Just to extend the conversation, the “right” choke has almost nothing to do with quality; if you treat each rail like a bipolar supply and rigorously enforce current draw balance in each side of the balanced choke, you can basically take saturation out equation, get rid of the air gap (saving a lot of bulk in the core and diameter of copper) and the magnetic fields largely cancel as well so you avoid magnetic coupling into down steam circuits.
Seems like you know your way around tubes, it’s essentially similar to using a SE vs class A push-pull transformer; the push pull is much smaller, and simpler, because the choke only has to work at one frequency. In extremis all that matters is that the L(crit@i) - or vice-versa - relationship is satisfied to keep the diodes conducting.
Seems like you know your way around tubes, it’s essentially similar to using a SE vs class A push-pull transformer; the push pull is much smaller, and simpler, because the choke only has to work at one frequency. In extremis all that matters is that the L(crit@i) - or vice-versa - relationship is satisfied to keep the diodes conducting.
I used to try the quasimodo snubbing for HV(tubes)...I learned a couple of things there...how sensitive windings are seeing each other...even taps see each other...bottom line was that ideally you have one primary, one secondary, one core...which is a bit unpractical, but otherwise resonances from other windings /diodes noise transfer can happen...But in the end i did not like the sound of snubbers. I prefer Mercury-Rectifiers for endtriides and Mesh-Tube recitifiers for driver/signal. All DIrectly Heated, which brought the need to find good sounding low voltage PSU as you heat with DC the cathode, so very audible what you do there. The lt4320 was eyeopening as the Nichcon superthrough caps with some nano chokes from Alexander. And what is good for a DHT cathode is good here as well it seems...
Fifos I am using since Ian's first version...even with Pulsar clocks (still have a pair i believe) etc..so the Bufallos all were on Fifo already. Later Soekris dam1941 has a fifo on board and Ians RasperryPififo as well...its a long time ago that i listened to a dac without fifo...
Nevertheless with all the fifo expierence: A very good source / streamer sounds better feeding a fifo than a normal one. The fifo does not rule that out...so I can see that a SD card player with an external clock like shown earlier might be the ultimate source....but uffff...i like to have some convenience as well...playing at least from my NAS.
Your idea on the choke, i will discuss with Alexander...we love to try new ideas and he is an artist in winding, even his power transformer sound different than the usual stuff...see dhtrob and ale bartolas reports (we all share the same observatiins).
Fifos I am using since Ian's first version...even with Pulsar clocks (still have a pair i believe) etc..so the Bufallos all were on Fifo already. Later Soekris dam1941 has a fifo on board and Ians RasperryPififo as well...its a long time ago that i listened to a dac without fifo...
Nevertheless with all the fifo expierence: A very good source / streamer sounds better feeding a fifo than a normal one. The fifo does not rule that out...so I can see that a SD card player with an external clock like shown earlier might be the ultimate source....but uffff...i like to have some convenience as well...playing at least from my NAS.
Your idea on the choke, i will discuss with Alexander...we love to try new ideas and he is an artist in winding, even his power transformer sound different than the usual stuff...see dhtrob and ale bartolas reports (we all share the same observatiins).
Thanks for the insights! Dios mio, i thought I was nearly done, and then I started asking questions…
I’m very slowly making my may up the mountain behind you, thanks again for posting this.
It seems to be the case that the Pulsar clocks are entirely bettered by Andrea’s clocking solution, fair to say?
I haven’t listened to the Pulsars yet, if you are interested in repurposing the funds invested in them for new projects please let me know (pm me!) as I don’t have the space to integrate the full Mori clock. I’ve contented myself with the Crystek, NDK and Si570(!) clocks but I understand the Pulsars/Mori clock are in a different league, one day I’d like to see if my efforts have the discrimination to reveal the virtues of a better clock…
I’m very slowly making my may up the mountain behind you, thanks again for posting this.
It seems to be the case that the Pulsar clocks are entirely bettered by Andrea’s clocking solution, fair to say?
