Reflektor-D builds

[Salas]

Hi! I built a Reflektor for my breaglebone black that is serving music to my dac in the same chassis ("chassis" being used liberally to describe an mdf board... ).

Before using the Reflektor I powered the bbb with a 2A/5vdc smps so I could get and idea of the current draw using the music I usually play. At turn on it draws a bit over 400mA. After it stabilizes and plays the average seems to be about 350mA, with the max recorded for a two hour listening session being 410mA.

I built the Reflektor with a 1R Caddock MP930, thinking that it could draw close to 500mA, but it fell a bit short. Should I leave it as is or try to get a 1.2R instead of the 1R? I ask because the heatsinks during this two hour period stabilized at 50-51 degrees F (max was 50.7dF). It's using a 9v trafo. I live in Florida so summer is long and hot and the thing is not yet inside a case...

My second question is regarding the startup. When I hit the on button, sometimes the bbb does not start! The leds are on in the Reflektor board, but the bbb does not boot (no lights come on, as if it had no power). When I was testing the setup with smps it always started. Pulling the bbb's 5vdc plug in/out makes it start. From the measurements above it doesn't seem to drawing, even for a brief moment at startup, anywheres near the max current, so I'm not sure why this is so.

Thanks in advance.

Degs F or C? Sounds too cool if in F. 51F=10.5C.

1 Ohm is about 600mA. Can go down to 450mA in your case. Check voltage across R1 and divide by R1 Ohmic value to define exactly. Check DC raw across the big filter capacitor. If its more than 10.5VDC with your mains and Tx maybe you can use a lower secondary VAC one to avoid some more heat. Or try power resistors to the primary until 10.5VDC rectified.

Does it do the missing a start thing with or without remote sensing? Check both occasions. I.e. as in the guide or by lifting the sense lines at the Ref-D connector and closing them locally with short jumper wires to the relative F+ and F0 screws.

 

[palmito]

Degs F or C? Sounds too cool if in F. 51F=10.5C.

Sorry, I think in F... it is centrigrade, 50-51 deg C!

I'll do the measurements tomorrow and report. My mains are on the high side too.

I tried it both ways, full sense cables (as photo) and jumpered at reg and single cables out to bbb. I'll test some more the start up current surge just to make sure and report.

Thanks Salas.

 

[Salas]

You may insulate the MOSFETS and mount them to the box's bottom plate via short cable extensions (move the gate stopper resistors on the gate pins) if in the end its a lot of peak current you need and the extra dissipation must be handled better. On the other hand if its not a matter of sudden peak limiting see if a small size decoupling polymer or electrolytic helps that mini Linux computer start always when installed at its DC input jack. Maybe the small delay of C2 to charge up catches it out sometimes, hmm. Smaller C2 to test? 470uF?

https://learn.adafruit.com/embedded-linux-board-comparison/power-usage

 

[polaron]

Hello!
Salas, if this subjet is not the most appropriate to this thread, I apologise, please remove my post.

I have two OHM FF-XAL-AL caps, one has 0,01 and other is 0,02.
Now each Reflek has one OHM soldered in paralell with the 0,1 wima.

I was not expecting a difference and I was surprised with the "finesse" that they add to the music. Lots of nuance and space. These psu really react, I just wanted to share this experience, it may interest someone.

Would a 22000uf cap to replace the 10000uf, for ex silmlic II, improve the sound?


If I introduce a 22000uf cap do I have to change something? Do I have to replace a resistor or something else?
How could I improve this psu? Not that I'm not happy, but it's exciting...

Thank you.

 

[Salas]

You mean C1 R5 termination values I believe. Many such changes for C1 quality and value had been done in the past by the few early experimenting builders on proto-boards on different set Reflektor for analogue and digital. Each one had a favorite combination. 0.1u Wima was an agreeable middle ground. Experimentation with value can take it to oscillation depending on the load circuit reactions though, so no guarantee given to those playing for surely not burning something in the end. That's a disclaimer! Experimentation with quality only is safer. C1 place on PCB offers various distance pads for that matter.
Bigger filter capacitor can tax the transfo and bridge diodes more, but with 30VA and TO-220 diodes it won't break anything for a try.

