That was also the conclusion of lots of dudes in basements when battery biasing valves was still a thing.
diyiggy is on to something here. Today I removed the ceramic caps first on the 45mHz adapter, later on the 49. In both cases a BG HiQ .68uF cap is in place on the top. I must say to my ear sound is improved. Like any clock improvement it just gets clearer, cleaner. Perceived volume goes up which I think might be result of lower noise floor.
The other thing that I think is helping the sound is vibration dampening. I added sand in ziplock bags around the Pi. To me its like a clock upgrade. In my case the Pi is mounted on its standoffs which are in silicon in a block of wood. The whole thing is in attached to a plate within the chassis. I just pile sandbags around it and sound improves. It is really very good with these simple tweaks and a few days of run in.
Must admit I really enjoy this rig. I like that its easy to remove and replace the clocks. Andrea's GB for SC cut crystals is bringing a well tempered master clock in the future.
Do you guys think a clock upgrade is a must? The manual somehow implies it. And at least some folks over here seem to confirm it.
From what I understand I would have to order the adapter boards from Ian and the clocks from DigiKey/Mouser. And then I'd have to put it all somehow smd-style together. Sounds very much like a DIYchallenge.
And the cost for it will at least be another $80.Almost as much as the board.
That was always Ian's intention as I understand it, have fun with clocks!
I skipped the clock adapters in the end. Extra trace length, extra inductance, more junctions. The board has footprints that will fit NDK's and Crysteks directly, I removed the DIL adapters for access. Be warned though you'll need a magnifier for NDK's and confidence with SMD... they are tiny! I popped a 10nF film cap over my NDK's across the supply for good luck. Even though my rig isn't run in yet, the step up in resolution was not subtle.
Lorien can do you a pair of NDK's for 30 Euro, just email him.
Last edited:
clock upgrade
If you are happy with HiFi. Then you can get away with the clocks supplied. But they were only intended for testing. If he had supplied expensive clocks, then that would have to be built into the price. Most of us have clocks already from another project. I would not want to have paid for NDK clocks when I already have some at home. I think that Ian's approach is excellent. You provide tremendous flexibility to play around with different clock solutions and understand what the impacts are. You tune your system to your personal taste.
If you are happy with HiFi. Then you can get away with the clocks supplied. But they were only intended for testing. If he had supplied expensive clocks, then that would have to be built into the price. Most of us have clocks already from another project. I would not want to have paid for NDK clocks when I already have some at home. I think that Ian's approach is excellent. You provide tremendous flexibility to play around with different clock solutions and understand what the impacts are. You tune your system to your personal taste.
He could have simply offered a finished set of clock upgrades.
The cheap default clocks are obviously wasted.
He could have offered a device probably at around 120 (-10+30) with NDKs then.
Good to see that somebody confirms my opinion about these clock sockets and adapters.
We all know from playing with opamps what impact these can have.
Not sure if Ian really intends to serve just a tiny market of super tweakers.
To me that'd be more then wasted efforts.
Especially when looking at the great efforts he puts into it and when looking at his excellent looking
finished and highly complex boards.
Luckily, the board I have here is borrowed...
...and I2S-master DACs don't need the FiFo and still achieve femto-sec jitter levels.
Perhaps next time.
Enjoy.
The cheap default clocks are obviously wasted.
He could have offered a device probably at around 120 (-10+30) with NDKs then.
Good to see that somebody confirms my opinion about these clock sockets and adapters.
We all know from playing with opamps what impact these can have.
Not sure if Ian really intends to serve just a tiny market of super tweakers.
To me that'd be more then wasted efforts.
Especially when looking at the great efforts he puts into it and when looking at his excellent looking
finished and highly complex boards.
Luckily, the board I have here is borrowed...
...and I2S-master DACs don't need the FiFo and still achieve femto-sec jitter levels.
Perhaps next time.
Enjoy.
Last edited:
don't give up th eidea to test smd acrylic cap here
that please me to see I'm not complety deaf yet.
Thank you for your tests testimonial.
Be carefull on clonclusions : it also could be because of the total amount of decoupling to the ground. Less being better in your case. But it amazed me to hear indeed a difference on the two sampling rates when working on the decoupling of only an area.
You should this time keep of one BG NX only then listen. Then populate the acrylic you have on hand to the bottom of the previous uncaped adaptator.
Iterative path. I agree it's about tunning it's own hifi, but for mine had always an improvment when I swapped X7R (and also NPO which is stranger!) by acrylic equivalent (but the inductance as th eacrylic are often larger case size than ceramic class II).
that please me to see I'm not complety deaf yet
.Thank you for your tests testimonial.
Be carefull on clonclusions : it also could be because of the total amount of decoupling to the ground. Less being better in your case. But it amazed me to hear indeed a difference on the two sampling rates when working on the decoupling of only an area.
