Hi Passive420,
Sure - I started with a BOM that Jwarch made for me. I have some notes and corrections but will take me a few days to get together for an update. One of my difficulties is that I have so much in my on hand supplies already it’s tough to see what was in the BOM - but I did order from the BOM to the extent possible (even for basic SMT resistors that I knew I had). Glad I did because several items that should be high accuracy low ppm metal thin film were not. Also discovered that for the crossfeed probably better to use film caps vs ceramic NP0’s. The biggest find was the inexpensive eBay LDO regulators (adjustable). Those work great!
It’s a formidable project and I hope you guys who ordered boards make it. I know build to ordering PCBs follow through rate is maybe on average 1/4 if I am not mistaken.
Sure - I started with a BOM that Jwarch made for me. I have some notes and corrections but will take me a few days to get together for an update. One of my difficulties is that I have so much in my on hand supplies already it’s tough to see what was in the BOM - but I did order from the BOM to the extent possible (even for basic SMT resistors that I knew I had). Glad I did because several items that should be high accuracy low ppm metal thin film were not. Also discovered that for the crossfeed probably better to use film caps vs ceramic NP0’s. The biggest find was the inexpensive eBay LDO regulators (adjustable). Those work great!
It’s a formidable project and I hope you guys who ordered boards make it. I know build to ordering PCBs follow through rate is maybe on average 1/4 if I am not mistaken.
Here is the case I am using in case anyone wants to use same one. Great case and excellent value.
PH I Headphone Amplifier All aluminum Chassis Preamplifier Enclosure Mini Case DIY Box-in Speaker Accessories from Consumer Electronics on Aliexpress.com | Alibaba Group
PH I Headphone Amplifier All aluminum Chassis Preamplifier Enclosure Mini Case DIY Box-in Speaker Accessories from Consumer Electronics on Aliexpress.com | Alibaba Group
I have never used crossfeed with my headphones before, so I am trying to decide whether it is worthwhile for me to use that board in my build. X, can you comment more on the defeat feature not fully working? I had imagined a simple bypass switch would be used. (Speaking of which--how are you toggling xfeed on and off? That handsome case appears to have only a power switch on the front.)
It’s hard to see but there is a DPST push button switch in the middle of the PCB. It goes through a hole drilled in the back panel. The filter is based on the Danyuk crossfeed with -4dB attenuation. So it sounds noticeably softer when crossfeed is activated. The problem is that in order to have a fully defeatable switch, another set of wires needs to be switched as well. So in defeat mode, the volume is not attenuated and there is a small amount of crosstalk. I think two small DPST relays can be used effect the defeat effect. I may try that.
Hi X!
Any progress with the BOM? I'm waiting with a finger on the trigger to order the parts...
Will work on it soon!
Here is a measurement of 2.83Vpp into 600ohms metal thin film 1W resistor dummy load. My cabling from the 600ohm resistor to the ADC is unshielded so picking up a little ambient mains EMI. The harmonic profile and THD remain good at 0.0016%:
Here is the stress test for maximum voltage swing before onset of clipping to achieve 15.25Vpp into 600ohms, Harmonic profile and THD still look excellent with monotonically fallin higher orders and dominant 2nd and less 3rd harmonic. Overall THD is still only 0.0076%:
If I max out the DAC output, I can squeeze 24.22Vpp into 600ohms and higher orders appear but it is still only 0.03%THD (quite listenable probably but perhaps higher orders would sound fatiguing?):
So now everyone can see what the limits are of this amp - no cherry picking the data here. Running a higher bias current would improve performance at these extreme drive loads of course. These are all for 115mA (selected using 2.7ohm Drain resistors on the MOSFETs). Also, increasing PSU rails to higher value would help as well - this is for +/-18v rails. There is the 145mA option and one can go as high as +/-20v I think.
Here is the stress test for maximum voltage swing before onset of clipping to achieve 15.25Vpp into 600ohms, Harmonic profile and THD still look excellent with monotonically fallin higher orders and dominant 2nd and less 3rd harmonic. Overall THD is still only 0.0076%:
If I max out the DAC output, I can squeeze 24.22Vpp into 600ohms and higher orders appear but it is still only 0.03%THD (quite listenable probably but perhaps higher orders would sound fatiguing?):
So now everyone can see what the limits are of this amp - no cherry picking the data here. Running a higher bias current would improve performance at these extreme drive loads of course. These are all for 115mA (selected using 2.7ohm Drain resistors on the MOSFETs). Also, increasing PSU rails to higher value would help as well - this is for +/-18v rails. There is the 145mA option and one can go as high as +/-20v I think.
