Heyyy SYN08..... Result of your gracious sharing of gerbers
One question though, where do you source those 3900uF polymer cap´s??
I´m having a really hard time finding anything larger than 1500uF (if that big) over here.
DigiKey 493-3103-ND
With 4 of these, the LF bump is less than 0.05dB, I guess you could use lower values like 1500uF without much damage.
syn08, this is an excellent and accurate description of a my wonky virtual earth.Guru Hans shows Duraglit is equivalent to this ... as is #375
Thanks, but I'm afraid this description doesn't fit the Duraglit. To start with, the Duraglit has the gain dependent on the bias. I don't like the Duraglit because of the beta dependent bias and gain.
Interesting data point, currently #375 is fed by a 24V wall wart, through an extra 7815 regulator as shown in the photos. I ordered and received today a MeanWell good quality 18V/6W switching power supply, to lower as much as possible the power dissipation (keeping the 7815 in place, 3V across, another 3V across the 317 current sources).
The experiment was a disaster. The level of hum (60Hz and harmonics) increased one order of magnitude, and was clearly audible even at moderate sound levels. I would expect some switching crud from a SMPS, perhaps affecting the noise, but mains hum??? The MeanWell AC/DC is far away (over 3ft) from the shielded pre, the hum effect is also very measurable on the bench. Very strange.
I'll stay with the classic AC/DC 24V adapter for the time being...
I used lately JLCPCB, 10 such boards would cost $6 plus shipping. Since it's under our canadian max $20 value, no fees or taxes in the custom (which I know could be a PITA in Europe). I usually pay for DHL shipping, no outrageous CODs from the courier. A total of about $3/board delivered at the door, I can live with that
. Quality is also very good as far as I can tell, they also have 4 layers for cheap (up to 10x10cm) and the controlled impedance traces (in two versions, for my microwave fun) are free!Thanks for (all) the info
. Just found them at Mouser too, around same price, and have also floating around somewhere 2700 Sanyo Oscon SEPC to try out.Hans.... got them made at PCBWay. 10 pieces. 24 hours from confirmation of gerbers to "ready to ship". 27USD including shipping. No extra taxes. 48 hours from China to Denmark with DHL Worldwide Express. I do not believe, that any other company can beat that
Thanks for (all) the info
Hans.... got them made at PCBWay. 10 pieces. 24 hours from confirmation of gerbers to "ready to ship". 27USD including shipping. No extra taxes. 48 hours from China to Denmark with DHL Worldwide Express. I do not believe, that any other company can beat that
JLCPCB does beat that
. I was using PCBWay myself, until they crapped the quality. I got twice boards for 0201 SMD with the solder mask so badly misaligned they were unusable. I had to file complaints, they redid the orders to an acceptable quality, but of course the delays were on me. No problem for boards like these you just got, of course.A practical advice, it would be very difficult to solder the LEDs using a regular soldering iron without melting the LED plastic casing. After tinning the LED PCB pads, I placed the LEDs in position and used a hot air soldering tool from the bottom of the board to heat up to the solder melting point. The LEDs will slide and self aligh right in the position. 100% success! Therefore, start assembling the boards with the LEDs.
Last edited:
Thanks for this Karl.
.. which makes me even more curious about Marik's 'ferrite' cores
____________________
Just a reminder to LISTEN to the important base caps. syn08 had good experiences with polymers but not gerhad. Problem will appear as 'greater than 1/f noise' if you use a spectrum analyser ... and usually sounds 'crackly'.gerhad said:
You'll have to dredge the thread to find chapter, verse & other info from the cap gurus .. but IIRC gerhad found cheapo Rubycons to be trouble free & quiet. This was my experience with tantalums from da 70s up to this Millenium.
JLCPCB does beat that
Will let them have a go next time I´m i need of PCB´s
Yeppppssss
Try make a common mode filter for the dc output. Ferrite core run through several times.
I have to say, every so often a thread like this comes along where, despite my inability to contribute anything useful something new and interesting emerges and gets built. Two thumbs up.
Ovidiu: Would you mind if I posted the pic of your completed unit on another forum I vist. I think will fuse the brains of some of the people there
Ovidiu: Would you mind if I posted the pic of your completed unit on another forum I vist. I think will fuse the brains of some of the people there
That would be $1000, thank you very muchI have to say, every so often a thread like this comes along where, despite my inability to contribute anything useful something new and interesting emerges and gets built. Two thumbs up.
Ovidiu: Would you mind if I posted the pic of your completed unit on another forum I vist. I think will fuse the brains of some of the people there
.Sure, go ahead.
This is the first thing I tried, no cigar. Once again this is about plain vanilla 60Hz hum, not some RF crud collected from the mains.
All true, but I still don't understand what has this to do with the mains frequency harmonics. The filtering is very good (and unchanged from when using a regular AC/DC wall wart), the extra linear regulator has very good line regulation at 60Hz, nothing seems to justify the observed behavior.
I had the same experience with +/- 24 v meanwell smps units. I used them to power a 60dB 1nV/rtHz flat amp to analyze sweeps from a flat recorded LP’s. Results were rather disappointing. When replacing the meanwells by 2x 24V lead batteries, noise went down drastically and FR became nice and flat. The only way I could tame the meanwells was with TPS7A4700 voltage regulators.
Hans
P.s. The 60dB amp had internal Jung Didden regulators to bring the supply to +/18 V
Last edited:
Interesting that this comes up today. Last week I got some IRM20-xx from a local distributor and found them quite OK at first sight. I used them for powering a digital phase noise measuring machine on 5 and 12V. They ran somewhat warm at 2/3 rated load. I need some more to power the SMA coax relays of a delay line (3*2 relays 1P6T at 24/28V, which to my surprise turned out to be the biggest consumer). And another 12V converter for the ovenized reference crystal oscillator that will run 24/7.
I saw them at DK and noted that they are the cheapest ones I could get. I also found that there is a MRM30 series with 50% more power and with specs for medical devices, i.e. controlled crosstalk etc; I put them on my DK list that I'll trigger on Monday with the other missing parts I'll find over the weekend. The medical version does not cost too much more.
< MPM-30-24 MEAN WELL USA Inc. | Spannungsversorgungen - Kartenmontage | DigiKey >
Last edited:
That 100uA would be at large signal (0VU?) into a heavy load. Hawksford discussed the case of small signals and conventional light loading. What midband output at say -60dBVU could we agree on? 0.2mV at 0dBVU would be 200nV RMS (back to MC phono stages again).
200nV across 2000 Ohms is 100pA RMS. Still a goodly number of electrons jostling around, but the question is still in my thoughts. Hawksford actually framed the question in terms of the number of charge carriers injected into an input transistor, much smaller still.
Much thanks,
Chris
Like the wire articles the fuzzy distortion was fuzzy thinking, he was out of his element but his position allows him to pontificate with no control on the output.
Just plain wrong sense, like the wire articles the physics is fundamentally flawed. You cannot separate the particle/wave properties, EM wave propagation can not be thought of as quantized electrons.