Jump to content

GGW

Members
  • Posts

    19
  • Joined

  • Last visited

Posts posted by GGW

  1. 9 hours ago, jamesmking said:

    Golden Reference LV Build Guide

    Schematic for the golden reference LV:

    parts in green are new additions compared to the published schematic pdf in joamats post above. Parts in purple set the output voltage, parts in brown are the optional fine output adjust and parts in dark blue are the optional power led.

    2087031191_goldenreference6dLV.thumb.JPG.a9e8ca560ebca1caa71d0c25c9df184b.JPG

    Component Layout

     

    481539799_goldenreferencelv.thumb.JPG.1e9480701ef04045862a731a6566717d.JPG

     

    Optional parts and Options:

    1. if you don't want to have a power led fed from the LV board omit the 2K resistors R13, R14 and the led D3

    2. if you don't want very fine adjustment of the output voltage omit the 1M resistors R17, R18, R23, R24 and the trimmers RV2 and RV1. If you want more adjustment range reduce the value of the  resistors. With 1M ohm resistors the fine adjust can only change the output voltage by a few mV.  Replacing with 810K resistors provides a little more adjustment range ~ 20mV but at the expense of decreasing the temperature stability. A cheaper alternative for good output accuracy is to use 1/4W 0.1% low ppm e.g. 15ppm resistors for the voltage set resistors and omit the fine adjust completely - which is cheaper and improves the temperature stability.

    3. if you have the dual output board (which has both positive and negative output rails) either you populate the middle diode bridge only and use a transformer which is centre tapped OR you populate the two outer diode bridges and use a transformer with two separate output windings. If use use synchronous rectifiers instead of diode bridges use must use the second option and use separate windings for positive and negative rails. Do not populate all three bridges. If you use a centre tapped transformer you must connect the centre tap to the middle screw terminal (which is connected to the GRLV ground plane) of the 5 terminal AC input block . Do not use a synchronous rectifier to the centre bridge it will not work correctly and can result in the driver mosfets of the synchronous rectifier burning. 

    Note the GRLV cant output less than the voltage of the voltage reference, ideally for proper regulation the output needs to be more than about 2V higher than the LT voltage reference. So for about 12V or less output it is recommended you change the 10V reference for a 7V and the zener diodes for a 7V. For 12V output 10V reference or 7V reference can be used but the 7V reference version may have less noise. Maximum output is constrained by the voltage rating of the caps and is about 30V without modifications.

     

    _MG_6214_DxO.thumb.jpg.8ea42af854ab0bdfefd40879a36f76c7.jpg

     

     

    Setting the Output Voltage

    For the positive output R8 and R7 control the output voltage along with the voltage reference,

    the output voltage is ((R7+R8)/R7)*voltage reference output D5. Assuming you keep R7 at the stock value of 1.5K:

    output voltage change R8 to
    3k = 30V
    1.5K = 20V
    750 = 15V
    300 = 12V

    In general to calculate R8: ((Voltage required * R7)/voltage reference D5 output) - R7 = R8

     

    For the negative output R10 and R9 control the output voltage along with the voltage reference,

    the output voltage is ((R10+R9)/R9)*voltage reference output D7. Assuming you keep R10 at the stock value of 1.5K:

    output voltage change R9 to
    3k = 30V
    1.5K = 20V
    750 = 15V
    300 = 12V

    In general to calculate R9: ((Voltage required * R10)/voltage reference D7 output) - R10 = R9

     

    Photos of finished boards for reference:

    This version uses a single diode bridge so this is setup for a centre tapped transformer, also has the trimmers and resistors for the power led populated but no power led installed. The build uses 4700uF reservoir caps and silmic 220uf output caps and 1/4w resistors from vishay and dale. This is what you get using the BOM link in this post (see bellow) and is very close to the original BOM published near the beginning of this thread.

     

    _MG_6115.thumb.jpg.a939bcdff885836dd38acf0f595e7e1b.jpg

     

    This version is setup for dual separate transformer windings, and has no power led or trimmers implemented. It also uses synchronous rectifiers instead of diode bridges, nippon chemicon 10000uF reservoir caps, Panasonic FR series 220uF caps and 1/2W 50ppm 1% koa resistors. The resistors that set the output voltage are 1/4W TE 0.1% 15ppm for increased temperature stability and output accuracy. This is my current default build.

