Folks,
Bare with me, I am a beginner.... and I am trying to design a Gomes (supertotempole) i/v stage for a tda1541a dac. I am messing around in LTSpice, and am rather disappointed with the THD figures (2.12% !).
This is the circuit I am using, how can I improve the performance and/or simulation? (is my simulation wrong, or is the circuit wrong?)
Thanks 😀
Bare with me, I am a beginner.... and I am trying to design a Gomes (supertotempole) i/v stage for a tda1541a dac. I am messing around in LTSpice, and am rather disappointed with the THD figures (2.12% !).
This is the circuit I am using, how can I improve the performance and/or simulation? (is my simulation wrong, or is the circuit wrong?)
Thanks 😀
Edit: yes, i think it does... When going to FFT, it mentions "Fourier components are normalized to correspond to the time domain RMS amplitude"
Why exactly a Gomes stage ?
Without any kind of feedback or servo circuits (like a css) you will of course get high distortion. It is really not complicated.
In terms of math I think nobody here can tell you how to improve the circuit you posted. Even it looks simple, it is yet to complicated. You just have to do more simulating. I hope you know John Broskie's issue about the gomes configuration: Gomez Vs XPP Amplifier
Without any kind of feedback or servo circuits (like a css) you will of course get high distortion. It is really not complicated.
In terms of math I think nobody here can tell you how to improve the circuit you posted. Even it looks simple, it is yet to complicated. You just have to do more simulating. I hope you know John Broskie's issue about the gomes configuration: Gomez Vs XPP Amplifier
LTSpice shows both Fourier and normalized in the same table, so no problems with that. But there are other faults in the Spice setup that makes the THD figures go wrong.
But whats the use of making a Gomez? Don´t think it will come anything good out of that.
But whats the use of making a Gomez? Don´t think it will come anything good out of that.
Thanks... I now get to 0.049%.
As for the reason why to use this stage, a very successful commercial player (AMR CD-77) uses this topology, and it's designer T. Loesch (active on this forum) has been promoting the circuit for some time here. 🙂
I thought why not try it for myself, but of course I would like to get optimum performance before I decide to build it.
Here's a simple iv stage, if you dont mind transformers. The output level is a little low, but still not so low most preamps can't handle it. My pre is a moderate gain type, and using the volume potentiometer on it's high setting is only a good thing (less potentiometer in the signal).
Sorry this isn't tube, but cant resist...
Sorry this isn't tube, but cant resist...
Attachments
Why on earth a transformer there? What would you gain? If there is DC on the DAC output it will be completely useless as the Lundahl core will saturate. With a 22ohm I/U resistor and 1541, output will be just a few multiples of 10mV. So a 1:1 on a current DAC you can live without. A 1:1 to with primary CT will work great on a balanced voltage DAC though.
Last edited:
Well that reminds me 🙂 the dac is balanced! Lol.
I did this mod a long time ago and forgot that little detail.
Works great. ( the CDplayer is one of those higher-end Sony with the weight that goes on top of the cd, cant remember which model number).
I did this mod a long time ago and forgot that little detail.
Works great. ( the CDplayer is one of those higher-end Sony with the weight that goes on top of the cd, cant remember which model number).
I have tried several output stages, both discrete and tubes. As I said, I would like to try this one due to the fact it's used in one of the best sounding tda1541a implementations known to man...
Are there any ways to further optimize the performance?
I did some more simulations, and got THD down to 0.000543%. 😛
-removed the Rk bypass cap. Of course I lose gain, but I dont really need it... what else do I lose by removing the cap? Slew rate?
-different operating point
-voltage divider between the left triodes. I can't really rationalize why this is beneficial though, but it simulates better... The top-right triode now gets a out-of-phase signal which is smaller in amplitude.... maybe better match with the bottom-right triode plus its plate resistor?
-removed the Rk bypass cap. Of course I lose gain, but I dont really need it... what else do I lose by removing the cap? Slew rate?
-different operating point
-voltage divider between the left triodes. I can't really rationalize why this is beneficial though, but it simulates better... The top-right triode now gets a out-of-phase signal which is smaller in amplitude.... maybe better match with the bottom-right triode plus its plate resistor?
If you are looking for extremely low distortion and still maximum gain you should exchange the upper left triode for a depletion MOSFET biased for equal current through both input and output pair. To keep Zout as low as possible you can exchange the bottom resistor/cap for diodes.
Just as much Gomes as the one you propose but even better😉.
Also note that the one your propose is the one with highest uneven harmonics and THD in my sims.
Actually I don´t think the .four is reliable.
Just as much Gomes as the one you propose but even better😉.
Also note that the one your propose is the one with highest uneven harmonics and THD in my sims.
Actually I don´t think the .four is reliable.
Last edited:
Hey S,
Just noted your .four is useless. Add the Spice-directive .options plotwinsize=0 to get going.
Just noted your .four is useless. Add the Spice-directive .options plotwinsize=0 to get going.
Honestly, I know even less of LTspice than I know of EE. So thanks for your comment... I need it.
I added it, same THD performance while simulating.
I added it, same THD performance while simulating.
The .four isn´t reliable.
A SRPP, that I dislike, sims a lot better(even if the harmless h2 is somewhat higher) as all the higher hamonics are a lot lower.
This indicates you can forget about these sims.
Could be something with the tubemodel. Try many different like E88CC etc. to get a better picture.
A SRPP, that I dislike, sims a lot better(even if the harmless h2 is somewhat higher) as all the higher hamonics are a lot lower.
This indicates you can forget about these sims.
Could be something with the tubemodel. Try many different like E88CC etc. to get a better picture.
Do you have any suitable depletion MOSFET models that work in LTspice? Cant get the supertex library to work. Also not sure whether in contains MOSFETs that can handle this voltage...
- Status
- Not open for further replies.