Last edited:
1021 is slightly easier to build. 1121 should be even better in dual-mono with synchronized clocks!
Great question and always important to check. I don't have a laser thermometer so I can't say for sure. But in room temperature the chassis is only slightly warm; the transformer vent is also relatively cool. But the MOSFET area is warmer at maybe 40-50C. I ended up using silicone insulators instead of the mica provided by Teabag since it is advertised to conduct heat better and doesn't require using messy thermal compound (which probably would make the build even more difficult if not impossible). Maybe that's useful info. I'll open up the box in a few hours after things stabilize to finalize the VRR and I'll report in more detail on the temps then. Btw, it doesn't actually matter much for +/- rails to have the exact same voltage in dam1021 right?... Maybe it could help the onboard LDOs be more accurate?
Sounds like a good thermal situation already. Hotter spots near the Mosfets mounting area on the chassis would not make them break a sweat. It means they transfer their heat successfully nonetheless. When those spots are not spread away from the edges far underneath the board with high intensity its alright. We prefer a logical level of air ambient temperature inside a box for long life of all parts, mainly the electrolytics, is the goal in any build.
You remind me of the thermal situation on my laptop. I use the bottom aluminum chassis on Dell Precision 5520 to dissipate heat (as much as I can with stacked thermal pads though there's a mylar layer on the inside...). I override fan control to keep things silent under 50C (Dell fans make horrible noise and vibrations...). Anyways, a few hours of up time on my laptop and half the chassis would be hot to touch. I'm glad my DAC fares slightly better
You remind me of the thermal situation on my laptop. I use the bottom aluminum chassis on Dell Precision 5520 to dissipate heat (as much as I can with stacked thermal pads though there's a mylar layer on the inside...). I override fan control to keep things silent under 50C (Dell fans make horrible noise and vibrations...). Anyways, a few hours of up time on my laptop and half the chassis would be hot to touch. I'm glad my DAC fares slightly better
Be happy, you would be horrified at what CPU temperature levels I allow my NUC Skull Canyon to work for staying relatively silent
Its a mini furnace No surprise on diyAudio that people are allergic to noise! I'm very thankful that most of my heavy computation tasks I dump off to remote machines. It's hard to imagine living in proximity to any kind of hard-working computer... So I guess it doesn't matter much for me to get both rails to be near perfect 9V after stabilization... It's confusing what voltage I should feed the Soekris - I do use the op-amps to drive my headphones so it seems more voltage is better, yet more voltage is just more power dissipated in the LDOs and more heat for the DAC...
So I guess it doesn't matter much for me to get both rails to be near perfect 9V after stabilization... It's confusing what voltage I should feed the Soekris - I do use the op-amps to drive my headphones so it seems more voltage is better, yet more voltage is just more power dissipated in the LDOs and more heat for the DAC...
Most interesting is that nearly all builders find subjective differences on what pre-reg is used, and even have preferences if its shunt or other, no matter there are modern local low drop out regulator chips on the modules. Don't know what would have happened in double blind test but there seems to be an anecdotal consensus.
Sigma22 worked well for me previously, before the mosfets shorted due to my oversight in insulation, so I don't have too much to share. Even though this is not my field and I sometimes just prefer to go with popular opinion, i.e. UltraBiB , I think it would be super interesting to see more theory in audio. I believe that subjective impressions can reflect subtle underlying differences in the equipment sometimes, and though it may be hard to measure all aspects involved, it would be really cool if we manage to. As it is I'm very happy with the UltraBiB - now I just need more listening sessions to tune my ears to the recordings
, I think it would be super interesting to see more theory in audio. I believe that subjective impressions can reflect subtle underlying differences in the equipment sometimes, and though it may be hard to measure all aspects involved, it would be really cool if we manage to. As it is I'm very happy with the UltraBiB - now I just need more listening sessions to tune my ears to the recordings Temps are actually lower than expected according to my thermometer. It only measures up to 42C but shows that the hottest parts of the chassis are around 40C. C1 feels hotter, maybe around 45C; C2 C3 are cool. The MOSFETs are not hot to touch so maybe the heat sinks are working well. R1 is much hotter. The voltage dropped almost 0.3V to 8.72V when I opened it and I adjusted both rails back to 9.00V. VRR on the +/- rails are 1932Ohms and 1945Ohms respectively in case it's helpful. I suspect that with a recalibrated voltage setting, the temps will stabilize to a bit higher (and the voltage will probably drop a little bit again). Will check again later
Edit: Just realized I've been measuring across pin 1 & 3 of the trimmer recommended in the group buy, but the wiper should be 2 according to the datasheet... Strange that the resistance measured this way does respond very sensitively to adjustments and definitely fooled me. Will double check it later
Edit: Just realized I've been measuring across pin 1 & 3 of the trimmer recommended in the group buy, but the wiper should be 2 according to the datasheet... Strange that the resistance measured this way does respond very sensitively to adjustments and definitely fooled me. Will double check it later
Last edited:
Those sound like very safe temps in most places. What Ohmic value and nominal power R1s you installed? The module's LDOs should feel easier with less Vin (if there's low Vout space) so the voltage drop across them goes less and their dissipation drops accordingly. That depends on why +/- 9V pre-reg setting is optimum or there are other comparable lower rail options and what they weigh against. You Soekris aficionados surely know about.
Negative Rail (1945Ohm VRR): 2.4Ohm 1W 50ppm: RS01A2R400FE12 Vishay / Dale | Mouser
Positive Rail (1932Ohm VRR): 1.2Ohm 3W 50ppm: RS02B1R200FE12 Vishay / Dale | Mouser
I just realized that measuring VRR across 1/3 shouldn't be a problem as 2 should be shorted with 1 or 3 anyways on the circuit?... Recollection of middle school physics seems to suggest that...