Attachments
Last edited:
BOM for Verification Build
For those wishing to get a start on the BOM, here is what I have so far. Note that not all parts are needed as there are duplicates for certain options. You will have to figure out which to pick when you see something is doubled up.
My main points are this:
1. For the resistors involved in the feedback loop R106/107/108 and R126/127/128 please use high quality metal thin film if you can find it. I only have metal thin film on R108/128 as that was all I had. I believe distortion can be reduced even further if the other two were also metal thin film.
2. For the main MOSFET Drain resistors R117/118 and R137/138, please use high quality metal thin film as well. I only had thick film on hand, so once again, you may see significant improvement in distortion over my verification build if you choose this. I would start with the 2R7 values to get 115mA. If you really want 145mA then use 2R2 and that will be the maximum capability of the Traco Dc-DC (315mA for both channels).
3. Don't use LM7818 and LM7918 voltage regulators as they have no ability to adjust the final voltage after the cap multiplier. Due to the different Vgs across the Pch and Nch MOSFETs for the cap Mx's, you will have different rail values. To get started, it's ok to use, but I recommend the eBay LDO TP7A regulators which have a pot. The pot is very sensitive so it will take some fine fine nudges to get the circa 18v to 19v rails.
4. The Xfeed has an error in the wiring - it was my fault as I thought all we needed was a DPDT switch to implement a cross-bar - I forgot it takes two DPDTs. As it is, the Xfeed can never be completely turned off as there is a small residual crossfeed even in "defeat mode". There is also an error on the C162 and C172, they should be tied to the voltage divider middle point and the other end to GND (not the end of the divider). My recommendation would be to skip the Xfeed unless you feel like debugging it. I can post detailed schematics notes and photos on how to fix if you really want the Xfeed. There is also a silkscreen error, R171 and R172 need to be swapped. As you can see, the one part where I put the least effort into double checking has the most errors. Also, I would recommend using the optional film caps vs Np0 ceramics as it sounds better.
5. The Volume pot board for the HP option has an error. One pad for the smaller RK09 pot (right channel input) is not connected to anything. It needs a small wire jumper to the input trace 5 mm away.
6. There are a lot of on board PCB JST connectors. You are welcome to hardwire it with flying leads. Or use a connector only on one side to allow disconnections yet reduce number of connector crimp wire leads that need to be made.
7. I made a recommendation for the case earlier - it is an excellent fit and great value at $36 shipping included for an all CNC aluminum case with most holes drilled out already. You may need to futz around with the hole for the pot shaft and to get it secure and correct alignment but otherwise it works well. There is an IEC input jack that you will need to make a small plate with a hole for an external DC jack (like typical 5.5mm x 2.1mm barrel jack).
If you notice any inconsistencies, please let everyone know and Jwjarch and I will try to revise the BOM for a clean one down the road. This BOM is still in what I called Beta tester stage and experienced builders will know what to do. I would not recommend using the BOM if you are not an experienced builder who is used to ordering and deciding what the best options are if certain ones not in stock from your favorite online store.
Cheers,
X
For those wishing to get a start on the BOM, here is what I have so far. Note that not all parts are needed as there are duplicates for certain options. You will have to figure out which to pick when you see something is doubled up.
My main points are this:
1. For the resistors involved in the feedback loop R106/107/108 and R126/127/128 please use high quality metal thin film if you can find it. I only have metal thin film on R108/128 as that was all I had. I believe distortion can be reduced even further if the other two were also metal thin film.
2. For the main MOSFET Drain resistors R117/118 and R137/138, please use high quality metal thin film as well. I only had thick film on hand, so once again, you may see significant improvement in distortion over my verification build if you choose this. I would start with the 2R7 values to get 115mA. If you really want 145mA then use 2R2 and that will be the maximum capability of the Traco Dc-DC (315mA for both channels).
