Measurements in post 149 is with dead time RCD network as per schematic from post #1 (100 ohm , 100pF) , now I have changed capacitor value to 200pF and this is a result. In your opinion which is beeter, with 100pF or with 200pF? Dsd64 and dsd128 is supported and recommended.
edit: added another ddpd5V 200pF sample (picture 2 and 3). In my opinion better result is with 100pF in high frequency end, but with 200pF a bit smoother but with a bit highr noise in high frequency end.
edit: added another ddpd5V 200pF sample (picture 2 and 3). In my opinion better result is with 100pF in high frequency end, but with 200pF a bit smoother but with a bit highr noise in high frequency end.
Attachments
Last edited:
I agree! It's a fact, measurements have shown it without error. The subjective feeling is different, the sound stage is realy huge. I will continue tests on https://www.diyaudio.com/community/threads/ct7302pl.357501/page-5, currently I'm waiting for two month Aliexpress CC,CV SMPS power supply and I will need to modify those ganFet concept a bit to be able to play music from zero volts, just few modifications is need
Mini ddpd on this concept have 32 gates per channel to form an average current with their own output resistor and it is quite possible that each gate has its own difference and not work idealy at the same time. I'm believing ganFet will get much bettr measurement result. The sound would be subjectively not much diferent. I will open ddpd mini v2 opensource, probably I will not sell mini ddpd product, sorry to say that, 250eur is a lot for this with a fact of high measured thd, and I can't sell for less. I will continue on something with better measurement result with a lot of more power. If you need ddpd mini v2 I will put gerbers, schematic and things on this thread for those who need one! JlcPcb can solder things for you with a resonable price. In day or two V2 will be fully open source! I think it's fairer than if I sell to someone for 250 euros so that they are not satisfied at the end. I will put my focus on ganFet ddpd, its on hold for almost one year, I will modify those pcb, will remove MAX flybuck and put something better, an simple push pull which will not give me noise problem as I have using MAX flybuck. The seccond thing will be to ganFet make half brigde independent, right now it was designed to share their +HV and PGND with MAX flybuck, and I had problem with noise. Frontent is also on hold for almost a year, I have received all the pcbs and things, just need to solder them. It will be ideal for further experiments. : )
Last edited:
Stay tuned! I think I found cause of high thd and resolution lost, my diy amanero is bad definitelly, see how bad square wave it have, its the 10x our ddpd mini dead time and thats why ddpd mini have only 9 bits in Multitone : )
DDPD have very very clean and nice sine wave at their output, expecialy at dsd128, so this must be I'm believing at least 24bits at their output when input square vawe get fixed. And one more thing I found in relation to amanero, it didn't stop when my music stop, I still see theit squares at theit output. I think I will continue after frontend get finished.
The strange thing DATA pin on amanero is so short and there is no capacitors at the pin, its realy strange why square wave is that bad, do you have idea? It have almost 100ns propagation delay, very strange!
DDPD have very very clean and nice sine wave at their output, expecialy at dsd128, so this must be I'm believing at least 24bits at their output when input square vawe get fixed. And one more thing I found in relation to amanero, it didn't stop when my music stop, I still see theit squares at theit output. I think I will continue after frontend get finished.
The strange thing DATA pin on amanero is so short and there is no capacitors at the pin, its realy strange why square wave is that bad, do you have idea? It have almost 100ns propagation delay, very strange!
Attachments
Last edited:
No, this will go opensource! Just want to figure out things. The fact is ddpd have very clean sine wave while dsd128 and from the start I was shocked about why Multitone see only 9bits. In comparation to my measurement using UCD180, sine wave at their ouptut gives me 17.9 bits but sine wave looks like an form of sine inside sine, that inside sine contain an constant 300-400mVpkpk sine, in comparation to clean ddpd sine wave at their output bad measurement result shocked me. Today I was surprised and got very optimistic after I saw bad square vawe from my amanero : )
U2 on our diy amanero design is LP5907MFX-3.3/NOPB and U2 also supply that U3 and all the things here, migh be that U2 is simply not enought to supply all the things here? OMG! : ) Looks like square wave I have got seems nicely track figure 15 and 16. I will try to add some big boiler capacitors to the 3.3V line for next test
Attachments
Last edited:
Yes, amanero is measured when no load at DATA pin! Also I have measured 3.3V line on amanero while amanero play dsd track, LP is definitelly ok! Picture 1 is while amanero is play dsd track,, measured on 3.3V line. Picture 2 is when I added an boiler capacitor 2200uF to 3.3V pinout of amanero, notice no changes on DATA out square. Hmm than what migh cause bad square wave I have realy no idea : ( Might be that cpld track clock source wave? I don't think it work that way on amanero. Osciloscope definitelly is able to display 10ns with no problem, notice around 150ns propagation delay on amanero data out pin when no load is there, 150ns rising edge and 100ns failing edge, this is realy strange. I will try to add some resistor load next day to data out pin!
Attachments
Last edited:
I'm sorry for wasting your time and this whole page! The truth is my probe x1 displays it wrong, its almost ten times the wrong from the real. Probe 10x looks ok, after calibrating it its close to +-10ns error, and this is the truth about my amanero after calibration and switching probe to 10x . Picture 1 is calibration process and seccond picture is measured amanero square wave. So Amanero is ok!
Attachments
I think problem on ddpd is buffer stage! One single bufer for too many number of gates in full bridge might be a bad idea because of paralel capacitance! I don't see anything else wrong besides that? 74lvc1g99 which is on dsd input have schmit triger and the only remaining problem might be only bufer and the number of gates inside full bridge, in real - number of gates inside full bridge definitelly it not work at the same time since its not paired, or is it? I will do further work in osciloscope mesurement in next few days.
Last edited: