bi-circleton

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Two figures from one source.

A little funny result from my Arta. It was taken at same settings.
At low resolution, my THD is 0.03%
At high resolution, 0.004%
If you like THD percentage, here they are
 

Attachments

  • 6075F5B7-274E-4244-8A16-FCB5EC6E62E5.jpg
    6075F5B7-274E-4244-8A16-FCB5EC6E62E5.jpg
    690.7 KB · Views: 315
  • E9A5117E-6B00-4BFC-BFDB-1510BB86F3D3.jpg
    E9A5117E-6B00-4BFC-BFDB-1510BB86F3D3.jpg
    705.3 KB · Views: 294
Mum, no 50Hz hum!


Second disappointment, 50Hz is -40dB at the output, it was reduced to -60dB with C-R add on (10mF and 2.7R)

Add-on another stage 10mF and 3.9R (because they are nearby, don’t know how to justify for values chosen), 50Hz dropped down to -74dB
 

Attachments

  • 56CD3DA5-2A76-4E6F-A603-034BD2D6FFCE.jpg
    56CD3DA5-2A76-4E6F-A603-034BD2D6FFCE.jpg
    949.8 KB · Views: 258
  • 0E01E906-A1F2-4A84-BEA2-111108007FE1.jpg
    0E01E906-A1F2-4A84-BEA2-111108007FE1.jpg
    732.7 KB · Views: 239
Hi Moss, keep posting on!

Hi Pad,
I meant “quiet” for any comparison with 300B amp.
So far, one channel had been rat raced wiring, fortunately working well with an exception of excessive filtering at the driver circle that had to be added for lowering 50Hz ripple, will do some fiddling with the help of any “calculator on line” to work it out the correct RC values (My background is not related to EE). By the way I discovered that the phase of 9v coil is equally important in the ripple reduction, I swapped coil terminals to diodes bridge, miraculously I got another 12 dB ripple less to an agréable value of -88dB for 50Hz noise.
Photos included, please note at the red and blue wires go to the bridge and Arta FFT result (yellow and green).
The biases for my Lateral MOSFET are different from Mr K, due to mine is a 5 pins AllFet 08NP160.
Hope that I will finish the whole amp next week with a presentable pcb and casing.
Thanks for making this thread less monotonous to my monologue.

Cheers from Pascal
 

Attachments

  • 03A14582-02E9-4E90-82D5-E027C50E61A1.jpeg
    03A14582-02E9-4E90-82D5-E027C50E61A1.jpeg
    374 KB · Views: 59
  • 1688AD51-544D-4F47-9004-953FD3EDEBA4.jpeg
    1688AD51-544D-4F47-9004-953FD3EDEBA4.jpeg
    356.5 KB · Views: 80
  • 8AC8A989-E9FB-4DCE-871C-BE4D0D1B7964.jpg
    8AC8A989-E9FB-4DCE-871C-BE4D0D1B7964.jpg
    813.6 KB · Views: 78
  • A6EA416E-5303-44CB-8DAE-6CC85971C371.jpg
    A6EA416E-5303-44CB-8DAE-6CC85971C371.jpg
    806.6 KB · Views: 65
  • EB4C6E60-19B0-417C-9A2A-47BEBDE788FC.jpg
    EB4C6E60-19B0-417C-9A2A-47BEBDE788FC.jpg
    874.4 KB · Views: 210
Nota Bene

The amp got his umbilical cord ectomy, it is running on a 50VA toroïdal transformer 240 / 18x2 / ES shield / 9.
ES shield is a must for a single transformer PS, it can be omitted if 9v is ressourced from another transformer (wall wart).
Now i am toying around with the odd and even harmonics by moving the window of biasing, looking for any sweet spot.
The amp is very stable, no pop at power on or off, thermal stability as well, due to lateral mosfets and clever design of driving stage, I mean circle.

