My (very) mini review of the Singxer SU-1 DDC

[scan80269]

3 hours ago, Duy Le said:

Is it the original cap of SU-1 ?

The original cap is a Panasonic FC 220uF electrolytic.  I replaced it with Panasonic FR 100uF 50V, following Sparkos Labs recommendation on their website for output capacitor type.

 

[Duy Le]

9 hours ago, michaeltay5871 said:

- (2) Panasonic FR EEU-FR1E331 330μF 25V - 56mΩ

Should we use Panasonic EEU-FR1V221 220 uF 35 VDC 56 mOhms  to replace or not ?

[Duy Le]

On 5/18/2017 at 9:51 PM, Energy said:

You're welcome @tboooe. The expanded sound stage and deafening quietness sure helps to make each produced sound that much more intimate and inviting for the listener.

 

By the way, I had some time to kill today and read up on the Linear Technologies LT3045 LDO Regulator and have some things to say that will clarify on how it performs opposed to the Sparkos Labs SS1117.

 

OUTPUT NOISE: (the lower the better)

At 3.3V the LT3045 is specified to 2.8μV (output noise) whereas the SS1117 is 3.2μV.

The LT3045 is only specified/advertised at 0.8μV if you are using voltages 1.3V or below.

 

RIPPLE REJECTION: (the higher the better)

At 120Hz the LT3045 is specified 117dB (input rejection) whereas the SS1117 is 124dB.

At 10KHz the LT3045 is specified 90dB (input rejection) whereas the SS1117 is 108dB.

 

Having said that, If you currently have the SS1117 Installed, It wouldn't make a lot of sense to replace it with the LT3045. Even though it gains a hair in output noise, it loses in input rejection. If one were to start the mod from scratch, I would recommend the Linear Technologies LT3045 for $22.95 as it's half the price and as better load regulation as well as spot noise. But just to say, I doubt you will be able to hear a difference between the two..

 

Cheers,

Timothy

Which one is better ?

[Energy]

Both should sound similar. You can't go wrong with either.

[Duy Le]

12 hours ago, Energy said:

Both should sound similar. You can't go wrong with either.

How do you think about replacing Panasonic FR EEU-FR1E331 330μF 25V - 56mΩ by Panasonic EEU-FR1V221 220 uF 35 V 56mΩ or Panasonic FR 100uF 50V following Sparkos recommendation ?

[scan80269]

19 minutes ago, Duy Le said:

How do you think about replacing Panasonic FR EEU-FR1E331 330μF 25V - 56mΩ by Panasonic EEU-FR1V221 220 uF 35 V 56mΩ or Panasonic FR 100uF 50V following Sparkos recommendation ?

I followed Sparkos recommendation and installed a Panasonic FR 100uF 50V cap at the location closest to the regulator output.  For a series pass regulator like the SS1117, output capacitance is not "the more the better" and output cap ESR is not "the lower the better".  The Panasonic FR 100uF 50V cap has capacitance and ESR to help optimize regulator transient response.  The Panasonic FR 330uF and 220uF caps have similar ESR as the 100uF, but the larger capacitance will slow down the transient response of the regulator.  Polymer (or ceramic or film) caps with very low ESR at the regulator output can degrade the phase margin of the regulator and potentially impact stability, so I use these only on the regulator input side.  I also removed 4 surface-mount 10uF MLCC caps sitting near the regulator to avoid degrading regulator phase margin.  Those MLCC caps may be OK with the stock LD1086 regulator, but they are not optimal for the SS1117.

 

[Duy Le]

26 minutes ago, scan80269 said:

I followed Sparkos recommendation and installed a Panasonic FR 100uF 50V cap at the location closest to the regulator.  For a series pass regulator like the SS1117, output capacitance is not "the more the better" and output cap ESR is not "the lower the better".  The Panasonic FR 100uF 50V cap has capacitance and ESR to help optimize regulator transient response.  The Panasonic FR 330uF and 220uF caps have similar ESR as the 100uF, but the larger capacitance will slow down the transient response of the regulator.  Polymer (or ceramic or film) caps with very low ESR at the regulator output can degrade the phase margin of the regulator and potentially impact stability, so I use these only on the regulator input side.  I also removed 4 surface-mount 10uF MLCC caps sitting near the regulator to avoid degrading regulator phase margin.  Those MLCC caps may be OK with the stock LD1086 regulator, but they are not optimal for the SS1117.

 

Could you please mark them (MLCC caps) in the SU-1 board photo because I don't know where they are placed ? If you remove them from the circuit path, how to keep electric transport ? Thanks.

And what caps do you use for the regulator input and nearest the FPGA chip ? And why do you do that ?

p/s: sorry if I ask too much but I don't know anything about digital electric

[scan80269]

11 minutes ago, Duy Le said:

Could you please mark them (MLCC caps) in the SU-1 board photo because I don't know where they are placed ? Thanks.

And what caps do you use for the regulator input and nearest the FCGA chip ? And why do you do that ?

I used a Nichicon FPCAP (R7) 220uF 6.3V polymer cap at the regulator input.  Low ESR cap is OK to use at regulator input side.  Near the Xilinx FPGA, I installed a Panasonic FR 330uF 25V cap.  It serves as bulk capacitance for the FPGA chip, and is located further away from the regulator output thus reduced risk of degrading regulator phase margin.

 

[Duy Le]

I also read that it's necessary to isolate the regulator in power board but I cannot image how to do that. Could someone show me if you already did it or know how to do?

[Duy Le]

@scan80269 Do you have any idea for caps on power board ?The small cap is output cap, isn't it? Should it be the Panasonic FR 100uF 50V too if I replace the stock LM2940CT-5 by Sparkos SS7805 ?

[Duy Le]

May I use Elna caps (Cerafine or Silmic II) instead of Panasonic and Nichicon? They are very easy to buy in Vietnam.

p/s: sorry because I cannot edit my previous post to merge them into one.

[bluesaint]

15 minutes ago, Duy Le said:

May I use Elna caps (Cerafine or Silmic II) instead of Panasonic and Nichicon? They are very easy to buy in Vietnam.

p/s: sorry because I cannot edit my previous post to merge them into one.

 

Don't because those were made for analog audiophile grade caps, not digital circuitry.

[Energy]

On 6/6/2017 at 3:51 AM, scan80269 said:

I used a Nichicon FPCAP (R7) 220uF 6.3V polymer cap at the regulator input.  Low ESR cap is OK to use at regulator input side.  Near the Xilinx FPGA, I installed a Panasonic FR 330uF 25V cap.  It serves as bulk capacitance for the FPGA chip, and is located further away from the regulator output thus reduced risk of degrading regulator phase margin.

 

 

Last time I checked, the Panasonic FR was recommended as capacitors for both input and output near the SS1117. Personally I would have gone with 220µF in either 25V/35V but Digi-Key only offered 330µF. I'll probably replace them in the future with 220µF but just thinking about opening the unit all over again feels so tiresome.

 

Also, shouldn't you add lead wires to the those MLCC pads to short them? That or an 0Ω SMD resistor. Your picture shows them unconnected so maybe this is a before photo?

 

I am curious as to why the MLCC SMD capacitors would degrade phase margin.. 

If they truly do, I will be really inclined to remove them from my own unit.

 

On 6/6/2017 at 4:04 AM, Duy Le said:

I also read that it's necessary to isolate the regulator in power board but I cannot image how to do that. Could someone show me if you already did it or know how to do?

 

Just attach a TO-220 mounting kit's thermal pad in between the discrete voltage regulator and the heatsink and you should be good. That or use MICA film with non-conductive thermal paste instead of a thermal pad which does the same job to isolate the two surfaces.

 

22 hours ago, Duy Le said:

@scan80269 Do you have any idea for caps on power board ?The small cap is output cap, isn't it? Should it be the Panasonic FR 100uF 50V too if I replace the stock LM2940CT-5 by Sparkos SS7805 ?

 

There are no 100µF capacitors on the power supply board only 470µF and 1000µF.

Even if you replace the LM2940CT-5 with the SS7805 you still keep the same capacitor values and rating everywhere else. I would keep the same capacitance level just to be safe.

 

Elna Silmic II is my personal favorite capacitor but only in analog circuits such as amplifiers where material science matters. As for digital ones, Panasonic FR and Nichicon R7 is my preferred choice.

 

It's like bikes. You don't bring a track bike to go mountain climbing. Some things just work better elsewhere.

[Duy Le]

25 minutes ago, Energy said:

Last time I checked, the Panasonic FR was recommended as capacitors for both input and output near the SS1117. Personally I would have gone with 220µF in either 25V/35V but Digi-Key only offered 330µF. I'll probably replace them in the future with 220µF but just thinking about opening the unit all over again feels so tiresome.

Why don't you go with Panasonic FR 100µF 50V as Sparkos recommendation for output cap?

[Energy]

1 minute ago, Duy Le said:

Why don't you go with Panasonic 100µF 50V as Sparkos recommendation ?

 

The stock capacitors are 220µF so who knows how much the nearby components needs to draw in terms of capacitance. I wouldn't go lower as it might not be sufficient but who really knows. I'm playing it safe.

 

I only got 330µF because DiGi-Key only had that value otherwise I would have gotten 220µF in 25V or 35V. Either way, they all have the same or similar ESR so it's really not a big deal.

[scan80269]

1 hour ago, Energy said:

Also, shouldn't you add lead wires to the those MLCC pads to short them? That or an 0Ω SMD resistor. Your picture shows them unconnected so maybe this is a before photo?

 

I am curious as to why the MLCC SMD capacitors would degrade phase margin.. 

If they truly do, I will be really inclined to remove them from my own unit.

 

Those four MLCC caps are in parallel with the electrolytic cap sitting at the regulator output rail.  Shorting the pads or installing a 0Ω SMD resistor will short the regulator output to ground!  I'm very sure you shouldn't do that.

 

I measured approx. 90Ω of ESR from each of those 10µF MLCC caps.  Four of them in parallel yields a group ESR of about 22Ω which is low enough to risk degrading regulator phase margin.  Per Andrew Sparks, it is beneficial for the output cap to have some amount of ESR to place a zero in the open loop response of the regulator to help optimize its transient response and phase margin.  Low-ESR (<20mΩ) ceramic or film or polymer caps connected in close proximity to the regulator output are therefore not recommended.  Based on these recommendations, I changed the electrolytic at the regulator output to a Panasonic FR 100µF 50V, and removed the four MLCC caps.

 

The above is also why I previous said for a series pass regulator like the SS1117, the output cap value is not "the higher the better" and the cap ESR is not "the lower the better".

 

Low-ESR caps are good to use at the regulator input.  Thus, I use 3 low-ESR polymer caps and 2 Panasonic FR caps to mod my SU-1.

 

[Duy Le]

9 minutes ago, scan80269 said:

Those four MLCC caps are in parallel with the electrolytic cap sitting at the regulator output rail.  Shorting the pads or installing a 0Ω SMD resistor will short the regulator output to ground!  I'm very sure you shouldn't do that.

 

I measured approx. 90Ω of ESR from each of those 10µF MLCC caps.  Four of them in parallel yields a group ESR of about 22Ω which is low enough to risk degrading regulator phase margin.  Per Andrew Sparks, it is beneficial for the output cap to have some amount of ESR to place a zero in the open loop response of the regulator to help optimize its transient response and phase margin.  Low-ESR (<20mΩ) ceramic or film or polymer caps connected in close proximity to the regulator output are therefore not recommended.  Based on these recommendations, I changed the electrolytic at the regulator output to a Panasonic FR 100µF 50V, and removed the four MLCC caps.

 

The above is also why I previous said for a series pass regulator like the SS1117, the output cap value is not "the higher the better" and the cap ESR is not "the lower the better".

 

Low-ESR caps are good to use at the regulator input.  Thus, I use 3 low-ESR polymer caps and 2 Panasonic FR caps to mod my SU-1.

 

@scan80269 in your opinion, does SS7805 need Panasonic FR 100µF 50V ?

[scan80269]

27 minutes ago, Duy Le said:

@scan80269 in your opinion, does SS7805 need Panasonic FR 100µF 50V ?

For the SS7805 output cap (C5) on the linear power supply, I would use a Panasonic FR 100µF 63V cap (EEU-FR1J101) with 63mΩ ESR which has the same 5mm lead spacing as the original cap at that location.  You can use Panasonic FR 100µF 50V cap but its lead spacing is 3.5mm so won't fit as nicely.

 

[Duy Le]

5 hours ago, Energy said:

Just attach a TO-220 mounting kit's thermal pad in between the discrete voltage regulator and the heatsink and you should be good. That or use MICA film with non-conductive thermal paste instead of a thermal pad which does the same job to isolate the two surfaces.

Which one will be be better?

4880MG (Mica) or 4880SG (Thermasill)

May I reuse the original heatsink kit ?

[Energy]

7 hours ago, scan80269 said:

Those four MLCC caps are in parallel with the electrolytic cap sitting at the regulator output rail.  Shorting the pads or installing a 0Ω SMD resistor will short the regulator output to ground!  I'm very sure you shouldn't do that.

 

I measured approx. 90Ω of ESR from each of those 10µF MLCC caps.  Four of them in parallel yields a group ESR of about 22Ω which is low enough to risk degrading regulator phase margin.  Per Andrew Sparks, it is beneficial for the output cap to have some amount of ESR to place a zero in the open loop response of the regulator to help optimize its transient response and phase margin.  Low-ESR (<20mΩ) ceramic or film or polymer caps connected in close proximity to the regulator output are therefore not recommended.  Based on these recommendations, I changed the electrolytic at the regulator output to a Panasonic FR 100µF 50V, and removed the four MLCC caps.

 

The above is also why I previous said for a series pass regulator like the SS1117, the output cap value is not "the higher the better" and the cap ESR is not "the lower the better".

 

Low-ESR caps are good to use at the regulator input.  Thus, I use 3 low-ESR polymer caps and 2 Panasonic FR caps to mod my SU-1.

 

 

I got a better look at the traces and see what you're talking about. They are tied from output to ground.

 

So you're saying that even if the original electrolytic output capacitors were replaced with Panasonic FR 100µF 50V, that these MLCC's lowers the ESR further which is undesirable when using the SS1117 as that ruins the phase margin and you're sure that removing them will only be beneficial. Sorry I'm just a little paranoid just in case it was used for something else in the circuit (surrounding components).

 

If what you're saying is the case, then those ceramic capacitors that were used to stabilize the LD1086 is working against the replacement discrete regulator then.

 

I know going down in value to 100µF has it's perks, but wouldn't it be best to keep the rating at 220µF?

Don't you think they would have picked that capacitance for a reason?

 

Just trying to clarify before I make the jump to 100µF 50V and remove the ceramic capacitors.

 

4 hours ago, Duy Le said:

Which one will be be better?

4880MG (Mica) or 4880SG (Thermasill)

May I reuse the original heatsink kit ?

 

Shouldn't really matter. Both does a good enough job keeping the regulator cool.

You can use the stock heatsink.

[scan80269]

1 hour ago, Energy said:

 

I got a better look at the traces and see what you're talking about. They are tied from output to ground.

 

So you're saying that even if the original electrolytic output capacitors were replaced with Panasonic FR 100µF 50V, that these MLCC's lowers the ESR further which is undesirable when using the SS1117 as that ruins the phase margin and you're sure that removing them will only be beneficial. Sorry I'm just a little paranoid just in case it was used for something else in the circuit (surrounding components).

