[xiamen]

I do not wish to go far off topic. But I must say my system was fine. While I appreciate your advice, I have been doing this stuff for a very, very long time. There was no edge to take off. Modern tube designs do not, or should not, soften the sound. What I have found using this preamp and a couple of (not very expensive) NOS tubes is an easily noticed increase in the three dimensionality of instruments within an enlarged sound-stage. Also, more of a you-are-there/it-is-here sensation. I have a sense that I might also find this with a very expensive modern SS device, but what I have found with a relatively low expenditure was IMO worth reporting in context. On topic, it reveals that the LKS had more to give than I had supposed.

I don't have the Freya but I think they use 6SN7 tubes. My preamp section uses both 6SN7 and 6SL7. Have rolled many different NOS tubes over the years right up to expensive ones like the Russian 1578. Problem with buying NOS is fraught with fakes and NOS that are not really new. Right now I am using the Psvane CV-181 TII (I think they are already superseded by the Psvane 6SN7SE). These are new productions with decent quality control. You can really tell the difference with different tubes and the good ones take you to a whole new level.

 

[xiamen]

For those who are looking for interesting experiment to improve their sound system, I have been playing with Bug Head Emperor utilities from a Japanese developer Hiroyuki. First there is MinorityClean http://www.mics.ne.jp/~coolverse/MinorityClean/. You can run it as a service on your PC which supposedly "addresses noise generated by the CPU register being out of sync with the "internal circuit standard in memory". There is also Affeine Rewrite Data which I think does something like disk optimization and place your audio file in contiguous disk location to reduce output latency. Finally there is also an audio player Nontallion/Pink HQ. I think MinorityClean and Affeine does something good to my system but I haven't really had any serious listening comparisons. The audio player is interesting when compared to the likes of HQPlayer. It does take time to initialize and stabilize before playing anything. The UI is very primitive as well, may be suitable if you just want yo listen to one album at a time. The good news is it's all open source so you don't have to pay anything to try them out. If it doesn't do anything for you, just remove them.

 

[tdrood]

Hi

I didn´t read the spec sheet at detail, so i didn´t know about the diodes, so i didn´t installed it.

Anyway, i didn´t connect the NC pins to ground (Only one is connected but is more to secure the tcxo with four legs, one leg in which corner/pad).

I don´t have here with me any picture but it is a simple install only the wires soldered to the pads then the wires soldered to RHEA board pads.

Also it has been working fine for me like this for months, so i didn´t bother to perfect the TCXO instalation.My spaire time is limited....

Best

Tiago

D

Duh, I must have been tired when I wrote this post. Capacitors, not diodes!! Sometimes I worry that my brain is eroding...

 

[tdrood]

In case anyone is interested, here is what I have done...

Start off by saying - this is my 1st experience with SMD and it was really quite easy. Of course, I have not worked up the necessary testicular fortitude required to demount/liberate the existing XO from the main board yet, so my opinion may change before this is over...

I bought a pair of the Rhea boards from Twisted pair and mounted a 950X on one and a SiTime SIT5357 TCXO on the other. I used the bypass cap pads on the backside of the the Rhea boards to mount a 0.1mF Murata Low ESL 0306 cap as can be seen in the photos. I made a very high tech clamp to keep the cap from moving during the solder process. Yes, that is a twist tie... worked quite well. I used MG low temp solder paste.

I was able to mount the SiTime TCXO without the use of extra jumpers, but the pads on the XO do not overlap onto the pads on the Rhea board by very much (just a couple of mm at each corner), so I don't know how durable that installation will prove to be. I did use a little extra solder on those pads in the interest if improving the fillet at the connection points - probably a bad idea, definitely bad form - but hey, we're experimenting here, right? Not sure why those solder joints look so cold in the photos; they look pretty clean in real life. If necessary, I will install silver jumper wires, but I like do like the elegance of the jumperless installation and it seems like less is more from a noise standpoint. In any event, time will tell how well this configuration works out and I will post accordingly.

Like I said, I have not installed these in my machine yet - I probably will today and will post again once I have at least one of the clocks installed and burned (or burning) in.

I have a question for @b0bb - or anyone else who wants to chime in: well, actually, a statement and a question. I noticed after the fact that there are pads for a bypass cap on the upper surface of the Rhea board that also connects pin 2 to pin 4 of the adapter and is closer to the Vdd/Vcc pin of the clock than are the pads on the lower surface of the board. Should I move the 100nF cap to this location and install a 1mF cap on the bottom of the board? I bought the 1mF caps (which are 0508) in case the 0306s were too difficult to install, so I have both. Or, for that matter, should I use different values/types entirely? Thanks in advance!