Cheers
 

Attachments

  • EAE19BFE-3BA6-4DD5-86F4-20A77A12F2F3.jpg
    EAE19BFE-3BA6-4DD5-86F4-20A77A12F2F3.jpg
    922.4 KB · Views: 77
  • FCB75C78-7504-4011-B784-F1C89CDB4370.jpg
    FCB75C78-7504-4011-B784-F1C89CDB4370.jpg
    718.2 KB · Views: 66
Disabled Account
Joined 2015
power supply

The power supply of bi-circleton requires a toroide transformer of 300VA 2x24V for each channel . The transformers need to be customized to 20V +20V(unloaded) +a shielded 9.9v (unloaded) of 160ma .

The transformer is put to naked , the center tap split , and equal number of windings taken off until required voltage achieved . By the same way the manufacture has treated the wires ,are retreaded again and the transformer locked back .

The shielded winding will dwell on the opposite side of the secondary wires ,two hours right of the primary entrence . The 9volt secondary to shield is merely 1.5VA that is a very small volume of cooper .

The following explanation will result , a 9.9volt (unloaded) secondary, having 5mm wide footprint wound in two layers, sandwiched between two cooper shields of open loop rings ,wide 15mm , harvested by a 600 ohm (two wires) coaxial microphone cable ,bringing the 9v shielded till the outer circle's pc board .

The wire has 160ma current ,the gauge table shows No 32 SWG or 30 AWG . The diametre is 0.275 mm. This is the minimum . Thicker will not be a problem except that when loaded the 9.9v will be higher than 9v , which the amp can accept without problem .

For the length ,first must know the voltage per turn that the transformer gives . Ten turns wound with any wire can provide precise value , it is between 0.25 to 0.3 volt/turn . Dividing 9.9v by this number gives the number of turns required to wind ,about 32 . Measuring the length of one turn on the donut ,about 17cm and the length of wire needed is fixed ,7m , by allowing some extra 10% gift . The whole wire is wound around a piece of cardboard about 15cm x4cm,with U shaped slots on each end , the ends of the wire are scraped and soldered for voltage measurements while winding .

The cooper foil is 15mm wide having a length just one turn precise . It is isolated by adhesive tape starting from the middle with single continuos band ,isolating also the edges . The shield is fixed on the donut with the endings inner side ,If it has a gap of one mm ,it is not a problem . On the left edge ,from the middle to 5mm downwards solder a thin wire, to end at the middle ,bend it leftwise ,cut it about 2cm long naked wire . Bring the wire to be wound about 5 cm long twisted with the shield wire and fix them temporarily with tape and start winding from left to right , As you are counting so that the lower layer has 4or 5 turn more than the upper one . Once the voltage is checked 9.9v then the arriving wire twists with its beginnig one and ends with same length . The coil is protected with adhesive tape . The coaxial cable of desired length gets its wires and shields exposed on both sides , because the wires can slide in the cable which will allow you to adjust precisely the length. The coaxial shield is separated into two brines , one is soldered along the width of the outer shield foil at the middle and comes on the coil with the cable on the right side , the inner shield's wire now is soldered to the upper's on the left bottom . The two winding wires first overcross the outer shield on a protected edge and go to the right entrance to greet their mates and come back together back to the left side, The coaxial wires are striped a generous 1 cm while the winding ones are scraped and soldered to be wound around the receivers ,soldered and isolated . A thick cardboard covering the wires and the shield solders is firmly adhered with tapes to serve as anchore to the recieving wires , which soldered tips are folded upon the cardboard acting as hooks , prventing the wires to be pulled. A last cooper shield is added isolated only underside, covering every thing bad looking and gets the second brine of the coaxial shield soldered diagonally on it, as the cable is now plied vertically . Done.

The power supply propsed bellow has mutual inductor . I tried it out using 12v/12v 48va isolator transformer , the result is spectacular dynamic sound . This small 50mh will fall about 1mh at 2x1.26A but it is sufficient . The 10000uF capacitors I chose, have a footprint of 18mm only ,which permits denser packing ,hence less inductive .
1pcs/37*22*15 50mH 1.0 Line 10A Annular filter Magnetic ring Common mode inductors Choke coi-in Inductors from Home Improvement on Aliexpress.com | Alibaba Group
20pcs/lot 25V 10000UF 18x35mm Aluminum Electrolytic Capacitors-in Circuits from Consumer Electronics on Aliexpress.com | Alibaba Group

I'll post in few ,a hand sketch proposition for pcb layouts for inner and outer circles .

