I have also done some P2P Sens.
I had similar symptoms with one channel working when the other was not connected and both not working when both were connected when I made a mess of epoxying the jfets to the heatsink getting unset hardener bridging the jfet legs and heatsink.
I've also seen oscillation when I connected the L+R Sen grounds together with a common lead to the DAC ground.
I have the Sen instances (with bias servos for the pcm1794) and DAC chip in different enclosures and getting the grounding right for my set-up was a nightmare.
This is probably not appropriate for your set-up which is hopefully easier but for the record...
With balanced inputs, my servos need to connect back to the DAC ground for the DC bias.
I got the best results from connecting the L Sen ground to the L shield of the balanced output to the L amp monoblock and similarly for the R Sen instance. These shields are connected to chassis ground where the signals enter the chassis. This follows the rule for balanced circuits of one connection from circuit ground to chassis ground. L+R Sen grounds are not connected together except through the chassis.
The L+R servo grounds which are separate in the Sen enclosure go back to the DAC enclosure (which is a PC with a noisy SMPS) in a third wire in each of the L+R shielded cables from the DAC chip and connect to the sound card ground. The ground of the bias servo power supply in the Sen enclosure is connected to one of these grounds in the Sen enclosure and therefore to the other one via the sound card. The shields on these cables are telescoped from the PC chassis and not connected to the DAC enclosure.
I have tried a lot of other combinations including leaving the Sen instances floating and this way is one of a few good solutions.
So, 3 lessons learned: it's a mistake to put the bias circuit in a different enclosure from the DAC chip (having the Sen circuit in a different enclosure may be a mistake but can be made to work well enough though); it's important to keep the L+R circuits separate to avoid oscillation; for smooth sound it's important to avoid ground loops.
Thanks for the suggestions
I am single ended, and have keep the channels completely separate, sans the heat sink. There is no path between the heat sink and the lead of the Jfets. I was hoping to just connect the sen to the DAC chip analog out/grd directly without installing it inside the player to evaluate the sound before removing the class A opamp system in use. I did remove the feed to it however(obvious). I was mostly happy with it, but you know how we modding audio types are, something better???? The signal connection and battery leads are twisted for outside rejection and about 8 inches long. No cables within 12 inches of these. Will try the grounding strap to the sink today.
I am single ended, and have keep the channels completely separate, sans the heat sink. There is no path between the heat sink and the lead of the Jfets. I was hoping to just connect the sen to the DAC chip analog out/grd directly without installing it inside the player to evaluate the sound before removing the class A opamp system in use. I did remove the feed to it however(obvious). I was mostly happy with it, but you know how we modding audio types are, something better???? The signal connection and battery leads are twisted for outside rejection and about 8 inches long. No cables within 12 inches of these. Will try the grounding strap to the sink today.
We have been asked about how to connect the SEN IV for AKD12P for 2 channel single ended (ES9012 DAC).
I cannot find any manual of the AKD12P in the public domain, and I do not want to guess.
So I think the best thing to do is to post the question here and let those who has the same gives the answer.
My speculation is that it has 2 channel differential out and DC offset as in the ES9018 in stereo mode.
So you would need to use SEN-V18 with 16 2SK369Vs in total.
You will then have 2 channel balanced output.
These you can convert to single ended using a number of solutions, such as ICs, transformers, or the XCEN converter.
There are people who suggested using only one phase of the output.
I cannot find any information in the ES9012 datasheet regarding unconnected current outputs, or single phase performance figures.
All specs are quoted in differential mode, which are of course better than single ended due to common mode rejection and even harmonics cancellation.
So if I were to do it myself I would use a converter.
But you can also read Paul's experience with his SE solutions in this thread.
Do feel free to experiment and post your results for all to share.
Patrick
I cannot find any manual of the AKD12P in the public domain, and I do not want to guess.
So I think the best thing to do is to post the question here and let those who has the same gives the answer.
My speculation is that it has 2 channel differential out and DC offset as in the ES9018 in stereo mode.
So you would need to use SEN-V18 with 16 2SK369Vs in total.
You will then have 2 channel balanced output.
These you can convert to single ended using a number of solutions, such as ICs, transformers, or the XCEN converter.
There are people who suggested using only one phase of the output.
I cannot find any information in the ES9012 datasheet regarding unconnected current outputs, or single phase performance figures.
All specs are quoted in differential mode, which are of course better than single ended due to common mode rejection and even harmonics cancellation.
So if I were to do it myself I would use a converter.
But you can also read Paul's experience with his SE solutions in this thread.
Do feel free to experiment and post your results for all to share.
