voltage, allow me to be lazy and use this point, a +B of 250V and a current of at least 4mA which gives a 27-33KOhm . My curves suggest a 33K Anode load and around 4.5mA current, which leaves us some voltage (9V) for a decoupling RC circuit, 2K2 seem apropriate.

The 6DJ8 appears to require 2.3V Bias at the above operating conditions, this then means a cathode resistor of 10 Ohm. If we leave this resistor unbypassed you get the cathode resistors value "amplified" by the Mu of the valve and added to valves anode impedance, which is nearly 5k at the operating point chosen.

So, with an unbyassed cathode R of 510R our effective anode impedance becomes around 20KOhm, way too much to be usefull with a 10K load as we would loose ton's of gain. Therefore we would need to do one of two things. We can bypass the cathode resistor with a 100uF Capacitor OR we can increase the current, to say 15mA and reduce the cathode resistor.

15mA at 90V require 1.2V Bias and thus a 82R cathode resistor. Effective anode impedance with unbypassed Cathode resistor is now around 5K but our Anode load resistor needs to drop to 10K. With that we have an effetcive load on the Valve of 5K (10K Anode load in parallel with 10K RIAA Module) and a 5K source impedance, so our gain of the first stage into the RIAA Module will now be around Mu/2 or 33/2 = 16.5 = 24db. The RIAA Module attenuates 20db @ 1KHz and thus after the RIAA we have a remaining gain of 4db.

If we now add a second stage in identical arrangement after the 10K RIAA Module and it's load resistor we will for that stage have a gain of 22 (27db) into high impedance loads and around 3K Output impedance. In total our Phonostage would have a gain of 27db + 4db = 31db, which is low but adequate (the Zanden Phonostage has a gain of 33db).

The circuit would look like this:



We could now take some measures to bump up gain.

We could use of diodes to bias the Valves (you could use a pair of standard 1N4001 in series with the commoned cathodes of both valves) This would reduce the effective Anode impedance for both stages to around 2K5 and would thus bump up the gain of the first stage to 22 (27db) and of the second stage to 26 (28db) with an overall gain of 35db.

Alternatively you could battery bias the second stage grid from a 1.2V or 1.5V battery in series with the 10K resistor and ground the cathode. And you could even use the series grid bias methode shown variously on the net like this:



The next step could be to substitute current sources like the Bottlehead C4s (www.bottlehead.com) or the CCS Kit offered by K&K Audio (www.kandkaudio.com) for the 10K Anode Load resistors. This gives us back even more gain, if no external load is present the output stage will give basically the full Valves "Mu" at 33/30db and the first stage gets up to 26/28db, giving overall 38db gain.

It shall be left to the genteele reader to apply these current sources and different bias methodes to the circuit as an excercise.

If the excercises appear too difficult a little study of the materiel provided in the On Line Tube Learning for newbies.... Thread (sticky thread at the top of tubes board) may remedy the difficulty.

Sayonara

[Eli Duttman]

Thorsten,

THANKS for the detailed explanation.

I recall seeing some "chatter" to the effect that the 600 Ohm RIAA module is superior to the 10 KOhm version. Is the "chatter" correct?

I've been mulling over a 6DS4 loaded by a Gary Pimm CCS as the 1st gain block of a phono stage for some time. The impedance of the CCS' mu follower O/P is a few hundred Ohms. Am I correct in thinking that the 600 Ohm RIAA module will be fine driven by the CCS' mu follower O/P, so long as the grid leak resistor of the 2nd gain block is 604 Ohms?

If my thinking is reasonable, could you make a recommendation for the coupling cap. that goes between the CCS' O/P and the RIAA network?

TIA for a response.



Eli D.

[Kuei Yang Wang]

Konnichiwa,

Originally posted by Eli Duttman
I recall seeing some "chatter" to the effect that the 600 Ohm RIAA module is superior to the 10 KOhm version. Is the "chatter" correct?

I have not had a chance to compare, the 10K RIAA's are in sort supply and I have come up short again and again (well, lack of time let me tell JB "okay, send them off elsewhere")...

Originally posted by Eli Duttman
I've been mulling over a 6DS4 loaded by a Gary Pimm CCS as the 1st gain block of a phono stage for some time. The impedance of the CCS' mu follower O/P is a few hundred Ohms. Am I correct in thinking that the 600 Ohm RIAA module will be fine driven by the CCS' mu follower O/P, so long as the grid leak resistor of the 2nd gain block is 604 Ohms?