     

    _MG_6212_DxO.thumb.jpg.d6dc5d0140108dc590995b8a148b4b11.jpg

     

    BOM (based on 10V voltage reference and all options being populated)

    https://www.mouser.com/ProjectManager/ProjectDetail.aspx?AccessID=edda1bdbf7

     

    Component list with cap size information

    Br1,Br2,Br3              = 2 x 4A 100V RS402L  use 2 for dual winding transformer use 1 for centre tapped transformer

     C,C                            = 2 x film 4.7uF 50V 5mm lead spacing 0.5mm lead diameter

    C,C,C11,C12,C14,C15                   = 6 x film 0.01uF 100V  5mm lead spacing max size ~ 7.6mmx3mm 0.5mm lead diameter

    C1,C2                        = 2 x 50V 4700uF 10mm lead spacing max diameter 25mm, 22mm more comfortable fit main reservoir cap I also use Nippon Chemicon 63V 10000uF which just fit.

    C3,C16                     = 2 x Tantalum 35V 10uF ESR=2 Ohms 5mm lead spacing 0.5mm lead diameter

    C4,C5                        = 2 x Multilayer Ceramic 50V 47pF 5mm lead spacing 0.5mm lead diameter

    C6,C10                     = 2 x 220uF 35V silmic or Panasonic FR series 5mm lead spacing 12.5mm max diameter

    C7,C8,C9,C13         = 4 x tantalium 35V 47uF ESR=0.8ohms 5mm lead spacing 0.5mm lead diameter

     

    D,D,D,D                    = 4 x 1N914

    D1,D2,D3,D8,D9     = 5 x 1.7V red led        4 are required one is for optional power led leally all 4 main leds should be from the same batch and have similar characteristics.                            

    D10,D11                   = 2 x IN4007G 1A 1000V

    D4,D6                        = 2 x IN4739A 9.1V 1W    use 7V zener for <=12V output

    D6,D7                        = 2 x lt1021-10        10V voltage reference  use 7V for <=12V output

     

    Q4                              = 1 x MJW21194G

    Q5                              = 1 x MJF15031G

    Q7                              = 1 x MJW21193G

    Q9                              = 1 x MJF15030G

    Q1,Q3,Q6,Q13         = 4 x KSP92TA

    Q2,Q10,Q11,Q14    = 4 x KSP42TA

    Q8,Q12                     = 2 x DN2540N3-G

     

    R1,R4                        = 2 x 150 1/2W

    R13,R14                   = 2 x 2K

    R15,R16                   = 2 x 10K

    R17,R18,R23,R24                            = 4 x 1M, 810K = +-20mv adj range optional for fine adjustment

    R2,R5,R11,R12       = 4 x 1K

    R20,R21                   = 2 x 500

    R3,R6,R19,R22       = 4 x 10

    R7,R10                     = 2 x 1.5K

    R8,R9                        = 2 x 750  for 15V output change as necessary for your output

     

    RV1,RV2                  = 2 x 100K optional for fine adjustment

     

    U1,U2                        = 2 x OPA134PA

     

    Building

    Once you have decided the build options and output voltage, construction is straightforward given there are no high voltages. Depending upon the current draw the large power transistors on the side of the board may require heatsinking. The metal tab is live so they will need to be insulated from the heatsink/chassis.

    Socketing the reference and opamps is optional but it does make replacement and reuse easier.

    Make sure the caps are installed with the correct polarity. The line on the tantalums denotes the positive Terminal whereas the line on electrolytic caps denotes the negative terminal.... the film caps can be installed either way around as can the small ceramic caps.

    For the leds the longer leg is the + leg and this goes to the + mark on the pcb

    be careful to install the opamps and voltage reference chips the correct way around they will get very hot very quickly if installed the wrong way.

    make sure you don't place as ksp42 where a ksp92 should go of visa versa. Its easy to do since they look identical and its not easy to see all the markings once populating the board is finished.

     

    Testing

    If you have a means to control the input voltage to the board e.g. a variac, then with no load connected to the grlv the outer leds (closest to the LT voltage reference) should just start to glow at about 2.4 to 2.5VAC rms input voltage to the board and the output of the GRLV should be around 1V DC. If you cant get the outer leds to light with about 3VAC rms input, stop and disconnect, something is definitely wrong. By 3VAC rms the outer leds should be bright. 

    Increasing the variac output further should see proportional rises in the GRLV DC output voltage until the output reaches the expected output. At this point the inner leds closest to the zener diode may not have lit and although basic regulation has been achieved, the input voltage is too low for full low noise regulation. Increase the input voltage a little more and the inner leds should light. This should happen with after about an additional 1 volt AC rms is added to the input. The output voltage should not increase. 