9V has been brought up a few times and some people reported better results by going down to 9V from 12V. The DAC should work well either way, just different temps on the LDOs. Low voltage could be a problem for the opamps (OPA1602) though? Datasheet says +/- 20V input and someone on the Soekris thread said +/- 15V (maybe that's for the older ver with different opamps). In any case, lower DC input will result in lower opamp supply voltage, whether or not it matters or to what extent is a different question. Btw, it's probably best to remove all LDO and feed in 4V at the shift register with Lifepo4 batteries directly (or something radical like that...), but that seems too complex for my good...
The output voltage has actually stabilized to 9.05V. Maybe because I left it in 19C room temperature... Turning up the heat lowered it to 9.03V but I'm fine either way. Probably will just leave VRR as is. Component temperature didn't increase much as far as I can tell, probably even cooler now that I opened the top cover for 10 minutes. Can barely feel the warmth on the component and chassis now. Very happy with the results so far!
Positive Rail (1932Ohm VRR): 1.2Ohm 3W 50ppm: RS02B1R200FE12 Vishay / Dale | Mouser
I just realized that measuring VRR across 1/3 shouldn't be a problem as 2 should be shorted with 1 or 3 anyways on the circuit?... Recollection of middle school physics seems to suggest that...
9V has been brought up a few times and some people reported better results by going down to 9V from 12V. The DAC should work well either way, just different temps on the LDOs. Low voltage could be a problem for the opamps (OPA1602) though? Datasheet says +/- 20V input and someone on the Soekris thread said +/- 15V (maybe that's for the older ver with different opamps). In any case, lower DC input will result in lower opamp supply voltage, whether or not it matters or to what extent is a different question. Btw, it's probably best to remove all LDO and feed in 4V at the shift register with Lifepo4 batteries directly (or something radical like that...), but that seems too complex for my good...
The output voltage has actually stabilized to 9.05V. Maybe because I left it in 19C room temperature... Turning up the heat lowered it to 9.03V but I'm fine either way. Probably will just leave VRR as is. Component temperature didn't increase much as far as I can tell, probably even cooler now that I opened the top cover for 10 minutes. Can barely feel the warmth on the component and chassis now. Very happy with the results so far!
In the Soekris dam1021 user's manual I can see
"Power Consumption
Positive Rail: 0.18A @ 10V
•
Negative Rail: 0.06A @ 10V
•
Total: 2.4W
•
The positive supply draw about 3 times as much current as the negative; the current is
almost independent of input voltage."
So they can alternatively be 2.0 Ohm in the Positive rail for 0.29A CC and 3.3 Ohm in the Negative rail for 0.175A CC (when assuming 0.58V average drop across R1s) and still have about 110mA spare at max consumption for both rails. 100mA to 150mA spare current is a good enough range for the UltraBiBs performance. That would bring the R1s and Mosfets dissipation even lower.
"Power Consumption
Positive Rail: 0.18A @ 10V
•
Negative Rail: 0.06A @ 10V
•
Total: 2.4W
•
The positive supply draw about 3 times as much current as the negative; the current is
almost independent of input voltage."
So they can alternatively be 2.0 Ohm in the Positive rail for 0.29A CC and 3.3 Ohm in the Negative rail for 0.175A CC (when assuming 0.58V average drop across R1s) and still have about 110mA spare at max consumption for both rails. 100mA to 150mA spare current is a good enough range for the UltraBiBs performance. That would bring the R1s and Mosfets dissipation even lower.
I think we were trying to account for an additional 50% peak power draw but it might've been unnecessary. I'm just super glad that the 25W toroidal seems to work. Fits much better in the chassis than a 35W.
Do you know what the expected temp drift is for UltraBiB? Looking back it seems there's a non-negligible 3% drift from cold to stable and a 0.2% across a few degrees centigrade. I used recommended components in the group buy. Thanks!
Do you know what the expected temp drift is for UltraBiB? Looking back it seems there's a non-negligible 3% drift from cold to stable and a 0.2% across a few degrees centigrade. I used recommended components in the group buy. Thanks!
Someone's got to measure the dam in full tilt to see if there is additional peak draw chance but according to what stated in its manual 2R in the positive and 3.3R in the negative should cover it with ample spare current for the UBiB's own use. Since with your R1 selections works fine and the Mosfets heat is still in check there should be no issue.
Drift until thermal settling is inherent in UBiB because of the Vbe junction's TC of Q2 (going down 2mV/deg C). That drift translates to current drift via R5 (ICCS=VbeQ2/R5). The trimmer multiplies that to voltage reference drift. Because Vref=Vbe(Q2+Q3)+VVR1+VFLED1+VFLED2 you can calculate the expected drift for any temperature delta.
If you don't want it you can use a Zener instead of RR and no trimmer as described in the guide. But a Zener is generally much noisier and harder to filter. When having post-regs there's no point to be bothered with the pre-reg's settling drift not even academically. Their job is to keep a steady Vout for any voltage input above it plus their drop-out spec. In modern LDOs that spec is especially low, below 1V or much less.
Drift until thermal settling is inherent in UBiB because of the Vbe junction's TC of Q2 (going down 2mV/deg C). That drift translates to current drift via R5 (ICCS=VbeQ2/R5). The trimmer multiplies that to voltage reference drift. Because Vref=Vbe(Q2+Q3)+VVR1+VFLED1+VFLED2 you can calculate the expected drift for any temperature delta.
If you don't want it you can use a Zener instead of RR and no trimmer as described in the guide. But a Zener is generally much noisier and harder to filter. When having post-regs there's no point to be bothered with the pre-reg's settling drift not even academically. Their job is to keep a steady Vout for any voltage input above it plus their drop-out spec. In modern LDOs that spec is especially low, below 1V or much less.