[piaseczek]

Inductor 1: 0.381uH Inductor 2: 0.412uH
Inductor 3: 0.383uH Inudctor 4: 0.385uH

Thanks for measurements, did you measure also the inductor (5) for the clock?

 

[Monolithic]

Thanks for measurements, did you measure also the inductor (5) for the clock?

I have not measured #5 yet. I suspect it will be in the same range (370nH to 410nH or +/-5% of 390nH).

 

[b0bb]

Most inductors at 1A will be wirewound. Is there a particular core material I should be looking for as well (powdered, ferrite, etc)?

Edit - inductance/size/shielded/>1A are fairly limiting regardless of core material which will be most likely be ferrite. Not a common set of inductor parameters. Height may also be an issue due to the proximity of the capacitors. Still looking...

Closest match so far - Shielded, 10%, 390nH, @840ma, 1008 package, Height may be an issue API Delevan

Murata has an LQM series with 20% and 30% tolerances for shielded 470nH inductors, possible source for values under 400nH with 1.1/1.2/1.3/1.4A ratings in 0805 size

The 1.1A Murata LQM is fine for the task.

 

[piaseczek]

The 1.1A Murata LQM is fine for the task.

B0bb,
Did you change inductors in your DAC? Did it help with the big 820uF FPCAPs on the input side of the regulators? Thank you.

 

[ESL57]

I bought these, are shielded 470nH:
https://www.mouser.es/ProductDetail/Murata-Electronics/LQM21PNR47MG0D?qs=Gufeu08L/l1%2BjocillWQvQ==
I have not installed them yet and I cannot speak of their effect, although Xoverman did do interesting tests with inductors. By the way, what brand are the original inductors and what saturation current?

 

[b0bb]

B0bb,
Did you change inductors in your DAC? Did it help with the big 820uF FPCAPs on the input side of the regulators? Thank you.

Linking the 820uF and the inductor is incorrect if you are still using the stock LM317.
Inductor saturation is the bottleneck after the LM317 is replaced.

I switched over to the LT3083 before adding the 820uF so I never had the issue.
Modern regulators like the LT3083 do not have the LM317 ESR limitation.
(LM317 was designed 40+ years ago)


LM317s performance degrades if presented with ESR is that too low at its output, this was discussed in this thread about 18 months ago.
National Semiconductor published an app note on this.

The 820uF FP-RNL cap ESR is low enough to cause this behavior in the LM317
(@Xoverman got bitten twice by this problem, initially by a conventional 1000uF Low ESR electrolytic and then by the 820uF polymer)

Adding extra inductance in the 5V supply line increases the ESR and makes the issue less stressful for the LM317, that is the "improvement"
The cost is that the 9038s are starved for current, especially the AVDD supplies with needs a lot of current.
This will limit the performance of the DAC.

When switching to a modern regulator, all the changes to "improve" performance of the LM317 must be removed before doing a baseline to determine performance of the new regulator.

 

[piaseczek]

Linking the 820uF and the inductor is incorrect if you are still using the stock LM317.
Inductor saturation is the bottleneck after the the LM317 is replaced.

I switched over to the LT3083 before adding the 820uF so I never had the issue.
Modern regulators like the LT3083 do not have the LM317 ESR limitation.
(LM317 was designed 40+ years ago)


LM317s performance degrades if presented with ESR is that too low at its output, this was discussed in this thread about 18 months ago.
National Semiconductor published an app note on this.

The 820uF FP-RNL cap ESR is low enough to cause this behavior in the LM317
(@Xoverman got bitten twice by this problem, initially by a conventional 1000uF Low ESR electrolytic and then by the 820uF polymer)

Adding extra inductance in the 5V supply line increases the ESR and makes the issue less stressful for the LM317, that is the "improvement"
The cost is that the 9038s are starved for current, especially the AVDD supplies with needs a lot of current.
This will limit the performance of the DAC.

When switching to a modern regulator, all the changes to "improve" performance of the LM317 must be removed before doing a baseline to determine performance of the new regulator.

Bobb,
That is really a great explanation, big thanks for all your effords to share with us all that advanced knowledge. Really appreciated.