Assuming the Mu-Follower operates at sufficient current to sustain the neccesary output voltage into 600 Ohm - Yes.

Originally posted by Eli Duttman
If my thinking is reasonable, could you make a recommendation for the coupling cap. that goes between the CCS' O/P and the RIAA network?

You will need a VERY large value capacitor. If you accept -3db @ 20Hz (which I do not) you need around 12 - 15uF, for a -3db point of 4Hz (which I aim for) you need 68uF.

You COULD use another trick.

The Mu-Follower output of your circuit has a low and fairly well defined impedance. The S&B Riaa contain Capacitors with 630V or more DC rating in all shunt path, so it will withstand quite a bit of DC offset, as long as no DC current flows.

Therefore, if your source can be made to have 600 Ohm net impedance you can couple the S&B RIAA (and ONLY the S&B one) directly and place the coupling cap AFTER the 600R RIAA, making it's value uncritical.

Sayonara

[fdegrove]

Hi,

If you accept -3db @ 20Hz (which I do not) you need around 12 - 15uF, for a -3db point of 4Hz (which I aim for) you need 68uF.

May I ask what kind of musical information you're hoping to find on a record at 4 to 20 Hz?

Information that, if at all present, will be filtered out by the RIAA correction anyway.

To the best of my knowledge the only thing to be found in that frequency range are subsonic resonances stemming from cartridge/tonearm interaction.
Amplifying those will only give you a false sense of bass response much the same way speaker cabinet resonances appear to be part of the music to some (most actually) people...........

EDIT: There should be no reason whatsoever for a 600 Ohm LCR
RIAA filter to be "better" than a 10K one provided both are properly designed and implemented.

Cheers,

[gingertube]

Trying to get too much low frequency response from a RIAA preamp is asking for trouble.

The RIAA Spec is:

75us low pass (2122Hz)
318us high pass (500Hz)
3180us low pass (50Hz)

This spec. was later modified by IEC to add a 10ms High Pass (16Hz).

The problem with the straight RIAA spec. is turntable rumble at approx 0.5Hz (33 r.p.m.) together with the 3180us (50Hz) low pass. This means response is increasing (at 20 dB/decade) below 50Hz. The response at 0.5Hz can therefore be +40dB above the 50Hz response. Even if your power amp /speaker etc can't reproduce the low frequencies you run the risk of saturating the preamp active stage(s).

IEC added the extra pole at 16Hz to make sure response was well down at turntable rumble frequencies.

My own experience was that I couldn't tell any difference with this extra pole anywhere between the recommended 16 Hz and an experimental version built with this extra pole at 32 Hz instead of 16Hz - I left it at 32Hz.

Of course this experience was with a solid state phono preamp with a mix of active and passive compensation and the LCR MAY not exibit this problem.

It is something to be extremely wary of when designing phono stages.

Cheers,
Ian

[jlsem]

The problem with the straight RIAA spec. is turntable rumble at approx 0.5Hz (33 r.p.m.) together with the 3180us (50Hz) low pass. This means response is increasing (at 20 dB/decade) below 50Hz. The response at 0.5Hz can therefore be +40dB above the 50Hz response. Even if your power amp /speaker etc can't reproduce the low frequencies you run the risk of saturating the preamp active stage(s).

This is where current hogging by unmatched tubes sharing a common cathode resistor in a push-pull output stage can get real ugly.

John

[rdf]

Originally posted by gingertube
The problem with the straight RIAA spec. is turntable rumble at approx 0.5Hz (33 r.p.m.)

If I recall the RIAA curve was ammended to deal with the arm mass/cartridge compliance resonance, rumble from mechanical sources such as vinyl surface and bearing noise, and reduce susceptibility to footfalls and structural and airborne feedback. Disc eccentricity is far below the high pass created by most cartidge/arm combinations and doesn't manifest itself as a .5 Hz signal of any significance but as a cyclic speed variation. The RIAA curve won't help fix this but it is effective on the rest.

[Kuei Yang Wang]

Konnichiwa,

Originally posted by fdegrove
May I ask what kind of musical information you're hoping to find on a record at 4 to 20 Hz?

You may ask.

You may also ask yourself what result multiple 3db rolloffs at 20Hz daisychained have further up in the audio band.

Originally posted by fdegrove
Information that, if at all present, will be filtered out by the RIAA correction anyway.