T

 

[ti5002000]

Nice Job!

Unfortunatelly, i don´t have the technical skils to help with your question.

A question myself :

At which temperature did you use heat gun to solder the sitime?

In case anyone is interested, here is what I have done...

Start off by saying - this is my 1st experience with SMD and it was really quite easy. Of course, I have not worked up the necessary testicular fortitude required to demount/liberate the existing XO from the main board yet, so my opinion may change before this is over...

I bought a pair of the Rhea boards from Twisted pair and mounted a 950X on one and a SiTime SIT5357 TCXO on the other. I used the bypass cap pads on the backside of the the Rhea boards to mount a 0.1mF Murata Low ESL 0306 cap as can be seen in the photos. I made a very high tech clamp to keep the cap from moving during the solder process. Yes, that is a twist tie... worked quite well. I used MG low temp solder paste.



I was able to mount the SiTime TCXO without the use of extra jumpers, but the pads on the XO do not overlap onto the pads on the Rhea board by very much (just a couple of mm at each corner), so I don't know how durable that installation will prove to be. I did use a little extra solder on those pads in the interest if improving the fillet at the connection points - probably a bad idea, definitely bad form - but hey, we're experimenting here, right? Not sure why those solder joints look so cold in the photos; they look pretty clean in real life. If necessary, I will install silver jumper wires, but I like do like the elegance of the jumperless installation and it seems like less is more from a noise standpoint. In any event, time will tell how well this configuration works out and I will post accordingly.

Like I said, I have not installed these in my machine yet - I probably will today and will post again once I have at least one of the clocks installed and burned (or burning) in.

I have a question for @b0bb - or anyone else who wants to chime in: well, actually, a statement and a question. I noticed after the fact that there are pads for a bypass cap on the upper surface of the Rhea board that also connects pin 2 to pin 4 of the adapter and is closer to the Vdd/Vcc pin of the clock than are the pads on the lower surface of the board. Should I move the 100nF cap to this location and install a 1mF cap on the bottom of the board? I bought the 1mF caps (which are 0508) in case the 0306s were too difficult to install, so I have both. Or, for that matter, should I use different values/types entirely? Thanks in advance!

T

 

[tdrood]

Nice Job!

Unfortunatelly, i don´t have the technical skils to help with your question.

A question myself :

At which temperature did you use heat gun to solder the sitime?

Thank you! I actually used a hot plate/IR heat pad set to 250F to pre-heat the whole board and used a fine tip soldering iron to actually do the soldering. As good as my clamp is, I was worried about dislodging the device if I used hot air to solder. A rookie mistake, I am sure... But ignorance is bliss and it seemed to work pretty well.

 

[oldearwax]

In case anyone is interested, here is what I have done...

...

I have a question for @b0bb - or anyone else who wants to chime in: well, actually, a statement and a question. I noticed after the fact that there are pads for a bypass cap on the upper surface of the Rhea board that also connects pin 2 to pin 4 of the adapter and is closer to the Vdd/Vcc pin of the clock than are the pads on the lower surface of the board. Should I move the 100nF cap to this location and install a 1mF cap on the bottom of the board? I bought the 1mF caps (which are 0508) in case the 0306s were too difficult to install, so I have both. Or, for that matter, should I use different values/types entirely? Thanks in advance!

T

Good job. The SiTime App Note suggests 0.1uF (not 1mF) decoupling cap (Fig. 1). I think 0.1uF on the top is preferred.

 

[tdrood]

Good job. The SiTime App Note suggests 0.1uF (not 1mF) decoupling cap (Fig. 1). I think 0.1uF on the top is preferred.

Yes the datasheet does indicate 0.1mF and I would have put it on the top had I noticed the pads up there (or if I had read instructions that I am sure exist but I did not!). So I guess I will move one up there. The bigger question is for me is whether or not I should bother to add a 1mF on the bottom. Seems like std practice is to use multiple caps of ascending value as you move further away from the Vcc pin.

 

[tdrood]

Ok, old XO removed, socket installed and Si TCXO populated. I decided to try it with the 0.1mF cap on the bottom and once broken in, I will change it oout and see if there is a discernible difference. As of now, it is up and running and breaking in.

 

[b0bb]

In case anyone is interested, here is what I have done...