    As a very rough rule of thumb, without load, the grlv needs about 2V less input AC rms than its DC output. So for example for 12v output expect regulation to just happen at about 10VAC rms input but the inner led not to light until about 11VAC rms. Note this figures depend on tolerances, the diode bridge voltage drop, led characteristics etc and are a rough guide only. Adding a load to the GRLV will mean it needs a higher input voltage in order to regulate since some of the sources of voltage drops are dependant on current draw e.g. the diode bridges. So if you adjust the variac to just get the inner led lit with no load don't be surprised if it goes out when you draw a few hundred milliamps.

    Another way to test is to connect the AC in of one rail to a current limited DC power supply. Set the current limit to say 0.005A (12mA) and slowly increase the voltage. Bellow about 3.3VDC input to the GRLV there should be low current draw <10mA. At about 3.4VDC the outer led will start to light and current draw should still be <10mA. Around 8.5VDC input current draw should reach about 10mA. The inner led close to the main cap should begin to light at about 2.6VDC input above the expected output e.g. 14.6VDC input for 12V output. At this point current draw should be about 20mA and you may need to increase your DC power supply current limit. Increasing the DC power supply output past this point should not result in increasing current draw or increasing GRLV output. Note when you increase the DC power supply voltage you will get an initial current draw spike as the resouviour cap charges to the new voltage level but the current draw should quickly subside to the figures shown.

    If the GRLv regulates and behaves as expected with no load and has no warm components or varying output then you can continue to load testing. The easiest way to do this is with an electronic DC load. I setup my load so that it will abort the test if the output voltage of the GRLV exceeds 0.2V above or bellow the no load output. A properly working GRLV with enough input voltage will vary very very little with current draw - much less than a traditional 78xx/79xx voltage regulator. I then load at 50mA and check for hot components, and then in 100mA steps up to 1A. Above about 400mA the large main transistors will get hot if they are not heatsinked... this is normal as the current draw increases there is more power dissipation in the transistor. Note transformers and diode bridges etc do drop more voltage as the current draw increases so if the inner led goes out at higher current draws check the input voltage to the GRLV has not decreased too much for the GRLV to regulate. Don't load test about above 300mA for long periods without heatsinking the main transistors.

     

    Additional Checks

    The voltage reference output voltage should be present between pins 4 and 6 of the LT reference and be stable and very close to the spec sheet voltage. Use fine tip probes and be careful not to short any pins together when probing:

     

    _MG_6220_DxO.thumb.jpg.f6ceeb38dbca1b9fbf4bbacfb6833b64.jpg

    The opamp compares the voltage reference with the output voltage from the voltage divider (the resistors which set the output voltage) and creates a correction signal. Pin 2 of the opamp is connected to the voltage reference and pin 3 is connected to the voltage divider. When working correctly, the voltages with respect to ground to pin 2 and from ground to pin 3 should be identical and be the same as the output of the voltage reference measured above.

     

    _MG_6221_DxO.thumb.jpg.72e157a68968b7f4cb88e075dbd41441.jpg

     

     

     

    Excellent post, James - thank you!!

  2. On 5/7/2021 at 1:35 PM, starcat said:

    Just go 2x CFA3 boards and 2x GRLV (on one or even two transformers). For the amp you do not need to separate the + and - per channel, so 2x CFA3 boards is more than good to go. No need to put the CFA3 boards in separate cases, if you want to separate the L and R channels, put an aluminium divider in the mid of the amp case and use per channel umbilical on the back (off separate GRLVs and transformers in the PSU case), so basically going strict dual mono in two chassis in total. That's already more than overkill. 

    You can have the headphone protector as well (for dynamic headphones, stats don't need that) and eventually the ZF/SS selector (which I didn't do). 

    Thanks, @starcat - very helpful. 
     

    Do you know where I can find the Gerbers for the ZF/SS selector and the headphone protector?

  3. 13 hours ago, Kerry said:

    SMD version of the CFA3 dynamic

    Ah Awesome!

    @Kerry @kevin gilmore - what would you gents consider to be the ultimate overkill build for this CFP/A?  4x CFA2 boards or 2x CFA3 boards in 2 separate chassis, with 2 separately enclosed GRLV psu? Or is this new SMD animal you have designed the way to go, assuming the same multiple separate chassis & PSU overkill plan?