 

If what you're saying is the case, then those ceramic capacitors that were used to stabilize the LD1086 is working against the replacement discrete regulator then.

 

I know going down in value to 100µF has it's perks, but wouldn't it be best to keep the rating at 220µF?

Don't you think they would have picked that capacitance for a reason?

 

Just trying to clarify before I make the jump to 100µF 50V and remove the ceramic capacitors.

 

Since the SU-1 board has one electrolytic and four SMD caps all sitting fairly close to the regulator output, I would say the board designer may have gone too far in output capacitor selection for the regulator.

 

ST LD1086 regulator datasheet only mentions a minimum of 10uF output capacitance needed for regulator stability.  TI LM1086 datahsheet recommends 10uF tantalum or 50uF aluminum, and has some extra wording:

 

Stability consideration primarily concerns the phase response of the feedback loop.  In order for stable operation, the loop must maintain negative feedback.  The LM1086 requires a certain amount series resistance with capacitive loads.  The series resistance introduces a zero within the loop to increase phase margin and thus increase stability.  The equivalent series resistance (ESR) of solid tantalum or aluminum electrolytic capacitors is used to provide the appropriate zero (approximately 150kHz).

 

It is specifically output caps that are connected close to the regulator output pin that can affect phase margin.  Since the Xilinx FPGA and Crystek oscillators each have their local decoupling caps, I would say those four 10uF ceramic caps sitting near the regulator output are doing more harm than good.  Not sure what the SU-1 designer line of thinking was, but I decided to follow Andrew Spark's recommendation to optimize output capacitance for the SS1117 to a single cap with the proper ESR, which meant removing those four SMD caps.

 

There is a tradeoff between transient response time and voltage overshoot magnitude.  For example, a larger output cap value will prolong the settling time from a transient event (current step), but shorten the amplitude of the overshoot caused by the step.  Smaller values such as 100uF will have more overshoot, but recover more quickly.  Andrew has evaluated many caps and communicated the Panasonic FR 100uF 50V cap delivers the best transient response with his discrete regulators.  The Panasonic FR 220uF and 330uF caps are likely good as well.

 

[Duy Le]

Quote

Superpower are compensated with enough capacitance for stability under any load condition. You may add as much or as little as you wish. Best dynamic response is achieved with a low inductance 100µF capacitor near the load. Capacitors distributed across a PCB, including decoupling caps near digital components are also good. If you are replacing regulators on a commercial PCB, this will likely be the case already, and Superpower works well in these applications.

 I also found this recommendation in Belleson FAQ.

@scan80269 why do you use the Panasonic cap closed to FPGA chip instead of Nichicon cap ?

[roberto2]

After 3,3V regulator I made internal power supply upgrade with Sparkos 5V regulator and Panasonic FR capacitors.

It took more time for the 5V regulator because of the heatsink but finally it works fine!

On Singxer card all capacitors are 220uF 25V.

And there is a Sbooster ultra between 5V output and 3,3V regulator. Sbooster input is 5V and 4,5V output.

 

All I can say is that my system has never played music as now!

 

[pkane2001]

On 6/8/2017 at 5:53 AM, roberto2 said:

And there is a Sbooster ultra between 5V output and 3,3V regulator. Sbooster input is 5V and 4,5V output.

 

Good job! So what does Sbooster do that the Sparkos regulators don't do already? Is the thought that two levels of regulation are more stable than one?

[roberto2]

22 hours ago, pkane2001 said:

 

Good job! So what does Sbooster do that the Sparkos regulators don't do already? Is the thought that two levels of regulation are more stable than one?

I cannot tell you exactly. I purchased the Booster ultra first, before Sparkos Labs regulators upgrade and I keep it in place after the upgrade.

I"m too lazy to take the Sbooster ultra off to see if there is a difference!