Kokoriantz
 

Attachments

  • MODEL PSU.JPG
    MODEL PSU.JPG
    118.8 KB · Views: 154
Disabled Account
Joined 2015
MossFed , Your amp has Booing sound ,probably the transistors you are using need to be biased higher current . Try it between 1.6A to 1.8A . The bass must sound Paah not boo. Instead of 9volt higher it to 11volt. The two DNs must function at temperature about 50°C . The heat sink I use is just a sheet of U plied aluminum with shredded edges can install one TO220 inside and the other on the back . Yours is gigantic .
 
… it is very easy . You have done it wrong .

Thank you for show me the right approach to the PS, I see what you mean by very easy.
I realised the 50VA toroid was for fathoming the circuit and for my learning curve of how to reduce the EMI.
Somehow I had modified the amp, with SMPS for output and the 50VA toroid for driver stage, my shielding is working very fine in comparison with battery DC supply.
I had tried all my possibilities for shielding the toroid, secondary coil grounded one end, mu-metal, GOSS band ...idc flat ribbon.
I found the best (on my bench) is copper screen shield that I had done and RC filter, a comparison spectrum wit batteries had confirmed.

Many thanks and my appreciation for your help

Pascal Nguyen
 

Attachments

  • 92D55F47-1B36-4163-8B1D-96855AA02F34.jpg
    92D55F47-1B36-4163-8B1D-96855AA02F34.jpg
    759.7 KB · Views: 314
  • B9DE0E14-0977-4312-820F-C52809D7310B.jpeg
    B9DE0E14-0977-4312-820F-C52809D7310B.jpeg
    670.2 KB · Views: 298
MossFed ,… The two DNs must function at temperature about 50°C . The heat sink I use is just a sheet of U plied aluminum with shredded edges can install one TO220 inside and the other on the back .
… Yours is gigantic .

I mounted the DNs on main heatsink next to power transistors with a purpose aiming for a quick start with less warm up time.
At cold stage less current flows through DNs, bias to power transistor is higher via R shunt Drain from one DN to Source of the other.
When it reaches 50 deg C, DNS current is higher then at cold stage, so at both ends of R shunt, V bias drops.

Cheers
 
I forgot to mention 50°C is channel temperature . Can someone explain please, how to edit a post .

With the hope that I understood well the transfer curve of the DN2540, my driver is working in the window choosen from cold start 22 deg C to the plateau of 50 deg C.

Cheers
 

Attachments

  • FCA56612-829A-4C43-B5F2-217AABF5DAEC.jpeg
    FCA56612-829A-4C43-B5F2-217AABF5DAEC.jpeg
    405.5 KB · Views: 283
...it is very easy . You have done it wrong .
Follow your instruction, I reshield the 9V coil, I realised that my previous shield was wrong, the foot print of the 9V coil projected on nearly half the core, now it is reduced to lesser than 10%.
 

Attachments

  • 0B8BF020-962E-4374-9D72-DC387F0170B0.jpg
    0B8BF020-962E-4374-9D72-DC387F0170B0.jpg
    737.3 KB · Views: 240
  • 0A726A90-0DC8-4634-BD4E-4AB2BCD194AF.jpg
    0A726A90-0DC8-4634-BD4E-4AB2BCD194AF.jpg
    645.2 KB · Views: 234
  • EBFC76FF-6424-42DA-B12E-8AD560B29DF9.jpg
    EBFC76FF-6424-42DA-B12E-8AD560B29DF9.jpg
    806.3 KB · Views: 84
... it is very easy . You have done it wrong .

What I have done wrong this time, perhaps some nasty ground loops are witching me. The FFT result is disappointing.

Rondino
Samuel Sanders and Kreisler collection, my favorite, from the rat nest.

YouTube
 

Attachments

  • 5DDC524D-F3C9-4260-8140-22BED3020BDE.jpeg
    5DDC524D-F3C9-4260-8140-22BED3020BDE.jpeg
    797.7 KB · Views: 75
  • 2548B29A-B46F-40CF-A0E4-586B67A3814F.jpeg
    2548B29A-B46F-40CF-A0E4-586B67A3814F.jpeg
    794.6 KB · Views: 50
Disabled Account
Joined 2015
600ohms coaxial cable is very thick used by professional fandom 48 v microphones ,for long distance ,as on stages . It has two wires and of course the shield .