 

[polaron]

Yes Salas, I was talking about C1.
I'll try another cap in C0, with more capacitance.
Can someone suggest a cap with 22000 uf max for this situation? Thanks.
Toroid will have at least 30VA.
I wonder if 50VA would make a difference but I bet that more than 30 is not necessary to feed these digital circuits.

 

[Salas]

Try Teflon and Silver Mica for C1 too. 0.022uF to 0.1uF should be safe.
The big filter capacitor is designated as CM (capacitor main) on the PCB. Usual 10mm snap in suspects from Nichicon, ELNA, Mundorf, Panasonic, should be fine. 16V-25V so to fit.
Even 15VA is enough for full 7V 600mA setting, but we use higher so to keep voltage sag and impedance better, as long as low heat in the Tx. Rest is subjective. Maybe someone comes with 300VA build and says wow, but you will never find why or prove anything.

 

[clsidxxl]

You may insulate the MOSFETS and mount them to the box's bottom plate via short cable extensions (move the gate stopper resistors on the gate pins)

Salas,short how,2-3 centimeters maxi.

"move the gate stopper resistors on the gate pins" I do not understand,you can detail please

 

[Salas]

As short as needed so the MOSFET pins stay very near the board sides bolted down on their backs to the floor. Silpads and insulating grommets are essential. R3 & R4 remove from board and stick them to the gate pins directly. Jumper their board places. If you got an oscilloscope and the rail is still a flat line with no oscillations before you try move them, then its OK let them on the board. Watch the wiring, the MOSFETS are gonna turn 180deg now than original position.

 

[polaron]

Well, I choosed Mundorf for CM- 22000uf-25volts
SCR teflon etain, 0,1 uf witha very good price for C1 16 euros each
and I wonder if I would have a benefict in replacing the 1000uf-25v by elna silmic II or Cerafine.
As for the toroids, I'll keep it with 30VA max, my litle Amanero and dac will be happy with it!


scr are for me excellent caps, very cheap for their performance.
A cheap scr 2euros gave me better results than AMPOHM and at least as good as Hovland, in signal.

PS : one ampohm I soldered with C1 has 0,02uf - it added 0,2 volts to the output. It is ok, but I will replace it by the SCR, 0,1uf.
The other reflektor has 0,01f with C1 and the voltage didn't increase.

Thanks Salas!

 

[Salas]

Suspicious was the voltage increase. Could have been instability.
For C2 1000uF its down to taste. There are also good solid polymer with great HF characteristics to try there too since you like cooking.
The stock kit components are a very good value mix, carefully selected. For the curious to throw money in and listen to parts, its another story.

 

[polaron]

I think I'll give some peace to C2 and keep it there. Besides, it is the same as the caps that populate my dac. CM and C1 will cost 38 euros. As far as I can understand until now, they are the most important parts to tweak.

 

[m a r k o]

My local electronic store had only a couple MTP3055s and they gave me a "substitute", IRF530 from International rectifier. I've built my third Reflektor, with IRF530 instead of the MTP and the current through the LEDs is ridiculously high so I get a higher than expected output voltage. I double checked the soldering and layout, finally switched the IRF to MTP from a working Reflektor and all was fine. I assumed that the pinout for the two mosfets are the same, however the IR datasheet of the IRF530 doesn't show the pinout, only the internal diagram. I'm kind of stuck.

Sent from my C6903 using Tapatalk

 

[Salas]

The MTP3055VL is a logic level MOSFET with low VGS threshold. You should not use a normal MOSFET in its place. Its not the pin-out, its the VGS. Normal MOSFET may be used for higher voltage Vref. If the Vref is lower than the VGS itself in this circuit then the current goes crazy through it and the open loop gain suffers, so output Z suffers too.

 

[palmito]

Can go down to 450mA in your case. Check voltage across R1 and divide by R1 Ohmic value to define exactly.

I did several measurements of current draw at boot time and the highest was 420mA, so you think I can drop the shunt to 450mA?

I changed the R1 from 1R to a 1.1R and measure 0.61vdc (550mA). After playing for about an hour the temp is now 46-47C. Still too high, since it's not in a case yet?