    Separately, I've read the entire thread and can't seem to get my head around the difference between using 4x CFA2 boards or 2x CFA3 boards - are there additional PCBs that need to be printed if I go the 4xCFA2 route (scheming going down this road and 2 dissipante cases such that each pcb has its own heatsink). 

    Or do I just wait and go for the SMD boards you're working on?

    Many thanks in advance for the help and guidance

    Cheers

  4. Thanks, @jamesmking - very helpful.

    I've read this entire thread and there are 3 updates to the version .45 board (see file names including the version on the silkscreen, in the snip below):

    image.png.f3debc7e85031ea716f2ec106f37c9fc.png

    For context, the descriptions Kevin uploaded with each:

    version 0.46 "rflip"

    because soren is running the grlv supply at 4.5 amperes, here is a new version with the feedback resistor locations flipped so that the sense is not one inch away from the output connector.

    sense is now tied directly to the inner power supply connector. So the 20mv change in voltage due to current should be completely canceled out.

    for higher currents, best if the board is made with 3 oz copper.
     

    version 0.47 "rflip to220"

    here is the version with the to220 sic diodes, someone really needs to check this

    version 0.48 "largecap"

    same size board, up to 35mm caps, had to move the pass transistors, needs checking silicon carbide rectifiers 100mm x 109mm

    I think @Pars mentioned the v0.48 is basically the v.45(?) with fatter, shorter (?) caps for when height is a constraint - is that correct?

    *******

    My questions -

    (1.) are these all improvements (is version 0.48 the best in all use cases?) or are each of versions 0.46-0.48 just tweaks for specific use cases ?

    (2.) Which board is best suited to (a.) power the SS Dynalo (mk2 "multiamp"); and (b.) power balanced CFA boards (i.e. 2x CFA2 or 1x CFA3?

     

    Many thanks as always for the guidance!  (For context - I'm trying to pull the trigger once on a PCB order, so trying to figure out the right versions of all the boards I need to avoid multiple shipping charges... i.e. GRLV, Dynalo, CFA2 or CFA3, etc.)

     

    • Like 1
  5. Hi Gents,

     

    1.) Please could someone point me in the direction of the version .47 of the GRLV board? I can't see it on the share drive 😕

    Separately, I have 2 additional questions if anyone could please help me:

    2.) What is the latest / greatest / preferred GRLV version for each of the Dynalo and CFA2 x 4 (dual / balanced craziness;   or 2 x CFA3 boards) ?

    3.) I noticed there is another little PCB called "GR78xx" - would I need to get a few these printed as well? Or is it optional? If optional, what's the benefit?

     

    Thanks in advance for the guidance!!

     

  6. 16 hours ago, Pars said:

    Adapter boards that I layed out and were in one of the GBs.

    I think you could just plug them into a 14 pin socket with no lead bending, but if I recall correctly, one of the JFETs would be backwards. No big deal on the Toshibas, at least according to John Curl. I presume the Linear Systems would be OK as well.

    Cheers, Pars - will go swim through the Dynalobal gerbers until I find it :) 

  7. Noob question but do you have to match across N and P channels for the input stage? Or just within N and within P if you go the single JFET route?

    I.e. if I go for, say, 2 x LSJ74 and 2x LSK170 - do they ALL have to be matched or only the 74s with each other and the 170s with each other?  If the latter, then no need to buy plenty  LSK489 / LSJ689 and match them?

    Thanks!

  8. Hi All,

    After reading this entire thread, it seems there are a number of options for transistors to use in place of THAT340 on the input stage (if you can source them).

    Please could someone help rank the below in order of preference?

     

    - 2SJ109 / 2SK389 appear impossible to source

    - LSJ109 / LSK389 (LSJ109 also looks hopeless unless you roll the dice on ebay / aliexpress)

    - 2x Matched pairs of LSJ74 and LSK170

    - 2SC3381 / 2SA1349

    - LSK489 / LSJ689

    - Anything else?

     

    Or just go with THAT340 and live with the low input impedance that seems to be a minor issue?

    Thanks!

  9. On 12/13/2020 at 10:50 PM, Youwin0125 said:

    Please help to upload the file again. 

    The web connection of file is loss.

    Thanks a lot ^^

    Hi Youwin0125

    I'm also new to the forum and with KG's help I found the share drive below with all the gerber files:

    ** LINK REMOVED ** GILMORE CAN POST IT IF HE WANTS **

    There are a ton of Dynalo zips in there... no clue which is the latest and greatest! 😕 

    image.png.f81c866c751d0518fc7ad4052b568175.png

×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use.