You should this time keep of one BG NX only then listen. Then populate the acrylic you have on hand to the bottom of the previous uncaped adaptator.
Iterative path. I agree it's about tunning it's own hifi, but for mine had always an improvment when I swapped X7R (and also NPO which is stranger!) by acrylic equivalent (but the inductance as th eacrylic are often larger case size than ceramic class II).
clock decoupling
agreed, diyiggy
I will take the next step to install the .68uF FCA cap I have on hand. Just moving a little slowly. As many will point out, its easy to think you hear something and then later wonder if you imagined it.
But what I believe I heard clearly over a couple of sessions was:
1. a clear improvement after removing the ceramic caps on the 45mHz clock which is the clock used predominately. But after a fair bit of listening and clearly letting the system warm up it still felt like there was a little edge that should not be there.
2. so I followed your advice and removed the caps on the 49mhz clock. I believe the sound cleared up at both speeds. But now we are in the range of easily imagined improvement so I look forward to hearing it again and then trying the FCA cap.
agreed, diyiggy
I will take the next step to install the .68uF FCA cap I have on hand. Just moving a little slowly. As many will point out, its easy to think you hear something and then later wonder if you imagined it.
But what I believe I heard clearly over a couple of sessions was:
1. a clear improvement after removing the ceramic caps on the 45mHz clock which is the clock used predominately. But after a fair bit of listening and clearly letting the system warm up it still felt like there was a little edge that should not be there.
2. so I followed your advice and removed the caps on the 49mhz clock. I believe the sound cleared up at both speeds. But now we are in the range of easily imagined improvement so I look forward to hearing it again and then trying the FCA cap.
@Soundcheck : it doesn't harm if you have a first reclocking, fifo and isolation before your femto-seconde dac-board. It has the advantage to reduce a first jitter before your femto embeded clock on the dac board you have and also provide a gnd isolation (then the signal clock showed to your femto-board is cleaner). In your case not sure a better crystal on the Fifo is usefull (but who knows).
I still like adaptators because it permitts me to tweak the sound I like both with crystal and decoupling on it. I also sure you understand another manufactering process as soldering NDKs on a further standalone adaptator is more expensive than buying the two cheap Crysteq provided with the FifoPi. Maybe also he uses a standalone order for the parts and the NDK-sa is not avaliable (I don't see it on the usual certified distributors we all use here.)
Maybe the only improvment possible in term of win/win is to avoid the dil12 adaptator in order to solder directly the smd SDK-A without the dil-12 adaptator so the manufacturing cost will be very close. But then still leaving the 12 vias for people for further dil-12 adaptator soldering (easier for them to desolder the NDK-SA than soldering it). Also direct crystal on the pcb is way better than adaptator with decoupling.
Maybe a next time, iancanada always improved his offers to the communauty and always with versatil choices for the diyers.
I still like adaptators because it permitts me to tweak the sound I like both with crystal and decoupling on it. I also sure you understand another manufactering process as soldering NDKs on a further standalone adaptator is more expensive than buying the two cheap Crysteq provided with the FifoPi. Maybe also he uses a standalone order for the parts and the NDK-sa is not avaliable (I don't see it on the usual certified distributors we all use here.)
Maybe the only improvment possible in term of win/win is to avoid the dil12 adaptator in order to solder directly the smd SDK-A without the dil-12 adaptator so the manufacturing cost will be very close. But then still leaving the 12 vias for people for further dil-12 adaptator soldering (easier for them to desolder the NDK-SA than soldering it). Also direct crystal on the pcb is way better than adaptator with decoupling.
Maybe a next time, iancanada always improved his offers to the communauty and always with versatil choices for the diyers.
IMHO that is the very definition of a diyaudio GB. A production run of 100 pieces at almost zero gross margin is not exactly a commercial business model. I think it is more about sharing with the side benefit of group testing and a small price break on volume.
NDK SDA clock caps
continuing to follow diyiggy recommendation to remove 3 ceramic caps from bottom of clock adapter PCB, I took the next step and added a 0.68uF FCA cap on the bottom of the board. I now have the .68 FCA on the solder pads on the bottom and a BG HiQ NX .68uf on the top. No question it brings out the full potential of the NDK SDA clock. I'll say more once I have listened for a few days, but initial impression is this is well worth the effort.
continuing to follow diyiggy recommendation to remove 3 ceramic caps from bottom of clock adapter PCB, I took the next step and added a 0.68uF FCA cap on the bottom of the board. I now have the .68 FCA on the solder pads on the bottom and a BG HiQ NX .68uf on the top. No question it brings out the full potential of the NDK SDA clock. I'll say more once I have listened for a few days, but initial impression is this is well worth the effort.
