An RIAA phono preamp

www.hifisonix.com
Joined 2003
Paid Member
I think the production variability would have been 1 dB for these types of EQ amps - ok for the day. If the active devices were selected and 1% components used in critical places it would have been manageable.

Looking back, the best way to have tested these things would have been with a square wave and inverse RIAA - but as I mentioned earlier, it never crossed my mind in 1979. I want to also make it clear my original design was a real hack - breadboard, scope calculator affair and the conformance measurements were done by injecting an input signal and looking at the output level on a scope at 10 or 15 points on the responose. It ended up on a PCB and I used it for about 10 yrs.
 
  • Like
Reactions: 1 user
I’m a bit amazed that we take the 3 to 4 dB deviation for granted.
I still have three amps from those days, Accuphase, Sansui and B&O.
All three Riaa’s measure with margin within +/- 1dB.
So those large deviations are probably for cheap TT’s with integrated (tube) amps like my wife had in the sixties.

Hans
 
www.hifisonix.com
Joined 2003
Paid Member
The 3~4 dB figure was quoted by Lipshitz. I agree it is a big spread but +- 1 dB from the era was I think quite the norm on good products - bit like 0.1% distortion being acceptable whereas now we tend to feel we can do better than that.

I have an old Pioneer receiver (from 1967) with a MM phono input under my bench I want to restore. It will be interesting to see what that is. I am not expecting anything better than 1-2 dB
 
Here's one I designed and built in 1979. I tweeked the EQ network a few yrs ago in LTspice, so its quite accurate now

View attachment 1165365

Heres the conformance. Its about 0.5 dB 20 - 20kHz

View attachment 1165366

Heres the square wave performance and loop gain - not as good as an opamp implementation, but this type of performance was pretty much standard back in the 1960's and 1970's. If you can reproduce a good square wave after passing the stimulus through and inverse RIAA, you can rest assured that the conformance of the equalizer if good.
I guess that input impedance is far below 47kOhm due to neg feedback resistor to the input base. With the quite low output impedance of your reverse RIAA-network certainly this will not show up.
 
  • Like
Reactions: 1 user
I'm pretty sure we had SPICE at SUNY Buffalo in '79 - input was hollerith punch card. You'd hand your deck to the lady through a window, then wait for your printout to be placed in a cubby. I vaguely remember someone was doing speaker crossover sims at the time. I wonder what ever happened with that?

Unfortunately, I didnt recognize the significance right then and just stuck with programming the EE class examples. I'd use it off and on over the next 40 years in my career.

In a couple-few years we'll literally say "Computer - create an electronic filter, 10k input impedance, 1k output impedance, low pass 3 db down at 20Hz, high pass 3 db down at 18kHz. Use discrete transistors. Use component values you can actually buy; provide a BOM from Digikey". A bit of an upgrade, from handing someone a punched card deck.
 
  • Like
Reactions: 2 users
Account Closed
Joined 2018
@jjasniew "In a couple-few years we'll literally say "Computer - create an electronic filter, 10k input impedance, 1k output impedance, low pass 3 db down at 20Hz, high pass 3 db down at 18kHz. Use discrete transistors. Use component values you can actually buy; provide a BOM from Digikey". A bit of an upgrade, from handing someone a punched card deck."

I have to say, that reminds me of the 1960's song by Zager & Evans "In the year, 2525"
We've (well, a lot of humans) have gotten to the point of barely using our mouths and limbs.
Just sit there, and tell that tabletop Alexa gadget what you want to do.
God forbid you'd have to use your own brain and limbs to do something.

Try to tell me that's a normal way to live..... try.
 
  • Like
Reactions: 1 user
Administrator
Joined 2007
Paid Member
In a couple-few years we'll literally say "Computer - create an electronic filter, 10k input impedance, 1k output impedance, low pass 3 db down at 20Hz, high pass 3 db down at 18kHz. Use discrete transistors. Use component values you can actually buy; provide a BOM from Digikey". A bit of an upgrade, from handing someone a punched card deck.

We're not there yet but we are well on the way... if you don't mind opamps ;) It's not much of a leap to have something like this recognise speech.

https://tools.analog.com/en/filterwizard/
 
Member
Joined 2017
Paid Member
Ha! I‘m lost in holidays!
 

Attachments

  • IMG_4488.png
    IMG_4488.png
    85.1 KB · Views: 105
The answer of ChatGPT:
To design an electronic filter with the given specifications, we can use a combination of a high-pass filter and a low-pass filter, connected in series. We will use a discrete transistor circuit for each filter stage.
Here's the circuit diagram:

Filter Circuit Diagram


Low-Pass Filter Stage
For the low-pass filter stage, we can use a common-emitter amplifier configuration with a capacitor in the feedback path. The cutoff frequency of this stage is determined by the value of the capacitor.
The required cutoff frequency is 20Hz and the input impedance is 10kΩ. Using the formula:
Cutoff Frequency = 1 / (2 * π * R * C)
We can calculate the value of the capacitor (C) needed as:
C = 1 / (2 * π * R * Cutoff Frequency)
Plugging in the values, we get:
C = 1 / (2 * π * 10kΩ * 20Hz) = 795 nF
We can use a 820nF capacitor (Digi-Key Part Number: 493-11343-1-ND).
The transistor used for the low-pass filter stage is a NPN transistor (2N3904). The resistor values can be determined by assuming a quiescent collector current of 1mA and a collector-emitter voltage drop of 5V.
Using the formula:
R = (Vcc - Vce) / Ic
We can calculate the value of the resistor (Rc) needed as:
Rc = (12V - 5V) / 1mA = 7kΩ
We can use a 6.8kΩ resistor (Digi-Key Part Number: 311-6.8KFRCT-ND).
The bias resistor (Rb) can be calculated using the formula:
Rb = (Vcc - Vbe) / Ib
Assuming a base-emitter voltage drop of 0.7V and a base current of 10μA, we get:
Rb = (12V - 0.7V) / 10μA = 1.13MΩ
We can use a 1MΩ resistor (Digi-Key Part Number: 311-1.0MFRCT-ND).
The input coupling capacitor (Ci) is used to block DC and prevent the input signal from being affected by the DC bias at the transistor base. We can use a 10μF capacitor (Digi-Key Part Number: 445-1308-1-ND).

