This is something that I have been working on over the last week or so. The inspiration comes from an USB audio interface - a Tascam 2X2. The preamp section has always produced very subjectively satisfying results. It is also works well both balanced and single ended.
A while ago it one of the analogue inputs died and in order to fix it without a service manual I had to trace the circuit of the preamp section. For all of the Tascam marketing that made a big deal of the discrete, low noise, high gain, high dynamic range preamps it turned out that really it was a fairly standard mic preamp circuit. This inspired me to look into the idea of using a similar kind of circuit as the front end of a phono stage.
I have to openly draw attention to the fact that what I wanted was a relatively simple circuit. so I could have as much design input as possible. Also something quite flexible being both for balanced and un-balanced MC and MM cartridges without separate circuits for both. This would mean that I would have to have some compromise on outright lowest noise and distortion but I still wanted decent performance.
This is the circuit so far:
There a a few things that I still haven't wrapped my head around fully yet. One of these is he input transistors in parallel. I have gone to some effort to learn how to model the noise accurately in Tina and believe I have done this correctly. Paralleling 3 common garden 2N4403s in the LTP does seem to provide some small benefit and as it stands the circuit does seem to have decent S/N. I plan to look into other transistors in parallel, hopefully modelling these in Tina.
MC frequency response is pretty flat. The roll off at the extremes is intentional. Although these is content at these extremes I would argue that there is no musical content. I find that some roll off of lowest frequencies has a subjectively positive effect. Also I find that some roll off of the highest frequencies seems to help suppress pops and clicks a little.
MM response has a slight rise as you increase frequency but this is only around 192mdB and I doubt that it would be audible.
Distortion at VF2 and VF3 is 0.035 for MM and 0.045 for MC. What I have yet to work out is why it's around 9% at VF1. I have always had similar figures for other simulated phono amps that have worked fine. I guess I am not simulating the test properly.
The shown gains are not correct. This is caused by the inverse RIAA macro, which is based on a passive RC circuit.
Although I have designed the circuit overall I should mention that the input stage is based on an ESP Audio circuit which was uncannily similar to the Tascam circuit. However, I feel that I have done enough overall to call it my own.
I've yet to work out how to simulate the distortion properly for phono stages. It seems like the RIAA network throws everything off.
Also can't work out what caused the very slight deviation in the RIAA curve for MM cartridges.
A while ago it one of the analogue inputs died and in order to fix it without a service manual I had to trace the circuit of the preamp section. For all of the Tascam marketing that made a big deal of the discrete, low noise, high gain, high dynamic range preamps it turned out that really it was a fairly standard mic preamp circuit. This inspired me to look into the idea of using a similar kind of circuit as the front end of a phono stage.
I have to openly draw attention to the fact that what I wanted was a relatively simple circuit. so I could have as much design input as possible. Also something quite flexible being both for balanced and un-balanced MC and MM cartridges without separate circuits for both. This would mean that I would have to have some compromise on outright lowest noise and distortion but I still wanted decent performance.
This is the circuit so far:
There a a few things that I still haven't wrapped my head around fully yet. One of these is he input transistors in parallel. I have gone to some effort to learn how to model the noise accurately in Tina and believe I have done this correctly. Paralleling 3 common garden 2N4403s in the LTP does seem to provide some small benefit and as it stands the circuit does seem to have decent S/N. I plan to look into other transistors in parallel, hopefully modelling these in Tina.
MC frequency response is pretty flat. The roll off at the extremes is intentional. Although these is content at these extremes I would argue that there is no musical content. I find that some roll off of lowest frequencies has a subjectively positive effect. Also I find that some roll off of the highest frequencies seems to help suppress pops and clicks a little.
MM response has a slight rise as you increase frequency but this is only around 192mdB and I doubt that it would be audible.
Distortion at VF2 and VF3 is 0.035 for MM and 0.045 for MC. What I have yet to work out is why it's around 9% at VF1. I have always had similar figures for other simulated phono amps that have worked fine. I guess I am not simulating the test properly.
The shown gains are not correct. This is caused by the inverse RIAA macro, which is based on a passive RC circuit.
Although I have designed the circuit overall I should mention that the input stage is based on an ESP Audio circuit which was uncannily similar to the Tascam circuit. However, I feel that I have done enough overall to call it my own.
I've yet to work out how to simulate the distortion properly for phono stages. It seems like the RIAA network throws everything off.
Also can't work out what caused the very slight deviation in the RIAA curve for MM cartridges.
Neat! I use a focusrite scarlett 18i20 as a phono pre/adc into my raspberry pi. The riaa correction is all done digitally. Works quite well.
One thing I do have is a very small (20pf if memory serves) capacitor across the input and ground, which may be where you have some deviation in your frequency response.
One thing I do have is a very small (20pf if memory serves) capacitor across the input and ground, which may be where you have some deviation in your frequency response.
I'm not sure I would try and do a universal input given the differing needs of MM and MC. But all good fun.
I just done it so I would have the option, accepting that while convenient it may not be technically optimal.
Funny you should mention it though. Turned out that simulating noise of parallel transistor was easier than I thought providing that Tina already has the model. For other transistors I will have to build the noise model using the noise source macros I guess.
For MC, 6R series with 100R load to ground.
1 x 2N4403 - O/P noise = 78n
3 x 2N4403 - O/P noise = 67.5n
6 x 2N4403 - O/P noise = 45.3n
10 x 2N4403 - O/P noise = 37.2n / I/P noise 937p
For MM, 500r series with 47K load to ground.
1 x 2N4403 - O/P noise = 38.5n / I/P noise 3.7n
2 x 2N4403 - O/P noise = 38.5n
6 x 2N4403 - O/P noise = 43.4n
It may make some sense to separate the MM & MC or at least switch out the number of transistors for each type of cartridge.
Funny you should mention it though. Turned out that simulating noise of parallel transistor was easier than I thought providing that Tina already has the model. For other transistors I will have to build the noise model using the noise source macros I guess.
For MC, 6R series with 100R load to ground.
1 x 2N4403 - O/P noise = 78n
3 x 2N4403 - O/P noise = 67.5n
6 x 2N4403 - O/P noise = 45.3n
10 x 2N4403 - O/P noise = 37.2n / I/P noise 937p
For MM, 500r series with 47K load to ground.
1 x 2N4403 - O/P noise = 38.5n / I/P noise 3.7n
2 x 2N4403 - O/P noise = 38.5n
6 x 2N4403 - O/P noise = 43.4n
It may make some sense to separate the MM & MC or at least switch out the number of transistors for each type of cartridge.
If you want to get sensible results out of the noise simulation with a moving magnet cartridge, don't forget to include its 500 mH or so inductance. This makes the whole thing far more sensitive to the equivalent input noise current than with a 500 ohm source.
By the way, you can get a good noise performance for both MC and MM by using very high transconductance JFETs. With bipolars you will need to find some compromise tail current or switch it depending on the type of cartridge.
By the way, you can get a good noise performance for both MC and MM by using very high transconductance JFETs. With bipolars you will need to find some compromise tail current or switch it depending on the type of cartridge.
Last edited: