Best SMPS for Gainclone

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
next picture:
 

Attachments

  • 4-p6070008-150nf-no load-600.jpg
    4-p6070008-150nf-no load-600.jpg
    83.3 KB · Views: 1,079
Thank you for the wonderful tech info and pictures bichi. The molex connector specs are in the options pdf from Cosel. Do you think it would be beneficial to get some or remove the female sockets and wire directly to the board? The option on these supplies is the -J option for molex connectors. With the H-IN-1 and H-OU-21 leads from Cosel.
 
Molex Sample Order

Just for fun I requested some samples from Molex. Here are the part numbers for plugs and crimp pins needed for the PBA150F-24 (-J option) supplies. Should accept 18-24awg wire

Input Plug: 10-63-3044
Output Plug: 09-50-1063
Crimp Pins: 08-70-1030

Hope this helps.

Molex sample request page.
http://www.molex.com/molex/smp/amesamplereq?channel=Products&langPref=english

These supplies seem so sweet at this price point. I am gonna go sell some plasma and buy more for future projects.
 
Lets see, from an "audioqwakker" view, direct solder connections to switcher PCB is best.
- actually, both engineering and "audioqwakker" perspectives, "no connectors" is superior...

PBA150F Cosel surplus supplies from BGMicro came with both AC input and DC out connector/cable harnesses.
- at least mine did... thought you ordered some?

wtf? You're gonna make me look-up Cosel Options? - LOL
- Cosel options H-IN-1 / H-OUT-21 are not the same Molex connectors installed on the surplus supplies.
- surplus supply Molex MTC Family; PN: 51240 - drawing: http://www.molex.com/pdm_docs/sd/512400300_sd.pdf
- Cosel options Molex SPOX Family; PN: 5199 - drawing: http://www.molex.com/pdm_docs/sd/010633044_sd.pdf

- p.s. Newark one-off prices for Molex MTC are about $8~$12 USD...

Picture of surplus BGMicro Cosel PBA150F switcher Molex connectors:
 

Attachments

  • 7-bgmicro-cosel pba150f connector-600.jpg
    7-bgmicro-cosel pba150f connector-600.jpg
    87.4 KB · Views: 798
Ouch. I make assumptions based on the docs and look what happens! *** out of me!

I am glad they come with harnesses. I know audio quality might suffer but for my first time using SMPS easier is better. I can tweak later.

Thanks again for your data and efforts Bichi.

ps. I didn't buy any harnesses or connectors yet. Just a sample request from molex based on the cosel -j option info from the pdfs.
 
terranigma said:

What is the main pupose of this C515 cap? Can we decrease noise level to 4mv level with another SMPS?...

Depends on topology, switching frequency, final stage inductors and load characteristics, etc.
- probably best done experimentally, while watching switcher output with oscilloscope, unloaded and loaded.
- then, same tests as above, for sound quality.
- otherwise, to calculate optimal value, considering complex impedance, frequency and time domains, etc., without switcher schematics would be time consuming.

Thought process is a bit different:
- unregulated supply's ripple noise increase with load (100 or 120Hz, depending on line)
- unregulated supply voltage droop vs. load (impedance change)
- switcher high-frequency noise and amplifier PSRR typically decreasing over frequency.
- switcher less prone to voltage droop with better/build-in regulation. (stiffer)
- etc, etc, etc...

The Cosel surplus PBA150F supply is well designed and made:
- forward converter 120kHz, active filters 60~550kHz.
- Renesas/Mitsubishi M51995AFP switcher controller.
- SMT components lower radiated/coupled RF energy.
- very good ripple and noise, both Vrms and Vpp, considering VA rating, compared to transformer.


Still experimenting with different configurations, measureing and listening.

current config:
- 150nF metalized film internal C515.
- 10,000uF electrolytics and 150nF film capacitors across each rail.
- RC snubber 150nF + 1ohm.
- noticed if internal C511 and C515 are not present, increased noise and possible noise radiaton from power leads.

future tests:
- adding second internal film, from positve to ground, to make it symetrical with negative rail. (C511)
- lowering bulk electrolytic values.
- adding/changing internal output filter capacitors.
(beware installed Nippon Chemicon 470uF @ 50vdc KZE are rated for 2amp ripple @ 100kHz)

REF:
Renesas M51995AFP datasheet:
http://documentation.renesas.com/eng/products/linear/rej03d0835_m51995apds.pdf

Picture, PCB bottom:
 

Attachments

  • 8-bottom pcb-600.jpg
    8-bottom pcb-600.jpg
    86.7 KB · Views: 751
oh, forgot to mention:
- siblients (ess'ing) seem to be under MUCH better control, without loss of hi-mids.
- especially female "spitters," where "S" and "T" scrap at the ears...
- with this particular switcher, powering LM3886's and Beovox MC120.2's, "S" and "T" sound more "natural..."
 
Sixtoe,

Easy Answer:
- Cosel PBA150F Switcher:
- C511 and C515:
- Panasonic 150nF @ 250vdc - Metallized Polypropylene Film; ECW-F2154JB

Hard Answer:
Can't say what the "optimal" value is, and depends on your:
- grounding topology.
- power bulk capacitor values.
- RC snubber values.
- LM3886 bulk and bypass capacitor values.
- LM3886 input noise and bandwidth limiting capacitors.
- sound quality tastes... etc., etc., etc...

Current Config:

Cosel PBA150F-24 switching, surplus, forward @ 120kHz
- two (2) Nippon Chemicon 470uF @ 50vdc KZE, internal, parallel, across output.
- C511 and C515: 150nF @ 250vdc metal polypropylene film.

Bulk Filter(s):
- Bulk: 10kuF @ 35vdc PEH532 Rifa capacitors.
- 150nF @ 63vdc Thomson IR67 metal film capacitors, across rails.
- RC Snubber: 1ohm + 150nF @ 63vdc Thomson IR67 metal film.

LM3886:
- 2.7kuF @ 80vdc KMH Nippon Chemicon and 100nF @ 100vdc X7R Novacap 1210 SMT.
- 1uF @ 250vdc X7R Tecate 2225 SMT, across V+/V-.
- 11pf @ 100vdc AVX "Accufilm" 1210 SMT, parallel to LM3886 feedback resistor.
- 200pF @ 50vdc NPO CalChip 1206 SMT across LM3886 input pins.

Picture, showing possible power distribution stuff, relating to C511 and C515 value choice:
 

Attachments

  • switching supply-isolated output-002-800.jpg
    switching supply-isolated output-002-800.jpg
    83.5 KB · Views: 748
I am using Peter Daniel's 4780 board without Zobel and with 1500uf bypass caps on the amp board. No extra/other filters. Does this help refine choices or should I stick with "easy answer" as a starting point? You did report 4mv noise with only using a 150nF cap and measuring before your bulk filter? Which seems great before going for possible esoteric changes. I pretty much just need to get my GC off the ground and listen to it for a while before making changes outside of getting 4 caps for the power supplies. I figured if I am gonna modify them to have floating ground I could toss those caps in there while I am at it. What is your learned opinion?

BTW: You rock for helping! Can I send you a cookie for the efforts?
 
Just to clarify:
- 4mVrms, 34mVpeak-to-peak, @ 120kHz, measured with o'scope, not a simple DVM.
- no-load condition and terminated with 150uF @ 100vdc Nichicon HE capacitors, at the end of 12" of the supplied harness.

Sounds like a plan:
- quality 150nF films at C511 and C515 and 1500uF's on your chipamp boards.
- then, try slapping in some bulk filter caps and see how it performs...

Just a cookie?
- I'm not that cheap, LOL!
- how about reporting back detailed configurations, noise/voltage measurements and how you percieve it sounds? Deal?
 
How about a BIG cookie?

I make 450 a month on disability. Wife works part time at Best Buy. Rent is 600/mo. We barely scrape by. You supply the scope and I'll measure away! :) I will report back with how it sounds when I get the GC finished.

I am putting this GC together on plasma donations. No joke. I can only talk the wife into what is necessary and financially correlates to her Ball Jointed Doll habit. Japanese dolls are expensive just like DIY audio or any hifi for that matter so we haggle a lot. No phone, no TV. Even the internet access is umm borrowed wifi. My regular listening set is on the fritz and I had to bypass the worn out protection relays(on thump is not too bad...), replace the pot wiring, clean the pot and once the signal was strong to the main amp (its a JVC CA-MX70BK with bi-amping) the left channel stopped its hum, in fact it stopped entirely! So I now have to use the schems I bought for 4 bucks on the net to troubleshoot the left chan path thru both amp boards and chips. Please baby Jesus, Buddah, Allah, Vishnu and mom let it be a ground issue. I sure hope the amp chips are not shot because they are old (1991) and I am not sure I could source replacements for cheap or at all. Whew!

I just found out that Dale CMF55 are too big to easily fit in the spacing design and am now considering SFR16S from BC just to be able to fit well enough without uneven leads. They aren't the best at 100ppm but I am eager to see if I can do this thing for cheaper than Caddock and Rikens. Dang!
 
Bichi,
Are you using a snubber with the smps? I built up a 3875 chipamp board and am using just the 1500uf on the board hooked up to the smps. I thought that I saw a snubber board in one of your pictures. Have you tried it with the snubber or without? I was thinking it might be a good idea to add some capacitance between the supply and the board. What would you recommend?

Thank you for your work with this supply. I added the caps per your recommendation. I don't have a linear supply to compare right now, and my amp is a work in progress, but you can't beat it for the price.

Thank you for your time.

Best,
Aaron.
 
I just received four of the Cosel supplies from BG Micro. The landed cost of each was much less than half what a similar, but Chinese made supply would cost from local supplier Jaycar.

Thanks to bcrescen for pointing these out. And to bichi for his work with the supplies posted here. Saved me a bundle.
 
I just received a few of these Cosel PSUs. I want to install C511 and C515 filter caps to reduce switching noise, but my local electronics supplier does not stock 150 nF polypropylene caps with a raster size of 10 mm. Instead, I can choose between:
- 100 nF polypropylene, 250 VDC, 5 %
- 220 nF polypropylene, 250 VDC, 5 %
- 150 nF polyester, 250 VDC, 10 %
- 150 nF polyester, 100 VDC, 10 %

The SMPS will feed bridged LM4780 kits with 1500 uF bypass caps, no bulk caps or snubbers.

Which filter cap should I choose, and why?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.