Patrick
connecting just one phase is bound for fail... the entire internal routing and analogue performance is based on using a very symmetrical signal outputs, you will lose more than 3dB thats for sure. a bal-se conversion stage is a much better solution. i'm interested in your XEN actually for a friend, are you nearing enough interest for production? nvm i'll ask in the thread.
yes you are basically correct, just treat the AKD12 analogue output specifications as es9018 in stereo balanced mode but with all the dacs paralleled internally instead of externally.
the AVCC/2 voltage is taken from the AVCC flea type regs and divided, for best performance they should be increased to 3.6v with an increased input voltage for the regs of >+/-10vdc (I use 12). it doesnt make for a very neat connection with SEN, because the outputs are in the center very close to the dac, rather than the edges and the pinout +/- and G are all very tightly coupled together for best performance.
I think for best connection to Ackodac a DIY PCB or perfboard construction arranged to allow stacking more directly is best. otherwise I fear the performance of both devices will suffer more than needed.
the current PCB can be made to fit, but would require some custom made rigging to allow close enough current connections and avoid having to keep the power supply (batteries) so far away
yes you are basically correct, just treat the AKD12 analogue output specifications as es9018 in stereo balanced mode but with all the dacs paralleled internally instead of externally.
the AVCC/2 voltage is taken from the AVCC flea type regs and divided, for best performance they should be increased to 3.6v with an increased input voltage for the regs of >+/-10vdc (I use 12). it doesnt make for a very neat connection with SEN, because the outputs are in the center very close to the dac, rather than the edges and the pinout +/- and G are all very tightly coupled together for best performance.
I think for best connection to Ackodac a DIY PCB or perfboard construction arranged to allow stacking more directly is best. otherwise I fear the performance of both devices will suffer more than needed.
the current PCB can be made to fit, but would require some custom made rigging to allow close enough current connections and avoid having to keep the power supply (batteries) so far away
Last edited:
We have enough discussion about stacking. You can refer to those a while back.
I remain totally convinced that the current PCB close to a shield and away from the DAC is my preferred solution,
especially that the connection between DAC and IV is in current mode.
So I see no need to change anything.
But do what pleases you, which is what DIY is all about.
Best regards,
Patrick
I remain totally convinced that the current PCB close to a shield and away from the DAC is my preferred solution,
especially that the connection between DAC and IV is in current mode.
So I see no need to change anything.
But do what pleases you, which is what DIY is all about.
Best regards,
Patrick
Both are correct. 4mApkpk/Dac output, and there are 4 parallel dacs in stereo mod.
Someone at some fishy forum was looking for a Spice model of the SEN IV.
Of course we have; this one for SEN V18 together with a simulated ES9018 output.
This is meant for you to play around with.
The FFT is not accurate, so don't quote THD figures from this.
For that we have other methods / models.
Patrick
Of course we have; this one for SEN V18 together with a simulated ES9018 output.
This is meant for you to play around with.
The FFT is not accurate, so don't quote THD figures from this.
For that we have other methods / models.
Patrick
Many fishy thanks Patrick, pleased your still looking after me.
Pleased it works with LTSpice as its the only one I can use although only in a novice manor but it will help with my diy enjoyment and maybe teach me something also.
Loving SEN by the way and surprised more or not using it with there Buffalos.
Taking your advice and moving to differential inputs on power amps in the near future.
Pleased it works with LTSpice as its the only one I can use although only in a novice manor but it will help with my diy enjoyment and maybe teach me something also.
Loving SEN by the way and surprised more or not using it with there Buffalos.
Taking your advice and moving to differential inputs on power amps in the near future.
There were some discussions elsewhere about using the SEN circuit for very high current output DACs.
Examples include using multiple ES9018s in parallel.
For such application, it is perhaps worth considering that the circuit bias be at leat 4~8 times that of the total DAC output.
You can always using many many 2SK369Vs in parallel, e.g. 8x to get to 200mA bias.
But if you are looking at such bias current, you might also consider the LU1014 triode cell as used in the DAO.
http://www.diyaudio.com/forums/head...lass-zgf-headphone-amplifier.html#post3284265
I have not tried this myself, so just an idea.
Patrick
Examples include using multiple ES9018s in parallel.
For such application, it is perhaps worth considering that the circuit bias be at leat 4~8 times that of the total DAC output.
You can always using many many 2SK369Vs in parallel, e.g. 8x to get to 200mA bias.
But if you are looking at such bias current, you might also consider the LU1014 triode cell as used in the DAO.
http://www.diyaudio.com/forums/head...lass-zgf-headphone-amplifier.html#post3284265
I have not tried this myself, so just an idea.
Patrick