I haven’t listened to the Pulsars yet, if you are interested in repurposing the funds invested in them for new projects please let me know (pm me!) as I don’t have the space to integrate the full Mori clock. I’ve contented myself with the Crystek, NDK and Si570(!) clocks but I understand the Pulsars/Mori clock are in a different league, one day I’d like to see if my efforts have the discrimination to reveal the virtues of a better clock…
I guess this could be quiet interesting:
https://www.analog.com/media/en/technical-documentation/user-guides/DC2432B.pdf
https://www.analog.com/media/en/technical-documentation/user-guides/DC2432B.pdf
Sure is, I missed that one. Both quiet and interesting. I use the LTC6655, the LTC6658's single output predecessor as a voltage reference. Apparently the 6655 doesn't stay precise enough over time for the nuts and volts guys, (and bless them, my HP3456A has been spot on for longer than I've been alive) but it is a seriously low noise reference and we don't care so much about absolute voltage but rather 1/f low noise, no? My clock reads the same on the counter whether fed 3.6V or 3.0V, the working range of a lot of batteries. The LT6655 got the Jim Williams treatment, the app note on how he measured 775nv noise is an incredible read and a mountain to climb, in an of itself.
https://www.arrow.com/en/reference-...ge-reference/b8ea4e9471b25002c6f4a64dea508a67
https://www.manualslib.com/manual/1085810/Linear-An124-775.html?page=4
The LTC6655 and a LT3042 yields the lowest 1/f noise PSU I've ever seen, for feeding a clock. It's a lot of IC in some tiny little packages, though, conceptually a far cry of a hefty choke or two. I feed mine from 15 year old, 2000ma, 1.2V, NiCd cells (how quaint!).
https://www.arrow.com/en/reference-...ge-reference/b8ea4e9471b25002c6f4a64dea508a67
https://www.manualslib.com/manual/1085810/Linear-An124-775.html?page=4
The LTC6655 and a LT3042 yields the lowest 1/f noise PSU I've ever seen, for feeding a clock. It's a lot of IC in some tiny little packages, though, conceptually a far cry of a hefty choke or two. I feed mine from 15 year old, 2000ma, 1.2V, NiCd cells (how quaint!).
Oh, and if you want to go purist, all you need is a 3.6V NiCd charger; 3x 2000mah 1.2V cells charged to 3.6V (not standard) downrated to 1500mah, with a 30ma draw nets you 50 hours of runtime with, with an eight hour cell optimal recharge. Two strings gives constant operation, all about the size of a small choke.
https://tf.nist.gov/general/pdf/1133.pdf
I wouldn't use this for a full range audio-spectrum supply, but it works a treat for clocks. Zero junctions/actives in the supply path.
https://tf.nist.gov/general/pdf/1133.pdf
I wouldn't use this for a full range audio-spectrum supply, but it works a treat for clocks. Zero junctions/actives in the supply path.
No kidding? I thought the ultraBIB bottomed out at 5V, with a 20-something volt input. I still use Salas’ “Reflektor” core for a 150ma 3.3V super-shunt from a ~5V supply…
Could someone do me a favor and explain how Andrea’s FIFO is different from previous FIFOs? What distinguishes it? I admit I haven’t done the reading yet…
Could someone do me a favor and explain how Andrea’s FIFO is different from previous FIFOs? What distinguishes it? I admit I haven’t done the reading yet…
Attachments
@lasercut, regarding the Reflektor, Salas said:
"In some apps low Zo its not the priority. Only wide flat Zo and low 1/F."
https://www.diyaudio.com/community/threads/reflektor-d-builds.261031/post-4048966
"Better use them for general PSU at 5V and tap off to some super low noise LDO chip dedicated next to the osc."
https://www.diyaudio.com/community/threads/reflektor-d-builds.261031/post-4049015
Can't believe that was almost 10 years ago...
"In some apps low Zo its not the priority. Only wide flat Zo and low 1/F."
https://www.diyaudio.com/community/threads/reflektor-d-builds.261031/post-4048966
"Better use them for general PSU at 5V and tap off to some super low noise LDO chip dedicated next to the osc."
https://www.diyaudio.com/community/threads/reflektor-d-builds.261031/post-4049015
Can't believe that was almost 10 years ago...
Well, I use them now at 4.8V, but just got the replacement parts to go lower to 3.3V with Ultrabib...but I want to settle in the burn-in process first before changing transistors and than start all over with burn-in.
I am soldering the battery supply while we talk, trying to get the parts together in parallel...
Has anyone the battery supply running ? What I have understood it is basically designed to power everything with battery...clocks...fifo...dac...which is nice when listening.
...maybe Andrea can enlighten us ?
Here are my questions:
A. From https://www.thewellaudio.com/wp-content/uploads/TWRPS-LBS-P_User_Manual.pdf
"- linear supply to power the TWSAFB-LT FIFO Lite board
- regulated supply to power the oscillators while recharging the batteries, to never shut down the
oscillators"
So, I do understand that the FIFO is powered by a linear regulator ...BUT: "5: Auxiliary 9 VDC output to power digital devices. Connect to the TWSAFB-LT FIFO buffer." So now its 9VDC instead of 7 VDC (see FIFO manual)???
