No need to use 2oz copper.
Just look at the PCB design and you will know why.
It was designed specifically to make use of double layer 1oz in parallel.
And don't use ENIG.
1) Nickel is magnetic.
2) Gold will cause embrittelment when dissolved in solder, over time, especially under thermal cycles.
Just standard (lead-free) tinned 1oz copper is just fine.
Patrick
Just look at the PCB design and you will know why.
It was designed specifically to make use of double layer 1oz in parallel.
And don't use ENIG.
1) Nickel is magnetic.
2) Gold will cause embrittelment when dissolved in solder, over time, especially under thermal cycles.
Just standard (lead-free) tinned 1oz copper is just fine.
Patrick
thank you for sharing details
since you have 4 Ohm speakers why don't try adding additional pair output devices and use more bias, exactly as post 1 documented schematic 2
I am very Interested to do such an amp with rails at 25Vdc and bias 2.5A
since you have 4 Ohm speakers why don't try adding additional pair output devices and use more bias, exactly as post 1 documented schematic 2
I am very Interested to do such an amp with rails at 25Vdc and bias 2.5A
I managed to get both channels up and running with the new matched FQA mosfets. Rail voltages were +-21,7V with the First Watt type of CRC supply and 300VA/2x18V transformer. Bias was stable over one hour 1,35A for left and 1,36A for right. DC offsets were easily settable to few mV without servo.
Some listening impressions:
I had my Ian Canada dual ESS9038 DAC feeding the amp directly. Music was flac files ripped from CDs and also a couple of high-res files 24bit/96/192kHz played with Moode software on a RPi 2. I compared the UDNeSS against my Denon AVR-X3000 amp and Volume levels were matched using DMM and 1kHz sine wave. My speakers are Phonar Veritas P6 Next (4ohm, 87dB/W).
Bass control/separation is still the biggest culprit with my 4ohm speakers, but there has been some positive development since the first proto iteration. Very small difference to my Denon AVR-X3000. Bass depth is very good.
Vocals is where the UDNeSS really shines. Very good presence, details and musical feeling that gives you goosebumps.
Tweeter section was also somewhat smoother and more pleasant than the Denon.
Sound stage has improved since the last listening session with the earlier proto and is on par or better now than Denon.
Overall the UDNeSS has a bit warmer and more pleasant sound and it shines with acoustic/lighter/vocal music.
I would dare to say that this is really good value for money with easy-to-get parts and if you happen to have speakers with higher impedance/sensitivity it will shine even more. I’m a happy camper even with my present speakers.
With a bit of help, I'd be willing to organize a USA group-buy / group-effort for boards and matching parts.
I'd be willing to order the PCBs. It would be my first PCB order, but guidance has been plentiful, so I see a high likelihood of success.
I'd be willing to buy a sufficient multiple of parts and create matched pairs. For the bugs, I'll get the jig Patrick recommended at my cost. I've matched MOSFETs prior, but some of this will be new. Patrick posted some documents that were very helpful.
All costs (with the exception of the jig and my time) would be spread evenly across the group. Since yield seems to be low in matching and/or finding the correct range for some parts; I'd send the remainder of the parts to the group. I cannot offer matches to Patrick's precision with a curve tracer under ideal thermal conditions. I'd be willing to do simple measurements without a heatsink / heat table. If this circuit would benefit greatly or requires more tightly matched parts under correct thermal conditions, then I could not offer to match. All of my MOSFETs for previous projects were matched to circuit voltage and current without heatsinks after 30s of power with all parts starting at room temp. I'd be willing do do something similar. I've never matched JFETs, BJTs or any "bugs". I've only done 9240s et. al., but I've read how to do it, and I believe I'd be successful.
If that creates an interest - I'd be willing to do this for up to 20 boards / 10 pairs. Limit would be one set per person.
I'd also be willing to set up a new thread and/or contribute to this thread to document a build.
If this is not the appropriate place to post or if I've proposed anything outside the bounds of a normal group-buy, my apologies. Right now, I am gauging interest vs. offering a group buy. Mods can please relocate as needed.
Many people, including Patrick, have donated their time to make things easier for me as a noob and put out fantastic projects. I'm happy to do what I can to contribute here.
Plus, I'm going to build one anyway...
I'd be willing to order the PCBs. It would be my first PCB order, but guidance has been plentiful, so I see a high likelihood of success.