But I have one fundamental question: is it worth doing all that digital power supply mods with the crappy LP38798 in place? What do you think b0bb?

 

[Monolithic]

By the way, what brand are the original inductors and what saturation current?

Mine in the Gen2 had no identifying markings.

 

[b0bb]

I bought these, are shielded 470nH:
https://www.mouser.es/ProductDetail/Murata-Electronics/LQM21PNR47MG0D?qs=Gufeu08L/l1%2BjocillWQvQ==
I have not installed them yet and I cannot speak of their effect, although Xoverman did do interesting tests with inductors. By the way, what brand are the original inductors and what saturation current?

The DC resistance is quite high around 1ohm, it will get quite warm at full load of 0.6A --> 360mW
The one @Monolithic listed is 8x lower at 0.12 ohm

 

[Monolithic]

The DC resistance is quite high around 1ohm, it will get quite warm at full load of 0.6A --> 360mW
The one @Monolithic listed is 8x lower at 0.12 ohm

There are some (DFE Murata, Datasheet) that are less than 0.05 ohms but about 2-3 times the price over the 0.094-0.120 ohm versions. Is there a point of diminishing returns or should we go for the lowest DCR we can get?

 

[b0bb]

There are some (DFE Murata, Datasheet) that are less than 0.05 ohms but about 2-3 times the price over the 0.094-0.120 ohm versions. Is there a point of diminishing returns or should we go for the lowest DCR we can get?

Point of diminishing returns, 0.12 ohm@600mA is a 72mV voltage drop

 

[b0bb]

Bobb,
That is really a great explanation, big thanks for all your effords to share with us all that advanced knowledge. Really appreciated.

But I have one fundamental question: is it worth doing all that digital power supply mods with the crappy LP38798 in place? What do you think b0bb?

Yes, that is one of the main change LKS made when they rolled out the 3.2k bugfix version of the 004.
I would get rid of the LP38798, I am using the Bellesion SPLV78 outputting 1.3V

 

[b0bb]

Supercap buffer prototype, uses the LTC4425 linear supercap controller, no switched mode controller.
2x15F Eaton TV-series supercap

This setup contains a power sequencing relay, the load runs on normal regulator power while the supercap charges.
Once charging is complete relay cuts over, no long waits for the cap to charge.
Also means no funny power up delay issues.

This setup was created to understand how the LTC4425 functions with the aim of dropping this into the 004.

Currently on the Singxer SU-2 DDC as the power connections are easily accessible.

This board has additional mods
- NVE 711S GMR isolator
- TI TPS 7A20 low noise regulators
- Tantalum polymer supply bypass

Stock SU-2 performance is an embarrassment, barely better than the SU-1
The mods here transformed the box into a worthy SU-6 competitor.
With 705k/768k material its performance exceeds that of the SU-6

 

[piaseczek]

Yes, that is one of the main change LKS made when they rolled out the 3.2k bugfix version of the 004.
I would get rid of the LP38798, I am using the Bellesion SPLV78 outputting 1.3V

I know that replacing the LP38798 is the best option but I won't be able to do so without distroing the 004.
I will only conduct easier to do modifications.

 

[ESL57]

Supercap buffer prototype, uses the LTC4425 linear supercap controller, no switched mode controller.
2x15F Eaton TV-series supercap

This setup contains a power sequencing relay, the load runs on normal regulator power while the supercap charges.
Once charging is complete relay cuts over, no long waits for the cap to charge.
Also means no funny power up delay issues.

This setup was created to understand how the LTC4425 functions with the aim of dropping this into the 004.

Currently on the Singxer SU-2 DDC as the power connections are easily accessible.

This board has additional mods
- NVE 711S GMR isolator
- TI TPS 7A20 low noise regulators
- Tantalum polymer supply bypass

Stock SU-2 performance is an embarrassment, barely better than the SU-1
The mods here transformed the box into a worthy SU-6 competitor.
With 705k/768k material its performance exceeds that of the SU-6

B0bb you have a restless soul, you always seek to improve, thank you for sharing your new projects