B. From The https://www.thewellaudio.com/wp-content/uploads/TWRPS-LBS-M_User_Manual.pdf
"
J4: 3V3, 6V6, 9V9, 13V2 or 16V5 output rail (depending on the installed batteries). It can be used
to power the oscillators and the frequency doublers.
J5: same output rail as J4 with 13-13.5V backup from linear regulator to power the oscillators
during batteries recharge. It can be used instead of J4 to the oscillators always on."
...just to get this straight: actually J5 is what to use, so during listening we are on batteries and during charging we are on regulator, correct ? So Regulator is NOT used on J5 when listening ?
and...
C. this means the DAC is actually most of the time not powered when using the battery supply ? There is unfortunately no J4/J5 for the daughter board...we ll need to see if this is a disadvantage...no ? Normally I let my Soekris /Ultrabib just powered on 24h/365...
Last edited:
D
Deleted member 537459
J5 of TWRPS-LBS-P provides 7 to 9VDC (it depends on the load since it's LC filtered) to power the FIFO Lite.
This is the dirty part of the FIFO (FPGA and micro, J4 or J20) which does not need to be ultra low noise, since the clock part of the FIFO (master clock and dividers) is powered separately (J18 on the FIFO).
J5 never switches to batteries.
J4 of the TWRPS-LBS-M provides batteries supply to power DRIXO oscillators and DBMs, with linear regulator backup when recharging the batteries.
So the only critical power supply of the FIFO Lite is J18 (masterc clock, sine to square converter and dividers).
The most critical power supply of the DAC Lite is Vref (J12 and J13), we have tested it using LTO batteries but we will try shunt regulators soon.
Then it needs VDD supplies (J5 and J6).
If the above power supplies are always on the DAC never switches off.
Last edited by a moderator:
Hi Andrea,
So, I understand:
- Vinp of the Fifo can be between 7VDC to 9VDC, so 9VDC is not a problem for the FiFo.
- J5 vs. J4 on the battery (J5 of LBS-M-Board, not J5 of the LBS-P-Board): J4 for full battery, J5 for battery/reg when charging ? That is what was stated in the manual...here you wrote now J4 for battery/reg...? Confuses me.
- Well, sure if I continue to use my Ultrabibs for powering the DAC, the DAC never switches off...but that is not the point when using the battery supply instead of the Ultrabib for the DAC I am afraid...so in summary: if you want the DAC powered all time, dont use the battery supply...hmmm...surprising as for the oscillators you came up with a smart switching solution to power them all the time...but maybe this is only a theoretical disadvantage...
So, I understand:
- Vinp of the Fifo can be between 7VDC to 9VDC, so 9VDC is not a problem for the FiFo.
- J5 vs. J4 on the battery (J5 of LBS-M-Board, not J5 of the LBS-P-Board): J4 for full battery, J5 for battery/reg when charging ? That is what was stated in the manual...here you wrote now J4 for battery/reg...? Confuses me.
- Well, sure if I continue to use my Ultrabibs for powering the DAC, the DAC never switches off...but that is not the point when using the battery supply instead of the Ultrabib for the DAC I am afraid...so in summary: if you want the DAC powered all time, dont use the battery supply...hmmm...surprising as for the oscillators you came up with a smart switching solution to power them all the time...but maybe this is only a theoretical disadvantage...
Ok...i am building that one as well and will try...but dont you need two of them for two clocks ?
Two drixo or two exo ...? So, 2*60mA...i just remembered that the reg was originally speced for one clock...but should work I guess
Just curious guys, I have some sine to square converters, TWTMC-STS-DX and TWTMC-STS-SX. Both appear to be identical except for the name. Component values appear to be the same, layout is the same, etc. Recently found one of them oscillating is the absence of any input signal. Tried another one and found the same thing. Oscillation seems to vary with power supply voltage, if I put my thumb on the board in a certain place, power bypass caps, etc. Some oscillation noise can be seen on Vcc, although the oscillation is much higher level at the square wave output terminal. Putting the squarer in a steel box to attenuate any environmental EMI/RFI produced no change in oscillation behavior. Some pics of some of the observed waveforms attached below.
My question for anyone is whether anyone else has seen this issue, or is it unique to something about testing at my location or my test gear?
My question for anyone is whether anyone else has seen this issue, or is it unique to something about testing at my location or my test gear?
Already talked to Andrea. He said they did not see any oscillations with the boards when they were developing them. He also said the chips they used for that are no longer available. The reason they used the particular chip was because its phase noise was 20% better than any of the other brands they tried.