Hi everyone!
This is my first post on this great forum. Designing and building audio circuits is one of my past times and this forum has helped me a lot so far. Now I'm facing an issue with one of my designs and I'm hoping some of you can help. So here goes.
I made a design for a phono stage and after building it on PCB I find it to be a bit on the noisy side. The noise I'm referring to is hiss only, so no hum. Now the excessive noise is only present when a phono cartridge is connected to the stage without having a record on. When I ground the inputs, the stage is almost silent. Based on this, one could conclude the noise is coming from the cartridge.
However, if I connect the same cartridge to my Cambridge Audio 551P phono stage, the noise is considerably lower.
I'm puzzled what could be the cause of this. Both stages have the same gain, input resistance and input capacitance. And both have an equally low noise floor with grounded inputs.
So the bottom line question is: is there any other 'mechanism' that can cause noise (hiss) as soon as a cartridge is connected to a phono stage?
Thanks!
This is my first post on this great forum. Designing and building audio circuits is one of my past times and this forum has helped me a lot so far. Now I'm facing an issue with one of my designs and I'm hoping some of you can help. So here goes.
I made a design for a phono stage and after building it on PCB I find it to be a bit on the noisy side. The noise I'm referring to is hiss only, so no hum. Now the excessive noise is only present when a phono cartridge is connected to the stage without having a record on. When I ground the inputs, the stage is almost silent. Based on this, one could conclude the noise is coming from the cartridge.
However, if I connect the same cartridge to my Cambridge Audio 551P phono stage, the noise is considerably lower.
I'm puzzled what could be the cause of this. Both stages have the same gain, input resistance and input capacitance. And both have an equally low noise floor with grounded inputs.
So the bottom line question is: is there any other 'mechanism' that can cause noise (hiss) as soon as a cartridge is connected to a phono stage?
Thanks!
The cartridge is MM type. It's a Nagaoka JT-322 but I can't find any specs on its internal resistance or inductance.
The phono stage topology is a 2 stage design. First stage only amplifies the signal while the second provides additional amplication as well as RIAA equalization. Both stages are built using opamps. The first stage uses a LT1115 opamp, the second a NE5532. Both provide 20dB of gain for a total of 40dB at 1KHz.
Thanks for suggesting current noise! It makes sense and since the problem arises when a cartridge is connected it should be in the first stage, am I right? The only doubt I have is whether the LT1115's input current noise is high enough to cause it.
Also, is it correct to say the higher the source impedance, the more dominant current noise will be?
The phono stage topology is a 2 stage design. First stage only amplifies the signal while the second provides additional amplication as well as RIAA equalization. Both stages are built using opamps. The first stage uses a LT1115 opamp, the second a NE5532. Both provide 20dB of gain for a total of 40dB at 1KHz.
Thanks for suggesting current noise! It makes sense and since the problem arises when a cartridge is connected it should be in the first stage, am I right? The only doubt I have is whether the LT1115's input current noise is high enough to cause it.
Also, is it correct to say the higher the source impedance, the more dominant current noise will be?
Yes, with MM there is the issue of the L an C resonance along with the terminating resistance presenting a high impedance. Sometimes this can increase the hiss. You could try a swap with a 5534 in front just to see or deliberately mis-terminate the cart with like a 4.7k which will mess up the flatness at the high end but should change the noise if it's from the current noise. The LT1115 has more than you would want for MM in general but usually it's not enough to be outright obvious.
EDIT - The recommended load is 100k so this cart is different than any I've seen and points to possible current noise issues.
Last edited:
I will check for oscillation.
Your suggestion to swap with a 5534 is based on its lower current noise than the 1115? It has, however, higher voltage noise which might compromise the improvement.
Can you explain why lowering the input resistance would reveal current noise? I'm really learning now
Your suggestion to swap with a 5534 is based on its lower current noise than the 1115? It has, however, higher voltage noise which might compromise the improvement.
Can you explain why lowering the input resistance would reveal current noise? I'm really learning now
I will check for oscillation.
Your suggestion to swap with a 5534 is based on its lower current noise than the 1115? It has, however, higher voltage noise which might compromise the improvement.
Can you explain why lowering the input resistance would reveal current noise? I'm really learning now
If you read the data sheets you can see the specs for noise in voltage and current, so since V = R X I you can figure how much voltage noise the current noise looks like at the input. Unfortunately it's a little complicated with MM carts since they have a fairly low DC resistance but the inductance, capacitance and load resistor form a complex circuit that can present a resistance to the input much higher than the DC resistance.