To measure the noise captured , you can test the transformer alone by applying 10k ohm on the terminal of the single coaxial carrying both 9v wires, between the shield and one of the 9v . On scope or milli voltmeter you must read zero across the resistor ,if not, check from where it is capturing . Every volt you read, one tenth of it will appear on the output. The ground of this amp is created by the sources of the outputs . With my transistors having their source on the backplate they ground the heat sink naturally . So the coaxial's shield gets grounded by the heat sink. I am 100% offline a month out of two and it will be from 3 June to 5 July .

You are bringing the coaxial wire to the diodes and then a pair of non shielded twisted wires are serving a giant mass of capacitors with bazar of components . All these capture noise . My proto circuit is only 5cm wide 8 cm long carrying the four diodes, two capacitors, bias resistors, the two DNs on same heat sink ,a source resistor for one and an adjust for the other ,plus the volume . The shielded supply arrive ,and Gn Gp depart . That's all.

The noise spectrum looks like that of switching power supply . Earth the heat sink see if it decreases .
 
600ohms coaxial cable is very thick used by professional fandom 48 v microphones ,for long distance ,as on stages . It has two wires and of course the shield .

To measure the noise captured , you can test the transformer alone by applying 10k ohm on the terminal of the single coaxial carrying both 9v wires, between the shield and one of the 9v . On scope or milli voltmeter you must read zero across the resistor ,if not, check from where it is capturing ...

Follow your advice, two transformers with different kind of shield for 9V coil are measured with 10k Ohm termination from 100 Ohm twisted wires coaxial cable, will build a 600 Ohm câble for the next test.
Far away from zero volt target, the broad shield captures less 50Hz field then the narrow ring shaped.

Cheers
 

Attachments

  • 4841D9C4-58AF-4098-94D6-9B19E72A3BF5.jpeg
    4841D9C4-58AF-4098-94D6-9B19E72A3BF5.jpeg
    869.9 KB · Views: 78
  • CFA917E9-8AAD-4F98-B2DF-C8272F664F84.jpeg
    CFA917E9-8AAD-4F98-B2DF-C8272F664F84.jpeg
    865.9 KB · Views: 79
Follow your advice, two transformers with different kind of shield for 9V coil are measured with 10k Ohm termination from 100 Ohm twisted wires coaxial cable, will build a 600 Ohm câble for the next test.
Far away from zero volt target, the broad shield captures less 50Hz field then the narrow ring shaped.

Cheers

My apologies, the photos annotations are cheating, please read them as:
Broad shield = 0.06Vrms
Narrow shield = 0.14Vrms

Many thanks
 
...will build a 600 Ohm câble for the next test...

Cheers

Not easy, real 600 Ohm cable doesn’t exist. It seems to me that 600Ohm is attributed to the impedance of transformer.
I use an application from Agilent to see how much the radius is if centre core has 1mm diam.
The app can run up to 413 only, but it ran on parallel wires, for 1mm wire, it must be separated at 1.5m with polypropylene as dielectric to achieve 600 Ohm impedance @ 1kHz.

I build a third toroid 9V with single shield on broad footprint.
For zero V 50Hz field inductance as Mr K stated, I think that he uses a double shield “medical grade” (ES + mu métal) power transformer to mount a 9V coil sandwiched shield.

The newly built toroid has 0.008Vrms, single shield, -80dB @ 50Hz.

Bi-Circleton Single Shield 9V coil - YouTube

My apologies for swivel chair noise during the recording.

Cheers
 

Attachments

  • F8614116-DEFE-4B20-AE5B-D9E699369EEC.jpg
    F8614116-DEFE-4B20-AE5B-D9E699369EEC.jpg
    1,014.4 KB · Views: 71
  • E76245A5-7394-4ECB-AA4E-BCD578D3181D.jpg
    E76245A5-7394-4ECB-AA4E-BCD578D3181D.jpg
    825.4 KB · Views: 74
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.