High-Pass Filter Stage
For the high-pass filter stage, we can use a common-base amplifier configuration with a capacitor in the input path. The cutoff frequency of this stage is determined by the value of the capacitor.
The required cutoff frequency is 18kHz and the output impedance is 1kΩ. Using the formula:
Cutoff Frequency = 1 / (2 * π * R * C)
We can calculate the value of the capacitor (C) needed as:
C = 1 / (2 * π * R * Cutoff Frequency)
Plugging in the values, we get:
C = 1 / (2 * π * 1kΩ * 18kHz) = 8.83 nF
We can use a 8.2nF capacitor (Digi-Key Part Number: 399-1427-1-ND).
The transistor used for the high-pass filter stage is a PNP transistor (2N3906). The resistor values can be determined by assuming a quiescent collector current of 1mA and a collector-emitter voltage drop of (That's the end).
 
  • Like
Reactions: 2 users
I asked for an other image:
Here's an alternative circuit diagram for the filter:

Alternative Filter Circuit Diagram


This circuit diagram shows a Sallen-Key topology for the filter, which is a common topology for active filter circuits. This topology uses an operational amplifier (op-amp) and two resistors and two capacitors to form a second-order filter. The op-amp provides high input impedance and low output impedance, which meets the specifications given in the question.

The low-pass and high-pass filter stages are combined into one circuit using this topology. The cutoff frequencies of the low-pass and high-pass filters are determined by the values of the resistors and capacitors.

The resistors and capacitors can be chosen to meet the given specifications of a 10k input impedance, 1k output impedance, low-pass 3 db down at 20Hz, and high-pass 3 db down at 18kHz. Here are some possible component values:
  • R1 = R2 = 10kΩ (Digi-Key Part Number: 311-10KFRCT-ND)
  • C1 = C2 = 1.5nF (Digi-Key Part Number: 399-1465-1-ND)
  • R3 = 1.82kΩ (Digi-Key Part Number: 311-1.82KFRCT-ND)
  • R4 = 100Ω (Digi-Key Part Number: 311-100FRCT-ND)
These values give a low-pass cutoff frequency of approximately 20Hz and a high-pass cutoff frequency of approximately 18kHz. The input and output impedance values are also met with these component values.
 
Member
Joined 2014
Paid Member
We've (well, a lot of humans) have gotten to the point of barely using our mouths and limbs.
Just sit there, and tell that tabletop Alexa gadget what you want to do.
God forbid you'd have to use your own brain and limbs to do something.

Try to tell me that's a normal way to live..... try.
Ignoring the irony in your comment for now. I think it all went wrong with electric start and synchromesh gear boxes. My Grandfather had an austin seven. Hand cranked and 3 speed non-synchro box. You actually had to drive that unlike modern cars.
 
  • Like
Reactions: 2 users
Administrator
Joined 2007
Paid Member
While we are on about AI I had a genuine question I asked ChatGPT on how to configure LTspice to generate noise source. The results were impressive reading but didn't actually work. As I kept saying in the LT thread years ago, a lot of knowledge is assumed and that is where many resources fall down. It took me a while to figure out (using the web generally) that you needed to use a Behavioural Voltage Source. This was over a week ago but I saved one of the sessions.

Anyhooo:

To configure a voltage source in LTspice for white noise, you can follow these steps:

Open LTspice and create a new schematic.
Add a voltage source to the schematic by selecting the "Voltage Source" icon from the toolbar or by typing "V" in the command bar.
Double-click on the voltage source to open the "Edit Attributes" window.
In the "Edit Attributes" window, change the "SpiceLine" parameter to the following: "noise V=0 type=white".
Click "OK" to save the changes and close the "Edit Attributes" window.
Connect the voltage source to the rest of your circuit.
Run a simulation to see the effect of the white noise voltage source on your circuit.

Note: The "type" parameter in the SpiceLine specifies the type of noise, which in this case is white noise. You can also specify other types of noise, such as pink noise, by changing this parameter. The "V" parameter sets the DC voltage of the noise source, which in this case is set to zero.

Me:
Your responses continually omit the need to select BV or Behavioural Voltage Source.

You are correct, I apologize for the oversight in my previous response. To add a white noise voltage source in LTspice, you can use the Behavioral Voltage Source (BV) component. Here are the steps to configure a BV for white noise:

Open LTspice and create a new schematic.
Add a Behavioral Voltage Source (BV) component to the schematic by selecting the "BV" icon from the toolbar or by typing "BV" in the command bar.
Double-click on the BV component to open the "Edit Attributes" window.
In the "Edit Attributes" window, enter the following in the "Equation" field: "white(0, Vrms)".
Enter a value for Vrms, which represents the root mean square (RMS) voltage of the white noise source. For example, if you want a white noise source with an RMS voltage of 1V, enter "1" in the "Value" field next to "Vrms".
Click "OK" to save the changes and close the "Edit Attributes" window.
Connect the BV component to the rest of your circuit.
Run a simulation to see the effect of the white noise voltage source on your circuit.
Note: The "white" function in the BV equation generates white noise, and the first parameter of the function specifies the DC offset of the noise source. In this case, the DC offset is set to zero.
 
  • Like
Reactions: 1 user