I'm being a little sloppy here not differentiating between resistance and the real part of the complex impedance but that would confuse you at this point. The point is MM's usually have too much DC resistance to take full advantage of ultra-low voltage noise op-amps and at high frequencies can have problems with the current noise due to their rising impedance. Without all that carts parameters you can't sit down and compute the result.
Here's a handy Excel sheet to compute your noise.
RIAA Noise Calculator | SYclotron Audio
Thanks a lot! I know where to look now
I didn't pay attention to the frequency dependency of the source impedance of a MM cart. When I was designing this amplifier, I did multiple noise simulations in SPICE and they looked great because I didn't model the cartridge properly. I was aware of its DCR but totally forgot the inductance. With this knowlegde, I revisited the sims and the difference quite impressive.
So I modeled the cartridge as a series RL system. I don't have specs on my cart so I used the ones of the popular Ortofon 2M Red. These are the S/N results assuming 500mVrms at the output:
- Grounded inputs: 85dB
- Cart (R only): 76dB (what I did so far)
- Cart (R + L): 64.5dB (matches the sheet you referred to closely btw)
So then I swapped the 1115 for a 5532 in the simulation and the S/N went up to 73dB! Due to the fact it has somewhat lower input current noise, which dominates voltage noise in this case because of the inductance. Am I drawing the right conclusions here?
One question remains. Why is the 551P stage more silent with my cart? Perhaps because it has a discrete input stage that has lower current noise compared to a LT1115? Could that be realistic?
I didn't pay attention to the frequency dependency of the source impedance of a MM cart. When I was designing this amplifier, I did multiple noise simulations in SPICE and they looked great because I didn't model the cartridge properly. I was aware of its DCR but totally forgot the inductance. With this knowlegde, I revisited the sims and the difference quite impressive.
So I modeled the cartridge as a series RL system. I don't have specs on my cart so I used the ones of the popular Ortofon 2M Red. These are the S/N results assuming 500mVrms at the output:
- Grounded inputs: 85dB
- Cart (R only): 76dB (what I did so far)
- Cart (R + L): 64.5dB (matches the sheet you referred to closely btw)
So then I swapped the 1115 for a 5532 in the simulation and the S/N went up to 73dB! Due to the fact it has somewhat lower input current noise, which dominates voltage noise in this case because of the inductance. Am I drawing the right conclusions here?
One question remains. Why is the 551P stage more silent with my cart? Perhaps because it has a discrete input stage that has lower current noise compared to a LT1115? Could that be realistic?
Yes to everything, with discrete you can optimize with just noise in mind. If you went to JFET's you can approach 80dB SNR but now the surface noise of even the best LP dominates.
Thanks a lot! I know where to look now
I didn't pay attention to the frequency dependency of the source impedance of a MM cart. When I was designing this amplifier, I did multiple noise simulations in SPICE and they looked great because I didn't model the cartridge properly. I was aware of its DCR but totally forgot the inductance. With this knowlegde, I revisited the sims and the difference quite impressive.
So I modeled the cartridge as a series RL system. I don't have specs on my cart so I used the ones of the popular Ortofon 2M Red. These are the S/N results assuming 500mVrms at the output:
- Grounded inputs: 85dB
- Cart (R only): 76dB (what I did so far)
- Cart (R + L): 64.5dB (matches the sheet you referred to closely btw)
So then I swapped the 1115 for a 5532 in the simulation and the S/N went up to 73dB! Due to the fact it has somewhat lower input current noise, which dominates voltage noise in this case because of the inductance. Am I drawing the right conclusions here?
One question remains. Why is the 551P stage more silent with my cart? Perhaps because it has a discrete input stage that has lower current noise compared to a LT1115? Could that be realistic?
A number of comments:
1) S/N for an MM should not be calculated for 500mV but something like 5mV@1Khz, depending on the output of your cartridge at 5cm/sec 1Khz.
2) The LT1115 is a fine amp for a MC Cart but not for an MM, because voltage noise is lower than needed but current noise is far too high.
3) The Syclotron model computes the S/N for a non RIAA amp with zero current noise, since it uses cell W34 instead of cell W35 and lacks the input parameter for current noise.
5) No weighting has been applied to the Syclotron noise figure
4) Non A-weighted noise has no meaning. A weighted noise corresponds more or less to what we hear.
Hans
An OPA1641 having a Fet input is an excellent choice, but also a bipolar AD8675 or the NE5532 are well suited for the job.
Hans
SY added a column to the spreadsheet, he thanks you for pointing that out.
This all comes from an old National App note, it seems trivial to include all factors such as Ib noise and weighting why no one has done it I don't know. A nice project for a rainy day.
I ordered a pair of OPA134's for the input stage. I believe they are the predecessor to the OPA1641. I like the specs of the latter more but unfortunately it's SMD only.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.