!!?? I fail to see ANY logic in this statement.

The RIAA EQ causes a boost of 20db @ 20Hz compared to 1KHz and should show a flat response below 20Hz (unless you subscrive to the ill advised IEC Ammendment to the RIAA curve).

How would that lead to the low frequency information on the LP being filtered out?

Originally posted by fdegrove
To the best of my knowledge the only thing to be found in that frequency range are subsonic resonances stemming from cartridge/tonearm interaction.

Then, with all due respect, the best of your knowledge could use some improvement.

Sayonara

[Kuei Yang Wang]

Konnichiwa,

Originally posted by gingertube
Trying to get too much low frequency response from a RIAA preamp is asking for trouble.

Why exatcly is that?

Originally posted by gingertube
This spec. was later modified by IEC to add a 10ms High Pass (16Hz).

An ammendment most ill advised for use with quality equipment though clearly advisable when used with poor quality turntables (eg the $ 50 all plastic jobs with loads of resonances and ruble)/

Originally posted by gingertube
The problem with the straight RIAA spec. is turntable rumble at approx 0.5Hz (33 r.p.m.)

Hmmm. It is trivial as engineering problem to make a turntable essentially free from rumble (though not neccesarily cheap). Therefore, there is no problem with quality equipment.

One would, perhaps wrongly assume that someone tinkering on a phonostage with $US 480 worth of LCR RIAA Modules would combine said phonostage with a quality turntable.

Originally posted by gingertube
together with the 3180us (50Hz) low pass. This means response is increasing (at 20 dB/decade) below 50Hz.

YOU WHAT? Please retake EE101.

A 50Hz lowpass implies that anything below 50Hz passes (therefore LOW-PASS) while ABOVE 50Hz the response rolls off 20db/10ade or 6db/8ave.

Originally posted by gingertube
The response at 0.5Hz can therefore be +40dB above the 50Hz response.

No it cannot. it cannot be any higher than the 50Hz UNLESS you have arm cartridge resonance at around 0.5Hz.

HOWEVER, most arm/cartridge resonances are around 8-16Hz (for good reasons too) and therefore any output below the Arm/Cartridge resonance rolls off steeply, at least 40db/10ade..

Originally posted by gingertube
Even if your power amp /speaker etc can't reproduce the low frequencies you run the risk of saturating the preamp active stage(s).

Chemically pure horse excerement. Please get your facts right.

Originally posted by gingertube
IEC added the extra pole at 16Hz to make sure response was well down at turntable rumble frequencies.

They did, but primarily because cheap turntables had very high levels of rumble, not for the reasons you invented..

Originally posted by gingertube
It is something to be extremely wary of when designing phono stages.

NO, IT IS ABSOLUTELY NOT.

You may be weary of getting your arm/cartridge set up wrongly.

Adding an LF rolloff higher up in the audio band (even at the IEC recommended 16Hz) has a clearly audible impact if you system has sufficient bandwidth and is often audible even when not. Steeper slope filters (such as advocated by D. Self) are especially problematic.

Sayonara

[Kuei Yang Wang]

Konnichiwa,

Originally posted by rdf
If I recall the RIAA curve was ammended to deal with the arm mass/cartridge compliance resonance, rumble from mechanical sources such as vinyl surface and bearing noise, and reduce susceptibility to footfalls and structural and airborne feedback. Disc eccentricity is far below the high pass created by most cartidge/arm combinations and doesn't manifest itself as a .5 Hz signal of any significance but as a cyclic speed variation. The RIAA curve won't help fix this but it is effective on the rest.

You are absolutely correct. However, correctly set up arm/cartridge combo's have little effect in the audible range and a quality turntable setup should be substantially free from rumble, susceptibility to footfall and the like. Well cleaned records show little noise.

Therefore the IEC ammendments is suitable to low quality equipment but should be avoided in high quality equipment.

If we assume (for arguments sake) 4 daisychained open loop Valve stages (which is the case for my own system) with all dominant timeconstants set to 40mS/4Hz we find that with four times this we will attenuate 4Hz by a total of 12db (3db @ 4Hz X 4 with a 4th order slopw below 4Hz.

Each 4Hz pole will also attenuate 20Hz by 0.17db, therefore with four poles at 4Hz daisychained we find that 20Hz is attenuated 0.68db already. It would seem undesirable to add more unneccesary rolloff, at least for quality equipment.

Sayonara