I'd be willing to buy a sufficient multiple of parts and create matched pairs. For the bugs, I'll get the jig Patrick recommended at my cost. I've matched MOSFETs prior, but some of this will be new. Patrick posted some documents that were very helpful.
All costs (with the exception of the jig and my time) would be spread evenly across the group. Since yield seems to be low in matching and/or finding the correct range for some parts; I'd send the remainder of the parts to the group. I cannot offer matches to Patrick's precision with a curve tracer under ideal thermal conditions. I'd be willing to do simple measurements without a heatsink / heat table. If this circuit would benefit greatly or requires more tightly matched parts under correct thermal conditions, then I could not offer to match. All of my MOSFETs for previous projects were matched to circuit voltage and current without heatsinks after 30s of power with all parts starting at room temp. I'd be willing do do something similar. I've never matched JFETs, BJTs or any "bugs". I've only done 9240s et. al., but I've read how to do it, and I believe I'd be successful.
If that creates an interest - I'd be willing to do this for up to 20 boards / 10 pairs. Limit would be one set per person.
I'd also be willing to set up a new thread and/or contribute to this thread to document a build.
If this is not the appropriate place to post or if I've proposed anything outside the bounds of a normal group-buy, my apologies. Right now, I am gauging interest vs. offering a group buy. Mods can please relocate as needed.
Many people, including Patrick, have donated their time to make things easier for me as a noob and put out fantastic projects. I'm happy to do what I can to contribute here.
Plus, I'm going to build one anyway...
That is very kind and generous of you, I have no dou t there would be interest.
From a mosfet matching perspective, the Vgs does drop very slightly under thermal conditions. I made a very simple jig with a small heatsi k to allow for things to stabilize over a minute or so. No idea how much more accurate it is but it is simple and allows for quick swapping of mosfets under test.
.. dB
From a mosfet matching perspective, the Vgs does drop very slightly under thermal conditions. I made a very simple jig with a small heatsi k to allow for things to stabilize over a minute or so. No idea how much more accurate it is but it is simple and allows for quick swapping of mosfets under test.
.. dB
Thanks dB. Greatly appreciated. I'm flexible.
My constraints are: it should provide a benefit, does not increase my budget too dramatically (I can probably find a free or cheap small sink and tap it), and does not increase time-to-measure toooooo dramatically. From what I can tell early on - 30% yield or so can be expected for the MOSFETs. So, even if it takes me ~5 mins or so to mount, measure, unmount - vs. about a minute for my previous process - I can do that. Even if it's 10mins, I could do it. Mainly, I want to agree among anyone participating what seems reasonable.
My constraints are: it should provide a benefit, does not increase my budget too dramatically (I can probably find a free or cheap small sink and tap it), and does not increase time-to-measure toooooo dramatically. From what I can tell early on - 30% yield or so can be expected for the MOSFETs. So, even if it takes me ~5 mins or so to mount, measure, unmount - vs. about a minute for my previous process - I can do that. Even if it's 10mins, I could do it. Mainly, I want to agree among anyone participating what seems reasonable.
You should start a new thread at the GB board titled :
Interest for potential GB for UDNess PCBs and matched FETs.
Otherwise some moderator might move the whole thing to the GB board, which is the wrong place for this.
No need for perfect match in this case.
The match is only for Left and Roght channel, not top and bottom MOSFETs
If anything the 2SK209GRs want a tighter match.
You will get good yield if you buy > 100 pieces.
Same for the MOSFETS, 10 is way to few, 50~100 minimum.
With 10 subscribers 20 channels, you need 40 MOSFETs anyway.
And they are reallly cheap, compared to other components such as caps.
So don't save on something that has a significant impact on sound.
If there are good response with your GB, you can then consider a separate build thread.
Otherwise you are also most welcome to post your building experience here.
Most importantly, enjoy,
Patrick
Interest for potential GB for UDNess PCBs and matched FETs.
Otherwise some moderator might move the whole thing to the GB board, which is the wrong place for this.
No need for perfect match in this case.
The match is only for Left and Roght channel, not top and bottom MOSFETs
If anything the 2SK209GRs want a tighter match.
You will get good yield if you buy > 100 pieces.
Same for the MOSFETS, 10 is way to few, 50~100 minimum.
With 10 subscribers 20 channels, you need 40 MOSFETs anyway.
And they are reallly cheap, compared to other components such as caps.
So don't save on something that has a significant impact on sound.
If there are good response with your GB, you can then consider a separate build thread.
Otherwise you are also most welcome to post your building experience here.
Most importantly, enjoy,
Patrick