What I/V output stage are you guys using when doing your clock-cap-rolling experiments? Possibly you are ignoring the elephant in the living room just playing with the mouse...
I/V & output
Always a possibility. However the mouse is doing very well these days
First note I am not using battery, so my experience with clock caps likely more pronounced than those using Ian's battery PS. Still hoping someone confirms.
The chain is FIFOPi to I2SPCM to 1541a S2 on RyanJ pcb with slow DEM mod. I/V is 60r resistor (non inductive manganin). Output is 6n2p tube, tube rectifiers, full mono. I won't go into all the PS details but suffice to say separate linear supply for each consumer.. mostly clclc or more. Linestage is a slagle autoformer. amps are 6c33c OTL monos.
Always a possibility. However the mouse is doing very well these days
First note I am not using battery, so my experience with clock caps likely more pronounced than those using Ian's battery PS. Still hoping someone confirms.
The chain is FIFOPi to I2SPCM to 1541a S2 on RyanJ pcb with slow DEM mod. I/V is 60r resistor (non inductive manganin). Output is 6n2p tube, tube rectifiers, full mono. I won't go into all the PS details but suffice to say separate linear supply for each consumer.. mostly clclc or more. Linestage is a slagle autoformer. amps are 6c33c OTL monos.
I guess I do know a bit more about what's going on behind curtains in this context.
You do have a very narrow view on the subject.To get into successful business into a wider DIY market you need to do it the Allo way. All-In. That's what makes Allo successful and Allos customers happy - on price and performance level.
Let's see how things develop. Ian might offer different pre-made clocks to choose from. That'd be a no-brainer.
***
A technical FifoPi comment.
I tried to attach an Allo Boss to the FiFoPi. The Boss can also run in Slave mode. Unfortunately the FifoPi default clocks are too big and touching the Boss HAT. I can't get it pushed down.
Plus a pair of the Lundahl transformers, they are only £ 100 each. Crystek clocks £ 40 per pair and some NDK to compare (parts prices only).
Car manufactures should also supply vehicles with a spare set of winter tyres,after all they provide better performance in winter so surely everyone would want them, and be prepared to pay the extra £ 4/500.
Running a business is easy when it someone else's money you are spending.
If you think there is a big market for the clock boards why not get a few hundred made and run your own group buy ?.
GB for clocks
I am sure there might be some interest in a GB for NDK SDA clocks. They have a good price performance but are difficult to assemble by hand. Ian could test that idea in a future GB, or anyone else on the forum could do the same. Problem is, you need some volume to make the production run. And everyone wants a different frequency. I suspect it will be hard to get interest for more than 25 of any one part.. IMNV (in my narrow view)
I am sure there might be some interest in a GB for NDK SDA clocks. They have a good price performance but are difficult to assemble by hand. Ian could test that idea in a future GB, or anyone else on the forum could do the same. Problem is, you need some volume to make the production run. And everyone wants a different frequency. I suspect it will be hard to get interest for more than 25 of any one part.. IMNV (in my narrow view)
Agreed. I have soldered them many times and would not be a customer for a higher price preassembled module. I have used a soldering iron, a toaster oven, and a hot air gun. The latter was the easiest. pic shows result
Some may be interested in paying for this service, but even then I doubt there is enough to justify a group buy.
Some may be interested in paying for this service, but even then I doubt there is enough to justify a group buy.
Attachments
Taking it to the next "level"
Okay, time to go to the next "level". Pretty straightforward. Just need to triple-check all connections for shorts and polarity.
Wires are Duelund 20awg tinned-copper. Solid-core wire is probably easier to handle but I have the Duelund on-hand.
Need run-in before any observations can be made. Too many options too little time.
Okay, time to go to the next "level". Pretty straightforward. Just need to triple-check all connections for shorts and polarity.
Wires are Duelund 20awg tinned-copper. Solid-core wire is probably easier to handle but I have the Duelund on-hand.
Need run-in before any observations can be made. Too many options too little time.
Attachments
I guess I do know a bit more about what's going on behind curtains in this context.
To get into successful business into a wider DIY market you need to do it the Allo way. All-In. That's what makes Allo successful and Allos customers happy - on price and performance level.
Let's see how things develop. Ian might offer different pre-made clocks to choose from. That'd be a no-brainer.
***
A technical FifoPi comment.
I tried to attach an Allo Boss to the FiFoPi. The Boss can also run in Slave mode. Unfortunately the FifoPi default clocks are too big and touching the Boss HAT. I can't get it pushed down.
What??? To go to wider DIY market you need to do it Allo way? Man, I do not think you understand basic business priciples. Copuling your solution to just one platform and become depandant (RPI in Allo case) is the dumpest decsion you could ever make regardless of what kind of business you run.