3. Don't use LM7818 and LM7918 voltage regulators as they have no ability to adjust the final voltage after the cap multiplier. Due to the different Vgs across the Pch and Nch MOSFETs for the cap Mx's, you will have different rail values. To get started, it's ok to use, but I recommend the eBay LDO TP7A regulators which have a pot. The pot is very sensitive so it will take some fine fine nudges to get the circa 18v to 19v rails.
4. The Xfeed has an error in the wiring - it was my fault as I thought all we needed was a DPDT switch to implement a cross-bar - I forgot it takes two DPDTs. As it is, the Xfeed can never be completely turned off as there is a small residual crossfeed even in "defeat mode". There is also an error on the C162 and C172, they should be tied to the voltage divider middle point and the other end to GND (not the end of the divider). My recommendation would be to skip the Xfeed unless you feel like debugging it. I can post detailed schematics notes and photos on how to fix if you really want the Xfeed. There is also a silkscreen error, R171 and R172 need to be swapped. As you can see, the one part where I put the least effort into double checking has the most errors. Also, I would recommend using the optional film caps vs Np0 ceramics as it sounds better.
5. The Volume pot board for the HP option has an error. One pad for the smaller RK09 pot (right channel input) is not connected to anything. It needs a small wire jumper to the input trace 5 mm away.
6. There are a lot of on board PCB JST connectors. You are welcome to hardwire it with flying leads. Or use a connector only on one side to allow disconnections yet reduce number of connector crimp wire leads that need to be made.
7. I made a recommendation for the case earlier - it is an excellent fit and great value at $36 shipping included for an all CNC aluminum case with most holes drilled out already. You may need to futz around with the hole for the pot shaft and to get it secure and correct alignment but otherwise it works well. There is an IEC input jack that you will need to make a small plate with a hole for an external DC jack (like typical 5.5mm x 2.1mm barrel jack).
If you notice any inconsistencies, please let everyone know and Jwjarch and I will try to revise the BOM for a clean one down the road. This BOM is still in what I called Beta tester stage and experienced builders will know what to do. I would not recommend using the BOM if you are not an experienced builder who is used to ordering and deciding what the best options are if certain ones not in stock from your favorite online store.
Cheers,
X
The PSU is where the building should begin. Note that the positive voltage regulator board should be facing the Traco DC-DC converter. I had to swap that out after a mistake. The boards I received were color coded red for +ve and green for -ve.:
Note that I ended up using the shortest heatsink available in order to fit inside the case. The original one fitted below was too tall. With smaller ones the temperature was about 60C on the heatsink. This design could use a different layout with regulators hanging to the sides and underhung on board to allow external heatsinking to main bottom plate.
Revised PSU schematic here:
http://www.diyaudio.com/forums/atta...s-hpa-sub-ppm-thd-hpa-complete_001a_page2-pdf
Note that I ended up using the shortest heatsink available in order to fit inside the case. The original one fitted below was too tall. With smaller ones the temperature was about 60C on the heatsink. This design could use a different layout with regulators hanging to the sides and underhung on board to allow external heatsinking to main bottom plate.
Revised PSU schematic here:
http://www.diyaudio.com/forums/atta...s-hpa-sub-ppm-thd-hpa-complete_001a_page2-pdf
Last edited:
Hi Morde,
I have bought from several sellers on both eBay and Aliexpress, and have never had a problem with the parts being counterfeit. They are simply too low cost for anyone to bother. They will be marked typically 2AWxx. When I measure the Idss they are always in spec but to be sure, variations in batches on same roll of tape will occur. Typically, i get 50 or so to make sure I have plenty to select from for matching.
Here is one example of a vendor that I bought from:
50PCS BF862 ORIGINAL JFET N-CHAN 20V SOT-23 | eBay
I have bought from several sellers on both eBay and Aliexpress, and have never had a problem with the parts being counterfeit. They are simply too low cost for anyone to bother. They will be marked typically 2AWxx. When I measure the Idss they are always in spec but to be sure, variations in batches on same roll of tape will occur. Typically, i get 50 or so to make sure I have plenty to select from for matching.
Here is one example of a vendor that I bought from:
50PCS BF862 ORIGINAL JFET N-CHAN 20V SOT-23 | eBay