Check DC raw across the big filter capacitor. If its more than 10.5VDC with your mains and Tx maybe you can use a lower secondary VAC one to avoid some more heat. Or try power resistors to the primary until 10.5VDC rectified.

The raw dc across the big filter is 11.5vdc. I don't have any 8vac secondary TX, so try the resistor? I've never done this before to the primaries of a TX, any recommendations for the type of resistor to use?

...if its not a matter of sudden peak limiting see if a small size decoupling polymer or electrolytic helps that mini Linux computer start always when installed at its DC input jack. Maybe the small delay of C2 to charge up catches it out sometimes, hmm. Smaller C2 to test? 470uF?

Based on the current draw measurement with a voltmeter it doesn't seem to be a peak limiting problem, even at the 550mA it's set for now. I did change the C2 for a 470uF but it made no difference, still starts randomly. I was going to add the cap as you suggested to the BBB plug, but I looked at the schematic and there is a 10uF cap already there at the other end of the plug on the board (image attached, from BBB schematics). Should I still add the cap there?

Thanks again for all your help.

 

[Salas]

You can drop to 450mA since the 420mA peak is rare, so there will still be enough margin averagely. That and 1V less DCin should back down few more C.

Restore C2. Add 100uF to the plug so to see what happens.

That temperature poses no threat to the reg MOSFETS. If it is not detrimental to other stuff in a box by radiation then its no problem. If it is suspected to be so in the long term, you can remove the board level sinks which can be up to a practical size only and use the chassis floor via cable extensions to M1 & M2 as previously described. But insulate the MOSFETS.

All cement power resistors are several hundred Volt rated. But on second thought do it on the secondary side so its safe stuff, and you know the exact current draw. Predict with Duncan PSUD2.

 

[palmito]

Restore C2. Add 100uF to the plug so to see what happens.

Put back original C2, added 100uF bbb 5vdc input, no difference. It seems to randomly choose to boot... Pulling the 5vdc plug out of the bbb and put it in again always works to make it start.

That temperature poses no threat to the reg MOSFETS

So the original 50-52 deg C (with the 1R) is ok too? If so I'll go back to the Caddock 1R. If the answer is yes then I will not use the resistor in the TX secondaries.

I may have to go back to another PS since it will be very difficult to pull the plug out of the bbb when it's all in a sealed case!

Any other ideas as to what may be causing this? Adding a delay circuit between reflektor and bbb might help? But I would rather not do that if I can avoid it...

Thanks for all the support.

 

[Salas]

This is something related to what bugs the mini Linux Ti computer which I can not know without having it in hand to investigate. Using the scope etc. Maybe it does not like so low output impedance of such wide bandwidth, or it has some local buck converter chip(s) that interacts kickstart, can only theorize. I would try this practical thing though: Using a switch to turn on the PSU and a relay or hand-switch to connect the plug a few seconds later. Can be pre-tested with any handy secondary switch near the plug end. The PSU would take the short off-load time on a 470R dummy resistor I would put across the plug switch so to have a complete circuit for sense always. As a matter of fact I would try such a resistor at the plug as it is already wired, just in case it helps.

The manual says you don't want more than 65C on the sinks. That should be your target for a boxed system in the highest room ambient anticipated. You don't want to cook the rest of the components with more than 50C inside box ambient also. Using your thermal probe through some opening for touching sinks or hanging in space in a final closed assembly will give the data.

 

[clsidxxl]

. I would try this practical thing though: Using a switch to turn on the PSU and a relay or hand-switch to connect the plug a few seconds later.
.

Yes Salas I do like this with the Cubietruck.I start the Reflektor and after plug the connector of the Cubietruck,it is no a problem.
But a switch is another solution.

 

[palmito]

...do like this with the Cubietruck.I start the Reflektor and after plug the connector of the Cubietruck,it is no a problem

Cubitruck, another ARM device has this problem. I know Enrico had the same issues with an Edel renderer, which is also an ARM device. There must be something in common between all these one board computers that cause this?

...Using the scope etc.

I have one (just got it, learning how to use it as we speak...), what should I look for?

...470R dummy resistor I would put across the plug switch so to have a complete circuit for sense always. As a matter of fact I would try such a resistor at the plug as it is already wired, just in case it helps.

A 5w 470R?