I have a question for @b0bb - or anyone else who wants to chime in: well, actually, a statement and a question. I noticed after the fact that there are pads for a bypass cap on the upper surface of the Rhea board that also connects pin 2 to pin 4 of the adapter and is closer to the Vdd/Vcc pin of the clock than are the pads on the lower surface of the board. Should I move the 100nF cap to this location and install a 1mF cap on the bottom of the board? I bought the 1mF caps (which are 0508) in case the 0306s were too difficult to install, so I have both. Or, for that matter, should I use different values/types entirely? Thanks in advance!

The cap values you using are the minimum recommended, looks OK to me.

I generally use the largest value I can find for the given SMD footprint and voltage rating.

You do however need to keep an eye on ESR, the measured value in your photo is about 10x too high. I suspect that your meter is not compensating properly for the DC resistance of the probes.
If you can change the measurement frequency use the highest possible setting

If the 2.6 ohm measurement is indeed correct, suggest you try another bypass cap

 

[tdrood]

The cap values you using are the minimum recommended, looks OK to me.

I generally use the largest value I can find for the given SMD footprint and voltage rating.

You do however need to keep an eye on ESR, the measured value in your photo is about 10x too high. I suspect that your meter is not compensating properly for the DC resistance of the probes.
If you can change the measurement frequency use the highest possible setting

If the 2.6 ohm measurement is indeed correct, suggest you try another bypass cap

Thank you very much b0bb. It is indeed an order of magnitude lower @ 100KHz (at lease according to spec - should be about 0.10-ohm at 100KHz, 0.01-ohm @ 1MHz - see data below). If I can change the frequency on my el-cheapo ESR meter, I don't know how... I selected these caps specifically because they are low ESL (hence the form factor), because as I understand it, inductance is (or can be) the enemy to a large degree, for whatever that is worth.

The other question remains: do you think it is worth the effort to move the 0.1mF cap to the upper pad location and install a 1mF cap on the lower surface - or any combination thereof?

I added the data for the 1mF cap below the 0.1mF data.

 

[b0bb]

Thank you very much b0bb. It is indeed an order of magnitude lower @ 100KHz (at lease according to spec - should be about 0.10-ohm at 100KHz, 0.01-ohm @ 1MHz - see data below). If I can change the frequency on my el-cheapo ESR meter, I don't know how... I selected these caps specifically because they are low ESL (hence the form factor), because as I understand it, inductance is (or can be) the enemy to a large degree, for whatever that is worth.

The other question remains: do you think it is worth the effort to move the 0.1mF cap to the upper pad location and install a 1mF cap on the lower surface - or any combination thereof?

I added the data for the 1mF cap below the 0.1mF data.

You have the minimum recommended capacitive bypass, I would leave it alone and move on.


If you really wanted to take this further, then take heed of what your instrument is trying to tell you.
2.6ohm ESR between the ground and power planes means that decoupling on the board is not effective, adding a bigger cap may not amount to much.

You could have a dry solder joint if you mixed leaded and unleaded solder or did not adequately heat the joint while using unleaded solder.

 

[oldearwax]

Nice Job!

...

At which temperature did you use heat gun to solder the sitime?

With shaky hands and aging eyes, I managed to mount the SiTime TCXO
on the Rhea board. The board's pads don't fit the small 5x3mm SiTime TCXO.
If enough people asks Brian, may be he will make another adapter board ?

I used MG Chemicals 4860P 63/37 lead paste. Melting point is 183C (~360F).
I used a thermal couple to calibrate the temperature of my hot air pencil.

A friend scoped the TCXO with a test board.
Ch1 is the clock, Ch2 is Vcc.

He said it is probably fine.

What do you guys think ?

 

[b0bb]

What do you guys think ?

Here is what my XO output looks like, limited by the 200Mhz Probe BW, which acts as a low pass filter, output looks like a near sinewave instead of a squarewave as a result.
Frequency is spot on 100Mhz


Noise waveform on the power supply pin, 200mV peak to peak at 100Mhz, could probably use additional decoupling

The output is quite distorted, try using the X10 probe setting and check the probe + scope capacitive loading is below 20pf

 

[b0bb]

A couple of pictures of the rectifier switching noise on the 004. Measured on the analog side of the DAC

The highspeed diodes used exhibit a relatively nasty transition characterisic.


A close up shows this causes ringing as it interacts with the transformer inductance


On the LKS003 I replaced these diodes with Vishay schottkys
Notice the spike is much less and the on-off transition is less abrupt.